Show More
@@ -1,3837 +1,3837 b'' | |||||
1 | # perf.py - performance test routines |
|
1 | # perf.py - performance test routines | |
2 | '''helper extension to measure performance |
|
2 | '''helper extension to measure performance | |
3 |
|
3 | |||
4 | Configurations |
|
4 | Configurations | |
5 | ============== |
|
5 | ============== | |
6 |
|
6 | |||
7 | ``perf`` |
|
7 | ``perf`` | |
8 | -------- |
|
8 | -------- | |
9 |
|
9 | |||
10 | ``all-timing`` |
|
10 | ``all-timing`` | |
11 | When set, additional statistics will be reported for each benchmark: best, |
|
11 | When set, additional statistics will be reported for each benchmark: best, | |
12 | worst, median average. If not set only the best timing is reported |
|
12 | worst, median average. If not set only the best timing is reported | |
13 | (default: off). |
|
13 | (default: off). | |
14 |
|
14 | |||
15 | ``presleep`` |
|
15 | ``presleep`` | |
16 | number of second to wait before any group of runs (default: 1) |
|
16 | number of second to wait before any group of runs (default: 1) | |
17 |
|
17 | |||
18 | ``pre-run`` |
|
18 | ``pre-run`` | |
19 | number of run to perform before starting measurement. |
|
19 | number of run to perform before starting measurement. | |
20 |
|
20 | |||
21 | ``profile-benchmark`` |
|
21 | ``profile-benchmark`` | |
22 | Enable profiling for the benchmarked section. |
|
22 | Enable profiling for the benchmarked section. | |
23 | (The first iteration is benchmarked) |
|
23 | (The first iteration is benchmarked) | |
24 |
|
24 | |||
25 | ``run-limits`` |
|
25 | ``run-limits`` | |
26 | Control the number of runs each benchmark will perform. The option value |
|
26 | Control the number of runs each benchmark will perform. The option value | |
27 | should be a list of `<time>-<numberofrun>` pairs. After each run the |
|
27 | should be a list of `<time>-<numberofrun>` pairs. After each run the | |
28 | conditions are considered in order with the following logic: |
|
28 | conditions are considered in order with the following logic: | |
29 |
|
29 | |||
30 | If benchmark has been running for <time> seconds, and we have performed |
|
30 | If benchmark has been running for <time> seconds, and we have performed | |
31 | <numberofrun> iterations, stop the benchmark, |
|
31 | <numberofrun> iterations, stop the benchmark, | |
32 |
|
32 | |||
33 | The default value is: `3.0-100, 10.0-3` |
|
33 | The default value is: `3.0-100, 10.0-3` | |
34 |
|
34 | |||
35 | ``stub`` |
|
35 | ``stub`` | |
36 | When set, benchmarks will only be run once, useful for testing |
|
36 | When set, benchmarks will only be run once, useful for testing | |
37 | (default: off) |
|
37 | (default: off) | |
38 | ''' |
|
38 | ''' | |
39 |
|
39 | |||
40 | # "historical portability" policy of perf.py: |
|
40 | # "historical portability" policy of perf.py: | |
41 | # |
|
41 | # | |
42 | # We have to do: |
|
42 | # We have to do: | |
43 | # - make perf.py "loadable" with as wide Mercurial version as possible |
|
43 | # - make perf.py "loadable" with as wide Mercurial version as possible | |
44 | # This doesn't mean that perf commands work correctly with that Mercurial. |
|
44 | # This doesn't mean that perf commands work correctly with that Mercurial. | |
45 | # BTW, perf.py itself has been available since 1.1 (or eb240755386d). |
|
45 | # BTW, perf.py itself has been available since 1.1 (or eb240755386d). | |
46 | # - make historical perf command work correctly with as wide Mercurial |
|
46 | # - make historical perf command work correctly with as wide Mercurial | |
47 | # version as possible |
|
47 | # version as possible | |
48 | # |
|
48 | # | |
49 | # We have to do, if possible with reasonable cost: |
|
49 | # We have to do, if possible with reasonable cost: | |
50 | # - make recent perf command for historical feature work correctly |
|
50 | # - make recent perf command for historical feature work correctly | |
51 | # with early Mercurial |
|
51 | # with early Mercurial | |
52 | # |
|
52 | # | |
53 | # We don't have to do: |
|
53 | # We don't have to do: | |
54 | # - make perf command for recent feature work correctly with early |
|
54 | # - make perf command for recent feature work correctly with early | |
55 | # Mercurial |
|
55 | # Mercurial | |
56 |
|
56 | |||
57 | from __future__ import absolute_import |
|
57 | from __future__ import absolute_import | |
58 | import contextlib |
|
58 | import contextlib | |
59 | import functools |
|
59 | import functools | |
60 | import gc |
|
60 | import gc | |
61 | import os |
|
61 | import os | |
62 | import random |
|
62 | import random | |
63 | import shutil |
|
63 | import shutil | |
64 | import struct |
|
64 | import struct | |
65 | import sys |
|
65 | import sys | |
66 | import tempfile |
|
66 | import tempfile | |
67 | import threading |
|
67 | import threading | |
68 | import time |
|
68 | import time | |
69 | from mercurial import ( |
|
69 | from mercurial import ( | |
70 | changegroup, |
|
70 | changegroup, | |
71 | cmdutil, |
|
71 | cmdutil, | |
72 | commands, |
|
72 | commands, | |
73 | copies, |
|
73 | copies, | |
74 | error, |
|
74 | error, | |
75 | extensions, |
|
75 | extensions, | |
76 | hg, |
|
76 | hg, | |
77 | mdiff, |
|
77 | mdiff, | |
78 | merge, |
|
78 | merge, | |
79 | revlog, |
|
79 | revlog, | |
80 | util, |
|
80 | util, | |
81 | ) |
|
81 | ) | |
82 |
|
82 | |||
83 | # for "historical portability": |
|
83 | # for "historical portability": | |
84 | # try to import modules separately (in dict order), and ignore |
|
84 | # try to import modules separately (in dict order), and ignore | |
85 | # failure, because these aren't available with early Mercurial |
|
85 | # failure, because these aren't available with early Mercurial | |
86 | try: |
|
86 | try: | |
87 | from mercurial import branchmap # since 2.5 (or bcee63733aad) |
|
87 | from mercurial import branchmap # since 2.5 (or bcee63733aad) | |
88 | except ImportError: |
|
88 | except ImportError: | |
89 | pass |
|
89 | pass | |
90 | try: |
|
90 | try: | |
91 | from mercurial import obsolete # since 2.3 (or ad0d6c2b3279) |
|
91 | from mercurial import obsolete # since 2.3 (or ad0d6c2b3279) | |
92 | except ImportError: |
|
92 | except ImportError: | |
93 | pass |
|
93 | pass | |
94 | try: |
|
94 | try: | |
95 | from mercurial import registrar # since 3.7 (or 37d50250b696) |
|
95 | from mercurial import registrar # since 3.7 (or 37d50250b696) | |
96 |
|
96 | |||
97 | dir(registrar) # forcibly load it |
|
97 | dir(registrar) # forcibly load it | |
98 | except ImportError: |
|
98 | except ImportError: | |
99 | registrar = None |
|
99 | registrar = None | |
100 | try: |
|
100 | try: | |
101 | from mercurial import repoview # since 2.5 (or 3a6ddacb7198) |
|
101 | from mercurial import repoview # since 2.5 (or 3a6ddacb7198) | |
102 | except ImportError: |
|
102 | except ImportError: | |
103 | pass |
|
103 | pass | |
104 | try: |
|
104 | try: | |
105 | from mercurial.utils import repoviewutil # since 5.0 |
|
105 | from mercurial.utils import repoviewutil # since 5.0 | |
106 | except ImportError: |
|
106 | except ImportError: | |
107 | repoviewutil = None |
|
107 | repoviewutil = None | |
108 | try: |
|
108 | try: | |
109 | from mercurial import scmutil # since 1.9 (or 8b252e826c68) |
|
109 | from mercurial import scmutil # since 1.9 (or 8b252e826c68) | |
110 | except ImportError: |
|
110 | except ImportError: | |
111 | pass |
|
111 | pass | |
112 | try: |
|
112 | try: | |
113 | from mercurial import setdiscovery # since 1.9 (or cb98fed52495) |
|
113 | from mercurial import setdiscovery # since 1.9 (or cb98fed52495) | |
114 | except ImportError: |
|
114 | except ImportError: | |
115 | pass |
|
115 | pass | |
116 |
|
116 | |||
117 | try: |
|
117 | try: | |
118 | from mercurial import profiling |
|
118 | from mercurial import profiling | |
119 | except ImportError: |
|
119 | except ImportError: | |
120 | profiling = None |
|
120 | profiling = None | |
121 |
|
121 | |||
122 |
|
122 | |||
123 | def identity(a): |
|
123 | def identity(a): | |
124 | return a |
|
124 | return a | |
125 |
|
125 | |||
126 |
|
126 | |||
127 | try: |
|
127 | try: | |
128 | from mercurial import pycompat |
|
128 | from mercurial import pycompat | |
129 |
|
129 | |||
130 | getargspec = pycompat.getargspec # added to module after 4.5 |
|
130 | getargspec = pycompat.getargspec # added to module after 4.5 | |
131 | _byteskwargs = pycompat.byteskwargs # since 4.1 (or fbc3f73dc802) |
|
131 | _byteskwargs = pycompat.byteskwargs # since 4.1 (or fbc3f73dc802) | |
132 | _sysstr = pycompat.sysstr # since 4.0 (or 2219f4f82ede) |
|
132 | _sysstr = pycompat.sysstr # since 4.0 (or 2219f4f82ede) | |
133 | _bytestr = pycompat.bytestr # since 4.2 (or b70407bd84d5) |
|
133 | _bytestr = pycompat.bytestr # since 4.2 (or b70407bd84d5) | |
134 | _xrange = pycompat.xrange # since 4.8 (or 7eba8f83129b) |
|
134 | _xrange = pycompat.xrange # since 4.8 (or 7eba8f83129b) | |
135 | fsencode = pycompat.fsencode # since 3.9 (or f4a5e0e86a7e) |
|
135 | fsencode = pycompat.fsencode # since 3.9 (or f4a5e0e86a7e) | |
136 | if pycompat.ispy3: |
|
136 | if pycompat.ispy3: | |
137 | _maxint = sys.maxsize # per py3 docs for replacing maxint |
|
137 | _maxint = sys.maxsize # per py3 docs for replacing maxint | |
138 | else: |
|
138 | else: | |
139 | _maxint = sys.maxint |
|
139 | _maxint = sys.maxint | |
140 | except (NameError, ImportError, AttributeError): |
|
140 | except (NameError, ImportError, AttributeError): | |
141 | import inspect |
|
141 | import inspect | |
142 |
|
142 | |||
143 | getargspec = inspect.getargspec |
|
143 | getargspec = inspect.getargspec | |
144 | _byteskwargs = identity |
|
144 | _byteskwargs = identity | |
145 | _bytestr = str |
|
145 | _bytestr = str | |
146 | fsencode = identity # no py3 support |
|
146 | fsencode = identity # no py3 support | |
147 | _maxint = sys.maxint # no py3 support |
|
147 | _maxint = sys.maxint # no py3 support | |
148 | _sysstr = lambda x: x # no py3 support |
|
148 | _sysstr = lambda x: x # no py3 support | |
149 | _xrange = xrange |
|
149 | _xrange = xrange | |
150 |
|
150 | |||
151 | try: |
|
151 | try: | |
152 | # 4.7+ |
|
152 | # 4.7+ | |
153 | queue = pycompat.queue.Queue |
|
153 | queue = pycompat.queue.Queue | |
154 | except (NameError, AttributeError, ImportError): |
|
154 | except (NameError, AttributeError, ImportError): | |
155 | # <4.7. |
|
155 | # <4.7. | |
156 | try: |
|
156 | try: | |
157 | queue = pycompat.queue |
|
157 | queue = pycompat.queue | |
158 | except (NameError, AttributeError, ImportError): |
|
158 | except (NameError, AttributeError, ImportError): | |
159 | import Queue as queue |
|
159 | import Queue as queue | |
160 |
|
160 | |||
161 | try: |
|
161 | try: | |
162 | from mercurial import logcmdutil |
|
162 | from mercurial import logcmdutil | |
163 |
|
163 | |||
164 | makelogtemplater = logcmdutil.maketemplater |
|
164 | makelogtemplater = logcmdutil.maketemplater | |
165 | except (AttributeError, ImportError): |
|
165 | except (AttributeError, ImportError): | |
166 | try: |
|
166 | try: | |
167 | makelogtemplater = cmdutil.makelogtemplater |
|
167 | makelogtemplater = cmdutil.makelogtemplater | |
168 | except (AttributeError, ImportError): |
|
168 | except (AttributeError, ImportError): | |
169 | makelogtemplater = None |
|
169 | makelogtemplater = None | |
170 |
|
170 | |||
171 | # for "historical portability": |
|
171 | # for "historical portability": | |
172 | # define util.safehasattr forcibly, because util.safehasattr has been |
|
172 | # define util.safehasattr forcibly, because util.safehasattr has been | |
173 | # available since 1.9.3 (or 94b200a11cf7) |
|
173 | # available since 1.9.3 (or 94b200a11cf7) | |
174 | _undefined = object() |
|
174 | _undefined = object() | |
175 |
|
175 | |||
176 |
|
176 | |||
177 | def safehasattr(thing, attr): |
|
177 | def safehasattr(thing, attr): | |
178 | return getattr(thing, _sysstr(attr), _undefined) is not _undefined |
|
178 | return getattr(thing, _sysstr(attr), _undefined) is not _undefined | |
179 |
|
179 | |||
180 |
|
180 | |||
181 | setattr(util, 'safehasattr', safehasattr) |
|
181 | setattr(util, 'safehasattr', safehasattr) | |
182 |
|
182 | |||
183 | # for "historical portability": |
|
183 | # for "historical portability": | |
184 | # define util.timer forcibly, because util.timer has been available |
|
184 | # define util.timer forcibly, because util.timer has been available | |
185 | # since ae5d60bb70c9 |
|
185 | # since ae5d60bb70c9 | |
186 | if safehasattr(time, 'perf_counter'): |
|
186 | if safehasattr(time, 'perf_counter'): | |
187 | util.timer = time.perf_counter |
|
187 | util.timer = time.perf_counter | |
188 | elif os.name == b'nt': |
|
188 | elif os.name == b'nt': | |
189 | util.timer = time.clock |
|
189 | util.timer = time.clock | |
190 | else: |
|
190 | else: | |
191 | util.timer = time.time |
|
191 | util.timer = time.time | |
192 |
|
192 | |||
193 | # for "historical portability": |
|
193 | # for "historical portability": | |
194 | # use locally defined empty option list, if formatteropts isn't |
|
194 | # use locally defined empty option list, if formatteropts isn't | |
195 | # available, because commands.formatteropts has been available since |
|
195 | # available, because commands.formatteropts has been available since | |
196 | # 3.2 (or 7a7eed5176a4), even though formatting itself has been |
|
196 | # 3.2 (or 7a7eed5176a4), even though formatting itself has been | |
197 | # available since 2.2 (or ae5f92e154d3) |
|
197 | # available since 2.2 (or ae5f92e154d3) | |
198 | formatteropts = getattr( |
|
198 | formatteropts = getattr( | |
199 | cmdutil, "formatteropts", getattr(commands, "formatteropts", []) |
|
199 | cmdutil, "formatteropts", getattr(commands, "formatteropts", []) | |
200 | ) |
|
200 | ) | |
201 |
|
201 | |||
202 | # for "historical portability": |
|
202 | # for "historical portability": | |
203 | # use locally defined option list, if debugrevlogopts isn't available, |
|
203 | # use locally defined option list, if debugrevlogopts isn't available, | |
204 | # because commands.debugrevlogopts has been available since 3.7 (or |
|
204 | # because commands.debugrevlogopts has been available since 3.7 (or | |
205 | # 5606f7d0d063), even though cmdutil.openrevlog() has been available |
|
205 | # 5606f7d0d063), even though cmdutil.openrevlog() has been available | |
206 | # since 1.9 (or a79fea6b3e77). |
|
206 | # since 1.9 (or a79fea6b3e77). | |
207 | revlogopts = getattr( |
|
207 | revlogopts = getattr( | |
208 | cmdutil, |
|
208 | cmdutil, | |
209 | "debugrevlogopts", |
|
209 | "debugrevlogopts", | |
210 | getattr( |
|
210 | getattr( | |
211 | commands, |
|
211 | commands, | |
212 | "debugrevlogopts", |
|
212 | "debugrevlogopts", | |
213 | [ |
|
213 | [ | |
214 | (b'c', b'changelog', False, b'open changelog'), |
|
214 | (b'c', b'changelog', False, b'open changelog'), | |
215 | (b'm', b'manifest', False, b'open manifest'), |
|
215 | (b'm', b'manifest', False, b'open manifest'), | |
216 | (b'', b'dir', False, b'open directory manifest'), |
|
216 | (b'', b'dir', False, b'open directory manifest'), | |
217 | ], |
|
217 | ], | |
218 | ), |
|
218 | ), | |
219 | ) |
|
219 | ) | |
220 |
|
220 | |||
221 | cmdtable = {} |
|
221 | cmdtable = {} | |
222 |
|
222 | |||
223 | # for "historical portability": |
|
223 | # for "historical portability": | |
224 | # define parsealiases locally, because cmdutil.parsealiases has been |
|
224 | # define parsealiases locally, because cmdutil.parsealiases has been | |
225 | # available since 1.5 (or 6252852b4332) |
|
225 | # available since 1.5 (or 6252852b4332) | |
226 | def parsealiases(cmd): |
|
226 | def parsealiases(cmd): | |
227 | return cmd.split(b"|") |
|
227 | return cmd.split(b"|") | |
228 |
|
228 | |||
229 |
|
229 | |||
230 | if safehasattr(registrar, 'command'): |
|
230 | if safehasattr(registrar, 'command'): | |
231 | command = registrar.command(cmdtable) |
|
231 | command = registrar.command(cmdtable) | |
232 | elif safehasattr(cmdutil, 'command'): |
|
232 | elif safehasattr(cmdutil, 'command'): | |
233 | command = cmdutil.command(cmdtable) |
|
233 | command = cmdutil.command(cmdtable) | |
234 | if b'norepo' not in getargspec(command).args: |
|
234 | if b'norepo' not in getargspec(command).args: | |
235 | # for "historical portability": |
|
235 | # for "historical portability": | |
236 | # wrap original cmdutil.command, because "norepo" option has |
|
236 | # wrap original cmdutil.command, because "norepo" option has | |
237 | # been available since 3.1 (or 75a96326cecb) |
|
237 | # been available since 3.1 (or 75a96326cecb) | |
238 | _command = command |
|
238 | _command = command | |
239 |
|
239 | |||
240 | def command(name, options=(), synopsis=None, norepo=False): |
|
240 | def command(name, options=(), synopsis=None, norepo=False): | |
241 | if norepo: |
|
241 | if norepo: | |
242 | commands.norepo += b' %s' % b' '.join(parsealiases(name)) |
|
242 | commands.norepo += b' %s' % b' '.join(parsealiases(name)) | |
243 | return _command(name, list(options), synopsis) |
|
243 | return _command(name, list(options), synopsis) | |
244 |
|
244 | |||
245 |
|
245 | |||
246 | else: |
|
246 | else: | |
247 | # for "historical portability": |
|
247 | # for "historical portability": | |
248 | # define "@command" annotation locally, because cmdutil.command |
|
248 | # define "@command" annotation locally, because cmdutil.command | |
249 | # has been available since 1.9 (or 2daa5179e73f) |
|
249 | # has been available since 1.9 (or 2daa5179e73f) | |
250 | def command(name, options=(), synopsis=None, norepo=False): |
|
250 | def command(name, options=(), synopsis=None, norepo=False): | |
251 | def decorator(func): |
|
251 | def decorator(func): | |
252 | if synopsis: |
|
252 | if synopsis: | |
253 | cmdtable[name] = func, list(options), synopsis |
|
253 | cmdtable[name] = func, list(options), synopsis | |
254 | else: |
|
254 | else: | |
255 | cmdtable[name] = func, list(options) |
|
255 | cmdtable[name] = func, list(options) | |
256 | if norepo: |
|
256 | if norepo: | |
257 | commands.norepo += b' %s' % b' '.join(parsealiases(name)) |
|
257 | commands.norepo += b' %s' % b' '.join(parsealiases(name)) | |
258 | return func |
|
258 | return func | |
259 |
|
259 | |||
260 | return decorator |
|
260 | return decorator | |
261 |
|
261 | |||
262 |
|
262 | |||
263 | try: |
|
263 | try: | |
264 | import mercurial.registrar |
|
264 | import mercurial.registrar | |
265 | import mercurial.configitems |
|
265 | import mercurial.configitems | |
266 |
|
266 | |||
267 | configtable = {} |
|
267 | configtable = {} | |
268 | configitem = mercurial.registrar.configitem(configtable) |
|
268 | configitem = mercurial.registrar.configitem(configtable) | |
269 | configitem( |
|
269 | configitem( | |
270 | b'perf', |
|
270 | b'perf', | |
271 | b'presleep', |
|
271 | b'presleep', | |
272 | default=mercurial.configitems.dynamicdefault, |
|
272 | default=mercurial.configitems.dynamicdefault, | |
273 | experimental=True, |
|
273 | experimental=True, | |
274 | ) |
|
274 | ) | |
275 | configitem( |
|
275 | configitem( | |
276 | b'perf', |
|
276 | b'perf', | |
277 | b'stub', |
|
277 | b'stub', | |
278 | default=mercurial.configitems.dynamicdefault, |
|
278 | default=mercurial.configitems.dynamicdefault, | |
279 | experimental=True, |
|
279 | experimental=True, | |
280 | ) |
|
280 | ) | |
281 | configitem( |
|
281 | configitem( | |
282 | b'perf', |
|
282 | b'perf', | |
283 | b'parentscount', |
|
283 | b'parentscount', | |
284 | default=mercurial.configitems.dynamicdefault, |
|
284 | default=mercurial.configitems.dynamicdefault, | |
285 | experimental=True, |
|
285 | experimental=True, | |
286 | ) |
|
286 | ) | |
287 | configitem( |
|
287 | configitem( | |
288 | b'perf', |
|
288 | b'perf', | |
289 | b'all-timing', |
|
289 | b'all-timing', | |
290 | default=mercurial.configitems.dynamicdefault, |
|
290 | default=mercurial.configitems.dynamicdefault, | |
291 | experimental=True, |
|
291 | experimental=True, | |
292 | ) |
|
292 | ) | |
293 | configitem( |
|
293 | configitem( | |
294 | b'perf', b'pre-run', default=mercurial.configitems.dynamicdefault, |
|
294 | b'perf', b'pre-run', default=mercurial.configitems.dynamicdefault, | |
295 | ) |
|
295 | ) | |
296 | configitem( |
|
296 | configitem( | |
297 | b'perf', |
|
297 | b'perf', | |
298 | b'profile-benchmark', |
|
298 | b'profile-benchmark', | |
299 | default=mercurial.configitems.dynamicdefault, |
|
299 | default=mercurial.configitems.dynamicdefault, | |
300 | ) |
|
300 | ) | |
301 | configitem( |
|
301 | configitem( | |
302 | b'perf', |
|
302 | b'perf', | |
303 | b'run-limits', |
|
303 | b'run-limits', | |
304 | default=mercurial.configitems.dynamicdefault, |
|
304 | default=mercurial.configitems.dynamicdefault, | |
305 | experimental=True, |
|
305 | experimental=True, | |
306 | ) |
|
306 | ) | |
307 | except (ImportError, AttributeError): |
|
307 | except (ImportError, AttributeError): | |
308 | pass |
|
308 | pass | |
309 | except TypeError: |
|
309 | except TypeError: | |
310 | # compatibility fix for a11fd395e83f |
|
310 | # compatibility fix for a11fd395e83f | |
311 | # hg version: 5.2 |
|
311 | # hg version: 5.2 | |
312 | configitem( |
|
312 | configitem( | |
313 | b'perf', b'presleep', default=mercurial.configitems.dynamicdefault, |
|
313 | b'perf', b'presleep', default=mercurial.configitems.dynamicdefault, | |
314 | ) |
|
314 | ) | |
315 | configitem( |
|
315 | configitem( | |
316 | b'perf', b'stub', default=mercurial.configitems.dynamicdefault, |
|
316 | b'perf', b'stub', default=mercurial.configitems.dynamicdefault, | |
317 | ) |
|
317 | ) | |
318 | configitem( |
|
318 | configitem( | |
319 | b'perf', b'parentscount', default=mercurial.configitems.dynamicdefault, |
|
319 | b'perf', b'parentscount', default=mercurial.configitems.dynamicdefault, | |
320 | ) |
|
320 | ) | |
321 | configitem( |
|
321 | configitem( | |
322 | b'perf', b'all-timing', default=mercurial.configitems.dynamicdefault, |
|
322 | b'perf', b'all-timing', default=mercurial.configitems.dynamicdefault, | |
323 | ) |
|
323 | ) | |
324 | configitem( |
|
324 | configitem( | |
325 | b'perf', b'pre-run', default=mercurial.configitems.dynamicdefault, |
|
325 | b'perf', b'pre-run', default=mercurial.configitems.dynamicdefault, | |
326 | ) |
|
326 | ) | |
327 | configitem( |
|
327 | configitem( | |
328 | b'perf', |
|
328 | b'perf', | |
329 | b'profile-benchmark', |
|
329 | b'profile-benchmark', | |
330 | default=mercurial.configitems.dynamicdefault, |
|
330 | default=mercurial.configitems.dynamicdefault, | |
331 | ) |
|
331 | ) | |
332 | configitem( |
|
332 | configitem( | |
333 | b'perf', b'run-limits', default=mercurial.configitems.dynamicdefault, |
|
333 | b'perf', b'run-limits', default=mercurial.configitems.dynamicdefault, | |
334 | ) |
|
334 | ) | |
335 |
|
335 | |||
336 |
|
336 | |||
337 | def getlen(ui): |
|
337 | def getlen(ui): | |
338 | if ui.configbool(b"perf", b"stub", False): |
|
338 | if ui.configbool(b"perf", b"stub", False): | |
339 | return lambda x: 1 |
|
339 | return lambda x: 1 | |
340 | return len |
|
340 | return len | |
341 |
|
341 | |||
342 |
|
342 | |||
343 | class noop(object): |
|
343 | class noop(object): | |
344 | """dummy context manager""" |
|
344 | """dummy context manager""" | |
345 |
|
345 | |||
346 | def __enter__(self): |
|
346 | def __enter__(self): | |
347 | pass |
|
347 | pass | |
348 |
|
348 | |||
349 | def __exit__(self, *args): |
|
349 | def __exit__(self, *args): | |
350 | pass |
|
350 | pass | |
351 |
|
351 | |||
352 |
|
352 | |||
353 | NOOPCTX = noop() |
|
353 | NOOPCTX = noop() | |
354 |
|
354 | |||
355 |
|
355 | |||
356 | def gettimer(ui, opts=None): |
|
356 | def gettimer(ui, opts=None): | |
357 | """return a timer function and formatter: (timer, formatter) |
|
357 | """return a timer function and formatter: (timer, formatter) | |
358 |
|
358 | |||
359 | This function exists to gather the creation of formatter in a single |
|
359 | This function exists to gather the creation of formatter in a single | |
360 | place instead of duplicating it in all performance commands.""" |
|
360 | place instead of duplicating it in all performance commands.""" | |
361 |
|
361 | |||
362 | # enforce an idle period before execution to counteract power management |
|
362 | # enforce an idle period before execution to counteract power management | |
363 | # experimental config: perf.presleep |
|
363 | # experimental config: perf.presleep | |
364 | time.sleep(getint(ui, b"perf", b"presleep", 1)) |
|
364 | time.sleep(getint(ui, b"perf", b"presleep", 1)) | |
365 |
|
365 | |||
366 | if opts is None: |
|
366 | if opts is None: | |
367 | opts = {} |
|
367 | opts = {} | |
368 | # redirect all to stderr unless buffer api is in use |
|
368 | # redirect all to stderr unless buffer api is in use | |
369 | if not ui._buffers: |
|
369 | if not ui._buffers: | |
370 | ui = ui.copy() |
|
370 | ui = ui.copy() | |
371 | uifout = safeattrsetter(ui, b'fout', ignoremissing=True) |
|
371 | uifout = safeattrsetter(ui, b'fout', ignoremissing=True) | |
372 | if uifout: |
|
372 | if uifout: | |
373 | # for "historical portability": |
|
373 | # for "historical portability": | |
374 | # ui.fout/ferr have been available since 1.9 (or 4e1ccd4c2b6d) |
|
374 | # ui.fout/ferr have been available since 1.9 (or 4e1ccd4c2b6d) | |
375 | uifout.set(ui.ferr) |
|
375 | uifout.set(ui.ferr) | |
376 |
|
376 | |||
377 | # get a formatter |
|
377 | # get a formatter | |
378 | uiformatter = getattr(ui, 'formatter', None) |
|
378 | uiformatter = getattr(ui, 'formatter', None) | |
379 | if uiformatter: |
|
379 | if uiformatter: | |
380 | fm = uiformatter(b'perf', opts) |
|
380 | fm = uiformatter(b'perf', opts) | |
381 | else: |
|
381 | else: | |
382 | # for "historical portability": |
|
382 | # for "historical portability": | |
383 | # define formatter locally, because ui.formatter has been |
|
383 | # define formatter locally, because ui.formatter has been | |
384 | # available since 2.2 (or ae5f92e154d3) |
|
384 | # available since 2.2 (or ae5f92e154d3) | |
385 | from mercurial import node |
|
385 | from mercurial import node | |
386 |
|
386 | |||
387 | class defaultformatter(object): |
|
387 | class defaultformatter(object): | |
388 | """Minimized composition of baseformatter and plainformatter |
|
388 | """Minimized composition of baseformatter and plainformatter | |
389 | """ |
|
389 | """ | |
390 |
|
390 | |||
391 | def __init__(self, ui, topic, opts): |
|
391 | def __init__(self, ui, topic, opts): | |
392 | self._ui = ui |
|
392 | self._ui = ui | |
393 | if ui.debugflag: |
|
393 | if ui.debugflag: | |
394 | self.hexfunc = node.hex |
|
394 | self.hexfunc = node.hex | |
395 | else: |
|
395 | else: | |
396 | self.hexfunc = node.short |
|
396 | self.hexfunc = node.short | |
397 |
|
397 | |||
398 | def __nonzero__(self): |
|
398 | def __nonzero__(self): | |
399 | return False |
|
399 | return False | |
400 |
|
400 | |||
401 | __bool__ = __nonzero__ |
|
401 | __bool__ = __nonzero__ | |
402 |
|
402 | |||
403 | def startitem(self): |
|
403 | def startitem(self): | |
404 | pass |
|
404 | pass | |
405 |
|
405 | |||
406 | def data(self, **data): |
|
406 | def data(self, **data): | |
407 | pass |
|
407 | pass | |
408 |
|
408 | |||
409 | def write(self, fields, deftext, *fielddata, **opts): |
|
409 | def write(self, fields, deftext, *fielddata, **opts): | |
410 | self._ui.write(deftext % fielddata, **opts) |
|
410 | self._ui.write(deftext % fielddata, **opts) | |
411 |
|
411 | |||
412 | def condwrite(self, cond, fields, deftext, *fielddata, **opts): |
|
412 | def condwrite(self, cond, fields, deftext, *fielddata, **opts): | |
413 | if cond: |
|
413 | if cond: | |
414 | self._ui.write(deftext % fielddata, **opts) |
|
414 | self._ui.write(deftext % fielddata, **opts) | |
415 |
|
415 | |||
416 | def plain(self, text, **opts): |
|
416 | def plain(self, text, **opts): | |
417 | self._ui.write(text, **opts) |
|
417 | self._ui.write(text, **opts) | |
418 |
|
418 | |||
419 | def end(self): |
|
419 | def end(self): | |
420 | pass |
|
420 | pass | |
421 |
|
421 | |||
422 | fm = defaultformatter(ui, b'perf', opts) |
|
422 | fm = defaultformatter(ui, b'perf', opts) | |
423 |
|
423 | |||
424 | # stub function, runs code only once instead of in a loop |
|
424 | # stub function, runs code only once instead of in a loop | |
425 | # experimental config: perf.stub |
|
425 | # experimental config: perf.stub | |
426 | if ui.configbool(b"perf", b"stub", False): |
|
426 | if ui.configbool(b"perf", b"stub", False): | |
427 | return functools.partial(stub_timer, fm), fm |
|
427 | return functools.partial(stub_timer, fm), fm | |
428 |
|
428 | |||
429 | # experimental config: perf.all-timing |
|
429 | # experimental config: perf.all-timing | |
430 | displayall = ui.configbool(b"perf", b"all-timing", False) |
|
430 | displayall = ui.configbool(b"perf", b"all-timing", False) | |
431 |
|
431 | |||
432 | # experimental config: perf.run-limits |
|
432 | # experimental config: perf.run-limits | |
433 | limitspec = ui.configlist(b"perf", b"run-limits", []) |
|
433 | limitspec = ui.configlist(b"perf", b"run-limits", []) | |
434 | limits = [] |
|
434 | limits = [] | |
435 | for item in limitspec: |
|
435 | for item in limitspec: | |
436 | parts = item.split(b'-', 1) |
|
436 | parts = item.split(b'-', 1) | |
437 | if len(parts) < 2: |
|
437 | if len(parts) < 2: | |
438 | ui.warn((b'malformatted run limit entry, missing "-": %s\n' % item)) |
|
438 | ui.warn((b'malformatted run limit entry, missing "-": %s\n' % item)) | |
439 | continue |
|
439 | continue | |
440 | try: |
|
440 | try: | |
441 | time_limit = float(_sysstr(parts[0])) |
|
441 | time_limit = float(_sysstr(parts[0])) | |
442 | except ValueError as e: |
|
442 | except ValueError as e: | |
443 | ui.warn( |
|
443 | ui.warn( | |
444 | ( |
|
444 | ( | |
445 | b'malformatted run limit entry, %s: %s\n' |
|
445 | b'malformatted run limit entry, %s: %s\n' | |
446 | % (_bytestr(e), item) |
|
446 | % (_bytestr(e), item) | |
447 | ) |
|
447 | ) | |
448 | ) |
|
448 | ) | |
449 | continue |
|
449 | continue | |
450 | try: |
|
450 | try: | |
451 | run_limit = int(_sysstr(parts[1])) |
|
451 | run_limit = int(_sysstr(parts[1])) | |
452 | except ValueError as e: |
|
452 | except ValueError as e: | |
453 | ui.warn( |
|
453 | ui.warn( | |
454 | ( |
|
454 | ( | |
455 | b'malformatted run limit entry, %s: %s\n' |
|
455 | b'malformatted run limit entry, %s: %s\n' | |
456 | % (_bytestr(e), item) |
|
456 | % (_bytestr(e), item) | |
457 | ) |
|
457 | ) | |
458 | ) |
|
458 | ) | |
459 | continue |
|
459 | continue | |
460 | limits.append((time_limit, run_limit)) |
|
460 | limits.append((time_limit, run_limit)) | |
461 | if not limits: |
|
461 | if not limits: | |
462 | limits = DEFAULTLIMITS |
|
462 | limits = DEFAULTLIMITS | |
463 |
|
463 | |||
464 | profiler = None |
|
464 | profiler = None | |
465 | if profiling is not None: |
|
465 | if profiling is not None: | |
466 | if ui.configbool(b"perf", b"profile-benchmark", False): |
|
466 | if ui.configbool(b"perf", b"profile-benchmark", False): | |
467 | profiler = profiling.profile(ui) |
|
467 | profiler = profiling.profile(ui) | |
468 |
|
468 | |||
469 | prerun = getint(ui, b"perf", b"pre-run", 0) |
|
469 | prerun = getint(ui, b"perf", b"pre-run", 0) | |
470 | t = functools.partial( |
|
470 | t = functools.partial( | |
471 | _timer, |
|
471 | _timer, | |
472 | fm, |
|
472 | fm, | |
473 | displayall=displayall, |
|
473 | displayall=displayall, | |
474 | limits=limits, |
|
474 | limits=limits, | |
475 | prerun=prerun, |
|
475 | prerun=prerun, | |
476 | profiler=profiler, |
|
476 | profiler=profiler, | |
477 | ) |
|
477 | ) | |
478 | return t, fm |
|
478 | return t, fm | |
479 |
|
479 | |||
480 |
|
480 | |||
481 | def stub_timer(fm, func, setup=None, title=None): |
|
481 | def stub_timer(fm, func, setup=None, title=None): | |
482 | if setup is not None: |
|
482 | if setup is not None: | |
483 | setup() |
|
483 | setup() | |
484 | func() |
|
484 | func() | |
485 |
|
485 | |||
486 |
|
486 | |||
487 | @contextlib.contextmanager |
|
487 | @contextlib.contextmanager | |
488 | def timeone(): |
|
488 | def timeone(): | |
489 | r = [] |
|
489 | r = [] | |
490 | ostart = os.times() |
|
490 | ostart = os.times() | |
491 | cstart = util.timer() |
|
491 | cstart = util.timer() | |
492 | yield r |
|
492 | yield r | |
493 | cstop = util.timer() |
|
493 | cstop = util.timer() | |
494 | ostop = os.times() |
|
494 | ostop = os.times() | |
495 | a, b = ostart, ostop |
|
495 | a, b = ostart, ostop | |
496 | r.append((cstop - cstart, b[0] - a[0], b[1] - a[1])) |
|
496 | r.append((cstop - cstart, b[0] - a[0], b[1] - a[1])) | |
497 |
|
497 | |||
498 |
|
498 | |||
499 | # list of stop condition (elapsed time, minimal run count) |
|
499 | # list of stop condition (elapsed time, minimal run count) | |
500 | DEFAULTLIMITS = ( |
|
500 | DEFAULTLIMITS = ( | |
501 | (3.0, 100), |
|
501 | (3.0, 100), | |
502 | (10.0, 3), |
|
502 | (10.0, 3), | |
503 | ) |
|
503 | ) | |
504 |
|
504 | |||
505 |
|
505 | |||
506 | def _timer( |
|
506 | def _timer( | |
507 | fm, |
|
507 | fm, | |
508 | func, |
|
508 | func, | |
509 | setup=None, |
|
509 | setup=None, | |
510 | title=None, |
|
510 | title=None, | |
511 | displayall=False, |
|
511 | displayall=False, | |
512 | limits=DEFAULTLIMITS, |
|
512 | limits=DEFAULTLIMITS, | |
513 | prerun=0, |
|
513 | prerun=0, | |
514 | profiler=None, |
|
514 | profiler=None, | |
515 | ): |
|
515 | ): | |
516 | gc.collect() |
|
516 | gc.collect() | |
517 | results = [] |
|
517 | results = [] | |
518 | begin = util.timer() |
|
518 | begin = util.timer() | |
519 | count = 0 |
|
519 | count = 0 | |
520 | if profiler is None: |
|
520 | if profiler is None: | |
521 | profiler = NOOPCTX |
|
521 | profiler = NOOPCTX | |
522 | for i in range(prerun): |
|
522 | for i in range(prerun): | |
523 | if setup is not None: |
|
523 | if setup is not None: | |
524 | setup() |
|
524 | setup() | |
525 | func() |
|
525 | func() | |
526 | keepgoing = True |
|
526 | keepgoing = True | |
527 | while keepgoing: |
|
527 | while keepgoing: | |
528 | if setup is not None: |
|
528 | if setup is not None: | |
529 | setup() |
|
529 | setup() | |
530 | with profiler: |
|
530 | with profiler: | |
531 | with timeone() as item: |
|
531 | with timeone() as item: | |
532 | r = func() |
|
532 | r = func() | |
533 | profiler = NOOPCTX |
|
533 | profiler = NOOPCTX | |
534 | count += 1 |
|
534 | count += 1 | |
535 | results.append(item[0]) |
|
535 | results.append(item[0]) | |
536 | cstop = util.timer() |
|
536 | cstop = util.timer() | |
537 | # Look for a stop condition. |
|
537 | # Look for a stop condition. | |
538 | elapsed = cstop - begin |
|
538 | elapsed = cstop - begin | |
539 | for t, mincount in limits: |
|
539 | for t, mincount in limits: | |
540 | if elapsed >= t and count >= mincount: |
|
540 | if elapsed >= t and count >= mincount: | |
541 | keepgoing = False |
|
541 | keepgoing = False | |
542 | break |
|
542 | break | |
543 |
|
543 | |||
544 | formatone(fm, results, title=title, result=r, displayall=displayall) |
|
544 | formatone(fm, results, title=title, result=r, displayall=displayall) | |
545 |
|
545 | |||
546 |
|
546 | |||
547 | def formatone(fm, timings, title=None, result=None, displayall=False): |
|
547 | def formatone(fm, timings, title=None, result=None, displayall=False): | |
548 |
|
548 | |||
549 | count = len(timings) |
|
549 | count = len(timings) | |
550 |
|
550 | |||
551 | fm.startitem() |
|
551 | fm.startitem() | |
552 |
|
552 | |||
553 | if title: |
|
553 | if title: | |
554 | fm.write(b'title', b'! %s\n', title) |
|
554 | fm.write(b'title', b'! %s\n', title) | |
555 | if result: |
|
555 | if result: | |
556 | fm.write(b'result', b'! result: %s\n', result) |
|
556 | fm.write(b'result', b'! result: %s\n', result) | |
557 |
|
557 | |||
558 | def display(role, entry): |
|
558 | def display(role, entry): | |
559 | prefix = b'' |
|
559 | prefix = b'' | |
560 | if role != b'best': |
|
560 | if role != b'best': | |
561 | prefix = b'%s.' % role |
|
561 | prefix = b'%s.' % role | |
562 | fm.plain(b'!') |
|
562 | fm.plain(b'!') | |
563 | fm.write(prefix + b'wall', b' wall %f', entry[0]) |
|
563 | fm.write(prefix + b'wall', b' wall %f', entry[0]) | |
564 | fm.write(prefix + b'comb', b' comb %f', entry[1] + entry[2]) |
|
564 | fm.write(prefix + b'comb', b' comb %f', entry[1] + entry[2]) | |
565 | fm.write(prefix + b'user', b' user %f', entry[1]) |
|
565 | fm.write(prefix + b'user', b' user %f', entry[1]) | |
566 | fm.write(prefix + b'sys', b' sys %f', entry[2]) |
|
566 | fm.write(prefix + b'sys', b' sys %f', entry[2]) | |
567 | fm.write(prefix + b'count', b' (%s of %%d)' % role, count) |
|
567 | fm.write(prefix + b'count', b' (%s of %%d)' % role, count) | |
568 | fm.plain(b'\n') |
|
568 | fm.plain(b'\n') | |
569 |
|
569 | |||
570 | timings.sort() |
|
570 | timings.sort() | |
571 | min_val = timings[0] |
|
571 | min_val = timings[0] | |
572 | display(b'best', min_val) |
|
572 | display(b'best', min_val) | |
573 | if displayall: |
|
573 | if displayall: | |
574 | max_val = timings[-1] |
|
574 | max_val = timings[-1] | |
575 | display(b'max', max_val) |
|
575 | display(b'max', max_val) | |
576 | avg = tuple([sum(x) / count for x in zip(*timings)]) |
|
576 | avg = tuple([sum(x) / count for x in zip(*timings)]) | |
577 | display(b'avg', avg) |
|
577 | display(b'avg', avg) | |
578 | median = timings[len(timings) // 2] |
|
578 | median = timings[len(timings) // 2] | |
579 | display(b'median', median) |
|
579 | display(b'median', median) | |
580 |
|
580 | |||
581 |
|
581 | |||
582 | # utilities for historical portability |
|
582 | # utilities for historical portability | |
583 |
|
583 | |||
584 |
|
584 | |||
585 | def getint(ui, section, name, default): |
|
585 | def getint(ui, section, name, default): | |
586 | # for "historical portability": |
|
586 | # for "historical portability": | |
587 | # ui.configint has been available since 1.9 (or fa2b596db182) |
|
587 | # ui.configint has been available since 1.9 (or fa2b596db182) | |
588 | v = ui.config(section, name, None) |
|
588 | v = ui.config(section, name, None) | |
589 | if v is None: |
|
589 | if v is None: | |
590 | return default |
|
590 | return default | |
591 | try: |
|
591 | try: | |
592 | return int(v) |
|
592 | return int(v) | |
593 | except ValueError: |
|
593 | except ValueError: | |
594 | raise error.ConfigError( |
|
594 | raise error.ConfigError( | |
595 | b"%s.%s is not an integer ('%s')" % (section, name, v) |
|
595 | b"%s.%s is not an integer ('%s')" % (section, name, v) | |
596 | ) |
|
596 | ) | |
597 |
|
597 | |||
598 |
|
598 | |||
599 | def safeattrsetter(obj, name, ignoremissing=False): |
|
599 | def safeattrsetter(obj, name, ignoremissing=False): | |
600 | """Ensure that 'obj' has 'name' attribute before subsequent setattr |
|
600 | """Ensure that 'obj' has 'name' attribute before subsequent setattr | |
601 |
|
601 | |||
602 | This function is aborted, if 'obj' doesn't have 'name' attribute |
|
602 | This function is aborted, if 'obj' doesn't have 'name' attribute | |
603 | at runtime. This avoids overlooking removal of an attribute, which |
|
603 | at runtime. This avoids overlooking removal of an attribute, which | |
604 | breaks assumption of performance measurement, in the future. |
|
604 | breaks assumption of performance measurement, in the future. | |
605 |
|
605 | |||
606 | This function returns the object to (1) assign a new value, and |
|
606 | This function returns the object to (1) assign a new value, and | |
607 | (2) restore an original value to the attribute. |
|
607 | (2) restore an original value to the attribute. | |
608 |
|
608 | |||
609 | If 'ignoremissing' is true, missing 'name' attribute doesn't cause |
|
609 | If 'ignoremissing' is true, missing 'name' attribute doesn't cause | |
610 | abortion, and this function returns None. This is useful to |
|
610 | abortion, and this function returns None. This is useful to | |
611 | examine an attribute, which isn't ensured in all Mercurial |
|
611 | examine an attribute, which isn't ensured in all Mercurial | |
612 | versions. |
|
612 | versions. | |
613 | """ |
|
613 | """ | |
614 | if not util.safehasattr(obj, name): |
|
614 | if not util.safehasattr(obj, name): | |
615 | if ignoremissing: |
|
615 | if ignoremissing: | |
616 | return None |
|
616 | return None | |
617 | raise error.Abort( |
|
617 | raise error.Abort( | |
618 | ( |
|
618 | ( | |
619 | b"missing attribute %s of %s might break assumption" |
|
619 | b"missing attribute %s of %s might break assumption" | |
620 | b" of performance measurement" |
|
620 | b" of performance measurement" | |
621 | ) |
|
621 | ) | |
622 | % (name, obj) |
|
622 | % (name, obj) | |
623 | ) |
|
623 | ) | |
624 |
|
624 | |||
625 | origvalue = getattr(obj, _sysstr(name)) |
|
625 | origvalue = getattr(obj, _sysstr(name)) | |
626 |
|
626 | |||
627 | class attrutil(object): |
|
627 | class attrutil(object): | |
628 | def set(self, newvalue): |
|
628 | def set(self, newvalue): | |
629 | setattr(obj, _sysstr(name), newvalue) |
|
629 | setattr(obj, _sysstr(name), newvalue) | |
630 |
|
630 | |||
631 | def restore(self): |
|
631 | def restore(self): | |
632 | setattr(obj, _sysstr(name), origvalue) |
|
632 | setattr(obj, _sysstr(name), origvalue) | |
633 |
|
633 | |||
634 | return attrutil() |
|
634 | return attrutil() | |
635 |
|
635 | |||
636 |
|
636 | |||
637 | # utilities to examine each internal API changes |
|
637 | # utilities to examine each internal API changes | |
638 |
|
638 | |||
639 |
|
639 | |||
640 | def getbranchmapsubsettable(): |
|
640 | def getbranchmapsubsettable(): | |
641 | # for "historical portability": |
|
641 | # for "historical portability": | |
642 | # subsettable is defined in: |
|
642 | # subsettable is defined in: | |
643 | # - branchmap since 2.9 (or 175c6fd8cacc) |
|
643 | # - branchmap since 2.9 (or 175c6fd8cacc) | |
644 | # - repoview since 2.5 (or 59a9f18d4587) |
|
644 | # - repoview since 2.5 (or 59a9f18d4587) | |
645 | # - repoviewutil since 5.0 |
|
645 | # - repoviewutil since 5.0 | |
646 | for mod in (branchmap, repoview, repoviewutil): |
|
646 | for mod in (branchmap, repoview, repoviewutil): | |
647 | subsettable = getattr(mod, 'subsettable', None) |
|
647 | subsettable = getattr(mod, 'subsettable', None) | |
648 | if subsettable: |
|
648 | if subsettable: | |
649 | return subsettable |
|
649 | return subsettable | |
650 |
|
650 | |||
651 | # bisecting in bcee63733aad::59a9f18d4587 can reach here (both |
|
651 | # bisecting in bcee63733aad::59a9f18d4587 can reach here (both | |
652 | # branchmap and repoview modules exist, but subsettable attribute |
|
652 | # branchmap and repoview modules exist, but subsettable attribute | |
653 | # doesn't) |
|
653 | # doesn't) | |
654 | raise error.Abort( |
|
654 | raise error.Abort( | |
655 | b"perfbranchmap not available with this Mercurial", |
|
655 | b"perfbranchmap not available with this Mercurial", | |
656 | hint=b"use 2.5 or later", |
|
656 | hint=b"use 2.5 or later", | |
657 | ) |
|
657 | ) | |
658 |
|
658 | |||
659 |
|
659 | |||
660 | def getsvfs(repo): |
|
660 | def getsvfs(repo): | |
661 | """Return appropriate object to access files under .hg/store |
|
661 | """Return appropriate object to access files under .hg/store | |
662 | """ |
|
662 | """ | |
663 | # for "historical portability": |
|
663 | # for "historical portability": | |
664 | # repo.svfs has been available since 2.3 (or 7034365089bf) |
|
664 | # repo.svfs has been available since 2.3 (or 7034365089bf) | |
665 | svfs = getattr(repo, 'svfs', None) |
|
665 | svfs = getattr(repo, 'svfs', None) | |
666 | if svfs: |
|
666 | if svfs: | |
667 | return svfs |
|
667 | return svfs | |
668 | else: |
|
668 | else: | |
669 | return getattr(repo, 'sopener') |
|
669 | return getattr(repo, 'sopener') | |
670 |
|
670 | |||
671 |
|
671 | |||
672 | def getvfs(repo): |
|
672 | def getvfs(repo): | |
673 | """Return appropriate object to access files under .hg |
|
673 | """Return appropriate object to access files under .hg | |
674 | """ |
|
674 | """ | |
675 | # for "historical portability": |
|
675 | # for "historical portability": | |
676 | # repo.vfs has been available since 2.3 (or 7034365089bf) |
|
676 | # repo.vfs has been available since 2.3 (or 7034365089bf) | |
677 | vfs = getattr(repo, 'vfs', None) |
|
677 | vfs = getattr(repo, 'vfs', None) | |
678 | if vfs: |
|
678 | if vfs: | |
679 | return vfs |
|
679 | return vfs | |
680 | else: |
|
680 | else: | |
681 | return getattr(repo, 'opener') |
|
681 | return getattr(repo, 'opener') | |
682 |
|
682 | |||
683 |
|
683 | |||
684 | def repocleartagscachefunc(repo): |
|
684 | def repocleartagscachefunc(repo): | |
685 | """Return the function to clear tags cache according to repo internal API |
|
685 | """Return the function to clear tags cache according to repo internal API | |
686 | """ |
|
686 | """ | |
687 | if util.safehasattr(repo, b'_tagscache'): # since 2.0 (or 9dca7653b525) |
|
687 | if util.safehasattr(repo, b'_tagscache'): # since 2.0 (or 9dca7653b525) | |
688 | # in this case, setattr(repo, '_tagscache', None) or so isn't |
|
688 | # in this case, setattr(repo, '_tagscache', None) or so isn't | |
689 | # correct way to clear tags cache, because existing code paths |
|
689 | # correct way to clear tags cache, because existing code paths | |
690 | # expect _tagscache to be a structured object. |
|
690 | # expect _tagscache to be a structured object. | |
691 | def clearcache(): |
|
691 | def clearcache(): | |
692 | # _tagscache has been filteredpropertycache since 2.5 (or |
|
692 | # _tagscache has been filteredpropertycache since 2.5 (or | |
693 | # 98c867ac1330), and delattr() can't work in such case |
|
693 | # 98c867ac1330), and delattr() can't work in such case | |
694 | if '_tagscache' in vars(repo): |
|
694 | if '_tagscache' in vars(repo): | |
695 | del repo.__dict__['_tagscache'] |
|
695 | del repo.__dict__['_tagscache'] | |
696 |
|
696 | |||
697 | return clearcache |
|
697 | return clearcache | |
698 |
|
698 | |||
699 | repotags = safeattrsetter(repo, b'_tags', ignoremissing=True) |
|
699 | repotags = safeattrsetter(repo, b'_tags', ignoremissing=True) | |
700 | if repotags: # since 1.4 (or 5614a628d173) |
|
700 | if repotags: # since 1.4 (or 5614a628d173) | |
701 | return lambda: repotags.set(None) |
|
701 | return lambda: repotags.set(None) | |
702 |
|
702 | |||
703 | repotagscache = safeattrsetter(repo, b'tagscache', ignoremissing=True) |
|
703 | repotagscache = safeattrsetter(repo, b'tagscache', ignoremissing=True) | |
704 | if repotagscache: # since 0.6 (or d7df759d0e97) |
|
704 | if repotagscache: # since 0.6 (or d7df759d0e97) | |
705 | return lambda: repotagscache.set(None) |
|
705 | return lambda: repotagscache.set(None) | |
706 |
|
706 | |||
707 | # Mercurial earlier than 0.6 (or d7df759d0e97) logically reaches |
|
707 | # Mercurial earlier than 0.6 (or d7df759d0e97) logically reaches | |
708 | # this point, but it isn't so problematic, because: |
|
708 | # this point, but it isn't so problematic, because: | |
709 | # - repo.tags of such Mercurial isn't "callable", and repo.tags() |
|
709 | # - repo.tags of such Mercurial isn't "callable", and repo.tags() | |
710 | # in perftags() causes failure soon |
|
710 | # in perftags() causes failure soon | |
711 | # - perf.py itself has been available since 1.1 (or eb240755386d) |
|
711 | # - perf.py itself has been available since 1.1 (or eb240755386d) | |
712 | raise error.Abort(b"tags API of this hg command is unknown") |
|
712 | raise error.Abort(b"tags API of this hg command is unknown") | |
713 |
|
713 | |||
714 |
|
714 | |||
715 | # utilities to clear cache |
|
715 | # utilities to clear cache | |
716 |
|
716 | |||
717 |
|
717 | |||
718 | def clearfilecache(obj, attrname): |
|
718 | def clearfilecache(obj, attrname): | |
719 | unfiltered = getattr(obj, 'unfiltered', None) |
|
719 | unfiltered = getattr(obj, 'unfiltered', None) | |
720 | if unfiltered is not None: |
|
720 | if unfiltered is not None: | |
721 | obj = obj.unfiltered() |
|
721 | obj = obj.unfiltered() | |
722 | if attrname in vars(obj): |
|
722 | if attrname in vars(obj): | |
723 | delattr(obj, attrname) |
|
723 | delattr(obj, attrname) | |
724 | obj._filecache.pop(attrname, None) |
|
724 | obj._filecache.pop(attrname, None) | |
725 |
|
725 | |||
726 |
|
726 | |||
727 | def clearchangelog(repo): |
|
727 | def clearchangelog(repo): | |
728 | if repo is not repo.unfiltered(): |
|
728 | if repo is not repo.unfiltered(): | |
729 | object.__setattr__(repo, '_clcachekey', None) |
|
729 | object.__setattr__(repo, '_clcachekey', None) | |
730 | object.__setattr__(repo, '_clcache', None) |
|
730 | object.__setattr__(repo, '_clcache', None) | |
731 | clearfilecache(repo.unfiltered(), 'changelog') |
|
731 | clearfilecache(repo.unfiltered(), 'changelog') | |
732 |
|
732 | |||
733 |
|
733 | |||
734 | # perf commands |
|
734 | # perf commands | |
735 |
|
735 | |||
736 |
|
736 | |||
737 | @command(b'perfwalk', formatteropts) |
|
737 | @command(b'perfwalk', formatteropts) | |
738 | def perfwalk(ui, repo, *pats, **opts): |
|
738 | def perfwalk(ui, repo, *pats, **opts): | |
739 | opts = _byteskwargs(opts) |
|
739 | opts = _byteskwargs(opts) | |
740 | timer, fm = gettimer(ui, opts) |
|
740 | timer, fm = gettimer(ui, opts) | |
741 | m = scmutil.match(repo[None], pats, {}) |
|
741 | m = scmutil.match(repo[None], pats, {}) | |
742 | timer( |
|
742 | timer( | |
743 | lambda: len( |
|
743 | lambda: len( | |
744 | list( |
|
744 | list( | |
745 | repo.dirstate.walk(m, subrepos=[], unknown=True, ignored=False) |
|
745 | repo.dirstate.walk(m, subrepos=[], unknown=True, ignored=False) | |
746 | ) |
|
746 | ) | |
747 | ) |
|
747 | ) | |
748 | ) |
|
748 | ) | |
749 | fm.end() |
|
749 | fm.end() | |
750 |
|
750 | |||
751 |
|
751 | |||
752 | @command(b'perfannotate', formatteropts) |
|
752 | @command(b'perfannotate', formatteropts) | |
753 | def perfannotate(ui, repo, f, **opts): |
|
753 | def perfannotate(ui, repo, f, **opts): | |
754 | opts = _byteskwargs(opts) |
|
754 | opts = _byteskwargs(opts) | |
755 | timer, fm = gettimer(ui, opts) |
|
755 | timer, fm = gettimer(ui, opts) | |
756 | fc = repo[b'.'][f] |
|
756 | fc = repo[b'.'][f] | |
757 | timer(lambda: len(fc.annotate(True))) |
|
757 | timer(lambda: len(fc.annotate(True))) | |
758 | fm.end() |
|
758 | fm.end() | |
759 |
|
759 | |||
760 |
|
760 | |||
761 | @command( |
|
761 | @command( | |
762 | b'perfstatus', |
|
762 | b'perfstatus', | |
763 | [ |
|
763 | [ | |
764 | (b'u', b'unknown', False, b'ask status to look for unknown files'), |
|
764 | (b'u', b'unknown', False, b'ask status to look for unknown files'), | |
765 | (b'', b'dirstate', False, b'benchmark the internal dirstate call'), |
|
765 | (b'', b'dirstate', False, b'benchmark the internal dirstate call'), | |
766 | ] |
|
766 | ] | |
767 | + formatteropts, |
|
767 | + formatteropts, | |
768 | ) |
|
768 | ) | |
769 | def perfstatus(ui, repo, **opts): |
|
769 | def perfstatus(ui, repo, **opts): | |
770 | """benchmark the performance of a single status call |
|
770 | """benchmark the performance of a single status call | |
771 |
|
771 | |||
772 | The repository data are preserved between each call. |
|
772 | The repository data are preserved between each call. | |
773 |
|
773 | |||
774 | By default, only the status of the tracked file are requested. If |
|
774 | By default, only the status of the tracked file are requested. If | |
775 | `--unknown` is passed, the "unknown" files are also tracked. |
|
775 | `--unknown` is passed, the "unknown" files are also tracked. | |
776 | """ |
|
776 | """ | |
777 | opts = _byteskwargs(opts) |
|
777 | opts = _byteskwargs(opts) | |
778 | # m = match.always(repo.root, repo.getcwd()) |
|
778 | # m = match.always(repo.root, repo.getcwd()) | |
779 | # timer(lambda: sum(map(len, repo.dirstate.status(m, [], False, False, |
|
779 | # timer(lambda: sum(map(len, repo.dirstate.status(m, [], False, False, | |
780 | # False)))) |
|
780 | # False)))) | |
781 | timer, fm = gettimer(ui, opts) |
|
781 | timer, fm = gettimer(ui, opts) | |
782 | if opts[b'dirstate']: |
|
782 | if opts[b'dirstate']: | |
783 | dirstate = repo.dirstate |
|
783 | dirstate = repo.dirstate | |
784 | m = scmutil.matchall(repo) |
|
784 | m = scmutil.matchall(repo) | |
785 | unknown = opts[b'unknown'] |
|
785 | unknown = opts[b'unknown'] | |
786 |
|
786 | |||
787 | def status_dirstate(): |
|
787 | def status_dirstate(): | |
788 | s = dirstate.status( |
|
788 | s = dirstate.status( | |
789 | m, subrepos=[], ignored=False, clean=False, unknown=unknown |
|
789 | m, subrepos=[], ignored=False, clean=False, unknown=unknown | |
790 | ) |
|
790 | ) | |
791 | sum(map(len, s)) |
|
791 | sum(map(len, s)) | |
792 |
|
792 | |||
793 | timer(status_dirstate) |
|
793 | timer(status_dirstate) | |
794 | else: |
|
794 | else: | |
795 | timer(lambda: sum(map(len, repo.status(unknown=opts[b'unknown'])))) |
|
795 | timer(lambda: sum(map(len, repo.status(unknown=opts[b'unknown'])))) | |
796 | fm.end() |
|
796 | fm.end() | |
797 |
|
797 | |||
798 |
|
798 | |||
799 | @command(b'perfaddremove', formatteropts) |
|
799 | @command(b'perfaddremove', formatteropts) | |
800 | def perfaddremove(ui, repo, **opts): |
|
800 | def perfaddremove(ui, repo, **opts): | |
801 | opts = _byteskwargs(opts) |
|
801 | opts = _byteskwargs(opts) | |
802 | timer, fm = gettimer(ui, opts) |
|
802 | timer, fm = gettimer(ui, opts) | |
803 | try: |
|
803 | try: | |
804 | oldquiet = repo.ui.quiet |
|
804 | oldquiet = repo.ui.quiet | |
805 | repo.ui.quiet = True |
|
805 | repo.ui.quiet = True | |
806 | matcher = scmutil.match(repo[None]) |
|
806 | matcher = scmutil.match(repo[None]) | |
807 | opts[b'dry_run'] = True |
|
807 | opts[b'dry_run'] = True | |
808 | if b'uipathfn' in getargspec(scmutil.addremove).args: |
|
808 | if b'uipathfn' in getargspec(scmutil.addremove).args: | |
809 | uipathfn = scmutil.getuipathfn(repo) |
|
809 | uipathfn = scmutil.getuipathfn(repo) | |
810 | timer(lambda: scmutil.addremove(repo, matcher, b"", uipathfn, opts)) |
|
810 | timer(lambda: scmutil.addremove(repo, matcher, b"", uipathfn, opts)) | |
811 | else: |
|
811 | else: | |
812 | timer(lambda: scmutil.addremove(repo, matcher, b"", opts)) |
|
812 | timer(lambda: scmutil.addremove(repo, matcher, b"", opts)) | |
813 | finally: |
|
813 | finally: | |
814 | repo.ui.quiet = oldquiet |
|
814 | repo.ui.quiet = oldquiet | |
815 | fm.end() |
|
815 | fm.end() | |
816 |
|
816 | |||
817 |
|
817 | |||
818 | def clearcaches(cl): |
|
818 | def clearcaches(cl): | |
819 | # behave somewhat consistently across internal API changes |
|
819 | # behave somewhat consistently across internal API changes | |
820 | if util.safehasattr(cl, b'clearcaches'): |
|
820 | if util.safehasattr(cl, b'clearcaches'): | |
821 | cl.clearcaches() |
|
821 | cl.clearcaches() | |
822 | elif util.safehasattr(cl, b'_nodecache'): |
|
822 | elif util.safehasattr(cl, b'_nodecache'): | |
823 | from mercurial.node import nullid, nullrev |
|
823 | from mercurial.node import nullid, nullrev | |
824 |
|
824 | |||
825 | cl._nodecache = {nullid: nullrev} |
|
825 | cl._nodecache = {nullid: nullrev} | |
826 | cl._nodepos = None |
|
826 | cl._nodepos = None | |
827 |
|
827 | |||
828 |
|
828 | |||
829 | @command(b'perfheads', formatteropts) |
|
829 | @command(b'perfheads', formatteropts) | |
830 | def perfheads(ui, repo, **opts): |
|
830 | def perfheads(ui, repo, **opts): | |
831 | """benchmark the computation of a changelog heads""" |
|
831 | """benchmark the computation of a changelog heads""" | |
832 | opts = _byteskwargs(opts) |
|
832 | opts = _byteskwargs(opts) | |
833 | timer, fm = gettimer(ui, opts) |
|
833 | timer, fm = gettimer(ui, opts) | |
834 | cl = repo.changelog |
|
834 | cl = repo.changelog | |
835 |
|
835 | |||
836 | def s(): |
|
836 | def s(): | |
837 | clearcaches(cl) |
|
837 | clearcaches(cl) | |
838 |
|
838 | |||
839 | def d(): |
|
839 | def d(): | |
840 | len(cl.headrevs()) |
|
840 | len(cl.headrevs()) | |
841 |
|
841 | |||
842 | timer(d, setup=s) |
|
842 | timer(d, setup=s) | |
843 | fm.end() |
|
843 | fm.end() | |
844 |
|
844 | |||
845 |
|
845 | |||
846 | @command( |
|
846 | @command( | |
847 | b'perftags', |
|
847 | b'perftags', | |
848 | formatteropts |
|
848 | formatteropts | |
849 | + [(b'', b'clear-revlogs', False, b'refresh changelog and manifest'),], |
|
849 | + [(b'', b'clear-revlogs', False, b'refresh changelog and manifest'),], | |
850 | ) |
|
850 | ) | |
851 | def perftags(ui, repo, **opts): |
|
851 | def perftags(ui, repo, **opts): | |
852 | opts = _byteskwargs(opts) |
|
852 | opts = _byteskwargs(opts) | |
853 | timer, fm = gettimer(ui, opts) |
|
853 | timer, fm = gettimer(ui, opts) | |
854 | repocleartagscache = repocleartagscachefunc(repo) |
|
854 | repocleartagscache = repocleartagscachefunc(repo) | |
855 | clearrevlogs = opts[b'clear_revlogs'] |
|
855 | clearrevlogs = opts[b'clear_revlogs'] | |
856 |
|
856 | |||
857 | def s(): |
|
857 | def s(): | |
858 | if clearrevlogs: |
|
858 | if clearrevlogs: | |
859 | clearchangelog(repo) |
|
859 | clearchangelog(repo) | |
860 | clearfilecache(repo.unfiltered(), 'manifest') |
|
860 | clearfilecache(repo.unfiltered(), 'manifest') | |
861 | repocleartagscache() |
|
861 | repocleartagscache() | |
862 |
|
862 | |||
863 | def t(): |
|
863 | def t(): | |
864 | return len(repo.tags()) |
|
864 | return len(repo.tags()) | |
865 |
|
865 | |||
866 | timer(t, setup=s) |
|
866 | timer(t, setup=s) | |
867 | fm.end() |
|
867 | fm.end() | |
868 |
|
868 | |||
869 |
|
869 | |||
870 | @command(b'perfancestors', formatteropts) |
|
870 | @command(b'perfancestors', formatteropts) | |
871 | def perfancestors(ui, repo, **opts): |
|
871 | def perfancestors(ui, repo, **opts): | |
872 | opts = _byteskwargs(opts) |
|
872 | opts = _byteskwargs(opts) | |
873 | timer, fm = gettimer(ui, opts) |
|
873 | timer, fm = gettimer(ui, opts) | |
874 | heads = repo.changelog.headrevs() |
|
874 | heads = repo.changelog.headrevs() | |
875 |
|
875 | |||
876 | def d(): |
|
876 | def d(): | |
877 | for a in repo.changelog.ancestors(heads): |
|
877 | for a in repo.changelog.ancestors(heads): | |
878 | pass |
|
878 | pass | |
879 |
|
879 | |||
880 | timer(d) |
|
880 | timer(d) | |
881 | fm.end() |
|
881 | fm.end() | |
882 |
|
882 | |||
883 |
|
883 | |||
884 | @command(b'perfancestorset', formatteropts) |
|
884 | @command(b'perfancestorset', formatteropts) | |
885 | def perfancestorset(ui, repo, revset, **opts): |
|
885 | def perfancestorset(ui, repo, revset, **opts): | |
886 | opts = _byteskwargs(opts) |
|
886 | opts = _byteskwargs(opts) | |
887 | timer, fm = gettimer(ui, opts) |
|
887 | timer, fm = gettimer(ui, opts) | |
888 | revs = repo.revs(revset) |
|
888 | revs = repo.revs(revset) | |
889 | heads = repo.changelog.headrevs() |
|
889 | heads = repo.changelog.headrevs() | |
890 |
|
890 | |||
891 | def d(): |
|
891 | def d(): | |
892 | s = repo.changelog.ancestors(heads) |
|
892 | s = repo.changelog.ancestors(heads) | |
893 | for rev in revs: |
|
893 | for rev in revs: | |
894 | rev in s |
|
894 | rev in s | |
895 |
|
895 | |||
896 | timer(d) |
|
896 | timer(d) | |
897 | fm.end() |
|
897 | fm.end() | |
898 |
|
898 | |||
899 |
|
899 | |||
900 | @command(b'perfdiscovery', formatteropts, b'PATH') |
|
900 | @command(b'perfdiscovery', formatteropts, b'PATH') | |
901 | def perfdiscovery(ui, repo, path, **opts): |
|
901 | def perfdiscovery(ui, repo, path, **opts): | |
902 | """benchmark discovery between local repo and the peer at given path |
|
902 | """benchmark discovery between local repo and the peer at given path | |
903 | """ |
|
903 | """ | |
904 | repos = [repo, None] |
|
904 | repos = [repo, None] | |
905 | timer, fm = gettimer(ui, opts) |
|
905 | timer, fm = gettimer(ui, opts) | |
906 | path = ui.expandpath(path) |
|
906 | path = ui.expandpath(path) | |
907 |
|
907 | |||
908 | def s(): |
|
908 | def s(): | |
909 | repos[1] = hg.peer(ui, opts, path) |
|
909 | repos[1] = hg.peer(ui, opts, path) | |
910 |
|
910 | |||
911 | def d(): |
|
911 | def d(): | |
912 | setdiscovery.findcommonheads(ui, *repos) |
|
912 | setdiscovery.findcommonheads(ui, *repos) | |
913 |
|
913 | |||
914 | timer(d, setup=s) |
|
914 | timer(d, setup=s) | |
915 | fm.end() |
|
915 | fm.end() | |
916 |
|
916 | |||
917 |
|
917 | |||
918 | @command( |
|
918 | @command( | |
919 | b'perfbookmarks', |
|
919 | b'perfbookmarks', | |
920 | formatteropts |
|
920 | formatteropts | |
921 | + [(b'', b'clear-revlogs', False, b'refresh changelog and manifest'),], |
|
921 | + [(b'', b'clear-revlogs', False, b'refresh changelog and manifest'),], | |
922 | ) |
|
922 | ) | |
923 | def perfbookmarks(ui, repo, **opts): |
|
923 | def perfbookmarks(ui, repo, **opts): | |
924 | """benchmark parsing bookmarks from disk to memory""" |
|
924 | """benchmark parsing bookmarks from disk to memory""" | |
925 | opts = _byteskwargs(opts) |
|
925 | opts = _byteskwargs(opts) | |
926 | timer, fm = gettimer(ui, opts) |
|
926 | timer, fm = gettimer(ui, opts) | |
927 |
|
927 | |||
928 | clearrevlogs = opts[b'clear_revlogs'] |
|
928 | clearrevlogs = opts[b'clear_revlogs'] | |
929 |
|
929 | |||
930 | def s(): |
|
930 | def s(): | |
931 | if clearrevlogs: |
|
931 | if clearrevlogs: | |
932 | clearchangelog(repo) |
|
932 | clearchangelog(repo) | |
933 | clearfilecache(repo, b'_bookmarks') |
|
933 | clearfilecache(repo, b'_bookmarks') | |
934 |
|
934 | |||
935 | def d(): |
|
935 | def d(): | |
936 | repo._bookmarks |
|
936 | repo._bookmarks | |
937 |
|
937 | |||
938 | timer(d, setup=s) |
|
938 | timer(d, setup=s) | |
939 | fm.end() |
|
939 | fm.end() | |
940 |
|
940 | |||
941 |
|
941 | |||
942 | @command(b'perfbundleread', formatteropts, b'BUNDLE') |
|
942 | @command(b'perfbundleread', formatteropts, b'BUNDLE') | |
943 | def perfbundleread(ui, repo, bundlepath, **opts): |
|
943 | def perfbundleread(ui, repo, bundlepath, **opts): | |
944 | """Benchmark reading of bundle files. |
|
944 | """Benchmark reading of bundle files. | |
945 |
|
945 | |||
946 | This command is meant to isolate the I/O part of bundle reading as |
|
946 | This command is meant to isolate the I/O part of bundle reading as | |
947 | much as possible. |
|
947 | much as possible. | |
948 | """ |
|
948 | """ | |
949 | from mercurial import ( |
|
949 | from mercurial import ( | |
950 | bundle2, |
|
950 | bundle2, | |
951 | exchange, |
|
951 | exchange, | |
952 | streamclone, |
|
952 | streamclone, | |
953 | ) |
|
953 | ) | |
954 |
|
954 | |||
955 | opts = _byteskwargs(opts) |
|
955 | opts = _byteskwargs(opts) | |
956 |
|
956 | |||
957 | def makebench(fn): |
|
957 | def makebench(fn): | |
958 | def run(): |
|
958 | def run(): | |
959 | with open(bundlepath, b'rb') as fh: |
|
959 | with open(bundlepath, b'rb') as fh: | |
960 | bundle = exchange.readbundle(ui, fh, bundlepath) |
|
960 | bundle = exchange.readbundle(ui, fh, bundlepath) | |
961 | fn(bundle) |
|
961 | fn(bundle) | |
962 |
|
962 | |||
963 | return run |
|
963 | return run | |
964 |
|
964 | |||
965 | def makereadnbytes(size): |
|
965 | def makereadnbytes(size): | |
966 | def run(): |
|
966 | def run(): | |
967 | with open(bundlepath, b'rb') as fh: |
|
967 | with open(bundlepath, b'rb') as fh: | |
968 | bundle = exchange.readbundle(ui, fh, bundlepath) |
|
968 | bundle = exchange.readbundle(ui, fh, bundlepath) | |
969 | while bundle.read(size): |
|
969 | while bundle.read(size): | |
970 | pass |
|
970 | pass | |
971 |
|
971 | |||
972 | return run |
|
972 | return run | |
973 |
|
973 | |||
974 | def makestdioread(size): |
|
974 | def makestdioread(size): | |
975 | def run(): |
|
975 | def run(): | |
976 | with open(bundlepath, b'rb') as fh: |
|
976 | with open(bundlepath, b'rb') as fh: | |
977 | while fh.read(size): |
|
977 | while fh.read(size): | |
978 | pass |
|
978 | pass | |
979 |
|
979 | |||
980 | return run |
|
980 | return run | |
981 |
|
981 | |||
982 | # bundle1 |
|
982 | # bundle1 | |
983 |
|
983 | |||
984 | def deltaiter(bundle): |
|
984 | def deltaiter(bundle): | |
985 | for delta in bundle.deltaiter(): |
|
985 | for delta in bundle.deltaiter(): | |
986 | pass |
|
986 | pass | |
987 |
|
987 | |||
988 | def iterchunks(bundle): |
|
988 | def iterchunks(bundle): | |
989 | for chunk in bundle.getchunks(): |
|
989 | for chunk in bundle.getchunks(): | |
990 | pass |
|
990 | pass | |
991 |
|
991 | |||
992 | # bundle2 |
|
992 | # bundle2 | |
993 |
|
993 | |||
994 | def forwardchunks(bundle): |
|
994 | def forwardchunks(bundle): | |
995 | for chunk in bundle._forwardchunks(): |
|
995 | for chunk in bundle._forwardchunks(): | |
996 | pass |
|
996 | pass | |
997 |
|
997 | |||
998 | def iterparts(bundle): |
|
998 | def iterparts(bundle): | |
999 | for part in bundle.iterparts(): |
|
999 | for part in bundle.iterparts(): | |
1000 | pass |
|
1000 | pass | |
1001 |
|
1001 | |||
1002 | def iterpartsseekable(bundle): |
|
1002 | def iterpartsseekable(bundle): | |
1003 | for part in bundle.iterparts(seekable=True): |
|
1003 | for part in bundle.iterparts(seekable=True): | |
1004 | pass |
|
1004 | pass | |
1005 |
|
1005 | |||
1006 | def seek(bundle): |
|
1006 | def seek(bundle): | |
1007 | for part in bundle.iterparts(seekable=True): |
|
1007 | for part in bundle.iterparts(seekable=True): | |
1008 | part.seek(0, os.SEEK_END) |
|
1008 | part.seek(0, os.SEEK_END) | |
1009 |
|
1009 | |||
1010 | def makepartreadnbytes(size): |
|
1010 | def makepartreadnbytes(size): | |
1011 | def run(): |
|
1011 | def run(): | |
1012 | with open(bundlepath, b'rb') as fh: |
|
1012 | with open(bundlepath, b'rb') as fh: | |
1013 | bundle = exchange.readbundle(ui, fh, bundlepath) |
|
1013 | bundle = exchange.readbundle(ui, fh, bundlepath) | |
1014 | for part in bundle.iterparts(): |
|
1014 | for part in bundle.iterparts(): | |
1015 | while part.read(size): |
|
1015 | while part.read(size): | |
1016 | pass |
|
1016 | pass | |
1017 |
|
1017 | |||
1018 | return run |
|
1018 | return run | |
1019 |
|
1019 | |||
1020 | benches = [ |
|
1020 | benches = [ | |
1021 | (makestdioread(8192), b'read(8k)'), |
|
1021 | (makestdioread(8192), b'read(8k)'), | |
1022 | (makestdioread(16384), b'read(16k)'), |
|
1022 | (makestdioread(16384), b'read(16k)'), | |
1023 | (makestdioread(32768), b'read(32k)'), |
|
1023 | (makestdioread(32768), b'read(32k)'), | |
1024 | (makestdioread(131072), b'read(128k)'), |
|
1024 | (makestdioread(131072), b'read(128k)'), | |
1025 | ] |
|
1025 | ] | |
1026 |
|
1026 | |||
1027 | with open(bundlepath, b'rb') as fh: |
|
1027 | with open(bundlepath, b'rb') as fh: | |
1028 | bundle = exchange.readbundle(ui, fh, bundlepath) |
|
1028 | bundle = exchange.readbundle(ui, fh, bundlepath) | |
1029 |
|
1029 | |||
1030 | if isinstance(bundle, changegroup.cg1unpacker): |
|
1030 | if isinstance(bundle, changegroup.cg1unpacker): | |
1031 | benches.extend( |
|
1031 | benches.extend( | |
1032 | [ |
|
1032 | [ | |
1033 | (makebench(deltaiter), b'cg1 deltaiter()'), |
|
1033 | (makebench(deltaiter), b'cg1 deltaiter()'), | |
1034 | (makebench(iterchunks), b'cg1 getchunks()'), |
|
1034 | (makebench(iterchunks), b'cg1 getchunks()'), | |
1035 | (makereadnbytes(8192), b'cg1 read(8k)'), |
|
1035 | (makereadnbytes(8192), b'cg1 read(8k)'), | |
1036 | (makereadnbytes(16384), b'cg1 read(16k)'), |
|
1036 | (makereadnbytes(16384), b'cg1 read(16k)'), | |
1037 | (makereadnbytes(32768), b'cg1 read(32k)'), |
|
1037 | (makereadnbytes(32768), b'cg1 read(32k)'), | |
1038 | (makereadnbytes(131072), b'cg1 read(128k)'), |
|
1038 | (makereadnbytes(131072), b'cg1 read(128k)'), | |
1039 | ] |
|
1039 | ] | |
1040 | ) |
|
1040 | ) | |
1041 | elif isinstance(bundle, bundle2.unbundle20): |
|
1041 | elif isinstance(bundle, bundle2.unbundle20): | |
1042 | benches.extend( |
|
1042 | benches.extend( | |
1043 | [ |
|
1043 | [ | |
1044 | (makebench(forwardchunks), b'bundle2 forwardchunks()'), |
|
1044 | (makebench(forwardchunks), b'bundle2 forwardchunks()'), | |
1045 | (makebench(iterparts), b'bundle2 iterparts()'), |
|
1045 | (makebench(iterparts), b'bundle2 iterparts()'), | |
1046 | ( |
|
1046 | ( | |
1047 | makebench(iterpartsseekable), |
|
1047 | makebench(iterpartsseekable), | |
1048 | b'bundle2 iterparts() seekable', |
|
1048 | b'bundle2 iterparts() seekable', | |
1049 | ), |
|
1049 | ), | |
1050 | (makebench(seek), b'bundle2 part seek()'), |
|
1050 | (makebench(seek), b'bundle2 part seek()'), | |
1051 | (makepartreadnbytes(8192), b'bundle2 part read(8k)'), |
|
1051 | (makepartreadnbytes(8192), b'bundle2 part read(8k)'), | |
1052 | (makepartreadnbytes(16384), b'bundle2 part read(16k)'), |
|
1052 | (makepartreadnbytes(16384), b'bundle2 part read(16k)'), | |
1053 | (makepartreadnbytes(32768), b'bundle2 part read(32k)'), |
|
1053 | (makepartreadnbytes(32768), b'bundle2 part read(32k)'), | |
1054 | (makepartreadnbytes(131072), b'bundle2 part read(128k)'), |
|
1054 | (makepartreadnbytes(131072), b'bundle2 part read(128k)'), | |
1055 | ] |
|
1055 | ] | |
1056 | ) |
|
1056 | ) | |
1057 | elif isinstance(bundle, streamclone.streamcloneapplier): |
|
1057 | elif isinstance(bundle, streamclone.streamcloneapplier): | |
1058 | raise error.Abort(b'stream clone bundles not supported') |
|
1058 | raise error.Abort(b'stream clone bundles not supported') | |
1059 | else: |
|
1059 | else: | |
1060 | raise error.Abort(b'unhandled bundle type: %s' % type(bundle)) |
|
1060 | raise error.Abort(b'unhandled bundle type: %s' % type(bundle)) | |
1061 |
|
1061 | |||
1062 | for fn, title in benches: |
|
1062 | for fn, title in benches: | |
1063 | timer, fm = gettimer(ui, opts) |
|
1063 | timer, fm = gettimer(ui, opts) | |
1064 | timer(fn, title=title) |
|
1064 | timer(fn, title=title) | |
1065 | fm.end() |
|
1065 | fm.end() | |
1066 |
|
1066 | |||
1067 |
|
1067 | |||
1068 | @command( |
|
1068 | @command( | |
1069 | b'perfchangegroupchangelog', |
|
1069 | b'perfchangegroupchangelog', | |
1070 | formatteropts |
|
1070 | formatteropts | |
1071 | + [ |
|
1071 | + [ | |
1072 | (b'', b'cgversion', b'02', b'changegroup version'), |
|
1072 | (b'', b'cgversion', b'02', b'changegroup version'), | |
1073 | (b'r', b'rev', b'', b'revisions to add to changegroup'), |
|
1073 | (b'r', b'rev', b'', b'revisions to add to changegroup'), | |
1074 | ], |
|
1074 | ], | |
1075 | ) |
|
1075 | ) | |
1076 | def perfchangegroupchangelog(ui, repo, cgversion=b'02', rev=None, **opts): |
|
1076 | def perfchangegroupchangelog(ui, repo, cgversion=b'02', rev=None, **opts): | |
1077 | """Benchmark producing a changelog group for a changegroup. |
|
1077 | """Benchmark producing a changelog group for a changegroup. | |
1078 |
|
1078 | |||
1079 | This measures the time spent processing the changelog during a |
|
1079 | This measures the time spent processing the changelog during a | |
1080 | bundle operation. This occurs during `hg bundle` and on a server |
|
1080 | bundle operation. This occurs during `hg bundle` and on a server | |
1081 | processing a `getbundle` wire protocol request (handles clones |
|
1081 | processing a `getbundle` wire protocol request (handles clones | |
1082 | and pull requests). |
|
1082 | and pull requests). | |
1083 |
|
1083 | |||
1084 | By default, all revisions are added to the changegroup. |
|
1084 | By default, all revisions are added to the changegroup. | |
1085 | """ |
|
1085 | """ | |
1086 | opts = _byteskwargs(opts) |
|
1086 | opts = _byteskwargs(opts) | |
1087 | cl = repo.changelog |
|
1087 | cl = repo.changelog | |
1088 | nodes = [cl.lookup(r) for r in repo.revs(rev or b'all()')] |
|
1088 | nodes = [cl.lookup(r) for r in repo.revs(rev or b'all()')] | |
1089 | bundler = changegroup.getbundler(cgversion, repo) |
|
1089 | bundler = changegroup.getbundler(cgversion, repo) | |
1090 |
|
1090 | |||
1091 | def d(): |
|
1091 | def d(): | |
1092 | state, chunks = bundler._generatechangelog(cl, nodes) |
|
1092 | state, chunks = bundler._generatechangelog(cl, nodes) | |
1093 | for chunk in chunks: |
|
1093 | for chunk in chunks: | |
1094 | pass |
|
1094 | pass | |
1095 |
|
1095 | |||
1096 | timer, fm = gettimer(ui, opts) |
|
1096 | timer, fm = gettimer(ui, opts) | |
1097 |
|
1097 | |||
1098 | # Terminal printing can interfere with timing. So disable it. |
|
1098 | # Terminal printing can interfere with timing. So disable it. | |
1099 | with ui.configoverride({(b'progress', b'disable'): True}): |
|
1099 | with ui.configoverride({(b'progress', b'disable'): True}): | |
1100 | timer(d) |
|
1100 | timer(d) | |
1101 |
|
1101 | |||
1102 | fm.end() |
|
1102 | fm.end() | |
1103 |
|
1103 | |||
1104 |
|
1104 | |||
1105 | @command(b'perfdirs', formatteropts) |
|
1105 | @command(b'perfdirs', formatteropts) | |
1106 | def perfdirs(ui, repo, **opts): |
|
1106 | def perfdirs(ui, repo, **opts): | |
1107 | opts = _byteskwargs(opts) |
|
1107 | opts = _byteskwargs(opts) | |
1108 | timer, fm = gettimer(ui, opts) |
|
1108 | timer, fm = gettimer(ui, opts) | |
1109 | dirstate = repo.dirstate |
|
1109 | dirstate = repo.dirstate | |
1110 | b'a' in dirstate |
|
1110 | b'a' in dirstate | |
1111 |
|
1111 | |||
1112 | def d(): |
|
1112 | def d(): | |
1113 | dirstate.hasdir(b'a') |
|
1113 | dirstate.hasdir(b'a') | |
1114 | del dirstate._map._dirs |
|
1114 | del dirstate._map._dirs | |
1115 |
|
1115 | |||
1116 | timer(d) |
|
1116 | timer(d) | |
1117 | fm.end() |
|
1117 | fm.end() | |
1118 |
|
1118 | |||
1119 |
|
1119 | |||
1120 | @command( |
|
1120 | @command( | |
1121 | b'perfdirstate', |
|
1121 | b'perfdirstate', | |
1122 | [ |
|
1122 | [ | |
1123 | ( |
|
1123 | ( | |
1124 | b'', |
|
1124 | b'', | |
1125 | b'iteration', |
|
1125 | b'iteration', | |
1126 | None, |
|
1126 | None, | |
1127 | b'benchmark a full iteration for the dirstate', |
|
1127 | b'benchmark a full iteration for the dirstate', | |
1128 | ), |
|
1128 | ), | |
1129 | ( |
|
1129 | ( | |
1130 | b'', |
|
1130 | b'', | |
1131 | b'contains', |
|
1131 | b'contains', | |
1132 | None, |
|
1132 | None, | |
1133 | b'benchmark a large amount of `nf in dirstate` calls', |
|
1133 | b'benchmark a large amount of `nf in dirstate` calls', | |
1134 | ), |
|
1134 | ), | |
1135 | ] |
|
1135 | ] | |
1136 | + formatteropts, |
|
1136 | + formatteropts, | |
1137 | ) |
|
1137 | ) | |
1138 | def perfdirstate(ui, repo, **opts): |
|
1138 | def perfdirstate(ui, repo, **opts): | |
1139 | """benchmap the time of various distate operations |
|
1139 | """benchmap the time of various distate operations | |
1140 |
|
1140 | |||
1141 | By default benchmark the time necessary to load a dirstate from scratch. |
|
1141 | By default benchmark the time necessary to load a dirstate from scratch. | |
1142 | The dirstate is loaded to the point were a "contains" request can be |
|
1142 | The dirstate is loaded to the point were a "contains" request can be | |
1143 | answered. |
|
1143 | answered. | |
1144 | """ |
|
1144 | """ | |
1145 | opts = _byteskwargs(opts) |
|
1145 | opts = _byteskwargs(opts) | |
1146 | timer, fm = gettimer(ui, opts) |
|
1146 | timer, fm = gettimer(ui, opts) | |
1147 | b"a" in repo.dirstate |
|
1147 | b"a" in repo.dirstate | |
1148 |
|
1148 | |||
1149 | if opts[b'iteration'] and opts[b'contains']: |
|
1149 | if opts[b'iteration'] and opts[b'contains']: | |
1150 | msg = b'only specify one of --iteration or --contains' |
|
1150 | msg = b'only specify one of --iteration or --contains' | |
1151 | raise error.Abort(msg) |
|
1151 | raise error.Abort(msg) | |
1152 |
|
1152 | |||
1153 | if opts[b'iteration']: |
|
1153 | if opts[b'iteration']: | |
1154 | setup = None |
|
1154 | setup = None | |
1155 | dirstate = repo.dirstate |
|
1155 | dirstate = repo.dirstate | |
1156 |
|
1156 | |||
1157 | def d(): |
|
1157 | def d(): | |
1158 | for f in dirstate: |
|
1158 | for f in dirstate: | |
1159 | pass |
|
1159 | pass | |
1160 |
|
1160 | |||
1161 | elif opts[b'contains']: |
|
1161 | elif opts[b'contains']: | |
1162 | setup = None |
|
1162 | setup = None | |
1163 | dirstate = repo.dirstate |
|
1163 | dirstate = repo.dirstate | |
1164 | allfiles = list(dirstate) |
|
1164 | allfiles = list(dirstate) | |
1165 | # also add file path that will be "missing" from the dirstate |
|
1165 | # also add file path that will be "missing" from the dirstate | |
1166 | allfiles.extend([f[::-1] for f in allfiles]) |
|
1166 | allfiles.extend([f[::-1] for f in allfiles]) | |
1167 |
|
1167 | |||
1168 | def d(): |
|
1168 | def d(): | |
1169 | for f in allfiles: |
|
1169 | for f in allfiles: | |
1170 | f in dirstate |
|
1170 | f in dirstate | |
1171 |
|
1171 | |||
1172 | else: |
|
1172 | else: | |
1173 |
|
1173 | |||
1174 | def setup(): |
|
1174 | def setup(): | |
1175 | repo.dirstate.invalidate() |
|
1175 | repo.dirstate.invalidate() | |
1176 |
|
1176 | |||
1177 | def d(): |
|
1177 | def d(): | |
1178 | b"a" in repo.dirstate |
|
1178 | b"a" in repo.dirstate | |
1179 |
|
1179 | |||
1180 | timer(d, setup=setup) |
|
1180 | timer(d, setup=setup) | |
1181 | fm.end() |
|
1181 | fm.end() | |
1182 |
|
1182 | |||
1183 |
|
1183 | |||
1184 | @command(b'perfdirstatedirs', formatteropts) |
|
1184 | @command(b'perfdirstatedirs', formatteropts) | |
1185 | def perfdirstatedirs(ui, repo, **opts): |
|
1185 | def perfdirstatedirs(ui, repo, **opts): | |
1186 | """benchmap a 'dirstate.hasdir' call from an empty `dirs` cache |
|
1186 | """benchmap a 'dirstate.hasdir' call from an empty `dirs` cache | |
1187 | """ |
|
1187 | """ | |
1188 | opts = _byteskwargs(opts) |
|
1188 | opts = _byteskwargs(opts) | |
1189 | timer, fm = gettimer(ui, opts) |
|
1189 | timer, fm = gettimer(ui, opts) | |
1190 | repo.dirstate.hasdir(b"a") |
|
1190 | repo.dirstate.hasdir(b"a") | |
1191 |
|
1191 | |||
1192 | def setup(): |
|
1192 | def setup(): | |
1193 | del repo.dirstate._map._dirs |
|
1193 | del repo.dirstate._map._dirs | |
1194 |
|
1194 | |||
1195 | def d(): |
|
1195 | def d(): | |
1196 | repo.dirstate.hasdir(b"a") |
|
1196 | repo.dirstate.hasdir(b"a") | |
1197 |
|
1197 | |||
1198 | timer(d, setup=setup) |
|
1198 | timer(d, setup=setup) | |
1199 | fm.end() |
|
1199 | fm.end() | |
1200 |
|
1200 | |||
1201 |
|
1201 | |||
1202 | @command(b'perfdirstatefoldmap', formatteropts) |
|
1202 | @command(b'perfdirstatefoldmap', formatteropts) | |
1203 | def perfdirstatefoldmap(ui, repo, **opts): |
|
1203 | def perfdirstatefoldmap(ui, repo, **opts): | |
1204 | """benchmap a `dirstate._map.filefoldmap.get()` request |
|
1204 | """benchmap a `dirstate._map.filefoldmap.get()` request | |
1205 |
|
1205 | |||
1206 | The dirstate filefoldmap cache is dropped between every request. |
|
1206 | The dirstate filefoldmap cache is dropped between every request. | |
1207 | """ |
|
1207 | """ | |
1208 | opts = _byteskwargs(opts) |
|
1208 | opts = _byteskwargs(opts) | |
1209 | timer, fm = gettimer(ui, opts) |
|
1209 | timer, fm = gettimer(ui, opts) | |
1210 | dirstate = repo.dirstate |
|
1210 | dirstate = repo.dirstate | |
1211 | dirstate._map.filefoldmap.get(b'a') |
|
1211 | dirstate._map.filefoldmap.get(b'a') | |
1212 |
|
1212 | |||
1213 | def setup(): |
|
1213 | def setup(): | |
1214 | del dirstate._map.filefoldmap |
|
1214 | del dirstate._map.filefoldmap | |
1215 |
|
1215 | |||
1216 | def d(): |
|
1216 | def d(): | |
1217 | dirstate._map.filefoldmap.get(b'a') |
|
1217 | dirstate._map.filefoldmap.get(b'a') | |
1218 |
|
1218 | |||
1219 | timer(d, setup=setup) |
|
1219 | timer(d, setup=setup) | |
1220 | fm.end() |
|
1220 | fm.end() | |
1221 |
|
1221 | |||
1222 |
|
1222 | |||
1223 | @command(b'perfdirfoldmap', formatteropts) |
|
1223 | @command(b'perfdirfoldmap', formatteropts) | |
1224 | def perfdirfoldmap(ui, repo, **opts): |
|
1224 | def perfdirfoldmap(ui, repo, **opts): | |
1225 | """benchmap a `dirstate._map.dirfoldmap.get()` request |
|
1225 | """benchmap a `dirstate._map.dirfoldmap.get()` request | |
1226 |
|
1226 | |||
1227 | The dirstate dirfoldmap cache is dropped between every request. |
|
1227 | The dirstate dirfoldmap cache is dropped between every request. | |
1228 | """ |
|
1228 | """ | |
1229 | opts = _byteskwargs(opts) |
|
1229 | opts = _byteskwargs(opts) | |
1230 | timer, fm = gettimer(ui, opts) |
|
1230 | timer, fm = gettimer(ui, opts) | |
1231 | dirstate = repo.dirstate |
|
1231 | dirstate = repo.dirstate | |
1232 | dirstate._map.dirfoldmap.get(b'a') |
|
1232 | dirstate._map.dirfoldmap.get(b'a') | |
1233 |
|
1233 | |||
1234 | def setup(): |
|
1234 | def setup(): | |
1235 | del dirstate._map.dirfoldmap |
|
1235 | del dirstate._map.dirfoldmap | |
1236 | del dirstate._map._dirs |
|
1236 | del dirstate._map._dirs | |
1237 |
|
1237 | |||
1238 | def d(): |
|
1238 | def d(): | |
1239 | dirstate._map.dirfoldmap.get(b'a') |
|
1239 | dirstate._map.dirfoldmap.get(b'a') | |
1240 |
|
1240 | |||
1241 | timer(d, setup=setup) |
|
1241 | timer(d, setup=setup) | |
1242 | fm.end() |
|
1242 | fm.end() | |
1243 |
|
1243 | |||
1244 |
|
1244 | |||
1245 | @command(b'perfdirstatewrite', formatteropts) |
|
1245 | @command(b'perfdirstatewrite', formatteropts) | |
1246 | def perfdirstatewrite(ui, repo, **opts): |
|
1246 | def perfdirstatewrite(ui, repo, **opts): | |
1247 | """benchmap the time it take to write a dirstate on disk |
|
1247 | """benchmap the time it take to write a dirstate on disk | |
1248 | """ |
|
1248 | """ | |
1249 | opts = _byteskwargs(opts) |
|
1249 | opts = _byteskwargs(opts) | |
1250 | timer, fm = gettimer(ui, opts) |
|
1250 | timer, fm = gettimer(ui, opts) | |
1251 | ds = repo.dirstate |
|
1251 | ds = repo.dirstate | |
1252 | b"a" in ds |
|
1252 | b"a" in ds | |
1253 |
|
1253 | |||
1254 | def setup(): |
|
1254 | def setup(): | |
1255 | ds._dirty = True |
|
1255 | ds._dirty = True | |
1256 |
|
1256 | |||
1257 | def d(): |
|
1257 | def d(): | |
1258 | ds.write(repo.currenttransaction()) |
|
1258 | ds.write(repo.currenttransaction()) | |
1259 |
|
1259 | |||
1260 | timer(d, setup=setup) |
|
1260 | timer(d, setup=setup) | |
1261 | fm.end() |
|
1261 | fm.end() | |
1262 |
|
1262 | |||
1263 |
|
1263 | |||
1264 | def _getmergerevs(repo, opts): |
|
1264 | def _getmergerevs(repo, opts): | |
1265 | """parse command argument to return rev involved in merge |
|
1265 | """parse command argument to return rev involved in merge | |
1266 |
|
1266 | |||
1267 | input: options dictionnary with `rev`, `from` and `bse` |
|
1267 | input: options dictionnary with `rev`, `from` and `bse` | |
1268 | output: (localctx, otherctx, basectx) |
|
1268 | output: (localctx, otherctx, basectx) | |
1269 | """ |
|
1269 | """ | |
1270 | if opts[b'from']: |
|
1270 | if opts[b'from']: | |
1271 | fromrev = scmutil.revsingle(repo, opts[b'from']) |
|
1271 | fromrev = scmutil.revsingle(repo, opts[b'from']) | |
1272 | wctx = repo[fromrev] |
|
1272 | wctx = repo[fromrev] | |
1273 | else: |
|
1273 | else: | |
1274 | wctx = repo[None] |
|
1274 | wctx = repo[None] | |
1275 | # we don't want working dir files to be stat'd in the benchmark, so |
|
1275 | # we don't want working dir files to be stat'd in the benchmark, so | |
1276 | # prime that cache |
|
1276 | # prime that cache | |
1277 | wctx.dirty() |
|
1277 | wctx.dirty() | |
1278 | rctx = scmutil.revsingle(repo, opts[b'rev'], opts[b'rev']) |
|
1278 | rctx = scmutil.revsingle(repo, opts[b'rev'], opts[b'rev']) | |
1279 | if opts[b'base']: |
|
1279 | if opts[b'base']: | |
1280 | fromrev = scmutil.revsingle(repo, opts[b'base']) |
|
1280 | fromrev = scmutil.revsingle(repo, opts[b'base']) | |
1281 | ancestor = repo[fromrev] |
|
1281 | ancestor = repo[fromrev] | |
1282 | else: |
|
1282 | else: | |
1283 | ancestor = wctx.ancestor(rctx) |
|
1283 | ancestor = wctx.ancestor(rctx) | |
1284 | return (wctx, rctx, ancestor) |
|
1284 | return (wctx, rctx, ancestor) | |
1285 |
|
1285 | |||
1286 |
|
1286 | |||
1287 | @command( |
|
1287 | @command( | |
1288 | b'perfmergecalculate', |
|
1288 | b'perfmergecalculate', | |
1289 | [ |
|
1289 | [ | |
1290 | (b'r', b'rev', b'.', b'rev to merge against'), |
|
1290 | (b'r', b'rev', b'.', b'rev to merge against'), | |
1291 | (b'', b'from', b'', b'rev to merge from'), |
|
1291 | (b'', b'from', b'', b'rev to merge from'), | |
1292 | (b'', b'base', b'', b'the revision to use as base'), |
|
1292 | (b'', b'base', b'', b'the revision to use as base'), | |
1293 | ] |
|
1293 | ] | |
1294 | + formatteropts, |
|
1294 | + formatteropts, | |
1295 | ) |
|
1295 | ) | |
1296 | def perfmergecalculate(ui, repo, **opts): |
|
1296 | def perfmergecalculate(ui, repo, **opts): | |
1297 | opts = _byteskwargs(opts) |
|
1297 | opts = _byteskwargs(opts) | |
1298 | timer, fm = gettimer(ui, opts) |
|
1298 | timer, fm = gettimer(ui, opts) | |
1299 |
|
1299 | |||
1300 | wctx, rctx, ancestor = _getmergerevs(repo, opts) |
|
1300 | wctx, rctx, ancestor = _getmergerevs(repo, opts) | |
1301 |
|
1301 | |||
1302 | def d(): |
|
1302 | def d(): | |
1303 | # acceptremote is True because we don't want prompts in the middle of |
|
1303 | # acceptremote is True because we don't want prompts in the middle of | |
1304 | # our benchmark |
|
1304 | # our benchmark | |
1305 | merge.calculateupdates( |
|
1305 | merge.calculateupdates( | |
1306 | repo, |
|
1306 | repo, | |
1307 | wctx, |
|
1307 | wctx, | |
1308 | rctx, |
|
1308 | rctx, | |
1309 | [ancestor], |
|
1309 | [ancestor], | |
1310 | branchmerge=False, |
|
1310 | branchmerge=False, | |
1311 | force=False, |
|
1311 | force=False, | |
1312 | acceptremote=True, |
|
1312 | acceptremote=True, | |
1313 | followcopies=True, |
|
1313 | followcopies=True, | |
1314 | ) |
|
1314 | ) | |
1315 |
|
1315 | |||
1316 | timer(d) |
|
1316 | timer(d) | |
1317 | fm.end() |
|
1317 | fm.end() | |
1318 |
|
1318 | |||
1319 |
|
1319 | |||
1320 | @command( |
|
1320 | @command( | |
1321 | b'perfmergecopies', |
|
1321 | b'perfmergecopies', | |
1322 | [ |
|
1322 | [ | |
1323 | (b'r', b'rev', b'.', b'rev to merge against'), |
|
1323 | (b'r', b'rev', b'.', b'rev to merge against'), | |
1324 | (b'', b'from', b'', b'rev to merge from'), |
|
1324 | (b'', b'from', b'', b'rev to merge from'), | |
1325 | (b'', b'base', b'', b'the revision to use as base'), |
|
1325 | (b'', b'base', b'', b'the revision to use as base'), | |
1326 | ] |
|
1326 | ] | |
1327 | + formatteropts, |
|
1327 | + formatteropts, | |
1328 | ) |
|
1328 | ) | |
1329 | def perfmergecopies(ui, repo, **opts): |
|
1329 | def perfmergecopies(ui, repo, **opts): | |
1330 | """measure runtime of `copies.mergecopies`""" |
|
1330 | """measure runtime of `copies.mergecopies`""" | |
1331 | opts = _byteskwargs(opts) |
|
1331 | opts = _byteskwargs(opts) | |
1332 | timer, fm = gettimer(ui, opts) |
|
1332 | timer, fm = gettimer(ui, opts) | |
1333 | wctx, rctx, ancestor = _getmergerevs(repo, opts) |
|
1333 | wctx, rctx, ancestor = _getmergerevs(repo, opts) | |
1334 |
|
1334 | |||
1335 | def d(): |
|
1335 | def d(): | |
1336 | # acceptremote is True because we don't want prompts in the middle of |
|
1336 | # acceptremote is True because we don't want prompts in the middle of | |
1337 | # our benchmark |
|
1337 | # our benchmark | |
1338 | copies.mergecopies(repo, wctx, rctx, ancestor) |
|
1338 | copies.mergecopies(repo, wctx, rctx, ancestor) | |
1339 |
|
1339 | |||
1340 | timer(d) |
|
1340 | timer(d) | |
1341 | fm.end() |
|
1341 | fm.end() | |
1342 |
|
1342 | |||
1343 |
|
1343 | |||
1344 | @command(b'perfpathcopies', [], b"REV REV") |
|
1344 | @command(b'perfpathcopies', [], b"REV REV") | |
1345 | def perfpathcopies(ui, repo, rev1, rev2, **opts): |
|
1345 | def perfpathcopies(ui, repo, rev1, rev2, **opts): | |
1346 | """benchmark the copy tracing logic""" |
|
1346 | """benchmark the copy tracing logic""" | |
1347 | opts = _byteskwargs(opts) |
|
1347 | opts = _byteskwargs(opts) | |
1348 | timer, fm = gettimer(ui, opts) |
|
1348 | timer, fm = gettimer(ui, opts) | |
1349 | ctx1 = scmutil.revsingle(repo, rev1, rev1) |
|
1349 | ctx1 = scmutil.revsingle(repo, rev1, rev1) | |
1350 | ctx2 = scmutil.revsingle(repo, rev2, rev2) |
|
1350 | ctx2 = scmutil.revsingle(repo, rev2, rev2) | |
1351 |
|
1351 | |||
1352 | def d(): |
|
1352 | def d(): | |
1353 | copies.pathcopies(ctx1, ctx2) |
|
1353 | copies.pathcopies(ctx1, ctx2) | |
1354 |
|
1354 | |||
1355 | timer(d) |
|
1355 | timer(d) | |
1356 | fm.end() |
|
1356 | fm.end() | |
1357 |
|
1357 | |||
1358 |
|
1358 | |||
1359 | @command( |
|
1359 | @command( | |
1360 | b'perfphases', |
|
1360 | b'perfphases', | |
1361 | [(b'', b'full', False, b'include file reading time too'),], |
|
1361 | [(b'', b'full', False, b'include file reading time too'),], | |
1362 | b"", |
|
1362 | b"", | |
1363 | ) |
|
1363 | ) | |
1364 | def perfphases(ui, repo, **opts): |
|
1364 | def perfphases(ui, repo, **opts): | |
1365 | """benchmark phasesets computation""" |
|
1365 | """benchmark phasesets computation""" | |
1366 | opts = _byteskwargs(opts) |
|
1366 | opts = _byteskwargs(opts) | |
1367 | timer, fm = gettimer(ui, opts) |
|
1367 | timer, fm = gettimer(ui, opts) | |
1368 | _phases = repo._phasecache |
|
1368 | _phases = repo._phasecache | |
1369 | full = opts.get(b'full') |
|
1369 | full = opts.get(b'full') | |
1370 |
|
1370 | |||
1371 | def d(): |
|
1371 | def d(): | |
1372 | phases = _phases |
|
1372 | phases = _phases | |
1373 | if full: |
|
1373 | if full: | |
1374 | clearfilecache(repo, b'_phasecache') |
|
1374 | clearfilecache(repo, b'_phasecache') | |
1375 | phases = repo._phasecache |
|
1375 | phases = repo._phasecache | |
1376 | phases.invalidate() |
|
1376 | phases.invalidate() | |
1377 | phases.loadphaserevs(repo) |
|
1377 | phases.loadphaserevs(repo) | |
1378 |
|
1378 | |||
1379 | timer(d) |
|
1379 | timer(d) | |
1380 | fm.end() |
|
1380 | fm.end() | |
1381 |
|
1381 | |||
1382 |
|
1382 | |||
1383 | @command(b'perfphasesremote', [], b"[DEST]") |
|
1383 | @command(b'perfphasesremote', [], b"[DEST]") | |
1384 | def perfphasesremote(ui, repo, dest=None, **opts): |
|
1384 | def perfphasesremote(ui, repo, dest=None, **opts): | |
1385 | """benchmark time needed to analyse phases of the remote server""" |
|
1385 | """benchmark time needed to analyse phases of the remote server""" | |
1386 | from mercurial.node import bin |
|
1386 | from mercurial.node import bin | |
1387 | from mercurial import ( |
|
1387 | from mercurial import ( | |
1388 | exchange, |
|
1388 | exchange, | |
1389 | hg, |
|
1389 | hg, | |
1390 | phases, |
|
1390 | phases, | |
1391 | ) |
|
1391 | ) | |
1392 |
|
1392 | |||
1393 | opts = _byteskwargs(opts) |
|
1393 | opts = _byteskwargs(opts) | |
1394 | timer, fm = gettimer(ui, opts) |
|
1394 | timer, fm = gettimer(ui, opts) | |
1395 |
|
1395 | |||
1396 | path = ui.paths.getpath(dest, default=(b'default-push', b'default')) |
|
1396 | path = ui.paths.getpath(dest, default=(b'default-push', b'default')) | |
1397 | if not path: |
|
1397 | if not path: | |
1398 | raise error.Abort( |
|
1398 | raise error.Abort( | |
1399 | b'default repository not configured!', |
|
1399 | b'default repository not configured!', | |
1400 | hint=b"see 'hg help config.paths'", |
|
1400 | hint=b"see 'hg help config.paths'", | |
1401 | ) |
|
1401 | ) | |
1402 | dest = path.pushloc or path.loc |
|
1402 | dest = path.pushloc or path.loc | |
1403 | ui.statusnoi18n(b'analysing phase of %s\n' % util.hidepassword(dest)) |
|
1403 | ui.statusnoi18n(b'analysing phase of %s\n' % util.hidepassword(dest)) | |
1404 | other = hg.peer(repo, opts, dest) |
|
1404 | other = hg.peer(repo, opts, dest) | |
1405 |
|
1405 | |||
1406 | # easier to perform discovery through the operation |
|
1406 | # easier to perform discovery through the operation | |
1407 | op = exchange.pushoperation(repo, other) |
|
1407 | op = exchange.pushoperation(repo, other) | |
1408 | exchange._pushdiscoverychangeset(op) |
|
1408 | exchange._pushdiscoverychangeset(op) | |
1409 |
|
1409 | |||
1410 | remotesubset = op.fallbackheads |
|
1410 | remotesubset = op.fallbackheads | |
1411 |
|
1411 | |||
1412 | with other.commandexecutor() as e: |
|
1412 | with other.commandexecutor() as e: | |
1413 | remotephases = e.callcommand( |
|
1413 | remotephases = e.callcommand( | |
1414 | b'listkeys', {b'namespace': b'phases'} |
|
1414 | b'listkeys', {b'namespace': b'phases'} | |
1415 | ).result() |
|
1415 | ).result() | |
1416 | del other |
|
1416 | del other | |
1417 | publishing = remotephases.get(b'publishing', False) |
|
1417 | publishing = remotephases.get(b'publishing', False) | |
1418 | if publishing: |
|
1418 | if publishing: | |
1419 | ui.statusnoi18n(b'publishing: yes\n') |
|
1419 | ui.statusnoi18n(b'publishing: yes\n') | |
1420 | else: |
|
1420 | else: | |
1421 | ui.statusnoi18n(b'publishing: no\n') |
|
1421 | ui.statusnoi18n(b'publishing: no\n') | |
1422 |
|
1422 | |||
1423 | nodemap = repo.changelog.nodemap |
|
1423 | nodemap = repo.changelog.nodemap | |
1424 | nonpublishroots = 0 |
|
1424 | nonpublishroots = 0 | |
1425 | for nhex, phase in remotephases.iteritems(): |
|
1425 | for nhex, phase in remotephases.iteritems(): | |
1426 | if nhex == b'publishing': # ignore data related to publish option |
|
1426 | if nhex == b'publishing': # ignore data related to publish option | |
1427 | continue |
|
1427 | continue | |
1428 | node = bin(nhex) |
|
1428 | node = bin(nhex) | |
1429 | if node in nodemap and int(phase): |
|
1429 | if node in nodemap and int(phase): | |
1430 | nonpublishroots += 1 |
|
1430 | nonpublishroots += 1 | |
1431 | ui.statusnoi18n(b'number of roots: %d\n' % len(remotephases)) |
|
1431 | ui.statusnoi18n(b'number of roots: %d\n' % len(remotephases)) | |
1432 | ui.statusnoi18n(b'number of known non public roots: %d\n' % nonpublishroots) |
|
1432 | ui.statusnoi18n(b'number of known non public roots: %d\n' % nonpublishroots) | |
1433 |
|
1433 | |||
1434 | def d(): |
|
1434 | def d(): | |
1435 | phases.remotephasessummary(repo, remotesubset, remotephases) |
|
1435 | phases.remotephasessummary(repo, remotesubset, remotephases) | |
1436 |
|
1436 | |||
1437 | timer(d) |
|
1437 | timer(d) | |
1438 | fm.end() |
|
1438 | fm.end() | |
1439 |
|
1439 | |||
1440 |
|
1440 | |||
1441 | @command( |
|
1441 | @command( | |
1442 | b'perfmanifest', |
|
1442 | b'perfmanifest', | |
1443 | [ |
|
1443 | [ | |
1444 | (b'm', b'manifest-rev', False, b'Look up a manifest node revision'), |
|
1444 | (b'm', b'manifest-rev', False, b'Look up a manifest node revision'), | |
1445 | (b'', b'clear-disk', False, b'clear on-disk caches too'), |
|
1445 | (b'', b'clear-disk', False, b'clear on-disk caches too'), | |
1446 | ] |
|
1446 | ] | |
1447 | + formatteropts, |
|
1447 | + formatteropts, | |
1448 | b'REV|NODE', |
|
1448 | b'REV|NODE', | |
1449 | ) |
|
1449 | ) | |
1450 | def perfmanifest(ui, repo, rev, manifest_rev=False, clear_disk=False, **opts): |
|
1450 | def perfmanifest(ui, repo, rev, manifest_rev=False, clear_disk=False, **opts): | |
1451 | """benchmark the time to read a manifest from disk and return a usable |
|
1451 | """benchmark the time to read a manifest from disk and return a usable | |
1452 | dict-like object |
|
1452 | dict-like object | |
1453 |
|
1453 | |||
1454 | Manifest caches are cleared before retrieval.""" |
|
1454 | Manifest caches are cleared before retrieval.""" | |
1455 | opts = _byteskwargs(opts) |
|
1455 | opts = _byteskwargs(opts) | |
1456 | timer, fm = gettimer(ui, opts) |
|
1456 | timer, fm = gettimer(ui, opts) | |
1457 | if not manifest_rev: |
|
1457 | if not manifest_rev: | |
1458 | ctx = scmutil.revsingle(repo, rev, rev) |
|
1458 | ctx = scmutil.revsingle(repo, rev, rev) | |
1459 | t = ctx.manifestnode() |
|
1459 | t = ctx.manifestnode() | |
1460 | else: |
|
1460 | else: | |
1461 | from mercurial.node import bin |
|
1461 | from mercurial.node import bin | |
1462 |
|
1462 | |||
1463 | if len(rev) == 40: |
|
1463 | if len(rev) == 40: | |
1464 | t = bin(rev) |
|
1464 | t = bin(rev) | |
1465 | else: |
|
1465 | else: | |
1466 | try: |
|
1466 | try: | |
1467 | rev = int(rev) |
|
1467 | rev = int(rev) | |
1468 |
|
1468 | |||
1469 | if util.safehasattr(repo.manifestlog, b'getstorage'): |
|
1469 | if util.safehasattr(repo.manifestlog, b'getstorage'): | |
1470 | t = repo.manifestlog.getstorage(b'').node(rev) |
|
1470 | t = repo.manifestlog.getstorage(b'').node(rev) | |
1471 | else: |
|
1471 | else: | |
1472 | t = repo.manifestlog._revlog.lookup(rev) |
|
1472 | t = repo.manifestlog._revlog.lookup(rev) | |
1473 | except ValueError: |
|
1473 | except ValueError: | |
1474 | raise error.Abort( |
|
1474 | raise error.Abort( | |
1475 | b'manifest revision must be integer or full node' |
|
1475 | b'manifest revision must be integer or full node' | |
1476 | ) |
|
1476 | ) | |
1477 |
|
1477 | |||
1478 | def d(): |
|
1478 | def d(): | |
1479 | repo.manifestlog.clearcaches(clear_persisted_data=clear_disk) |
|
1479 | repo.manifestlog.clearcaches(clear_persisted_data=clear_disk) | |
1480 | repo.manifestlog[t].read() |
|
1480 | repo.manifestlog[t].read() | |
1481 |
|
1481 | |||
1482 | timer(d) |
|
1482 | timer(d) | |
1483 | fm.end() |
|
1483 | fm.end() | |
1484 |
|
1484 | |||
1485 |
|
1485 | |||
1486 | @command(b'perfchangeset', formatteropts) |
|
1486 | @command(b'perfchangeset', formatteropts) | |
1487 | def perfchangeset(ui, repo, rev, **opts): |
|
1487 | def perfchangeset(ui, repo, rev, **opts): | |
1488 | opts = _byteskwargs(opts) |
|
1488 | opts = _byteskwargs(opts) | |
1489 | timer, fm = gettimer(ui, opts) |
|
1489 | timer, fm = gettimer(ui, opts) | |
1490 | n = scmutil.revsingle(repo, rev).node() |
|
1490 | n = scmutil.revsingle(repo, rev).node() | |
1491 |
|
1491 | |||
1492 | def d(): |
|
1492 | def d(): | |
1493 | repo.changelog.read(n) |
|
1493 | repo.changelog.read(n) | |
1494 | # repo.changelog._cache = None |
|
1494 | # repo.changelog._cache = None | |
1495 |
|
1495 | |||
1496 | timer(d) |
|
1496 | timer(d) | |
1497 | fm.end() |
|
1497 | fm.end() | |
1498 |
|
1498 | |||
1499 |
|
1499 | |||
1500 | @command(b'perfignore', formatteropts) |
|
1500 | @command(b'perfignore', formatteropts) | |
1501 | def perfignore(ui, repo, **opts): |
|
1501 | def perfignore(ui, repo, **opts): | |
1502 | """benchmark operation related to computing ignore""" |
|
1502 | """benchmark operation related to computing ignore""" | |
1503 | opts = _byteskwargs(opts) |
|
1503 | opts = _byteskwargs(opts) | |
1504 | timer, fm = gettimer(ui, opts) |
|
1504 | timer, fm = gettimer(ui, opts) | |
1505 | dirstate = repo.dirstate |
|
1505 | dirstate = repo.dirstate | |
1506 |
|
1506 | |||
1507 | def setupone(): |
|
1507 | def setupone(): | |
1508 | dirstate.invalidate() |
|
1508 | dirstate.invalidate() | |
1509 | clearfilecache(dirstate, b'_ignore') |
|
1509 | clearfilecache(dirstate, b'_ignore') | |
1510 |
|
1510 | |||
1511 | def runone(): |
|
1511 | def runone(): | |
1512 | dirstate._ignore |
|
1512 | dirstate._ignore | |
1513 |
|
1513 | |||
1514 | timer(runone, setup=setupone, title=b"load") |
|
1514 | timer(runone, setup=setupone, title=b"load") | |
1515 | fm.end() |
|
1515 | fm.end() | |
1516 |
|
1516 | |||
1517 |
|
1517 | |||
1518 | @command( |
|
1518 | @command( | |
1519 | b'perfindex', |
|
1519 | b'perfindex', | |
1520 | [ |
|
1520 | [ | |
1521 | (b'', b'rev', [], b'revision to be looked up (default tip)'), |
|
1521 | (b'', b'rev', [], b'revision to be looked up (default tip)'), | |
1522 | (b'', b'no-lookup', None, b'do not revision lookup post creation'), |
|
1522 | (b'', b'no-lookup', None, b'do not revision lookup post creation'), | |
1523 | ] |
|
1523 | ] | |
1524 | + formatteropts, |
|
1524 | + formatteropts, | |
1525 | ) |
|
1525 | ) | |
1526 | def perfindex(ui, repo, **opts): |
|
1526 | def perfindex(ui, repo, **opts): | |
1527 | """benchmark index creation time followed by a lookup |
|
1527 | """benchmark index creation time followed by a lookup | |
1528 |
|
1528 | |||
1529 | The default is to look `tip` up. Depending on the index implementation, |
|
1529 | The default is to look `tip` up. Depending on the index implementation, | |
1530 | the revision looked up can matters. For example, an implementation |
|
1530 | the revision looked up can matters. For example, an implementation | |
1531 | scanning the index will have a faster lookup time for `--rev tip` than for |
|
1531 | scanning the index will have a faster lookup time for `--rev tip` than for | |
1532 | `--rev 0`. The number of looked up revisions and their order can also |
|
1532 | `--rev 0`. The number of looked up revisions and their order can also | |
1533 | matters. |
|
1533 | matters. | |
1534 |
|
1534 | |||
1535 | Example of useful set to test: |
|
1535 | Example of useful set to test: | |
1536 | * tip |
|
1536 | * tip | |
1537 | * 0 |
|
1537 | * 0 | |
1538 | * -10: |
|
1538 | * -10: | |
1539 | * :10 |
|
1539 | * :10 | |
1540 | * -10: + :10 |
|
1540 | * -10: + :10 | |
1541 | * :10: + -10: |
|
1541 | * :10: + -10: | |
1542 | * -10000: |
|
1542 | * -10000: | |
1543 | * -10000: + 0 |
|
1543 | * -10000: + 0 | |
1544 |
|
1544 | |||
1545 | It is not currently possible to check for lookup of a missing node. For |
|
1545 | It is not currently possible to check for lookup of a missing node. For | |
1546 | deeper lookup benchmarking, checkout the `perfnodemap` command.""" |
|
1546 | deeper lookup benchmarking, checkout the `perfnodemap` command.""" | |
1547 | import mercurial.revlog |
|
1547 | import mercurial.revlog | |
1548 |
|
1548 | |||
1549 | opts = _byteskwargs(opts) |
|
1549 | opts = _byteskwargs(opts) | |
1550 | timer, fm = gettimer(ui, opts) |
|
1550 | timer, fm = gettimer(ui, opts) | |
1551 | mercurial.revlog._prereadsize = 2 ** 24 # disable lazy parser in old hg |
|
1551 | mercurial.revlog._prereadsize = 2 ** 24 # disable lazy parser in old hg | |
1552 | if opts[b'no_lookup']: |
|
1552 | if opts[b'no_lookup']: | |
1553 | if opts['rev']: |
|
1553 | if opts['rev']: | |
1554 | raise error.Abort('--no-lookup and --rev are mutually exclusive') |
|
1554 | raise error.Abort('--no-lookup and --rev are mutually exclusive') | |
1555 | nodes = [] |
|
1555 | nodes = [] | |
1556 | elif not opts[b'rev']: |
|
1556 | elif not opts[b'rev']: | |
1557 | nodes = [repo[b"tip"].node()] |
|
1557 | nodes = [repo[b"tip"].node()] | |
1558 | else: |
|
1558 | else: | |
1559 | revs = scmutil.revrange(repo, opts[b'rev']) |
|
1559 | revs = scmutil.revrange(repo, opts[b'rev']) | |
1560 | cl = repo.changelog |
|
1560 | cl = repo.changelog | |
1561 | nodes = [cl.node(r) for r in revs] |
|
1561 | nodes = [cl.node(r) for r in revs] | |
1562 |
|
1562 | |||
1563 | unfi = repo.unfiltered() |
|
1563 | unfi = repo.unfiltered() | |
1564 | # find the filecache func directly |
|
1564 | # find the filecache func directly | |
1565 | # This avoid polluting the benchmark with the filecache logic |
|
1565 | # This avoid polluting the benchmark with the filecache logic | |
1566 | makecl = unfi.__class__.changelog.func |
|
1566 | makecl = unfi.__class__.changelog.func | |
1567 |
|
1567 | |||
1568 | def setup(): |
|
1568 | def setup(): | |
1569 | # probably not necessary, but for good measure |
|
1569 | # probably not necessary, but for good measure | |
1570 | clearchangelog(unfi) |
|
1570 | clearchangelog(unfi) | |
1571 |
|
1571 | |||
1572 | def d(): |
|
1572 | def d(): | |
1573 | cl = makecl(unfi) |
|
1573 | cl = makecl(unfi) | |
1574 | for n in nodes: |
|
1574 | for n in nodes: | |
1575 | cl.rev(n) |
|
1575 | cl.rev(n) | |
1576 |
|
1576 | |||
1577 | timer(d, setup=setup) |
|
1577 | timer(d, setup=setup) | |
1578 | fm.end() |
|
1578 | fm.end() | |
1579 |
|
1579 | |||
1580 |
|
1580 | |||
1581 | @command( |
|
1581 | @command( | |
1582 | b'perfnodemap', |
|
1582 | b'perfnodemap', | |
1583 | [ |
|
1583 | [ | |
1584 | (b'', b'rev', [], b'revision to be looked up (default tip)'), |
|
1584 | (b'', b'rev', [], b'revision to be looked up (default tip)'), | |
1585 | (b'', b'clear-caches', True, b'clear revlog cache between calls'), |
|
1585 | (b'', b'clear-caches', True, b'clear revlog cache between calls'), | |
1586 | ] |
|
1586 | ] | |
1587 | + formatteropts, |
|
1587 | + formatteropts, | |
1588 | ) |
|
1588 | ) | |
1589 | def perfnodemap(ui, repo, **opts): |
|
1589 | def perfnodemap(ui, repo, **opts): | |
1590 | """benchmark the time necessary to look up revision from a cold nodemap |
|
1590 | """benchmark the time necessary to look up revision from a cold nodemap | |
1591 |
|
1591 | |||
1592 | Depending on the implementation, the amount and order of revision we look |
|
1592 | Depending on the implementation, the amount and order of revision we look | |
1593 | up can varies. Example of useful set to test: |
|
1593 | up can varies. Example of useful set to test: | |
1594 | * tip |
|
1594 | * tip | |
1595 | * 0 |
|
1595 | * 0 | |
1596 | * -10: |
|
1596 | * -10: | |
1597 | * :10 |
|
1597 | * :10 | |
1598 | * -10: + :10 |
|
1598 | * -10: + :10 | |
1599 | * :10: + -10: |
|
1599 | * :10: + -10: | |
1600 | * -10000: |
|
1600 | * -10000: | |
1601 | * -10000: + 0 |
|
1601 | * -10000: + 0 | |
1602 |
|
1602 | |||
1603 | The command currently focus on valid binary lookup. Benchmarking for |
|
1603 | The command currently focus on valid binary lookup. Benchmarking for | |
1604 | hexlookup, prefix lookup and missing lookup would also be valuable. |
|
1604 | hexlookup, prefix lookup and missing lookup would also be valuable. | |
1605 | """ |
|
1605 | """ | |
1606 | import mercurial.revlog |
|
1606 | import mercurial.revlog | |
1607 |
|
1607 | |||
1608 | opts = _byteskwargs(opts) |
|
1608 | opts = _byteskwargs(opts) | |
1609 | timer, fm = gettimer(ui, opts) |
|
1609 | timer, fm = gettimer(ui, opts) | |
1610 | mercurial.revlog._prereadsize = 2 ** 24 # disable lazy parser in old hg |
|
1610 | mercurial.revlog._prereadsize = 2 ** 24 # disable lazy parser in old hg | |
1611 |
|
1611 | |||
1612 | unfi = repo.unfiltered() |
|
1612 | unfi = repo.unfiltered() | |
1613 | clearcaches = opts['clear_caches'] |
|
1613 | clearcaches = opts['clear_caches'] | |
1614 | # find the filecache func directly |
|
1614 | # find the filecache func directly | |
1615 | # This avoid polluting the benchmark with the filecache logic |
|
1615 | # This avoid polluting the benchmark with the filecache logic | |
1616 | makecl = unfi.__class__.changelog.func |
|
1616 | makecl = unfi.__class__.changelog.func | |
1617 | if not opts[b'rev']: |
|
1617 | if not opts[b'rev']: | |
1618 | raise error.Abort('use --rev to specify revisions to look up') |
|
1618 | raise error.Abort('use --rev to specify revisions to look up') | |
1619 | revs = scmutil.revrange(repo, opts[b'rev']) |
|
1619 | revs = scmutil.revrange(repo, opts[b'rev']) | |
1620 | cl = repo.changelog |
|
1620 | cl = repo.changelog | |
1621 | nodes = [cl.node(r) for r in revs] |
|
1621 | nodes = [cl.node(r) for r in revs] | |
1622 |
|
1622 | |||
1623 | # use a list to pass reference to a nodemap from one closure to the next |
|
1623 | # use a list to pass reference to a nodemap from one closure to the next | |
1624 | nodeget = [None] |
|
1624 | nodeget = [None] | |
1625 |
|
1625 | |||
1626 | def setnodeget(): |
|
1626 | def setnodeget(): | |
1627 | # probably not necessary, but for good measure |
|
1627 | # probably not necessary, but for good measure | |
1628 | clearchangelog(unfi) |
|
1628 | clearchangelog(unfi) | |
1629 | nodeget[0] = makecl(unfi).nodemap.get |
|
1629 | nodeget[0] = makecl(unfi).nodemap.get | |
1630 |
|
1630 | |||
1631 | def d(): |
|
1631 | def d(): | |
1632 | get = nodeget[0] |
|
1632 | get = nodeget[0] | |
1633 | for n in nodes: |
|
1633 | for n in nodes: | |
1634 | get(n) |
|
1634 | get(n) | |
1635 |
|
1635 | |||
1636 | setup = None |
|
1636 | setup = None | |
1637 | if clearcaches: |
|
1637 | if clearcaches: | |
1638 |
|
1638 | |||
1639 | def setup(): |
|
1639 | def setup(): | |
1640 | setnodeget() |
|
1640 | setnodeget() | |
1641 |
|
1641 | |||
1642 | else: |
|
1642 | else: | |
1643 | setnodeget() |
|
1643 | setnodeget() | |
1644 | d() # prewarm the data structure |
|
1644 | d() # prewarm the data structure | |
1645 | timer(d, setup=setup) |
|
1645 | timer(d, setup=setup) | |
1646 | fm.end() |
|
1646 | fm.end() | |
1647 |
|
1647 | |||
1648 |
|
1648 | |||
1649 | @command(b'perfstartup', formatteropts) |
|
1649 | @command(b'perfstartup', formatteropts) | |
1650 | def perfstartup(ui, repo, **opts): |
|
1650 | def perfstartup(ui, repo, **opts): | |
1651 | opts = _byteskwargs(opts) |
|
1651 | opts = _byteskwargs(opts) | |
1652 | timer, fm = gettimer(ui, opts) |
|
1652 | timer, fm = gettimer(ui, opts) | |
1653 |
|
1653 | |||
1654 | def d(): |
|
1654 | def d(): | |
1655 | if os.name != 'nt': |
|
1655 | if os.name != 'nt': | |
1656 | os.system( |
|
1656 | os.system( | |
1657 | b"HGRCPATH= %s version -q > /dev/null" % fsencode(sys.argv[0]) |
|
1657 | b"HGRCPATH= %s version -q > /dev/null" % fsencode(sys.argv[0]) | |
1658 | ) |
|
1658 | ) | |
1659 | else: |
|
1659 | else: | |
1660 | os.environ['HGRCPATH'] = r' ' |
|
1660 | os.environ['HGRCPATH'] = r' ' | |
1661 | os.system("%s version -q > NUL" % sys.argv[0]) |
|
1661 | os.system("%s version -q > NUL" % sys.argv[0]) | |
1662 |
|
1662 | |||
1663 | timer(d) |
|
1663 | timer(d) | |
1664 | fm.end() |
|
1664 | fm.end() | |
1665 |
|
1665 | |||
1666 |
|
1666 | |||
1667 | @command(b'perfparents', formatteropts) |
|
1667 | @command(b'perfparents', formatteropts) | |
1668 | def perfparents(ui, repo, **opts): |
|
1668 | def perfparents(ui, repo, **opts): | |
1669 | """benchmark the time necessary to fetch one changeset's parents. |
|
1669 | """benchmark the time necessary to fetch one changeset's parents. | |
1670 |
|
1670 | |||
1671 | The fetch is done using the `node identifier`, traversing all object layers |
|
1671 | The fetch is done using the `node identifier`, traversing all object layers | |
1672 | from the repository object. The first N revisions will be used for this |
|
1672 | from the repository object. The first N revisions will be used for this | |
1673 | benchmark. N is controlled by the ``perf.parentscount`` config option |
|
1673 | benchmark. N is controlled by the ``perf.parentscount`` config option | |
1674 | (default: 1000). |
|
1674 | (default: 1000). | |
1675 | """ |
|
1675 | """ | |
1676 | opts = _byteskwargs(opts) |
|
1676 | opts = _byteskwargs(opts) | |
1677 | timer, fm = gettimer(ui, opts) |
|
1677 | timer, fm = gettimer(ui, opts) | |
1678 | # control the number of commits perfparents iterates over |
|
1678 | # control the number of commits perfparents iterates over | |
1679 | # experimental config: perf.parentscount |
|
1679 | # experimental config: perf.parentscount | |
1680 | count = getint(ui, b"perf", b"parentscount", 1000) |
|
1680 | count = getint(ui, b"perf", b"parentscount", 1000) | |
1681 | if len(repo.changelog) < count: |
|
1681 | if len(repo.changelog) < count: | |
1682 | raise error.Abort(b"repo needs %d commits for this test" % count) |
|
1682 | raise error.Abort(b"repo needs %d commits for this test" % count) | |
1683 | repo = repo.unfiltered() |
|
1683 | repo = repo.unfiltered() | |
1684 | nl = [repo.changelog.node(i) for i in _xrange(count)] |
|
1684 | nl = [repo.changelog.node(i) for i in _xrange(count)] | |
1685 |
|
1685 | |||
1686 | def d(): |
|
1686 | def d(): | |
1687 | for n in nl: |
|
1687 | for n in nl: | |
1688 | repo.changelog.parents(n) |
|
1688 | repo.changelog.parents(n) | |
1689 |
|
1689 | |||
1690 | timer(d) |
|
1690 | timer(d) | |
1691 | fm.end() |
|
1691 | fm.end() | |
1692 |
|
1692 | |||
1693 |
|
1693 | |||
1694 | @command(b'perfctxfiles', formatteropts) |
|
1694 | @command(b'perfctxfiles', formatteropts) | |
1695 | def perfctxfiles(ui, repo, x, **opts): |
|
1695 | def perfctxfiles(ui, repo, x, **opts): | |
1696 | opts = _byteskwargs(opts) |
|
1696 | opts = _byteskwargs(opts) | |
1697 | x = int(x) |
|
1697 | x = int(x) | |
1698 | timer, fm = gettimer(ui, opts) |
|
1698 | timer, fm = gettimer(ui, opts) | |
1699 |
|
1699 | |||
1700 | def d(): |
|
1700 | def d(): | |
1701 | len(repo[x].files()) |
|
1701 | len(repo[x].files()) | |
1702 |
|
1702 | |||
1703 | timer(d) |
|
1703 | timer(d) | |
1704 | fm.end() |
|
1704 | fm.end() | |
1705 |
|
1705 | |||
1706 |
|
1706 | |||
1707 | @command(b'perfrawfiles', formatteropts) |
|
1707 | @command(b'perfrawfiles', formatteropts) | |
1708 | def perfrawfiles(ui, repo, x, **opts): |
|
1708 | def perfrawfiles(ui, repo, x, **opts): | |
1709 | opts = _byteskwargs(opts) |
|
1709 | opts = _byteskwargs(opts) | |
1710 | x = int(x) |
|
1710 | x = int(x) | |
1711 | timer, fm = gettimer(ui, opts) |
|
1711 | timer, fm = gettimer(ui, opts) | |
1712 | cl = repo.changelog |
|
1712 | cl = repo.changelog | |
1713 |
|
1713 | |||
1714 | def d(): |
|
1714 | def d(): | |
1715 | len(cl.read(x)[3]) |
|
1715 | len(cl.read(x)[3]) | |
1716 |
|
1716 | |||
1717 | timer(d) |
|
1717 | timer(d) | |
1718 | fm.end() |
|
1718 | fm.end() | |
1719 |
|
1719 | |||
1720 |
|
1720 | |||
1721 | @command(b'perflookup', formatteropts) |
|
1721 | @command(b'perflookup', formatteropts) | |
1722 | def perflookup(ui, repo, rev, **opts): |
|
1722 | def perflookup(ui, repo, rev, **opts): | |
1723 | opts = _byteskwargs(opts) |
|
1723 | opts = _byteskwargs(opts) | |
1724 | timer, fm = gettimer(ui, opts) |
|
1724 | timer, fm = gettimer(ui, opts) | |
1725 | timer(lambda: len(repo.lookup(rev))) |
|
1725 | timer(lambda: len(repo.lookup(rev))) | |
1726 | fm.end() |
|
1726 | fm.end() | |
1727 |
|
1727 | |||
1728 |
|
1728 | |||
1729 | @command( |
|
1729 | @command( | |
1730 | b'perflinelogedits', |
|
1730 | b'perflinelogedits', | |
1731 | [ |
|
1731 | [ | |
1732 | (b'n', b'edits', 10000, b'number of edits'), |
|
1732 | (b'n', b'edits', 10000, b'number of edits'), | |
1733 | (b'', b'max-hunk-lines', 10, b'max lines in a hunk'), |
|
1733 | (b'', b'max-hunk-lines', 10, b'max lines in a hunk'), | |
1734 | ], |
|
1734 | ], | |
1735 | norepo=True, |
|
1735 | norepo=True, | |
1736 | ) |
|
1736 | ) | |
1737 | def perflinelogedits(ui, **opts): |
|
1737 | def perflinelogedits(ui, **opts): | |
1738 | from mercurial import linelog |
|
1738 | from mercurial import linelog | |
1739 |
|
1739 | |||
1740 | opts = _byteskwargs(opts) |
|
1740 | opts = _byteskwargs(opts) | |
1741 |
|
1741 | |||
1742 | edits = opts[b'edits'] |
|
1742 | edits = opts[b'edits'] | |
1743 | maxhunklines = opts[b'max_hunk_lines'] |
|
1743 | maxhunklines = opts[b'max_hunk_lines'] | |
1744 |
|
1744 | |||
1745 | maxb1 = 100000 |
|
1745 | maxb1 = 100000 | |
1746 | random.seed(0) |
|
1746 | random.seed(0) | |
1747 | randint = random.randint |
|
1747 | randint = random.randint | |
1748 | currentlines = 0 |
|
1748 | currentlines = 0 | |
1749 | arglist = [] |
|
1749 | arglist = [] | |
1750 | for rev in _xrange(edits): |
|
1750 | for rev in _xrange(edits): | |
1751 | a1 = randint(0, currentlines) |
|
1751 | a1 = randint(0, currentlines) | |
1752 | a2 = randint(a1, min(currentlines, a1 + maxhunklines)) |
|
1752 | a2 = randint(a1, min(currentlines, a1 + maxhunklines)) | |
1753 | b1 = randint(0, maxb1) |
|
1753 | b1 = randint(0, maxb1) | |
1754 | b2 = randint(b1, b1 + maxhunklines) |
|
1754 | b2 = randint(b1, b1 + maxhunklines) | |
1755 | currentlines += (b2 - b1) - (a2 - a1) |
|
1755 | currentlines += (b2 - b1) - (a2 - a1) | |
1756 | arglist.append((rev, a1, a2, b1, b2)) |
|
1756 | arglist.append((rev, a1, a2, b1, b2)) | |
1757 |
|
1757 | |||
1758 | def d(): |
|
1758 | def d(): | |
1759 | ll = linelog.linelog() |
|
1759 | ll = linelog.linelog() | |
1760 | for args in arglist: |
|
1760 | for args in arglist: | |
1761 | ll.replacelines(*args) |
|
1761 | ll.replacelines(*args) | |
1762 |
|
1762 | |||
1763 | timer, fm = gettimer(ui, opts) |
|
1763 | timer, fm = gettimer(ui, opts) | |
1764 | timer(d) |
|
1764 | timer(d) | |
1765 | fm.end() |
|
1765 | fm.end() | |
1766 |
|
1766 | |||
1767 |
|
1767 | |||
1768 | @command(b'perfrevrange', formatteropts) |
|
1768 | @command(b'perfrevrange', formatteropts) | |
1769 | def perfrevrange(ui, repo, *specs, **opts): |
|
1769 | def perfrevrange(ui, repo, *specs, **opts): | |
1770 | opts = _byteskwargs(opts) |
|
1770 | opts = _byteskwargs(opts) | |
1771 | timer, fm = gettimer(ui, opts) |
|
1771 | timer, fm = gettimer(ui, opts) | |
1772 | revrange = scmutil.revrange |
|
1772 | revrange = scmutil.revrange | |
1773 | timer(lambda: len(revrange(repo, specs))) |
|
1773 | timer(lambda: len(revrange(repo, specs))) | |
1774 | fm.end() |
|
1774 | fm.end() | |
1775 |
|
1775 | |||
1776 |
|
1776 | |||
1777 | @command(b'perfnodelookup', formatteropts) |
|
1777 | @command(b'perfnodelookup', formatteropts) | |
1778 | def perfnodelookup(ui, repo, rev, **opts): |
|
1778 | def perfnodelookup(ui, repo, rev, **opts): | |
1779 | opts = _byteskwargs(opts) |
|
1779 | opts = _byteskwargs(opts) | |
1780 | timer, fm = gettimer(ui, opts) |
|
1780 | timer, fm = gettimer(ui, opts) | |
1781 | import mercurial.revlog |
|
1781 | import mercurial.revlog | |
1782 |
|
1782 | |||
1783 | mercurial.revlog._prereadsize = 2 ** 24 # disable lazy parser in old hg |
|
1783 | mercurial.revlog._prereadsize = 2 ** 24 # disable lazy parser in old hg | |
1784 | n = scmutil.revsingle(repo, rev).node() |
|
1784 | n = scmutil.revsingle(repo, rev).node() | |
1785 | cl = mercurial.revlog.revlog(getsvfs(repo), b"00changelog.i") |
|
1785 | cl = mercurial.revlog.revlog(getsvfs(repo), b"00changelog.i") | |
1786 |
|
1786 | |||
1787 | def d(): |
|
1787 | def d(): | |
1788 | cl.rev(n) |
|
1788 | cl.rev(n) | |
1789 | clearcaches(cl) |
|
1789 | clearcaches(cl) | |
1790 |
|
1790 | |||
1791 | timer(d) |
|
1791 | timer(d) | |
1792 | fm.end() |
|
1792 | fm.end() | |
1793 |
|
1793 | |||
1794 |
|
1794 | |||
1795 | @command( |
|
1795 | @command( | |
1796 | b'perflog', |
|
1796 | b'perflog', | |
1797 | [(b'', b'rename', False, b'ask log to follow renames')] + formatteropts, |
|
1797 | [(b'', b'rename', False, b'ask log to follow renames')] + formatteropts, | |
1798 | ) |
|
1798 | ) | |
1799 | def perflog(ui, repo, rev=None, **opts): |
|
1799 | def perflog(ui, repo, rev=None, **opts): | |
1800 | opts = _byteskwargs(opts) |
|
1800 | opts = _byteskwargs(opts) | |
1801 | if rev is None: |
|
1801 | if rev is None: | |
1802 | rev = [] |
|
1802 | rev = [] | |
1803 | timer, fm = gettimer(ui, opts) |
|
1803 | timer, fm = gettimer(ui, opts) | |
1804 | ui.pushbuffer() |
|
1804 | ui.pushbuffer() | |
1805 | timer( |
|
1805 | timer( | |
1806 | lambda: commands.log( |
|
1806 | lambda: commands.log( | |
1807 | ui, repo, rev=rev, date=b'', user=b'', copies=opts.get(b'rename') |
|
1807 | ui, repo, rev=rev, date=b'', user=b'', copies=opts.get(b'rename') | |
1808 | ) |
|
1808 | ) | |
1809 | ) |
|
1809 | ) | |
1810 | ui.popbuffer() |
|
1810 | ui.popbuffer() | |
1811 | fm.end() |
|
1811 | fm.end() | |
1812 |
|
1812 | |||
1813 |
|
1813 | |||
1814 | @command(b'perfmoonwalk', formatteropts) |
|
1814 | @command(b'perfmoonwalk', formatteropts) | |
1815 | def perfmoonwalk(ui, repo, **opts): |
|
1815 | def perfmoonwalk(ui, repo, **opts): | |
1816 | """benchmark walking the changelog backwards |
|
1816 | """benchmark walking the changelog backwards | |
1817 |
|
1817 | |||
1818 | This also loads the changelog data for each revision in the changelog. |
|
1818 | This also loads the changelog data for each revision in the changelog. | |
1819 | """ |
|
1819 | """ | |
1820 | opts = _byteskwargs(opts) |
|
1820 | opts = _byteskwargs(opts) | |
1821 | timer, fm = gettimer(ui, opts) |
|
1821 | timer, fm = gettimer(ui, opts) | |
1822 |
|
1822 | |||
1823 | def moonwalk(): |
|
1823 | def moonwalk(): | |
1824 | for i in repo.changelog.revs(start=(len(repo) - 1), stop=-1): |
|
1824 | for i in repo.changelog.revs(start=(len(repo) - 1), stop=-1): | |
1825 | ctx = repo[i] |
|
1825 | ctx = repo[i] | |
1826 | ctx.branch() # read changelog data (in addition to the index) |
|
1826 | ctx.branch() # read changelog data (in addition to the index) | |
1827 |
|
1827 | |||
1828 | timer(moonwalk) |
|
1828 | timer(moonwalk) | |
1829 | fm.end() |
|
1829 | fm.end() | |
1830 |
|
1830 | |||
1831 |
|
1831 | |||
1832 | @command( |
|
1832 | @command( | |
1833 | b'perftemplating', |
|
1833 | b'perftemplating', | |
1834 | [(b'r', b'rev', [], b'revisions to run the template on'),] + formatteropts, |
|
1834 | [(b'r', b'rev', [], b'revisions to run the template on'),] + formatteropts, | |
1835 | ) |
|
1835 | ) | |
1836 | def perftemplating(ui, repo, testedtemplate=None, **opts): |
|
1836 | def perftemplating(ui, repo, testedtemplate=None, **opts): | |
1837 | """test the rendering time of a given template""" |
|
1837 | """test the rendering time of a given template""" | |
1838 | if makelogtemplater is None: |
|
1838 | if makelogtemplater is None: | |
1839 | raise error.Abort( |
|
1839 | raise error.Abort( | |
1840 | b"perftemplating not available with this Mercurial", |
|
1840 | b"perftemplating not available with this Mercurial", | |
1841 | hint=b"use 4.3 or later", |
|
1841 | hint=b"use 4.3 or later", | |
1842 | ) |
|
1842 | ) | |
1843 |
|
1843 | |||
1844 | opts = _byteskwargs(opts) |
|
1844 | opts = _byteskwargs(opts) | |
1845 |
|
1845 | |||
1846 | nullui = ui.copy() |
|
1846 | nullui = ui.copy() | |
1847 | nullui.fout = open(os.devnull, 'wb') |
|
1847 | nullui.fout = open(os.devnull, 'wb') | |
1848 | nullui.disablepager() |
|
1848 | nullui.disablepager() | |
1849 | revs = opts.get(b'rev') |
|
1849 | revs = opts.get(b'rev') | |
1850 | if not revs: |
|
1850 | if not revs: | |
1851 | revs = [b'all()'] |
|
1851 | revs = [b'all()'] | |
1852 | revs = list(scmutil.revrange(repo, revs)) |
|
1852 | revs = list(scmutil.revrange(repo, revs)) | |
1853 |
|
1853 | |||
1854 | defaulttemplate = ( |
|
1854 | defaulttemplate = ( | |
1855 | b'{date|shortdate} [{rev}:{node|short}]' |
|
1855 | b'{date|shortdate} [{rev}:{node|short}]' | |
1856 | b' {author|person}: {desc|firstline}\n' |
|
1856 | b' {author|person}: {desc|firstline}\n' | |
1857 | ) |
|
1857 | ) | |
1858 | if testedtemplate is None: |
|
1858 | if testedtemplate is None: | |
1859 | testedtemplate = defaulttemplate |
|
1859 | testedtemplate = defaulttemplate | |
1860 | displayer = makelogtemplater(nullui, repo, testedtemplate) |
|
1860 | displayer = makelogtemplater(nullui, repo, testedtemplate) | |
1861 |
|
1861 | |||
1862 | def format(): |
|
1862 | def format(): | |
1863 | for r in revs: |
|
1863 | for r in revs: | |
1864 | ctx = repo[r] |
|
1864 | ctx = repo[r] | |
1865 | displayer.show(ctx) |
|
1865 | displayer.show(ctx) | |
1866 | displayer.flush(ctx) |
|
1866 | displayer.flush(ctx) | |
1867 |
|
1867 | |||
1868 | timer, fm = gettimer(ui, opts) |
|
1868 | timer, fm = gettimer(ui, opts) | |
1869 | timer(format) |
|
1869 | timer(format) | |
1870 | fm.end() |
|
1870 | fm.end() | |
1871 |
|
1871 | |||
1872 |
|
1872 | |||
1873 | def _displaystats(ui, opts, entries, data): |
|
1873 | def _displaystats(ui, opts, entries, data): | |
1874 | pass |
|
1874 | pass | |
1875 | # use a second formatter because the data are quite different, not sure |
|
1875 | # use a second formatter because the data are quite different, not sure | |
1876 | # how it flies with the templater. |
|
1876 | # how it flies with the templater. | |
1877 | fm = ui.formatter(b'perf-stats', opts) |
|
1877 | fm = ui.formatter(b'perf-stats', opts) | |
1878 | for key, title in entries: |
|
1878 | for key, title in entries: | |
1879 | values = data[key] |
|
1879 | values = data[key] | |
1880 | nbvalues = len(data) |
|
1880 | nbvalues = len(data) | |
1881 | values.sort() |
|
1881 | values.sort() | |
1882 | stats = { |
|
1882 | stats = { | |
1883 | 'key': key, |
|
1883 | 'key': key, | |
1884 | 'title': title, |
|
1884 | 'title': title, | |
1885 | 'nbitems': len(values), |
|
1885 | 'nbitems': len(values), | |
1886 | 'min': values[0][0], |
|
1886 | 'min': values[0][0], | |
1887 | '10%': values[(nbvalues * 10) // 100][0], |
|
1887 | '10%': values[(nbvalues * 10) // 100][0], | |
1888 | '25%': values[(nbvalues * 25) // 100][0], |
|
1888 | '25%': values[(nbvalues * 25) // 100][0], | |
1889 | '50%': values[(nbvalues * 50) // 100][0], |
|
1889 | '50%': values[(nbvalues * 50) // 100][0], | |
1890 | '75%': values[(nbvalues * 75) // 100][0], |
|
1890 | '75%': values[(nbvalues * 75) // 100][0], | |
1891 | '80%': values[(nbvalues * 80) // 100][0], |
|
1891 | '80%': values[(nbvalues * 80) // 100][0], | |
1892 | '85%': values[(nbvalues * 85) // 100][0], |
|
1892 | '85%': values[(nbvalues * 85) // 100][0], | |
1893 | '90%': values[(nbvalues * 90) // 100][0], |
|
1893 | '90%': values[(nbvalues * 90) // 100][0], | |
1894 | '95%': values[(nbvalues * 95) // 100][0], |
|
1894 | '95%': values[(nbvalues * 95) // 100][0], | |
1895 | '99%': values[(nbvalues * 99) // 100][0], |
|
1895 | '99%': values[(nbvalues * 99) // 100][0], | |
1896 | 'max': values[-1][0], |
|
1896 | 'max': values[-1][0], | |
1897 | } |
|
1897 | } | |
1898 | fm.startitem() |
|
1898 | fm.startitem() | |
1899 | fm.data(**stats) |
|
1899 | fm.data(**stats) | |
1900 | # make node pretty for the human output |
|
1900 | # make node pretty for the human output | |
1901 | fm.plain('### %s (%d items)\n' % (title, len(values))) |
|
1901 | fm.plain('### %s (%d items)\n' % (title, len(values))) | |
1902 | lines = [ |
|
1902 | lines = [ | |
1903 | 'min', |
|
1903 | 'min', | |
1904 | '10%', |
|
1904 | '10%', | |
1905 | '25%', |
|
1905 | '25%', | |
1906 | '50%', |
|
1906 | '50%', | |
1907 | '75%', |
|
1907 | '75%', | |
1908 | '80%', |
|
1908 | '80%', | |
1909 | '85%', |
|
1909 | '85%', | |
1910 | '90%', |
|
1910 | '90%', | |
1911 | '95%', |
|
1911 | '95%', | |
1912 | '99%', |
|
1912 | '99%', | |
1913 | 'max', |
|
1913 | 'max', | |
1914 | ] |
|
1914 | ] | |
1915 | for l in lines: |
|
1915 | for l in lines: | |
1916 | fm.plain('%s: %s\n' % (l, stats[l])) |
|
1916 | fm.plain('%s: %s\n' % (l, stats[l])) | |
1917 | fm.end() |
|
1917 | fm.end() | |
1918 |
|
1918 | |||
1919 |
|
1919 | |||
1920 | @command( |
|
1920 | @command( | |
1921 | b'perfhelper-mergecopies', |
|
1921 | b'perfhelper-mergecopies', | |
1922 | formatteropts |
|
1922 | formatteropts | |
1923 | + [ |
|
1923 | + [ | |
1924 | (b'r', b'revs', [], b'restrict search to these revisions'), |
|
1924 | (b'r', b'revs', [], b'restrict search to these revisions'), | |
1925 | (b'', b'timing', False, b'provides extra data (costly)'), |
|
1925 | (b'', b'timing', False, b'provides extra data (costly)'), | |
1926 | (b'', b'stats', False, b'provides statistic about the measured data'), |
|
1926 | (b'', b'stats', False, b'provides statistic about the measured data'), | |
1927 | ], |
|
1927 | ], | |
1928 | ) |
|
1928 | ) | |
1929 | def perfhelpermergecopies(ui, repo, revs=[], **opts): |
|
1929 | def perfhelpermergecopies(ui, repo, revs=[], **opts): | |
1930 | """find statistics about potential parameters for `perfmergecopies` |
|
1930 | """find statistics about potential parameters for `perfmergecopies` | |
1931 |
|
1931 | |||
1932 | This command find (base, p1, p2) triplet relevant for copytracing |
|
1932 | This command find (base, p1, p2) triplet relevant for copytracing | |
1933 | benchmarking in the context of a merge. It reports values for some of the |
|
1933 | benchmarking in the context of a merge. It reports values for some of the | |
1934 | parameters that impact merge copy tracing time during merge. |
|
1934 | parameters that impact merge copy tracing time during merge. | |
1935 |
|
1935 | |||
1936 | If `--timing` is set, rename detection is run and the associated timing |
|
1936 | If `--timing` is set, rename detection is run and the associated timing | |
1937 | will be reported. The extra details come at the cost of slower command |
|
1937 | will be reported. The extra details come at the cost of slower command | |
1938 | execution. |
|
1938 | execution. | |
1939 |
|
1939 | |||
1940 | Since rename detection is only run once, other factors might easily |
|
1940 | Since rename detection is only run once, other factors might easily | |
1941 | affect the precision of the timing. However it should give a good |
|
1941 | affect the precision of the timing. However it should give a good | |
1942 | approximation of which revision triplets are very costly. |
|
1942 | approximation of which revision triplets are very costly. | |
1943 | """ |
|
1943 | """ | |
1944 | opts = _byteskwargs(opts) |
|
1944 | opts = _byteskwargs(opts) | |
1945 | fm = ui.formatter(b'perf', opts) |
|
1945 | fm = ui.formatter(b'perf', opts) | |
1946 | dotiming = opts[b'timing'] |
|
1946 | dotiming = opts[b'timing'] | |
1947 | dostats = opts[b'stats'] |
|
1947 | dostats = opts[b'stats'] | |
1948 |
|
1948 | |||
1949 | output_template = [ |
|
1949 | output_template = [ | |
1950 | ("base", "%(base)12s"), |
|
1950 | ("base", "%(base)12s"), | |
1951 | ("p1", "%(p1.node)12s"), |
|
1951 | ("p1", "%(p1.node)12s"), | |
1952 | ("p2", "%(p2.node)12s"), |
|
1952 | ("p2", "%(p2.node)12s"), | |
1953 | ("p1.nb-revs", "%(p1.nbrevs)12d"), |
|
1953 | ("p1.nb-revs", "%(p1.nbrevs)12d"), | |
1954 | ("p1.nb-files", "%(p1.nbmissingfiles)12d"), |
|
1954 | ("p1.nb-files", "%(p1.nbmissingfiles)12d"), | |
1955 | ("p1.renames", "%(p1.renamedfiles)12d"), |
|
1955 | ("p1.renames", "%(p1.renamedfiles)12d"), | |
1956 | ("p1.time", "%(p1.time)12.3f"), |
|
1956 | ("p1.time", "%(p1.time)12.3f"), | |
1957 | ("p2.nb-revs", "%(p2.nbrevs)12d"), |
|
1957 | ("p2.nb-revs", "%(p2.nbrevs)12d"), | |
1958 | ("p2.nb-files", "%(p2.nbmissingfiles)12d"), |
|
1958 | ("p2.nb-files", "%(p2.nbmissingfiles)12d"), | |
1959 | ("p2.renames", "%(p2.renamedfiles)12d"), |
|
1959 | ("p2.renames", "%(p2.renamedfiles)12d"), | |
1960 | ("p2.time", "%(p2.time)12.3f"), |
|
1960 | ("p2.time", "%(p2.time)12.3f"), | |
1961 | ("renames", "%(nbrenamedfiles)12d"), |
|
1961 | ("renames", "%(nbrenamedfiles)12d"), | |
1962 | ("total.time", "%(time)12.3f"), |
|
1962 | ("total.time", "%(time)12.3f"), | |
1963 | ] |
|
1963 | ] | |
1964 | if not dotiming: |
|
1964 | if not dotiming: | |
1965 | output_template = [ |
|
1965 | output_template = [ | |
1966 | i |
|
1966 | i | |
1967 | for i in output_template |
|
1967 | for i in output_template | |
1968 | if not ('time' in i[0] or 'renames' in i[0]) |
|
1968 | if not ('time' in i[0] or 'renames' in i[0]) | |
1969 | ] |
|
1969 | ] | |
1970 | header_names = [h for (h, v) in output_template] |
|
1970 | header_names = [h for (h, v) in output_template] | |
1971 | output = ' '.join([v for (h, v) in output_template]) + '\n' |
|
1971 | output = ' '.join([v for (h, v) in output_template]) + '\n' | |
1972 | header = ' '.join(['%12s'] * len(header_names)) + '\n' |
|
1972 | header = ' '.join(['%12s'] * len(header_names)) + '\n' | |
1973 | fm.plain(header % tuple(header_names)) |
|
1973 | fm.plain(header % tuple(header_names)) | |
1974 |
|
1974 | |||
1975 | if not revs: |
|
1975 | if not revs: | |
1976 | revs = ['all()'] |
|
1976 | revs = ['all()'] | |
1977 | revs = scmutil.revrange(repo, revs) |
|
1977 | revs = scmutil.revrange(repo, revs) | |
1978 |
|
1978 | |||
1979 | if dostats: |
|
1979 | if dostats: | |
1980 | alldata = { |
|
1980 | alldata = { | |
1981 | 'nbrevs': [], |
|
1981 | 'nbrevs': [], | |
1982 | 'nbmissingfiles': [], |
|
1982 | 'nbmissingfiles': [], | |
1983 | } |
|
1983 | } | |
1984 | if dotiming: |
|
1984 | if dotiming: | |
1985 | alldata['parentnbrenames'] = [] |
|
1985 | alldata['parentnbrenames'] = [] | |
1986 | alldata['totalnbrenames'] = [] |
|
1986 | alldata['totalnbrenames'] = [] | |
1987 | alldata['parenttime'] = [] |
|
1987 | alldata['parenttime'] = [] | |
1988 | alldata['totaltime'] = [] |
|
1988 | alldata['totaltime'] = [] | |
1989 |
|
1989 | |||
1990 | roi = repo.revs('merge() and %ld', revs) |
|
1990 | roi = repo.revs('merge() and %ld', revs) | |
1991 | for r in roi: |
|
1991 | for r in roi: | |
1992 | ctx = repo[r] |
|
1992 | ctx = repo[r] | |
1993 | p1 = ctx.p1() |
|
1993 | p1 = ctx.p1() | |
1994 | p2 = ctx.p2() |
|
1994 | p2 = ctx.p2() | |
1995 | bases = repo.changelog._commonancestorsheads(p1.rev(), p2.rev()) |
|
1995 | bases = repo.changelog._commonancestorsheads(p1.rev(), p2.rev()) | |
1996 | for b in bases: |
|
1996 | for b in bases: | |
1997 | b = repo[b] |
|
1997 | b = repo[b] | |
1998 | p1missing = copies._computeforwardmissing(b, p1) |
|
1998 | p1missing = copies._computeforwardmissing(b, p1) | |
1999 | p2missing = copies._computeforwardmissing(b, p2) |
|
1999 | p2missing = copies._computeforwardmissing(b, p2) | |
2000 | data = { |
|
2000 | data = { | |
2001 | b'base': b.hex(), |
|
2001 | b'base': b.hex(), | |
2002 | b'p1.node': p1.hex(), |
|
2002 | b'p1.node': p1.hex(), | |
2003 | b'p1.nbrevs': len(repo.revs('only(%d, %d)', p1.rev(), b.rev())), |
|
2003 | b'p1.nbrevs': len(repo.revs('only(%d, %d)', p1.rev(), b.rev())), | |
2004 | b'p1.nbmissingfiles': len(p1missing), |
|
2004 | b'p1.nbmissingfiles': len(p1missing), | |
2005 | b'p2.node': p2.hex(), |
|
2005 | b'p2.node': p2.hex(), | |
2006 | b'p2.nbrevs': len(repo.revs('only(%d, %d)', p2.rev(), b.rev())), |
|
2006 | b'p2.nbrevs': len(repo.revs('only(%d, %d)', p2.rev(), b.rev())), | |
2007 | b'p2.nbmissingfiles': len(p2missing), |
|
2007 | b'p2.nbmissingfiles': len(p2missing), | |
2008 | } |
|
2008 | } | |
2009 | if dostats: |
|
2009 | if dostats: | |
2010 | if p1missing: |
|
2010 | if p1missing: | |
2011 | alldata['nbrevs'].append( |
|
2011 | alldata['nbrevs'].append( | |
2012 | (data['p1.nbrevs'], b.hex(), p1.hex()) |
|
2012 | (data['p1.nbrevs'], b.hex(), p1.hex()) | |
2013 | ) |
|
2013 | ) | |
2014 | alldata['nbmissingfiles'].append( |
|
2014 | alldata['nbmissingfiles'].append( | |
2015 | (data['p1.nbmissingfiles'], b.hex(), p1.hex()) |
|
2015 | (data['p1.nbmissingfiles'], b.hex(), p1.hex()) | |
2016 | ) |
|
2016 | ) | |
2017 | if p2missing: |
|
2017 | if p2missing: | |
2018 | alldata['nbrevs'].append( |
|
2018 | alldata['nbrevs'].append( | |
2019 | (data['p2.nbrevs'], b.hex(), p2.hex()) |
|
2019 | (data['p2.nbrevs'], b.hex(), p2.hex()) | |
2020 | ) |
|
2020 | ) | |
2021 | alldata['nbmissingfiles'].append( |
|
2021 | alldata['nbmissingfiles'].append( | |
2022 | (data['p2.nbmissingfiles'], b.hex(), p2.hex()) |
|
2022 | (data['p2.nbmissingfiles'], b.hex(), p2.hex()) | |
2023 | ) |
|
2023 | ) | |
2024 | if dotiming: |
|
2024 | if dotiming: | |
2025 | begin = util.timer() |
|
2025 | begin = util.timer() | |
2026 | mergedata = copies.mergecopies(repo, p1, p2, b) |
|
2026 | mergedata = copies.mergecopies(repo, p1, p2, b) | |
2027 | end = util.timer() |
|
2027 | end = util.timer() | |
2028 | # not very stable timing since we did only one run |
|
2028 | # not very stable timing since we did only one run | |
2029 | data['time'] = end - begin |
|
2029 | data['time'] = end - begin | |
2030 | # mergedata contains five dicts: "copy", "movewithdir", |
|
2030 | # mergedata contains five dicts: "copy", "movewithdir", | |
2031 | # "diverge", "renamedelete" and "dirmove". |
|
2031 | # "diverge", "renamedelete" and "dirmove". | |
2032 | # The first 4 are about renamed file so lets count that. |
|
2032 | # The first 4 are about renamed file so lets count that. | |
2033 | renames = len(mergedata[0]) |
|
2033 | renames = len(mergedata[0]) | |
2034 | renames += len(mergedata[1]) |
|
2034 | renames += len(mergedata[1]) | |
2035 | renames += len(mergedata[2]) |
|
2035 | renames += len(mergedata[2]) | |
2036 | renames += len(mergedata[3]) |
|
2036 | renames += len(mergedata[3]) | |
2037 | data['nbrenamedfiles'] = renames |
|
2037 | data['nbrenamedfiles'] = renames | |
2038 | begin = util.timer() |
|
2038 | begin = util.timer() | |
2039 | p1renames = copies.pathcopies(b, p1) |
|
2039 | p1renames = copies.pathcopies(b, p1) | |
2040 | end = util.timer() |
|
2040 | end = util.timer() | |
2041 | data['p1.time'] = end - begin |
|
2041 | data['p1.time'] = end - begin | |
2042 | begin = util.timer() |
|
2042 | begin = util.timer() | |
2043 | p2renames = copies.pathcopies(b, p2) |
|
2043 | p2renames = copies.pathcopies(b, p2) | |
2044 | data['p2.time'] = end - begin |
|
2044 | data['p2.time'] = end - begin | |
2045 | end = util.timer() |
|
2045 | end = util.timer() | |
2046 | data['p1.renamedfiles'] = len(p1renames) |
|
2046 | data['p1.renamedfiles'] = len(p1renames) | |
2047 | data['p2.renamedfiles'] = len(p2renames) |
|
2047 | data['p2.renamedfiles'] = len(p2renames) | |
2048 |
|
2048 | |||
2049 | if dostats: |
|
2049 | if dostats: | |
2050 | if p1missing: |
|
2050 | if p1missing: | |
2051 | alldata['parentnbrenames'].append( |
|
2051 | alldata['parentnbrenames'].append( | |
2052 | (data['p1.renamedfiles'], b.hex(), p1.hex()) |
|
2052 | (data['p1.renamedfiles'], b.hex(), p1.hex()) | |
2053 | ) |
|
2053 | ) | |
2054 | alldata['parenttime'].append( |
|
2054 | alldata['parenttime'].append( | |
2055 | (data['p1.time'], b.hex(), p1.hex()) |
|
2055 | (data['p1.time'], b.hex(), p1.hex()) | |
2056 | ) |
|
2056 | ) | |
2057 | if p2missing: |
|
2057 | if p2missing: | |
2058 | alldata['parentnbrenames'].append( |
|
2058 | alldata['parentnbrenames'].append( | |
2059 | (data['p2.renamedfiles'], b.hex(), p2.hex()) |
|
2059 | (data['p2.renamedfiles'], b.hex(), p2.hex()) | |
2060 | ) |
|
2060 | ) | |
2061 | alldata['parenttime'].append( |
|
2061 | alldata['parenttime'].append( | |
2062 | (data['p2.time'], b.hex(), p2.hex()) |
|
2062 | (data['p2.time'], b.hex(), p2.hex()) | |
2063 | ) |
|
2063 | ) | |
2064 | if p1missing or p2missing: |
|
2064 | if p1missing or p2missing: | |
2065 | alldata['totalnbrenames'].append( |
|
2065 | alldata['totalnbrenames'].append( | |
2066 | ( |
|
2066 | ( | |
2067 | data['nbrenamedfiles'], |
|
2067 | data['nbrenamedfiles'], | |
2068 | b.hex(), |
|
2068 | b.hex(), | |
2069 | p1.hex(), |
|
2069 | p1.hex(), | |
2070 | p2.hex(), |
|
2070 | p2.hex(), | |
2071 | ) |
|
2071 | ) | |
2072 | ) |
|
2072 | ) | |
2073 | alldata['totaltime'].append( |
|
2073 | alldata['totaltime'].append( | |
2074 | (data['time'], b.hex(), p1.hex(), p2.hex()) |
|
2074 | (data['time'], b.hex(), p1.hex(), p2.hex()) | |
2075 | ) |
|
2075 | ) | |
2076 | fm.startitem() |
|
2076 | fm.startitem() | |
2077 | fm.data(**data) |
|
2077 | fm.data(**data) | |
2078 | # make node pretty for the human output |
|
2078 | # make node pretty for the human output | |
2079 | out = data.copy() |
|
2079 | out = data.copy() | |
2080 | out['base'] = fm.hexfunc(b.node()) |
|
2080 | out['base'] = fm.hexfunc(b.node()) | |
2081 | out['p1.node'] = fm.hexfunc(p1.node()) |
|
2081 | out['p1.node'] = fm.hexfunc(p1.node()) | |
2082 | out['p2.node'] = fm.hexfunc(p2.node()) |
|
2082 | out['p2.node'] = fm.hexfunc(p2.node()) | |
2083 | fm.plain(output % out) |
|
2083 | fm.plain(output % out) | |
2084 |
|
2084 | |||
2085 | fm.end() |
|
2085 | fm.end() | |
2086 | if dostats: |
|
2086 | if dostats: | |
2087 | # use a second formatter because the data are quite different, not sure |
|
2087 | # use a second formatter because the data are quite different, not sure | |
2088 | # how it flies with the templater. |
|
2088 | # how it flies with the templater. | |
2089 | entries = [ |
|
2089 | entries = [ | |
2090 | ('nbrevs', 'number of revision covered'), |
|
2090 | ('nbrevs', 'number of revision covered'), | |
2091 | ('nbmissingfiles', 'number of missing files at head'), |
|
2091 | ('nbmissingfiles', 'number of missing files at head'), | |
2092 | ] |
|
2092 | ] | |
2093 | if dotiming: |
|
2093 | if dotiming: | |
2094 | entries.append( |
|
2094 | entries.append( | |
2095 | ('parentnbrenames', 'rename from one parent to base') |
|
2095 | ('parentnbrenames', 'rename from one parent to base') | |
2096 | ) |
|
2096 | ) | |
2097 | entries.append(('totalnbrenames', 'total number of renames')) |
|
2097 | entries.append(('totalnbrenames', 'total number of renames')) | |
2098 | entries.append(('parenttime', 'time for one parent')) |
|
2098 | entries.append(('parenttime', 'time for one parent')) | |
2099 | entries.append(('totaltime', 'time for both parents')) |
|
2099 | entries.append(('totaltime', 'time for both parents')) | |
2100 | _displaystats(ui, opts, entries, alldata) |
|
2100 | _displaystats(ui, opts, entries, alldata) | |
2101 |
|
2101 | |||
2102 |
|
2102 | |||
2103 | @command( |
|
2103 | @command( | |
2104 | b'perfhelper-pathcopies', |
|
2104 | b'perfhelper-pathcopies', | |
2105 | formatteropts |
|
2105 | formatteropts | |
2106 | + [ |
|
2106 | + [ | |
2107 | (b'r', b'revs', [], b'restrict search to these revisions'), |
|
2107 | (b'r', b'revs', [], b'restrict search to these revisions'), | |
2108 | (b'', b'timing', False, b'provides extra data (costly)'), |
|
2108 | (b'', b'timing', False, b'provides extra data (costly)'), | |
2109 | (b'', b'stats', False, b'provides statistic about the measured data'), |
|
2109 | (b'', b'stats', False, b'provides statistic about the measured data'), | |
2110 | ], |
|
2110 | ], | |
2111 | ) |
|
2111 | ) | |
2112 | def perfhelperpathcopies(ui, repo, revs=[], **opts): |
|
2112 | def perfhelperpathcopies(ui, repo, revs=[], **opts): | |
2113 | """find statistic about potential parameters for the `perftracecopies` |
|
2113 | """find statistic about potential parameters for the `perftracecopies` | |
2114 |
|
2114 | |||
2115 | This command find source-destination pair relevant for copytracing testing. |
|
2115 | This command find source-destination pair relevant for copytracing testing. | |
2116 | It report value for some of the parameters that impact copy tracing time. |
|
2116 | It report value for some of the parameters that impact copy tracing time. | |
2117 |
|
2117 | |||
2118 | If `--timing` is set, rename detection is run and the associated timing |
|
2118 | If `--timing` is set, rename detection is run and the associated timing | |
2119 | will be reported. The extra details comes at the cost of a slower command |
|
2119 | will be reported. The extra details comes at the cost of a slower command | |
2120 | execution. |
|
2120 | execution. | |
2121 |
|
2121 | |||
2122 | Since the rename detection is only run once, other factors might easily |
|
2122 | Since the rename detection is only run once, other factors might easily | |
2123 | affect the precision of the timing. However it should give a good |
|
2123 | affect the precision of the timing. However it should give a good | |
2124 | approximation of which revision pairs are very costly. |
|
2124 | approximation of which revision pairs are very costly. | |
2125 | """ |
|
2125 | """ | |
2126 | opts = _byteskwargs(opts) |
|
2126 | opts = _byteskwargs(opts) | |
2127 | fm = ui.formatter(b'perf', opts) |
|
2127 | fm = ui.formatter(b'perf', opts) | |
2128 | dotiming = opts[b'timing'] |
|
2128 | dotiming = opts[b'timing'] | |
2129 | dostats = opts[b'stats'] |
|
2129 | dostats = opts[b'stats'] | |
2130 |
|
2130 | |||
2131 | if dotiming: |
|
2131 | if dotiming: | |
2132 | header = '%12s %12s %12s %12s %12s %12s\n' |
|
2132 | header = '%12s %12s %12s %12s %12s %12s\n' | |
2133 | output = ( |
|
2133 | output = ( | |
2134 | "%(source)12s %(destination)12s " |
|
2134 | "%(source)12s %(destination)12s " | |
2135 | "%(nbrevs)12d %(nbmissingfiles)12d " |
|
2135 | "%(nbrevs)12d %(nbmissingfiles)12d " | |
2136 | "%(nbrenamedfiles)12d %(time)18.5f\n" |
|
2136 | "%(nbrenamedfiles)12d %(time)18.5f\n" | |
2137 | ) |
|
2137 | ) | |
2138 | header_names = ( |
|
2138 | header_names = ( | |
2139 | "source", |
|
2139 | "source", | |
2140 | "destination", |
|
2140 | "destination", | |
2141 | "nb-revs", |
|
2141 | "nb-revs", | |
2142 | "nb-files", |
|
2142 | "nb-files", | |
2143 | "nb-renames", |
|
2143 | "nb-renames", | |
2144 | "time", |
|
2144 | "time", | |
2145 | ) |
|
2145 | ) | |
2146 | fm.plain(header % header_names) |
|
2146 | fm.plain(header % header_names) | |
2147 | else: |
|
2147 | else: | |
2148 | header = '%12s %12s %12s %12s\n' |
|
2148 | header = '%12s %12s %12s %12s\n' | |
2149 | output = ( |
|
2149 | output = ( | |
2150 | "%(source)12s %(destination)12s " |
|
2150 | "%(source)12s %(destination)12s " | |
2151 | "%(nbrevs)12d %(nbmissingfiles)12d\n" |
|
2151 | "%(nbrevs)12d %(nbmissingfiles)12d\n" | |
2152 | ) |
|
2152 | ) | |
2153 | fm.plain(header % ("source", "destination", "nb-revs", "nb-files")) |
|
2153 | fm.plain(header % ("source", "destination", "nb-revs", "nb-files")) | |
2154 |
|
2154 | |||
2155 | if not revs: |
|
2155 | if not revs: | |
2156 | revs = ['all()'] |
|
2156 | revs = ['all()'] | |
2157 | revs = scmutil.revrange(repo, revs) |
|
2157 | revs = scmutil.revrange(repo, revs) | |
2158 |
|
2158 | |||
2159 | if dostats: |
|
2159 | if dostats: | |
2160 | alldata = { |
|
2160 | alldata = { | |
2161 | 'nbrevs': [], |
|
2161 | 'nbrevs': [], | |
2162 | 'nbmissingfiles': [], |
|
2162 | 'nbmissingfiles': [], | |
2163 | } |
|
2163 | } | |
2164 | if dotiming: |
|
2164 | if dotiming: | |
2165 | alldata['nbrenames'] = [] |
|
2165 | alldata['nbrenames'] = [] | |
2166 | alldata['time'] = [] |
|
2166 | alldata['time'] = [] | |
2167 |
|
2167 | |||
2168 | roi = repo.revs('merge() and %ld', revs) |
|
2168 | roi = repo.revs('merge() and %ld', revs) | |
2169 | for r in roi: |
|
2169 | for r in roi: | |
2170 | ctx = repo[r] |
|
2170 | ctx = repo[r] | |
2171 | p1 = ctx.p1().rev() |
|
2171 | p1 = ctx.p1().rev() | |
2172 | p2 = ctx.p2().rev() |
|
2172 | p2 = ctx.p2().rev() | |
2173 | bases = repo.changelog._commonancestorsheads(p1, p2) |
|
2173 | bases = repo.changelog._commonancestorsheads(p1, p2) | |
2174 | for p in (p1, p2): |
|
2174 | for p in (p1, p2): | |
2175 | for b in bases: |
|
2175 | for b in bases: | |
2176 | base = repo[b] |
|
2176 | base = repo[b] | |
2177 | parent = repo[p] |
|
2177 | parent = repo[p] | |
2178 | missing = copies._computeforwardmissing(base, parent) |
|
2178 | missing = copies._computeforwardmissing(base, parent) | |
2179 | if not missing: |
|
2179 | if not missing: | |
2180 | continue |
|
2180 | continue | |
2181 | data = { |
|
2181 | data = { | |
2182 | b'source': base.hex(), |
|
2182 | b'source': base.hex(), | |
2183 | b'destination': parent.hex(), |
|
2183 | b'destination': parent.hex(), | |
2184 | b'nbrevs': len(repo.revs('only(%d, %d)', p, b)), |
|
2184 | b'nbrevs': len(repo.revs('only(%d, %d)', p, b)), | |
2185 | b'nbmissingfiles': len(missing), |
|
2185 | b'nbmissingfiles': len(missing), | |
2186 | } |
|
2186 | } | |
2187 | if dostats: |
|
2187 | if dostats: | |
2188 | alldata['nbrevs'].append( |
|
2188 | alldata['nbrevs'].append( | |
2189 | (data['nbrevs'], base.hex(), parent.hex(),) |
|
2189 | (data['nbrevs'], base.hex(), parent.hex(),) | |
2190 | ) |
|
2190 | ) | |
2191 | alldata['nbmissingfiles'].append( |
|
2191 | alldata['nbmissingfiles'].append( | |
2192 | (data['nbmissingfiles'], base.hex(), parent.hex(),) |
|
2192 | (data['nbmissingfiles'], base.hex(), parent.hex(),) | |
2193 | ) |
|
2193 | ) | |
2194 | if dotiming: |
|
2194 | if dotiming: | |
2195 | begin = util.timer() |
|
2195 | begin = util.timer() | |
2196 | renames = copies.pathcopies(base, parent) |
|
2196 | renames = copies.pathcopies(base, parent) | |
2197 | end = util.timer() |
|
2197 | end = util.timer() | |
2198 | # not very stable timing since we did only one run |
|
2198 | # not very stable timing since we did only one run | |
2199 | data['time'] = end - begin |
|
2199 | data['time'] = end - begin | |
2200 | data['nbrenamedfiles'] = len(renames) |
|
2200 | data['nbrenamedfiles'] = len(renames) | |
2201 | if dostats: |
|
2201 | if dostats: | |
2202 | alldata['time'].append( |
|
2202 | alldata['time'].append( | |
2203 | (data['time'], base.hex(), parent.hex(),) |
|
2203 | (data['time'], base.hex(), parent.hex(),) | |
2204 | ) |
|
2204 | ) | |
2205 | alldata['nbrenames'].append( |
|
2205 | alldata['nbrenames'].append( | |
2206 | (data['nbrenamedfiles'], base.hex(), parent.hex(),) |
|
2206 | (data['nbrenamedfiles'], base.hex(), parent.hex(),) | |
2207 | ) |
|
2207 | ) | |
2208 | fm.startitem() |
|
2208 | fm.startitem() | |
2209 | fm.data(**data) |
|
2209 | fm.data(**data) | |
2210 | out = data.copy() |
|
2210 | out = data.copy() | |
2211 | out['source'] = fm.hexfunc(base.node()) |
|
2211 | out['source'] = fm.hexfunc(base.node()) | |
2212 | out['destination'] = fm.hexfunc(parent.node()) |
|
2212 | out['destination'] = fm.hexfunc(parent.node()) | |
2213 | fm.plain(output % out) |
|
2213 | fm.plain(output % out) | |
2214 |
|
2214 | |||
2215 | fm.end() |
|
2215 | fm.end() | |
2216 | if dostats: |
|
2216 | if dostats: | |
2217 | # use a second formatter because the data are quite different, not sure |
|
2217 | # use a second formatter because the data are quite different, not sure | |
2218 | # how it flies with the templater. |
|
2218 | # how it flies with the templater. | |
2219 | fm = ui.formatter(b'perf', opts) |
|
2219 | fm = ui.formatter(b'perf', opts) | |
2220 | entries = [ |
|
2220 | entries = [ | |
2221 | ('nbrevs', 'number of revision covered'), |
|
2221 | ('nbrevs', 'number of revision covered'), | |
2222 | ('nbmissingfiles', 'number of missing files at head'), |
|
2222 | ('nbmissingfiles', 'number of missing files at head'), | |
2223 | ] |
|
2223 | ] | |
2224 | if dotiming: |
|
2224 | if dotiming: | |
2225 | entries.append(('nbrenames', 'renamed files')) |
|
2225 | entries.append(('nbrenames', 'renamed files')) | |
2226 | entries.append(('time', 'time')) |
|
2226 | entries.append(('time', 'time')) | |
2227 | _displaystats(ui, opts, entries, alldata) |
|
2227 | _displaystats(ui, opts, entries, alldata) | |
2228 |
|
2228 | |||
2229 |
|
2229 | |||
2230 | @command(b'perfcca', formatteropts) |
|
2230 | @command(b'perfcca', formatteropts) | |
2231 | def perfcca(ui, repo, **opts): |
|
2231 | def perfcca(ui, repo, **opts): | |
2232 | opts = _byteskwargs(opts) |
|
2232 | opts = _byteskwargs(opts) | |
2233 | timer, fm = gettimer(ui, opts) |
|
2233 | timer, fm = gettimer(ui, opts) | |
2234 | timer(lambda: scmutil.casecollisionauditor(ui, False, repo.dirstate)) |
|
2234 | timer(lambda: scmutil.casecollisionauditor(ui, False, repo.dirstate)) | |
2235 | fm.end() |
|
2235 | fm.end() | |
2236 |
|
2236 | |||
2237 |
|
2237 | |||
2238 | @command(b'perffncacheload', formatteropts) |
|
2238 | @command(b'perffncacheload', formatteropts) | |
2239 | def perffncacheload(ui, repo, **opts): |
|
2239 | def perffncacheload(ui, repo, **opts): | |
2240 | opts = _byteskwargs(opts) |
|
2240 | opts = _byteskwargs(opts) | |
2241 | timer, fm = gettimer(ui, opts) |
|
2241 | timer, fm = gettimer(ui, opts) | |
2242 | s = repo.store |
|
2242 | s = repo.store | |
2243 |
|
2243 | |||
2244 | def d(): |
|
2244 | def d(): | |
2245 | s.fncache._load() |
|
2245 | s.fncache._load() | |
2246 |
|
2246 | |||
2247 | timer(d) |
|
2247 | timer(d) | |
2248 | fm.end() |
|
2248 | fm.end() | |
2249 |
|
2249 | |||
2250 |
|
2250 | |||
2251 | @command(b'perffncachewrite', formatteropts) |
|
2251 | @command(b'perffncachewrite', formatteropts) | |
2252 | def perffncachewrite(ui, repo, **opts): |
|
2252 | def perffncachewrite(ui, repo, **opts): | |
2253 | opts = _byteskwargs(opts) |
|
2253 | opts = _byteskwargs(opts) | |
2254 | timer, fm = gettimer(ui, opts) |
|
2254 | timer, fm = gettimer(ui, opts) | |
2255 | s = repo.store |
|
2255 | s = repo.store | |
2256 | lock = repo.lock() |
|
2256 | lock = repo.lock() | |
2257 | s.fncache._load() |
|
2257 | s.fncache._load() | |
2258 | tr = repo.transaction(b'perffncachewrite') |
|
2258 | tr = repo.transaction(b'perffncachewrite') | |
2259 | tr.addbackup(b'fncache') |
|
2259 | tr.addbackup(b'fncache') | |
2260 |
|
2260 | |||
2261 | def d(): |
|
2261 | def d(): | |
2262 | s.fncache._dirty = True |
|
2262 | s.fncache._dirty = True | |
2263 | s.fncache.write(tr) |
|
2263 | s.fncache.write(tr) | |
2264 |
|
2264 | |||
2265 | timer(d) |
|
2265 | timer(d) | |
2266 | tr.close() |
|
2266 | tr.close() | |
2267 | lock.release() |
|
2267 | lock.release() | |
2268 | fm.end() |
|
2268 | fm.end() | |
2269 |
|
2269 | |||
2270 |
|
2270 | |||
2271 | @command(b'perffncacheencode', formatteropts) |
|
2271 | @command(b'perffncacheencode', formatteropts) | |
2272 | def perffncacheencode(ui, repo, **opts): |
|
2272 | def perffncacheencode(ui, repo, **opts): | |
2273 | opts = _byteskwargs(opts) |
|
2273 | opts = _byteskwargs(opts) | |
2274 | timer, fm = gettimer(ui, opts) |
|
2274 | timer, fm = gettimer(ui, opts) | |
2275 | s = repo.store |
|
2275 | s = repo.store | |
2276 | s.fncache._load() |
|
2276 | s.fncache._load() | |
2277 |
|
2277 | |||
2278 | def d(): |
|
2278 | def d(): | |
2279 | for p in s.fncache.entries: |
|
2279 | for p in s.fncache.entries: | |
2280 | s.encode(p) |
|
2280 | s.encode(p) | |
2281 |
|
2281 | |||
2282 | timer(d) |
|
2282 | timer(d) | |
2283 | fm.end() |
|
2283 | fm.end() | |
2284 |
|
2284 | |||
2285 |
|
2285 | |||
2286 | def _bdiffworker(q, blocks, xdiff, ready, done): |
|
2286 | def _bdiffworker(q, blocks, xdiff, ready, done): | |
2287 | while not done.is_set(): |
|
2287 | while not done.is_set(): | |
2288 | pair = q.get() |
|
2288 | pair = q.get() | |
2289 | while pair is not None: |
|
2289 | while pair is not None: | |
2290 | if xdiff: |
|
2290 | if xdiff: | |
2291 | mdiff.bdiff.xdiffblocks(*pair) |
|
2291 | mdiff.bdiff.xdiffblocks(*pair) | |
2292 | elif blocks: |
|
2292 | elif blocks: | |
2293 | mdiff.bdiff.blocks(*pair) |
|
2293 | mdiff.bdiff.blocks(*pair) | |
2294 | else: |
|
2294 | else: | |
2295 | mdiff.textdiff(*pair) |
|
2295 | mdiff.textdiff(*pair) | |
2296 | q.task_done() |
|
2296 | q.task_done() | |
2297 | pair = q.get() |
|
2297 | pair = q.get() | |
2298 | q.task_done() # for the None one |
|
2298 | q.task_done() # for the None one | |
2299 | with ready: |
|
2299 | with ready: | |
2300 | ready.wait() |
|
2300 | ready.wait() | |
2301 |
|
2301 | |||
2302 |
|
2302 | |||
2303 | def _manifestrevision(repo, mnode): |
|
2303 | def _manifestrevision(repo, mnode): | |
2304 | ml = repo.manifestlog |
|
2304 | ml = repo.manifestlog | |
2305 |
|
2305 | |||
2306 | if util.safehasattr(ml, b'getstorage'): |
|
2306 | if util.safehasattr(ml, b'getstorage'): | |
2307 | store = ml.getstorage(b'') |
|
2307 | store = ml.getstorage(b'') | |
2308 | else: |
|
2308 | else: | |
2309 | store = ml._revlog |
|
2309 | store = ml._revlog | |
2310 |
|
2310 | |||
2311 | return store.revision(mnode) |
|
2311 | return store.revision(mnode) | |
2312 |
|
2312 | |||
2313 |
|
2313 | |||
2314 | @command( |
|
2314 | @command( | |
2315 | b'perfbdiff', |
|
2315 | b'perfbdiff', | |
2316 | revlogopts |
|
2316 | revlogopts | |
2317 | + formatteropts |
|
2317 | + formatteropts | |
2318 | + [ |
|
2318 | + [ | |
2319 | ( |
|
2319 | ( | |
2320 | b'', |
|
2320 | b'', | |
2321 | b'count', |
|
2321 | b'count', | |
2322 | 1, |
|
2322 | 1, | |
2323 | b'number of revisions to test (when using --startrev)', |
|
2323 | b'number of revisions to test (when using --startrev)', | |
2324 | ), |
|
2324 | ), | |
2325 | (b'', b'alldata', False, b'test bdiffs for all associated revisions'), |
|
2325 | (b'', b'alldata', False, b'test bdiffs for all associated revisions'), | |
2326 | (b'', b'threads', 0, b'number of thread to use (disable with 0)'), |
|
2326 | (b'', b'threads', 0, b'number of thread to use (disable with 0)'), | |
2327 | (b'', b'blocks', False, b'test computing diffs into blocks'), |
|
2327 | (b'', b'blocks', False, b'test computing diffs into blocks'), | |
2328 | (b'', b'xdiff', False, b'use xdiff algorithm'), |
|
2328 | (b'', b'xdiff', False, b'use xdiff algorithm'), | |
2329 | ], |
|
2329 | ], | |
2330 | b'-c|-m|FILE REV', |
|
2330 | b'-c|-m|FILE REV', | |
2331 | ) |
|
2331 | ) | |
2332 | def perfbdiff(ui, repo, file_, rev=None, count=None, threads=0, **opts): |
|
2332 | def perfbdiff(ui, repo, file_, rev=None, count=None, threads=0, **opts): | |
2333 | """benchmark a bdiff between revisions |
|
2333 | """benchmark a bdiff between revisions | |
2334 |
|
2334 | |||
2335 | By default, benchmark a bdiff between its delta parent and itself. |
|
2335 | By default, benchmark a bdiff between its delta parent and itself. | |
2336 |
|
2336 | |||
2337 | With ``--count``, benchmark bdiffs between delta parents and self for N |
|
2337 | With ``--count``, benchmark bdiffs between delta parents and self for N | |
2338 | revisions starting at the specified revision. |
|
2338 | revisions starting at the specified revision. | |
2339 |
|
2339 | |||
2340 | With ``--alldata``, assume the requested revision is a changeset and |
|
2340 | With ``--alldata``, assume the requested revision is a changeset and | |
2341 | measure bdiffs for all changes related to that changeset (manifest |
|
2341 | measure bdiffs for all changes related to that changeset (manifest | |
2342 | and filelogs). |
|
2342 | and filelogs). | |
2343 | """ |
|
2343 | """ | |
2344 | opts = _byteskwargs(opts) |
|
2344 | opts = _byteskwargs(opts) | |
2345 |
|
2345 | |||
2346 | if opts[b'xdiff'] and not opts[b'blocks']: |
|
2346 | if opts[b'xdiff'] and not opts[b'blocks']: | |
2347 | raise error.CommandError(b'perfbdiff', b'--xdiff requires --blocks') |
|
2347 | raise error.CommandError(b'perfbdiff', b'--xdiff requires --blocks') | |
2348 |
|
2348 | |||
2349 | if opts[b'alldata']: |
|
2349 | if opts[b'alldata']: | |
2350 | opts[b'changelog'] = True |
|
2350 | opts[b'changelog'] = True | |
2351 |
|
2351 | |||
2352 | if opts.get(b'changelog') or opts.get(b'manifest'): |
|
2352 | if opts.get(b'changelog') or opts.get(b'manifest'): | |
2353 | file_, rev = None, file_ |
|
2353 | file_, rev = None, file_ | |
2354 | elif rev is None: |
|
2354 | elif rev is None: | |
2355 | raise error.CommandError(b'perfbdiff', b'invalid arguments') |
|
2355 | raise error.CommandError(b'perfbdiff', b'invalid arguments') | |
2356 |
|
2356 | |||
2357 | blocks = opts[b'blocks'] |
|
2357 | blocks = opts[b'blocks'] | |
2358 | xdiff = opts[b'xdiff'] |
|
2358 | xdiff = opts[b'xdiff'] | |
2359 | textpairs = [] |
|
2359 | textpairs = [] | |
2360 |
|
2360 | |||
2361 | r = cmdutil.openrevlog(repo, b'perfbdiff', file_, opts) |
|
2361 | r = cmdutil.openrevlog(repo, b'perfbdiff', file_, opts) | |
2362 |
|
2362 | |||
2363 | startrev = r.rev(r.lookup(rev)) |
|
2363 | startrev = r.rev(r.lookup(rev)) | |
2364 | for rev in range(startrev, min(startrev + count, len(r) - 1)): |
|
2364 | for rev in range(startrev, min(startrev + count, len(r) - 1)): | |
2365 | if opts[b'alldata']: |
|
2365 | if opts[b'alldata']: | |
2366 | # Load revisions associated with changeset. |
|
2366 | # Load revisions associated with changeset. | |
2367 | ctx = repo[rev] |
|
2367 | ctx = repo[rev] | |
2368 | mtext = _manifestrevision(repo, ctx.manifestnode()) |
|
2368 | mtext = _manifestrevision(repo, ctx.manifestnode()) | |
2369 | for pctx in ctx.parents(): |
|
2369 | for pctx in ctx.parents(): | |
2370 | pman = _manifestrevision(repo, pctx.manifestnode()) |
|
2370 | pman = _manifestrevision(repo, pctx.manifestnode()) | |
2371 | textpairs.append((pman, mtext)) |
|
2371 | textpairs.append((pman, mtext)) | |
2372 |
|
2372 | |||
2373 | # Load filelog revisions by iterating manifest delta. |
|
2373 | # Load filelog revisions by iterating manifest delta. | |
2374 | man = ctx.manifest() |
|
2374 | man = ctx.manifest() | |
2375 | pman = ctx.p1().manifest() |
|
2375 | pman = ctx.p1().manifest() | |
2376 | for filename, change in pman.diff(man).items(): |
|
2376 | for filename, change in pman.diff(man).items(): | |
2377 | fctx = repo.file(filename) |
|
2377 | fctx = repo.file(filename) | |
2378 | f1 = fctx.revision(change[0][0] or -1) |
|
2378 | f1 = fctx.revision(change[0][0] or -1) | |
2379 | f2 = fctx.revision(change[1][0] or -1) |
|
2379 | f2 = fctx.revision(change[1][0] or -1) | |
2380 | textpairs.append((f1, f2)) |
|
2380 | textpairs.append((f1, f2)) | |
2381 | else: |
|
2381 | else: | |
2382 | dp = r.deltaparent(rev) |
|
2382 | dp = r.deltaparent(rev) | |
2383 | textpairs.append((r.revision(dp), r.revision(rev))) |
|
2383 | textpairs.append((r.revision(dp), r.revision(rev))) | |
2384 |
|
2384 | |||
2385 | withthreads = threads > 0 |
|
2385 | withthreads = threads > 0 | |
2386 | if not withthreads: |
|
2386 | if not withthreads: | |
2387 |
|
2387 | |||
2388 | def d(): |
|
2388 | def d(): | |
2389 | for pair in textpairs: |
|
2389 | for pair in textpairs: | |
2390 | if xdiff: |
|
2390 | if xdiff: | |
2391 | mdiff.bdiff.xdiffblocks(*pair) |
|
2391 | mdiff.bdiff.xdiffblocks(*pair) | |
2392 | elif blocks: |
|
2392 | elif blocks: | |
2393 | mdiff.bdiff.blocks(*pair) |
|
2393 | mdiff.bdiff.blocks(*pair) | |
2394 | else: |
|
2394 | else: | |
2395 | mdiff.textdiff(*pair) |
|
2395 | mdiff.textdiff(*pair) | |
2396 |
|
2396 | |||
2397 | else: |
|
2397 | else: | |
2398 | q = queue() |
|
2398 | q = queue() | |
2399 | for i in _xrange(threads): |
|
2399 | for i in _xrange(threads): | |
2400 | q.put(None) |
|
2400 | q.put(None) | |
2401 | ready = threading.Condition() |
|
2401 | ready = threading.Condition() | |
2402 | done = threading.Event() |
|
2402 | done = threading.Event() | |
2403 | for i in _xrange(threads): |
|
2403 | for i in _xrange(threads): | |
2404 | threading.Thread( |
|
2404 | threading.Thread( | |
2405 | target=_bdiffworker, args=(q, blocks, xdiff, ready, done) |
|
2405 | target=_bdiffworker, args=(q, blocks, xdiff, ready, done) | |
2406 | ).start() |
|
2406 | ).start() | |
2407 | q.join() |
|
2407 | q.join() | |
2408 |
|
2408 | |||
2409 | def d(): |
|
2409 | def d(): | |
2410 | for pair in textpairs: |
|
2410 | for pair in textpairs: | |
2411 | q.put(pair) |
|
2411 | q.put(pair) | |
2412 | for i in _xrange(threads): |
|
2412 | for i in _xrange(threads): | |
2413 | q.put(None) |
|
2413 | q.put(None) | |
2414 | with ready: |
|
2414 | with ready: | |
2415 | ready.notify_all() |
|
2415 | ready.notify_all() | |
2416 | q.join() |
|
2416 | q.join() | |
2417 |
|
2417 | |||
2418 | timer, fm = gettimer(ui, opts) |
|
2418 | timer, fm = gettimer(ui, opts) | |
2419 | timer(d) |
|
2419 | timer(d) | |
2420 | fm.end() |
|
2420 | fm.end() | |
2421 |
|
2421 | |||
2422 | if withthreads: |
|
2422 | if withthreads: | |
2423 | done.set() |
|
2423 | done.set() | |
2424 | for i in _xrange(threads): |
|
2424 | for i in _xrange(threads): | |
2425 | q.put(None) |
|
2425 | q.put(None) | |
2426 | with ready: |
|
2426 | with ready: | |
2427 | ready.notify_all() |
|
2427 | ready.notify_all() | |
2428 |
|
2428 | |||
2429 |
|
2429 | |||
2430 | @command( |
|
2430 | @command( | |
2431 | b'perfunidiff', |
|
2431 | b'perfunidiff', | |
2432 | revlogopts |
|
2432 | revlogopts | |
2433 | + formatteropts |
|
2433 | + formatteropts | |
2434 | + [ |
|
2434 | + [ | |
2435 | ( |
|
2435 | ( | |
2436 | b'', |
|
2436 | b'', | |
2437 | b'count', |
|
2437 | b'count', | |
2438 | 1, |
|
2438 | 1, | |
2439 | b'number of revisions to test (when using --startrev)', |
|
2439 | b'number of revisions to test (when using --startrev)', | |
2440 | ), |
|
2440 | ), | |
2441 | (b'', b'alldata', False, b'test unidiffs for all associated revisions'), |
|
2441 | (b'', b'alldata', False, b'test unidiffs for all associated revisions'), | |
2442 | ], |
|
2442 | ], | |
2443 | b'-c|-m|FILE REV', |
|
2443 | b'-c|-m|FILE REV', | |
2444 | ) |
|
2444 | ) | |
2445 | def perfunidiff(ui, repo, file_, rev=None, count=None, **opts): |
|
2445 | def perfunidiff(ui, repo, file_, rev=None, count=None, **opts): | |
2446 | """benchmark a unified diff between revisions |
|
2446 | """benchmark a unified diff between revisions | |
2447 |
|
2447 | |||
2448 | This doesn't include any copy tracing - it's just a unified diff |
|
2448 | This doesn't include any copy tracing - it's just a unified diff | |
2449 | of the texts. |
|
2449 | of the texts. | |
2450 |
|
2450 | |||
2451 | By default, benchmark a diff between its delta parent and itself. |
|
2451 | By default, benchmark a diff between its delta parent and itself. | |
2452 |
|
2452 | |||
2453 | With ``--count``, benchmark diffs between delta parents and self for N |
|
2453 | With ``--count``, benchmark diffs between delta parents and self for N | |
2454 | revisions starting at the specified revision. |
|
2454 | revisions starting at the specified revision. | |
2455 |
|
2455 | |||
2456 | With ``--alldata``, assume the requested revision is a changeset and |
|
2456 | With ``--alldata``, assume the requested revision is a changeset and | |
2457 | measure diffs for all changes related to that changeset (manifest |
|
2457 | measure diffs for all changes related to that changeset (manifest | |
2458 | and filelogs). |
|
2458 | and filelogs). | |
2459 | """ |
|
2459 | """ | |
2460 | opts = _byteskwargs(opts) |
|
2460 | opts = _byteskwargs(opts) | |
2461 | if opts[b'alldata']: |
|
2461 | if opts[b'alldata']: | |
2462 | opts[b'changelog'] = True |
|
2462 | opts[b'changelog'] = True | |
2463 |
|
2463 | |||
2464 | if opts.get(b'changelog') or opts.get(b'manifest'): |
|
2464 | if opts.get(b'changelog') or opts.get(b'manifest'): | |
2465 | file_, rev = None, file_ |
|
2465 | file_, rev = None, file_ | |
2466 | elif rev is None: |
|
2466 | elif rev is None: | |
2467 | raise error.CommandError(b'perfunidiff', b'invalid arguments') |
|
2467 | raise error.CommandError(b'perfunidiff', b'invalid arguments') | |
2468 |
|
2468 | |||
2469 | textpairs = [] |
|
2469 | textpairs = [] | |
2470 |
|
2470 | |||
2471 | r = cmdutil.openrevlog(repo, b'perfunidiff', file_, opts) |
|
2471 | r = cmdutil.openrevlog(repo, b'perfunidiff', file_, opts) | |
2472 |
|
2472 | |||
2473 | startrev = r.rev(r.lookup(rev)) |
|
2473 | startrev = r.rev(r.lookup(rev)) | |
2474 | for rev in range(startrev, min(startrev + count, len(r) - 1)): |
|
2474 | for rev in range(startrev, min(startrev + count, len(r) - 1)): | |
2475 | if opts[b'alldata']: |
|
2475 | if opts[b'alldata']: | |
2476 | # Load revisions associated with changeset. |
|
2476 | # Load revisions associated with changeset. | |
2477 | ctx = repo[rev] |
|
2477 | ctx = repo[rev] | |
2478 | mtext = _manifestrevision(repo, ctx.manifestnode()) |
|
2478 | mtext = _manifestrevision(repo, ctx.manifestnode()) | |
2479 | for pctx in ctx.parents(): |
|
2479 | for pctx in ctx.parents(): | |
2480 | pman = _manifestrevision(repo, pctx.manifestnode()) |
|
2480 | pman = _manifestrevision(repo, pctx.manifestnode()) | |
2481 | textpairs.append((pman, mtext)) |
|
2481 | textpairs.append((pman, mtext)) | |
2482 |
|
2482 | |||
2483 | # Load filelog revisions by iterating manifest delta. |
|
2483 | # Load filelog revisions by iterating manifest delta. | |
2484 | man = ctx.manifest() |
|
2484 | man = ctx.manifest() | |
2485 | pman = ctx.p1().manifest() |
|
2485 | pman = ctx.p1().manifest() | |
2486 | for filename, change in pman.diff(man).items(): |
|
2486 | for filename, change in pman.diff(man).items(): | |
2487 | fctx = repo.file(filename) |
|
2487 | fctx = repo.file(filename) | |
2488 | f1 = fctx.revision(change[0][0] or -1) |
|
2488 | f1 = fctx.revision(change[0][0] or -1) | |
2489 | f2 = fctx.revision(change[1][0] or -1) |
|
2489 | f2 = fctx.revision(change[1][0] or -1) | |
2490 | textpairs.append((f1, f2)) |
|
2490 | textpairs.append((f1, f2)) | |
2491 | else: |
|
2491 | else: | |
2492 | dp = r.deltaparent(rev) |
|
2492 | dp = r.deltaparent(rev) | |
2493 | textpairs.append((r.revision(dp), r.revision(rev))) |
|
2493 | textpairs.append((r.revision(dp), r.revision(rev))) | |
2494 |
|
2494 | |||
2495 | def d(): |
|
2495 | def d(): | |
2496 | for left, right in textpairs: |
|
2496 | for left, right in textpairs: | |
2497 | # The date strings don't matter, so we pass empty strings. |
|
2497 | # The date strings don't matter, so we pass empty strings. | |
2498 | headerlines, hunks = mdiff.unidiff( |
|
2498 | headerlines, hunks = mdiff.unidiff( | |
2499 | left, b'', right, b'', b'left', b'right', binary=False |
|
2499 | left, b'', right, b'', b'left', b'right', binary=False | |
2500 | ) |
|
2500 | ) | |
2501 | # consume iterators in roughly the way patch.py does |
|
2501 | # consume iterators in roughly the way patch.py does | |
2502 | b'\n'.join(headerlines) |
|
2502 | b'\n'.join(headerlines) | |
2503 | b''.join(sum((list(hlines) for hrange, hlines in hunks), [])) |
|
2503 | b''.join(sum((list(hlines) for hrange, hlines in hunks), [])) | |
2504 |
|
2504 | |||
2505 | timer, fm = gettimer(ui, opts) |
|
2505 | timer, fm = gettimer(ui, opts) | |
2506 | timer(d) |
|
2506 | timer(d) | |
2507 | fm.end() |
|
2507 | fm.end() | |
2508 |
|
2508 | |||
2509 |
|
2509 | |||
2510 | @command(b'perfdiffwd', formatteropts) |
|
2510 | @command(b'perfdiffwd', formatteropts) | |
2511 | def perfdiffwd(ui, repo, **opts): |
|
2511 | def perfdiffwd(ui, repo, **opts): | |
2512 | """Profile diff of working directory changes""" |
|
2512 | """Profile diff of working directory changes""" | |
2513 | opts = _byteskwargs(opts) |
|
2513 | opts = _byteskwargs(opts) | |
2514 | timer, fm = gettimer(ui, opts) |
|
2514 | timer, fm = gettimer(ui, opts) | |
2515 | options = { |
|
2515 | options = { | |
2516 | 'w': 'ignore_all_space', |
|
2516 | 'w': 'ignore_all_space', | |
2517 | 'b': 'ignore_space_change', |
|
2517 | 'b': 'ignore_space_change', | |
2518 | 'B': 'ignore_blank_lines', |
|
2518 | 'B': 'ignore_blank_lines', | |
2519 | } |
|
2519 | } | |
2520 |
|
2520 | |||
2521 | for diffopt in ('', 'w', 'b', 'B', 'wB'): |
|
2521 | for diffopt in ('', 'w', 'b', 'B', 'wB'): | |
2522 | opts = dict((options[c], b'1') for c in diffopt) |
|
2522 | opts = dict((options[c], b'1') for c in diffopt) | |
2523 |
|
2523 | |||
2524 | def d(): |
|
2524 | def d(): | |
2525 | ui.pushbuffer() |
|
2525 | ui.pushbuffer() | |
2526 | commands.diff(ui, repo, **opts) |
|
2526 | commands.diff(ui, repo, **opts) | |
2527 | ui.popbuffer() |
|
2527 | ui.popbuffer() | |
2528 |
|
2528 | |||
2529 | diffopt = diffopt.encode('ascii') |
|
2529 | diffopt = diffopt.encode('ascii') | |
2530 | title = b'diffopts: %s' % (diffopt and (b'-' + diffopt) or b'none') |
|
2530 | title = b'diffopts: %s' % (diffopt and (b'-' + diffopt) or b'none') | |
2531 | timer(d, title=title) |
|
2531 | timer(d, title=title) | |
2532 | fm.end() |
|
2532 | fm.end() | |
2533 |
|
2533 | |||
2534 |
|
2534 | |||
2535 | @command(b'perfrevlogindex', revlogopts + formatteropts, b'-c|-m|FILE') |
|
2535 | @command(b'perfrevlogindex', revlogopts + formatteropts, b'-c|-m|FILE') | |
2536 | def perfrevlogindex(ui, repo, file_=None, **opts): |
|
2536 | def perfrevlogindex(ui, repo, file_=None, **opts): | |
2537 | """Benchmark operations against a revlog index. |
|
2537 | """Benchmark operations against a revlog index. | |
2538 |
|
2538 | |||
2539 | This tests constructing a revlog instance, reading index data, |
|
2539 | This tests constructing a revlog instance, reading index data, | |
2540 | parsing index data, and performing various operations related to |
|
2540 | parsing index data, and performing various operations related to | |
2541 | index data. |
|
2541 | index data. | |
2542 | """ |
|
2542 | """ | |
2543 |
|
2543 | |||
2544 | opts = _byteskwargs(opts) |
|
2544 | opts = _byteskwargs(opts) | |
2545 |
|
2545 | |||
2546 | rl = cmdutil.openrevlog(repo, b'perfrevlogindex', file_, opts) |
|
2546 | rl = cmdutil.openrevlog(repo, b'perfrevlogindex', file_, opts) | |
2547 |
|
2547 | |||
2548 | opener = getattr(rl, 'opener') # trick linter |
|
2548 | opener = getattr(rl, 'opener') # trick linter | |
2549 | indexfile = rl.indexfile |
|
2549 | indexfile = rl.indexfile | |
2550 | data = opener.read(indexfile) |
|
2550 | data = opener.read(indexfile) | |
2551 |
|
2551 | |||
2552 | header = struct.unpack(b'>I', data[0:4])[0] |
|
2552 | header = struct.unpack(b'>I', data[0:4])[0] | |
2553 | version = header & 0xFFFF |
|
2553 | version = header & 0xFFFF | |
2554 | if version == 1: |
|
2554 | if version == 1: | |
2555 | revlogio = revlog.revlogio() |
|
2555 | revlogio = revlog.revlogio() | |
2556 | inline = header & (1 << 16) |
|
2556 | inline = header & (1 << 16) | |
2557 | else: |
|
2557 | else: | |
2558 | raise error.Abort(b'unsupported revlog version: %d' % version) |
|
2558 | raise error.Abort(b'unsupported revlog version: %d' % version) | |
2559 |
|
2559 | |||
2560 | rllen = len(rl) |
|
2560 | rllen = len(rl) | |
2561 |
|
2561 | |||
2562 | node0 = rl.node(0) |
|
2562 | node0 = rl.node(0) | |
2563 | node25 = rl.node(rllen // 4) |
|
2563 | node25 = rl.node(rllen // 4) | |
2564 | node50 = rl.node(rllen // 2) |
|
2564 | node50 = rl.node(rllen // 2) | |
2565 | node75 = rl.node(rllen // 4 * 3) |
|
2565 | node75 = rl.node(rllen // 4 * 3) | |
2566 | node100 = rl.node(rllen - 1) |
|
2566 | node100 = rl.node(rllen - 1) | |
2567 |
|
2567 | |||
2568 | allrevs = range(rllen) |
|
2568 | allrevs = range(rllen) | |
2569 | allrevsrev = list(reversed(allrevs)) |
|
2569 | allrevsrev = list(reversed(allrevs)) | |
2570 | allnodes = [rl.node(rev) for rev in range(rllen)] |
|
2570 | allnodes = [rl.node(rev) for rev in range(rllen)] | |
2571 | allnodesrev = list(reversed(allnodes)) |
|
2571 | allnodesrev = list(reversed(allnodes)) | |
2572 |
|
2572 | |||
2573 | def constructor(): |
|
2573 | def constructor(): | |
2574 | revlog.revlog(opener, indexfile) |
|
2574 | revlog.revlog(opener, indexfile) | |
2575 |
|
2575 | |||
2576 | def read(): |
|
2576 | def read(): | |
2577 | with opener(indexfile) as fh: |
|
2577 | with opener(indexfile) as fh: | |
2578 | fh.read() |
|
2578 | fh.read() | |
2579 |
|
2579 | |||
2580 | def parseindex(): |
|
2580 | def parseindex(): | |
2581 | revlogio.parseindex(data, inline) |
|
2581 | revlogio.parseindex(data, inline) | |
2582 |
|
2582 | |||
2583 | def getentry(revornode): |
|
2583 | def getentry(revornode): | |
2584 | index = revlogio.parseindex(data, inline)[0] |
|
2584 | index = revlogio.parseindex(data, inline)[0] | |
2585 | index[revornode] |
|
2585 | index[revornode] | |
2586 |
|
2586 | |||
2587 | def getentries(revs, count=1): |
|
2587 | def getentries(revs, count=1): | |
2588 | index = revlogio.parseindex(data, inline)[0] |
|
2588 | index = revlogio.parseindex(data, inline)[0] | |
2589 |
|
2589 | |||
2590 | for i in range(count): |
|
2590 | for i in range(count): | |
2591 | for rev in revs: |
|
2591 | for rev in revs: | |
2592 | index[rev] |
|
2592 | index[rev] | |
2593 |
|
2593 | |||
2594 | def resolvenode(node): |
|
2594 | def resolvenode(node): | |
2595 |
nodemap = revlogio.parseindex(data, inline)[ |
|
2595 | nodemap = getattr(revlogio.parseindex(data, inline)[0], 'nodemap', None) | |
2596 | # This only works for the C code. |
|
2596 | # This only works for the C code. | |
2597 | if nodemap is None: |
|
2597 | if nodemap is None: | |
2598 | return |
|
2598 | return | |
2599 |
|
2599 | |||
2600 | try: |
|
2600 | try: | |
2601 | nodemap[node] |
|
2601 | nodemap[node] | |
2602 | except error.RevlogError: |
|
2602 | except error.RevlogError: | |
2603 | pass |
|
2603 | pass | |
2604 |
|
2604 | |||
2605 | def resolvenodes(nodes, count=1): |
|
2605 | def resolvenodes(nodes, count=1): | |
2606 |
nodemap = revlogio.parseindex(data, inline)[ |
|
2606 | nodemap = getattr(revlogio.parseindex(data, inline)[0], 'nodemap', None) | |
2607 | if nodemap is None: |
|
2607 | if nodemap is None: | |
2608 | return |
|
2608 | return | |
2609 |
|
2609 | |||
2610 | for i in range(count): |
|
2610 | for i in range(count): | |
2611 | for node in nodes: |
|
2611 | for node in nodes: | |
2612 | try: |
|
2612 | try: | |
2613 | nodemap[node] |
|
2613 | nodemap[node] | |
2614 | except error.RevlogError: |
|
2614 | except error.RevlogError: | |
2615 | pass |
|
2615 | pass | |
2616 |
|
2616 | |||
2617 | benches = [ |
|
2617 | benches = [ | |
2618 | (constructor, b'revlog constructor'), |
|
2618 | (constructor, b'revlog constructor'), | |
2619 | (read, b'read'), |
|
2619 | (read, b'read'), | |
2620 | (parseindex, b'create index object'), |
|
2620 | (parseindex, b'create index object'), | |
2621 | (lambda: getentry(0), b'retrieve index entry for rev 0'), |
|
2621 | (lambda: getentry(0), b'retrieve index entry for rev 0'), | |
2622 | (lambda: resolvenode(b'a' * 20), b'look up missing node'), |
|
2622 | (lambda: resolvenode(b'a' * 20), b'look up missing node'), | |
2623 | (lambda: resolvenode(node0), b'look up node at rev 0'), |
|
2623 | (lambda: resolvenode(node0), b'look up node at rev 0'), | |
2624 | (lambda: resolvenode(node25), b'look up node at 1/4 len'), |
|
2624 | (lambda: resolvenode(node25), b'look up node at 1/4 len'), | |
2625 | (lambda: resolvenode(node50), b'look up node at 1/2 len'), |
|
2625 | (lambda: resolvenode(node50), b'look up node at 1/2 len'), | |
2626 | (lambda: resolvenode(node75), b'look up node at 3/4 len'), |
|
2626 | (lambda: resolvenode(node75), b'look up node at 3/4 len'), | |
2627 | (lambda: resolvenode(node100), b'look up node at tip'), |
|
2627 | (lambda: resolvenode(node100), b'look up node at tip'), | |
2628 | # 2x variation is to measure caching impact. |
|
2628 | # 2x variation is to measure caching impact. | |
2629 | (lambda: resolvenodes(allnodes), b'look up all nodes (forward)'), |
|
2629 | (lambda: resolvenodes(allnodes), b'look up all nodes (forward)'), | |
2630 | (lambda: resolvenodes(allnodes, 2), b'look up all nodes 2x (forward)'), |
|
2630 | (lambda: resolvenodes(allnodes, 2), b'look up all nodes 2x (forward)'), | |
2631 | (lambda: resolvenodes(allnodesrev), b'look up all nodes (reverse)'), |
|
2631 | (lambda: resolvenodes(allnodesrev), b'look up all nodes (reverse)'), | |
2632 | ( |
|
2632 | ( | |
2633 | lambda: resolvenodes(allnodesrev, 2), |
|
2633 | lambda: resolvenodes(allnodesrev, 2), | |
2634 | b'look up all nodes 2x (reverse)', |
|
2634 | b'look up all nodes 2x (reverse)', | |
2635 | ), |
|
2635 | ), | |
2636 | (lambda: getentries(allrevs), b'retrieve all index entries (forward)'), |
|
2636 | (lambda: getentries(allrevs), b'retrieve all index entries (forward)'), | |
2637 | ( |
|
2637 | ( | |
2638 | lambda: getentries(allrevs, 2), |
|
2638 | lambda: getentries(allrevs, 2), | |
2639 | b'retrieve all index entries 2x (forward)', |
|
2639 | b'retrieve all index entries 2x (forward)', | |
2640 | ), |
|
2640 | ), | |
2641 | ( |
|
2641 | ( | |
2642 | lambda: getentries(allrevsrev), |
|
2642 | lambda: getentries(allrevsrev), | |
2643 | b'retrieve all index entries (reverse)', |
|
2643 | b'retrieve all index entries (reverse)', | |
2644 | ), |
|
2644 | ), | |
2645 | ( |
|
2645 | ( | |
2646 | lambda: getentries(allrevsrev, 2), |
|
2646 | lambda: getentries(allrevsrev, 2), | |
2647 | b'retrieve all index entries 2x (reverse)', |
|
2647 | b'retrieve all index entries 2x (reverse)', | |
2648 | ), |
|
2648 | ), | |
2649 | ] |
|
2649 | ] | |
2650 |
|
2650 | |||
2651 | for fn, title in benches: |
|
2651 | for fn, title in benches: | |
2652 | timer, fm = gettimer(ui, opts) |
|
2652 | timer, fm = gettimer(ui, opts) | |
2653 | timer(fn, title=title) |
|
2653 | timer(fn, title=title) | |
2654 | fm.end() |
|
2654 | fm.end() | |
2655 |
|
2655 | |||
2656 |
|
2656 | |||
2657 | @command( |
|
2657 | @command( | |
2658 | b'perfrevlogrevisions', |
|
2658 | b'perfrevlogrevisions', | |
2659 | revlogopts |
|
2659 | revlogopts | |
2660 | + formatteropts |
|
2660 | + formatteropts | |
2661 | + [ |
|
2661 | + [ | |
2662 | (b'd', b'dist', 100, b'distance between the revisions'), |
|
2662 | (b'd', b'dist', 100, b'distance between the revisions'), | |
2663 | (b's', b'startrev', 0, b'revision to start reading at'), |
|
2663 | (b's', b'startrev', 0, b'revision to start reading at'), | |
2664 | (b'', b'reverse', False, b'read in reverse'), |
|
2664 | (b'', b'reverse', False, b'read in reverse'), | |
2665 | ], |
|
2665 | ], | |
2666 | b'-c|-m|FILE', |
|
2666 | b'-c|-m|FILE', | |
2667 | ) |
|
2667 | ) | |
2668 | def perfrevlogrevisions( |
|
2668 | def perfrevlogrevisions( | |
2669 | ui, repo, file_=None, startrev=0, reverse=False, **opts |
|
2669 | ui, repo, file_=None, startrev=0, reverse=False, **opts | |
2670 | ): |
|
2670 | ): | |
2671 | """Benchmark reading a series of revisions from a revlog. |
|
2671 | """Benchmark reading a series of revisions from a revlog. | |
2672 |
|
2672 | |||
2673 | By default, we read every ``-d/--dist`` revision from 0 to tip of |
|
2673 | By default, we read every ``-d/--dist`` revision from 0 to tip of | |
2674 | the specified revlog. |
|
2674 | the specified revlog. | |
2675 |
|
2675 | |||
2676 | The start revision can be defined via ``-s/--startrev``. |
|
2676 | The start revision can be defined via ``-s/--startrev``. | |
2677 | """ |
|
2677 | """ | |
2678 | opts = _byteskwargs(opts) |
|
2678 | opts = _byteskwargs(opts) | |
2679 |
|
2679 | |||
2680 | rl = cmdutil.openrevlog(repo, b'perfrevlogrevisions', file_, opts) |
|
2680 | rl = cmdutil.openrevlog(repo, b'perfrevlogrevisions', file_, opts) | |
2681 | rllen = getlen(ui)(rl) |
|
2681 | rllen = getlen(ui)(rl) | |
2682 |
|
2682 | |||
2683 | if startrev < 0: |
|
2683 | if startrev < 0: | |
2684 | startrev = rllen + startrev |
|
2684 | startrev = rllen + startrev | |
2685 |
|
2685 | |||
2686 | def d(): |
|
2686 | def d(): | |
2687 | rl.clearcaches() |
|
2687 | rl.clearcaches() | |
2688 |
|
2688 | |||
2689 | beginrev = startrev |
|
2689 | beginrev = startrev | |
2690 | endrev = rllen |
|
2690 | endrev = rllen | |
2691 | dist = opts[b'dist'] |
|
2691 | dist = opts[b'dist'] | |
2692 |
|
2692 | |||
2693 | if reverse: |
|
2693 | if reverse: | |
2694 | beginrev, endrev = endrev - 1, beginrev - 1 |
|
2694 | beginrev, endrev = endrev - 1, beginrev - 1 | |
2695 | dist = -1 * dist |
|
2695 | dist = -1 * dist | |
2696 |
|
2696 | |||
2697 | for x in _xrange(beginrev, endrev, dist): |
|
2697 | for x in _xrange(beginrev, endrev, dist): | |
2698 | # Old revisions don't support passing int. |
|
2698 | # Old revisions don't support passing int. | |
2699 | n = rl.node(x) |
|
2699 | n = rl.node(x) | |
2700 | rl.revision(n) |
|
2700 | rl.revision(n) | |
2701 |
|
2701 | |||
2702 | timer, fm = gettimer(ui, opts) |
|
2702 | timer, fm = gettimer(ui, opts) | |
2703 | timer(d) |
|
2703 | timer(d) | |
2704 | fm.end() |
|
2704 | fm.end() | |
2705 |
|
2705 | |||
2706 |
|
2706 | |||
2707 | @command( |
|
2707 | @command( | |
2708 | b'perfrevlogwrite', |
|
2708 | b'perfrevlogwrite', | |
2709 | revlogopts |
|
2709 | revlogopts | |
2710 | + formatteropts |
|
2710 | + formatteropts | |
2711 | + [ |
|
2711 | + [ | |
2712 | (b's', b'startrev', 1000, b'revision to start writing at'), |
|
2712 | (b's', b'startrev', 1000, b'revision to start writing at'), | |
2713 | (b'', b'stoprev', -1, b'last revision to write'), |
|
2713 | (b'', b'stoprev', -1, b'last revision to write'), | |
2714 | (b'', b'count', 3, b'number of passes to perform'), |
|
2714 | (b'', b'count', 3, b'number of passes to perform'), | |
2715 | (b'', b'details', False, b'print timing for every revisions tested'), |
|
2715 | (b'', b'details', False, b'print timing for every revisions tested'), | |
2716 | (b'', b'source', b'full', b'the kind of data feed in the revlog'), |
|
2716 | (b'', b'source', b'full', b'the kind of data feed in the revlog'), | |
2717 | (b'', b'lazydeltabase', True, b'try the provided delta first'), |
|
2717 | (b'', b'lazydeltabase', True, b'try the provided delta first'), | |
2718 | (b'', b'clear-caches', True, b'clear revlog cache between calls'), |
|
2718 | (b'', b'clear-caches', True, b'clear revlog cache between calls'), | |
2719 | ], |
|
2719 | ], | |
2720 | b'-c|-m|FILE', |
|
2720 | b'-c|-m|FILE', | |
2721 | ) |
|
2721 | ) | |
2722 | def perfrevlogwrite(ui, repo, file_=None, startrev=1000, stoprev=-1, **opts): |
|
2722 | def perfrevlogwrite(ui, repo, file_=None, startrev=1000, stoprev=-1, **opts): | |
2723 | """Benchmark writing a series of revisions to a revlog. |
|
2723 | """Benchmark writing a series of revisions to a revlog. | |
2724 |
|
2724 | |||
2725 | Possible source values are: |
|
2725 | Possible source values are: | |
2726 | * `full`: add from a full text (default). |
|
2726 | * `full`: add from a full text (default). | |
2727 | * `parent-1`: add from a delta to the first parent |
|
2727 | * `parent-1`: add from a delta to the first parent | |
2728 | * `parent-2`: add from a delta to the second parent if it exists |
|
2728 | * `parent-2`: add from a delta to the second parent if it exists | |
2729 | (use a delta from the first parent otherwise) |
|
2729 | (use a delta from the first parent otherwise) | |
2730 | * `parent-smallest`: add from the smallest delta (either p1 or p2) |
|
2730 | * `parent-smallest`: add from the smallest delta (either p1 or p2) | |
2731 | * `storage`: add from the existing precomputed deltas |
|
2731 | * `storage`: add from the existing precomputed deltas | |
2732 |
|
2732 | |||
2733 | Note: This performance command measures performance in a custom way. As a |
|
2733 | Note: This performance command measures performance in a custom way. As a | |
2734 | result some of the global configuration of the 'perf' command does not |
|
2734 | result some of the global configuration of the 'perf' command does not | |
2735 | apply to it: |
|
2735 | apply to it: | |
2736 |
|
2736 | |||
2737 | * ``pre-run``: disabled |
|
2737 | * ``pre-run``: disabled | |
2738 |
|
2738 | |||
2739 | * ``profile-benchmark``: disabled |
|
2739 | * ``profile-benchmark``: disabled | |
2740 |
|
2740 | |||
2741 | * ``run-limits``: disabled use --count instead |
|
2741 | * ``run-limits``: disabled use --count instead | |
2742 | """ |
|
2742 | """ | |
2743 | opts = _byteskwargs(opts) |
|
2743 | opts = _byteskwargs(opts) | |
2744 |
|
2744 | |||
2745 | rl = cmdutil.openrevlog(repo, b'perfrevlogwrite', file_, opts) |
|
2745 | rl = cmdutil.openrevlog(repo, b'perfrevlogwrite', file_, opts) | |
2746 | rllen = getlen(ui)(rl) |
|
2746 | rllen = getlen(ui)(rl) | |
2747 | if startrev < 0: |
|
2747 | if startrev < 0: | |
2748 | startrev = rllen + startrev |
|
2748 | startrev = rllen + startrev | |
2749 | if stoprev < 0: |
|
2749 | if stoprev < 0: | |
2750 | stoprev = rllen + stoprev |
|
2750 | stoprev = rllen + stoprev | |
2751 |
|
2751 | |||
2752 | lazydeltabase = opts['lazydeltabase'] |
|
2752 | lazydeltabase = opts['lazydeltabase'] | |
2753 | source = opts['source'] |
|
2753 | source = opts['source'] | |
2754 | clearcaches = opts['clear_caches'] |
|
2754 | clearcaches = opts['clear_caches'] | |
2755 | validsource = ( |
|
2755 | validsource = ( | |
2756 | b'full', |
|
2756 | b'full', | |
2757 | b'parent-1', |
|
2757 | b'parent-1', | |
2758 | b'parent-2', |
|
2758 | b'parent-2', | |
2759 | b'parent-smallest', |
|
2759 | b'parent-smallest', | |
2760 | b'storage', |
|
2760 | b'storage', | |
2761 | ) |
|
2761 | ) | |
2762 | if source not in validsource: |
|
2762 | if source not in validsource: | |
2763 | raise error.Abort('invalid source type: %s' % source) |
|
2763 | raise error.Abort('invalid source type: %s' % source) | |
2764 |
|
2764 | |||
2765 | ### actually gather results |
|
2765 | ### actually gather results | |
2766 | count = opts['count'] |
|
2766 | count = opts['count'] | |
2767 | if count <= 0: |
|
2767 | if count <= 0: | |
2768 | raise error.Abort('invalide run count: %d' % count) |
|
2768 | raise error.Abort('invalide run count: %d' % count) | |
2769 | allresults = [] |
|
2769 | allresults = [] | |
2770 | for c in range(count): |
|
2770 | for c in range(count): | |
2771 | timing = _timeonewrite( |
|
2771 | timing = _timeonewrite( | |
2772 | ui, |
|
2772 | ui, | |
2773 | rl, |
|
2773 | rl, | |
2774 | source, |
|
2774 | source, | |
2775 | startrev, |
|
2775 | startrev, | |
2776 | stoprev, |
|
2776 | stoprev, | |
2777 | c + 1, |
|
2777 | c + 1, | |
2778 | lazydeltabase=lazydeltabase, |
|
2778 | lazydeltabase=lazydeltabase, | |
2779 | clearcaches=clearcaches, |
|
2779 | clearcaches=clearcaches, | |
2780 | ) |
|
2780 | ) | |
2781 | allresults.append(timing) |
|
2781 | allresults.append(timing) | |
2782 |
|
2782 | |||
2783 | ### consolidate the results in a single list |
|
2783 | ### consolidate the results in a single list | |
2784 | results = [] |
|
2784 | results = [] | |
2785 | for idx, (rev, t) in enumerate(allresults[0]): |
|
2785 | for idx, (rev, t) in enumerate(allresults[0]): | |
2786 | ts = [t] |
|
2786 | ts = [t] | |
2787 | for other in allresults[1:]: |
|
2787 | for other in allresults[1:]: | |
2788 | orev, ot = other[idx] |
|
2788 | orev, ot = other[idx] | |
2789 | assert orev == rev |
|
2789 | assert orev == rev | |
2790 | ts.append(ot) |
|
2790 | ts.append(ot) | |
2791 | results.append((rev, ts)) |
|
2791 | results.append((rev, ts)) | |
2792 | resultcount = len(results) |
|
2792 | resultcount = len(results) | |
2793 |
|
2793 | |||
2794 | ### Compute and display relevant statistics |
|
2794 | ### Compute and display relevant statistics | |
2795 |
|
2795 | |||
2796 | # get a formatter |
|
2796 | # get a formatter | |
2797 | fm = ui.formatter(b'perf', opts) |
|
2797 | fm = ui.formatter(b'perf', opts) | |
2798 | displayall = ui.configbool(b"perf", b"all-timing", False) |
|
2798 | displayall = ui.configbool(b"perf", b"all-timing", False) | |
2799 |
|
2799 | |||
2800 | # print individual details if requested |
|
2800 | # print individual details if requested | |
2801 | if opts['details']: |
|
2801 | if opts['details']: | |
2802 | for idx, item in enumerate(results, 1): |
|
2802 | for idx, item in enumerate(results, 1): | |
2803 | rev, data = item |
|
2803 | rev, data = item | |
2804 | title = 'revisions #%d of %d, rev %d' % (idx, resultcount, rev) |
|
2804 | title = 'revisions #%d of %d, rev %d' % (idx, resultcount, rev) | |
2805 | formatone(fm, data, title=title, displayall=displayall) |
|
2805 | formatone(fm, data, title=title, displayall=displayall) | |
2806 |
|
2806 | |||
2807 | # sorts results by median time |
|
2807 | # sorts results by median time | |
2808 | results.sort(key=lambda x: sorted(x[1])[len(x[1]) // 2]) |
|
2808 | results.sort(key=lambda x: sorted(x[1])[len(x[1]) // 2]) | |
2809 | # list of (name, index) to display) |
|
2809 | # list of (name, index) to display) | |
2810 | relevants = [ |
|
2810 | relevants = [ | |
2811 | ("min", 0), |
|
2811 | ("min", 0), | |
2812 | ("10%", resultcount * 10 // 100), |
|
2812 | ("10%", resultcount * 10 // 100), | |
2813 | ("25%", resultcount * 25 // 100), |
|
2813 | ("25%", resultcount * 25 // 100), | |
2814 | ("50%", resultcount * 70 // 100), |
|
2814 | ("50%", resultcount * 70 // 100), | |
2815 | ("75%", resultcount * 75 // 100), |
|
2815 | ("75%", resultcount * 75 // 100), | |
2816 | ("90%", resultcount * 90 // 100), |
|
2816 | ("90%", resultcount * 90 // 100), | |
2817 | ("95%", resultcount * 95 // 100), |
|
2817 | ("95%", resultcount * 95 // 100), | |
2818 | ("99%", resultcount * 99 // 100), |
|
2818 | ("99%", resultcount * 99 // 100), | |
2819 | ("99.9%", resultcount * 999 // 1000), |
|
2819 | ("99.9%", resultcount * 999 // 1000), | |
2820 | ("99.99%", resultcount * 9999 // 10000), |
|
2820 | ("99.99%", resultcount * 9999 // 10000), | |
2821 | ("99.999%", resultcount * 99999 // 100000), |
|
2821 | ("99.999%", resultcount * 99999 // 100000), | |
2822 | ("max", -1), |
|
2822 | ("max", -1), | |
2823 | ] |
|
2823 | ] | |
2824 | if not ui.quiet: |
|
2824 | if not ui.quiet: | |
2825 | for name, idx in relevants: |
|
2825 | for name, idx in relevants: | |
2826 | data = results[idx] |
|
2826 | data = results[idx] | |
2827 | title = '%s of %d, rev %d' % (name, resultcount, data[0]) |
|
2827 | title = '%s of %d, rev %d' % (name, resultcount, data[0]) | |
2828 | formatone(fm, data[1], title=title, displayall=displayall) |
|
2828 | formatone(fm, data[1], title=title, displayall=displayall) | |
2829 |
|
2829 | |||
2830 | # XXX summing that many float will not be very precise, we ignore this fact |
|
2830 | # XXX summing that many float will not be very precise, we ignore this fact | |
2831 | # for now |
|
2831 | # for now | |
2832 | totaltime = [] |
|
2832 | totaltime = [] | |
2833 | for item in allresults: |
|
2833 | for item in allresults: | |
2834 | totaltime.append( |
|
2834 | totaltime.append( | |
2835 | ( |
|
2835 | ( | |
2836 | sum(x[1][0] for x in item), |
|
2836 | sum(x[1][0] for x in item), | |
2837 | sum(x[1][1] for x in item), |
|
2837 | sum(x[1][1] for x in item), | |
2838 | sum(x[1][2] for x in item), |
|
2838 | sum(x[1][2] for x in item), | |
2839 | ) |
|
2839 | ) | |
2840 | ) |
|
2840 | ) | |
2841 | formatone( |
|
2841 | formatone( | |
2842 | fm, |
|
2842 | fm, | |
2843 | totaltime, |
|
2843 | totaltime, | |
2844 | title="total time (%d revs)" % resultcount, |
|
2844 | title="total time (%d revs)" % resultcount, | |
2845 | displayall=displayall, |
|
2845 | displayall=displayall, | |
2846 | ) |
|
2846 | ) | |
2847 | fm.end() |
|
2847 | fm.end() | |
2848 |
|
2848 | |||
2849 |
|
2849 | |||
2850 | class _faketr(object): |
|
2850 | class _faketr(object): | |
2851 | def add(s, x, y, z=None): |
|
2851 | def add(s, x, y, z=None): | |
2852 | return None |
|
2852 | return None | |
2853 |
|
2853 | |||
2854 |
|
2854 | |||
2855 | def _timeonewrite( |
|
2855 | def _timeonewrite( | |
2856 | ui, |
|
2856 | ui, | |
2857 | orig, |
|
2857 | orig, | |
2858 | source, |
|
2858 | source, | |
2859 | startrev, |
|
2859 | startrev, | |
2860 | stoprev, |
|
2860 | stoprev, | |
2861 | runidx=None, |
|
2861 | runidx=None, | |
2862 | lazydeltabase=True, |
|
2862 | lazydeltabase=True, | |
2863 | clearcaches=True, |
|
2863 | clearcaches=True, | |
2864 | ): |
|
2864 | ): | |
2865 | timings = [] |
|
2865 | timings = [] | |
2866 | tr = _faketr() |
|
2866 | tr = _faketr() | |
2867 | with _temprevlog(ui, orig, startrev) as dest: |
|
2867 | with _temprevlog(ui, orig, startrev) as dest: | |
2868 | dest._lazydeltabase = lazydeltabase |
|
2868 | dest._lazydeltabase = lazydeltabase | |
2869 | revs = list(orig.revs(startrev, stoprev)) |
|
2869 | revs = list(orig.revs(startrev, stoprev)) | |
2870 | total = len(revs) |
|
2870 | total = len(revs) | |
2871 | topic = 'adding' |
|
2871 | topic = 'adding' | |
2872 | if runidx is not None: |
|
2872 | if runidx is not None: | |
2873 | topic += ' (run #%d)' % runidx |
|
2873 | topic += ' (run #%d)' % runidx | |
2874 | # Support both old and new progress API |
|
2874 | # Support both old and new progress API | |
2875 | if util.safehasattr(ui, 'makeprogress'): |
|
2875 | if util.safehasattr(ui, 'makeprogress'): | |
2876 | progress = ui.makeprogress(topic, unit='revs', total=total) |
|
2876 | progress = ui.makeprogress(topic, unit='revs', total=total) | |
2877 |
|
2877 | |||
2878 | def updateprogress(pos): |
|
2878 | def updateprogress(pos): | |
2879 | progress.update(pos) |
|
2879 | progress.update(pos) | |
2880 |
|
2880 | |||
2881 | def completeprogress(): |
|
2881 | def completeprogress(): | |
2882 | progress.complete() |
|
2882 | progress.complete() | |
2883 |
|
2883 | |||
2884 | else: |
|
2884 | else: | |
2885 |
|
2885 | |||
2886 | def updateprogress(pos): |
|
2886 | def updateprogress(pos): | |
2887 | ui.progress(topic, pos, unit='revs', total=total) |
|
2887 | ui.progress(topic, pos, unit='revs', total=total) | |
2888 |
|
2888 | |||
2889 | def completeprogress(): |
|
2889 | def completeprogress(): | |
2890 | ui.progress(topic, None, unit='revs', total=total) |
|
2890 | ui.progress(topic, None, unit='revs', total=total) | |
2891 |
|
2891 | |||
2892 | for idx, rev in enumerate(revs): |
|
2892 | for idx, rev in enumerate(revs): | |
2893 | updateprogress(idx) |
|
2893 | updateprogress(idx) | |
2894 | addargs, addkwargs = _getrevisionseed(orig, rev, tr, source) |
|
2894 | addargs, addkwargs = _getrevisionseed(orig, rev, tr, source) | |
2895 | if clearcaches: |
|
2895 | if clearcaches: | |
2896 | dest.index.clearcaches() |
|
2896 | dest.index.clearcaches() | |
2897 | dest.clearcaches() |
|
2897 | dest.clearcaches() | |
2898 | with timeone() as r: |
|
2898 | with timeone() as r: | |
2899 | dest.addrawrevision(*addargs, **addkwargs) |
|
2899 | dest.addrawrevision(*addargs, **addkwargs) | |
2900 | timings.append((rev, r[0])) |
|
2900 | timings.append((rev, r[0])) | |
2901 | updateprogress(total) |
|
2901 | updateprogress(total) | |
2902 | completeprogress() |
|
2902 | completeprogress() | |
2903 | return timings |
|
2903 | return timings | |
2904 |
|
2904 | |||
2905 |
|
2905 | |||
2906 | def _getrevisionseed(orig, rev, tr, source): |
|
2906 | def _getrevisionseed(orig, rev, tr, source): | |
2907 | from mercurial.node import nullid |
|
2907 | from mercurial.node import nullid | |
2908 |
|
2908 | |||
2909 | linkrev = orig.linkrev(rev) |
|
2909 | linkrev = orig.linkrev(rev) | |
2910 | node = orig.node(rev) |
|
2910 | node = orig.node(rev) | |
2911 | p1, p2 = orig.parents(node) |
|
2911 | p1, p2 = orig.parents(node) | |
2912 | flags = orig.flags(rev) |
|
2912 | flags = orig.flags(rev) | |
2913 | cachedelta = None |
|
2913 | cachedelta = None | |
2914 | text = None |
|
2914 | text = None | |
2915 |
|
2915 | |||
2916 | if source == b'full': |
|
2916 | if source == b'full': | |
2917 | text = orig.revision(rev) |
|
2917 | text = orig.revision(rev) | |
2918 | elif source == b'parent-1': |
|
2918 | elif source == b'parent-1': | |
2919 | baserev = orig.rev(p1) |
|
2919 | baserev = orig.rev(p1) | |
2920 | cachedelta = (baserev, orig.revdiff(p1, rev)) |
|
2920 | cachedelta = (baserev, orig.revdiff(p1, rev)) | |
2921 | elif source == b'parent-2': |
|
2921 | elif source == b'parent-2': | |
2922 | parent = p2 |
|
2922 | parent = p2 | |
2923 | if p2 == nullid: |
|
2923 | if p2 == nullid: | |
2924 | parent = p1 |
|
2924 | parent = p1 | |
2925 | baserev = orig.rev(parent) |
|
2925 | baserev = orig.rev(parent) | |
2926 | cachedelta = (baserev, orig.revdiff(parent, rev)) |
|
2926 | cachedelta = (baserev, orig.revdiff(parent, rev)) | |
2927 | elif source == b'parent-smallest': |
|
2927 | elif source == b'parent-smallest': | |
2928 | p1diff = orig.revdiff(p1, rev) |
|
2928 | p1diff = orig.revdiff(p1, rev) | |
2929 | parent = p1 |
|
2929 | parent = p1 | |
2930 | diff = p1diff |
|
2930 | diff = p1diff | |
2931 | if p2 != nullid: |
|
2931 | if p2 != nullid: | |
2932 | p2diff = orig.revdiff(p2, rev) |
|
2932 | p2diff = orig.revdiff(p2, rev) | |
2933 | if len(p1diff) > len(p2diff): |
|
2933 | if len(p1diff) > len(p2diff): | |
2934 | parent = p2 |
|
2934 | parent = p2 | |
2935 | diff = p2diff |
|
2935 | diff = p2diff | |
2936 | baserev = orig.rev(parent) |
|
2936 | baserev = orig.rev(parent) | |
2937 | cachedelta = (baserev, diff) |
|
2937 | cachedelta = (baserev, diff) | |
2938 | elif source == b'storage': |
|
2938 | elif source == b'storage': | |
2939 | baserev = orig.deltaparent(rev) |
|
2939 | baserev = orig.deltaparent(rev) | |
2940 | cachedelta = (baserev, orig.revdiff(orig.node(baserev), rev)) |
|
2940 | cachedelta = (baserev, orig.revdiff(orig.node(baserev), rev)) | |
2941 |
|
2941 | |||
2942 | return ( |
|
2942 | return ( | |
2943 | (text, tr, linkrev, p1, p2), |
|
2943 | (text, tr, linkrev, p1, p2), | |
2944 | {'node': node, 'flags': flags, 'cachedelta': cachedelta}, |
|
2944 | {'node': node, 'flags': flags, 'cachedelta': cachedelta}, | |
2945 | ) |
|
2945 | ) | |
2946 |
|
2946 | |||
2947 |
|
2947 | |||
2948 | @contextlib.contextmanager |
|
2948 | @contextlib.contextmanager | |
2949 | def _temprevlog(ui, orig, truncaterev): |
|
2949 | def _temprevlog(ui, orig, truncaterev): | |
2950 | from mercurial import vfs as vfsmod |
|
2950 | from mercurial import vfs as vfsmod | |
2951 |
|
2951 | |||
2952 | if orig._inline: |
|
2952 | if orig._inline: | |
2953 | raise error.Abort('not supporting inline revlog (yet)') |
|
2953 | raise error.Abort('not supporting inline revlog (yet)') | |
2954 | revlogkwargs = {} |
|
2954 | revlogkwargs = {} | |
2955 | k = 'upperboundcomp' |
|
2955 | k = 'upperboundcomp' | |
2956 | if util.safehasattr(orig, k): |
|
2956 | if util.safehasattr(orig, k): | |
2957 | revlogkwargs[k] = getattr(orig, k) |
|
2957 | revlogkwargs[k] = getattr(orig, k) | |
2958 |
|
2958 | |||
2959 | origindexpath = orig.opener.join(orig.indexfile) |
|
2959 | origindexpath = orig.opener.join(orig.indexfile) | |
2960 | origdatapath = orig.opener.join(orig.datafile) |
|
2960 | origdatapath = orig.opener.join(orig.datafile) | |
2961 | indexname = 'revlog.i' |
|
2961 | indexname = 'revlog.i' | |
2962 | dataname = 'revlog.d' |
|
2962 | dataname = 'revlog.d' | |
2963 |
|
2963 | |||
2964 | tmpdir = tempfile.mkdtemp(prefix='tmp-hgperf-') |
|
2964 | tmpdir = tempfile.mkdtemp(prefix='tmp-hgperf-') | |
2965 | try: |
|
2965 | try: | |
2966 | # copy the data file in a temporary directory |
|
2966 | # copy the data file in a temporary directory | |
2967 | ui.debug('copying data in %s\n' % tmpdir) |
|
2967 | ui.debug('copying data in %s\n' % tmpdir) | |
2968 | destindexpath = os.path.join(tmpdir, 'revlog.i') |
|
2968 | destindexpath = os.path.join(tmpdir, 'revlog.i') | |
2969 | destdatapath = os.path.join(tmpdir, 'revlog.d') |
|
2969 | destdatapath = os.path.join(tmpdir, 'revlog.d') | |
2970 | shutil.copyfile(origindexpath, destindexpath) |
|
2970 | shutil.copyfile(origindexpath, destindexpath) | |
2971 | shutil.copyfile(origdatapath, destdatapath) |
|
2971 | shutil.copyfile(origdatapath, destdatapath) | |
2972 |
|
2972 | |||
2973 | # remove the data we want to add again |
|
2973 | # remove the data we want to add again | |
2974 | ui.debug('truncating data to be rewritten\n') |
|
2974 | ui.debug('truncating data to be rewritten\n') | |
2975 | with open(destindexpath, 'ab') as index: |
|
2975 | with open(destindexpath, 'ab') as index: | |
2976 | index.seek(0) |
|
2976 | index.seek(0) | |
2977 | index.truncate(truncaterev * orig._io.size) |
|
2977 | index.truncate(truncaterev * orig._io.size) | |
2978 | with open(destdatapath, 'ab') as data: |
|
2978 | with open(destdatapath, 'ab') as data: | |
2979 | data.seek(0) |
|
2979 | data.seek(0) | |
2980 | data.truncate(orig.start(truncaterev)) |
|
2980 | data.truncate(orig.start(truncaterev)) | |
2981 |
|
2981 | |||
2982 | # instantiate a new revlog from the temporary copy |
|
2982 | # instantiate a new revlog from the temporary copy | |
2983 | ui.debug('truncating adding to be rewritten\n') |
|
2983 | ui.debug('truncating adding to be rewritten\n') | |
2984 | vfs = vfsmod.vfs(tmpdir) |
|
2984 | vfs = vfsmod.vfs(tmpdir) | |
2985 | vfs.options = getattr(orig.opener, 'options', None) |
|
2985 | vfs.options = getattr(orig.opener, 'options', None) | |
2986 |
|
2986 | |||
2987 | dest = revlog.revlog( |
|
2987 | dest = revlog.revlog( | |
2988 | vfs, indexfile=indexname, datafile=dataname, **revlogkwargs |
|
2988 | vfs, indexfile=indexname, datafile=dataname, **revlogkwargs | |
2989 | ) |
|
2989 | ) | |
2990 | if dest._inline: |
|
2990 | if dest._inline: | |
2991 | raise error.Abort('not supporting inline revlog (yet)') |
|
2991 | raise error.Abort('not supporting inline revlog (yet)') | |
2992 | # make sure internals are initialized |
|
2992 | # make sure internals are initialized | |
2993 | dest.revision(len(dest) - 1) |
|
2993 | dest.revision(len(dest) - 1) | |
2994 | yield dest |
|
2994 | yield dest | |
2995 | del dest, vfs |
|
2995 | del dest, vfs | |
2996 | finally: |
|
2996 | finally: | |
2997 | shutil.rmtree(tmpdir, True) |
|
2997 | shutil.rmtree(tmpdir, True) | |
2998 |
|
2998 | |||
2999 |
|
2999 | |||
3000 | @command( |
|
3000 | @command( | |
3001 | b'perfrevlogchunks', |
|
3001 | b'perfrevlogchunks', | |
3002 | revlogopts |
|
3002 | revlogopts | |
3003 | + formatteropts |
|
3003 | + formatteropts | |
3004 | + [ |
|
3004 | + [ | |
3005 | (b'e', b'engines', b'', b'compression engines to use'), |
|
3005 | (b'e', b'engines', b'', b'compression engines to use'), | |
3006 | (b's', b'startrev', 0, b'revision to start at'), |
|
3006 | (b's', b'startrev', 0, b'revision to start at'), | |
3007 | ], |
|
3007 | ], | |
3008 | b'-c|-m|FILE', |
|
3008 | b'-c|-m|FILE', | |
3009 | ) |
|
3009 | ) | |
3010 | def perfrevlogchunks(ui, repo, file_=None, engines=None, startrev=0, **opts): |
|
3010 | def perfrevlogchunks(ui, repo, file_=None, engines=None, startrev=0, **opts): | |
3011 | """Benchmark operations on revlog chunks. |
|
3011 | """Benchmark operations on revlog chunks. | |
3012 |
|
3012 | |||
3013 | Logically, each revlog is a collection of fulltext revisions. However, |
|
3013 | Logically, each revlog is a collection of fulltext revisions. However, | |
3014 | stored within each revlog are "chunks" of possibly compressed data. This |
|
3014 | stored within each revlog are "chunks" of possibly compressed data. This | |
3015 | data needs to be read and decompressed or compressed and written. |
|
3015 | data needs to be read and decompressed or compressed and written. | |
3016 |
|
3016 | |||
3017 | This command measures the time it takes to read+decompress and recompress |
|
3017 | This command measures the time it takes to read+decompress and recompress | |
3018 | chunks in a revlog. It effectively isolates I/O and compression performance. |
|
3018 | chunks in a revlog. It effectively isolates I/O and compression performance. | |
3019 | For measurements of higher-level operations like resolving revisions, |
|
3019 | For measurements of higher-level operations like resolving revisions, | |
3020 | see ``perfrevlogrevisions`` and ``perfrevlogrevision``. |
|
3020 | see ``perfrevlogrevisions`` and ``perfrevlogrevision``. | |
3021 | """ |
|
3021 | """ | |
3022 | opts = _byteskwargs(opts) |
|
3022 | opts = _byteskwargs(opts) | |
3023 |
|
3023 | |||
3024 | rl = cmdutil.openrevlog(repo, b'perfrevlogchunks', file_, opts) |
|
3024 | rl = cmdutil.openrevlog(repo, b'perfrevlogchunks', file_, opts) | |
3025 |
|
3025 | |||
3026 | # _chunkraw was renamed to _getsegmentforrevs. |
|
3026 | # _chunkraw was renamed to _getsegmentforrevs. | |
3027 | try: |
|
3027 | try: | |
3028 | segmentforrevs = rl._getsegmentforrevs |
|
3028 | segmentforrevs = rl._getsegmentforrevs | |
3029 | except AttributeError: |
|
3029 | except AttributeError: | |
3030 | segmentforrevs = rl._chunkraw |
|
3030 | segmentforrevs = rl._chunkraw | |
3031 |
|
3031 | |||
3032 | # Verify engines argument. |
|
3032 | # Verify engines argument. | |
3033 | if engines: |
|
3033 | if engines: | |
3034 | engines = set(e.strip() for e in engines.split(b',')) |
|
3034 | engines = set(e.strip() for e in engines.split(b',')) | |
3035 | for engine in engines: |
|
3035 | for engine in engines: | |
3036 | try: |
|
3036 | try: | |
3037 | util.compressionengines[engine] |
|
3037 | util.compressionengines[engine] | |
3038 | except KeyError: |
|
3038 | except KeyError: | |
3039 | raise error.Abort(b'unknown compression engine: %s' % engine) |
|
3039 | raise error.Abort(b'unknown compression engine: %s' % engine) | |
3040 | else: |
|
3040 | else: | |
3041 | engines = [] |
|
3041 | engines = [] | |
3042 | for e in util.compengines: |
|
3042 | for e in util.compengines: | |
3043 | engine = util.compengines[e] |
|
3043 | engine = util.compengines[e] | |
3044 | try: |
|
3044 | try: | |
3045 | if engine.available(): |
|
3045 | if engine.available(): | |
3046 | engine.revlogcompressor().compress(b'dummy') |
|
3046 | engine.revlogcompressor().compress(b'dummy') | |
3047 | engines.append(e) |
|
3047 | engines.append(e) | |
3048 | except NotImplementedError: |
|
3048 | except NotImplementedError: | |
3049 | pass |
|
3049 | pass | |
3050 |
|
3050 | |||
3051 | revs = list(rl.revs(startrev, len(rl) - 1)) |
|
3051 | revs = list(rl.revs(startrev, len(rl) - 1)) | |
3052 |
|
3052 | |||
3053 | def rlfh(rl): |
|
3053 | def rlfh(rl): | |
3054 | if rl._inline: |
|
3054 | if rl._inline: | |
3055 | return getsvfs(repo)(rl.indexfile) |
|
3055 | return getsvfs(repo)(rl.indexfile) | |
3056 | else: |
|
3056 | else: | |
3057 | return getsvfs(repo)(rl.datafile) |
|
3057 | return getsvfs(repo)(rl.datafile) | |
3058 |
|
3058 | |||
3059 | def doread(): |
|
3059 | def doread(): | |
3060 | rl.clearcaches() |
|
3060 | rl.clearcaches() | |
3061 | for rev in revs: |
|
3061 | for rev in revs: | |
3062 | segmentforrevs(rev, rev) |
|
3062 | segmentforrevs(rev, rev) | |
3063 |
|
3063 | |||
3064 | def doreadcachedfh(): |
|
3064 | def doreadcachedfh(): | |
3065 | rl.clearcaches() |
|
3065 | rl.clearcaches() | |
3066 | fh = rlfh(rl) |
|
3066 | fh = rlfh(rl) | |
3067 | for rev in revs: |
|
3067 | for rev in revs: | |
3068 | segmentforrevs(rev, rev, df=fh) |
|
3068 | segmentforrevs(rev, rev, df=fh) | |
3069 |
|
3069 | |||
3070 | def doreadbatch(): |
|
3070 | def doreadbatch(): | |
3071 | rl.clearcaches() |
|
3071 | rl.clearcaches() | |
3072 | segmentforrevs(revs[0], revs[-1]) |
|
3072 | segmentforrevs(revs[0], revs[-1]) | |
3073 |
|
3073 | |||
3074 | def doreadbatchcachedfh(): |
|
3074 | def doreadbatchcachedfh(): | |
3075 | rl.clearcaches() |
|
3075 | rl.clearcaches() | |
3076 | fh = rlfh(rl) |
|
3076 | fh = rlfh(rl) | |
3077 | segmentforrevs(revs[0], revs[-1], df=fh) |
|
3077 | segmentforrevs(revs[0], revs[-1], df=fh) | |
3078 |
|
3078 | |||
3079 | def dochunk(): |
|
3079 | def dochunk(): | |
3080 | rl.clearcaches() |
|
3080 | rl.clearcaches() | |
3081 | fh = rlfh(rl) |
|
3081 | fh = rlfh(rl) | |
3082 | for rev in revs: |
|
3082 | for rev in revs: | |
3083 | rl._chunk(rev, df=fh) |
|
3083 | rl._chunk(rev, df=fh) | |
3084 |
|
3084 | |||
3085 | chunks = [None] |
|
3085 | chunks = [None] | |
3086 |
|
3086 | |||
3087 | def dochunkbatch(): |
|
3087 | def dochunkbatch(): | |
3088 | rl.clearcaches() |
|
3088 | rl.clearcaches() | |
3089 | fh = rlfh(rl) |
|
3089 | fh = rlfh(rl) | |
3090 | # Save chunks as a side-effect. |
|
3090 | # Save chunks as a side-effect. | |
3091 | chunks[0] = rl._chunks(revs, df=fh) |
|
3091 | chunks[0] = rl._chunks(revs, df=fh) | |
3092 |
|
3092 | |||
3093 | def docompress(compressor): |
|
3093 | def docompress(compressor): | |
3094 | rl.clearcaches() |
|
3094 | rl.clearcaches() | |
3095 |
|
3095 | |||
3096 | try: |
|
3096 | try: | |
3097 | # Swap in the requested compression engine. |
|
3097 | # Swap in the requested compression engine. | |
3098 | oldcompressor = rl._compressor |
|
3098 | oldcompressor = rl._compressor | |
3099 | rl._compressor = compressor |
|
3099 | rl._compressor = compressor | |
3100 | for chunk in chunks[0]: |
|
3100 | for chunk in chunks[0]: | |
3101 | rl.compress(chunk) |
|
3101 | rl.compress(chunk) | |
3102 | finally: |
|
3102 | finally: | |
3103 | rl._compressor = oldcompressor |
|
3103 | rl._compressor = oldcompressor | |
3104 |
|
3104 | |||
3105 | benches = [ |
|
3105 | benches = [ | |
3106 | (lambda: doread(), b'read'), |
|
3106 | (lambda: doread(), b'read'), | |
3107 | (lambda: doreadcachedfh(), b'read w/ reused fd'), |
|
3107 | (lambda: doreadcachedfh(), b'read w/ reused fd'), | |
3108 | (lambda: doreadbatch(), b'read batch'), |
|
3108 | (lambda: doreadbatch(), b'read batch'), | |
3109 | (lambda: doreadbatchcachedfh(), b'read batch w/ reused fd'), |
|
3109 | (lambda: doreadbatchcachedfh(), b'read batch w/ reused fd'), | |
3110 | (lambda: dochunk(), b'chunk'), |
|
3110 | (lambda: dochunk(), b'chunk'), | |
3111 | (lambda: dochunkbatch(), b'chunk batch'), |
|
3111 | (lambda: dochunkbatch(), b'chunk batch'), | |
3112 | ] |
|
3112 | ] | |
3113 |
|
3113 | |||
3114 | for engine in sorted(engines): |
|
3114 | for engine in sorted(engines): | |
3115 | compressor = util.compengines[engine].revlogcompressor() |
|
3115 | compressor = util.compengines[engine].revlogcompressor() | |
3116 | benches.append( |
|
3116 | benches.append( | |
3117 | ( |
|
3117 | ( | |
3118 | functools.partial(docompress, compressor), |
|
3118 | functools.partial(docompress, compressor), | |
3119 | b'compress w/ %s' % engine, |
|
3119 | b'compress w/ %s' % engine, | |
3120 | ) |
|
3120 | ) | |
3121 | ) |
|
3121 | ) | |
3122 |
|
3122 | |||
3123 | for fn, title in benches: |
|
3123 | for fn, title in benches: | |
3124 | timer, fm = gettimer(ui, opts) |
|
3124 | timer, fm = gettimer(ui, opts) | |
3125 | timer(fn, title=title) |
|
3125 | timer(fn, title=title) | |
3126 | fm.end() |
|
3126 | fm.end() | |
3127 |
|
3127 | |||
3128 |
|
3128 | |||
3129 | @command( |
|
3129 | @command( | |
3130 | b'perfrevlogrevision', |
|
3130 | b'perfrevlogrevision', | |
3131 | revlogopts |
|
3131 | revlogopts | |
3132 | + formatteropts |
|
3132 | + formatteropts | |
3133 | + [(b'', b'cache', False, b'use caches instead of clearing')], |
|
3133 | + [(b'', b'cache', False, b'use caches instead of clearing')], | |
3134 | b'-c|-m|FILE REV', |
|
3134 | b'-c|-m|FILE REV', | |
3135 | ) |
|
3135 | ) | |
3136 | def perfrevlogrevision(ui, repo, file_, rev=None, cache=None, **opts): |
|
3136 | def perfrevlogrevision(ui, repo, file_, rev=None, cache=None, **opts): | |
3137 | """Benchmark obtaining a revlog revision. |
|
3137 | """Benchmark obtaining a revlog revision. | |
3138 |
|
3138 | |||
3139 | Obtaining a revlog revision consists of roughly the following steps: |
|
3139 | Obtaining a revlog revision consists of roughly the following steps: | |
3140 |
|
3140 | |||
3141 | 1. Compute the delta chain |
|
3141 | 1. Compute the delta chain | |
3142 | 2. Slice the delta chain if applicable |
|
3142 | 2. Slice the delta chain if applicable | |
3143 | 3. Obtain the raw chunks for that delta chain |
|
3143 | 3. Obtain the raw chunks for that delta chain | |
3144 | 4. Decompress each raw chunk |
|
3144 | 4. Decompress each raw chunk | |
3145 | 5. Apply binary patches to obtain fulltext |
|
3145 | 5. Apply binary patches to obtain fulltext | |
3146 | 6. Verify hash of fulltext |
|
3146 | 6. Verify hash of fulltext | |
3147 |
|
3147 | |||
3148 | This command measures the time spent in each of these phases. |
|
3148 | This command measures the time spent in each of these phases. | |
3149 | """ |
|
3149 | """ | |
3150 | opts = _byteskwargs(opts) |
|
3150 | opts = _byteskwargs(opts) | |
3151 |
|
3151 | |||
3152 | if opts.get(b'changelog') or opts.get(b'manifest'): |
|
3152 | if opts.get(b'changelog') or opts.get(b'manifest'): | |
3153 | file_, rev = None, file_ |
|
3153 | file_, rev = None, file_ | |
3154 | elif rev is None: |
|
3154 | elif rev is None: | |
3155 | raise error.CommandError(b'perfrevlogrevision', b'invalid arguments') |
|
3155 | raise error.CommandError(b'perfrevlogrevision', b'invalid arguments') | |
3156 |
|
3156 | |||
3157 | r = cmdutil.openrevlog(repo, b'perfrevlogrevision', file_, opts) |
|
3157 | r = cmdutil.openrevlog(repo, b'perfrevlogrevision', file_, opts) | |
3158 |
|
3158 | |||
3159 | # _chunkraw was renamed to _getsegmentforrevs. |
|
3159 | # _chunkraw was renamed to _getsegmentforrevs. | |
3160 | try: |
|
3160 | try: | |
3161 | segmentforrevs = r._getsegmentforrevs |
|
3161 | segmentforrevs = r._getsegmentforrevs | |
3162 | except AttributeError: |
|
3162 | except AttributeError: | |
3163 | segmentforrevs = r._chunkraw |
|
3163 | segmentforrevs = r._chunkraw | |
3164 |
|
3164 | |||
3165 | node = r.lookup(rev) |
|
3165 | node = r.lookup(rev) | |
3166 | rev = r.rev(node) |
|
3166 | rev = r.rev(node) | |
3167 |
|
3167 | |||
3168 | def getrawchunks(data, chain): |
|
3168 | def getrawchunks(data, chain): | |
3169 | start = r.start |
|
3169 | start = r.start | |
3170 | length = r.length |
|
3170 | length = r.length | |
3171 | inline = r._inline |
|
3171 | inline = r._inline | |
3172 | iosize = r._io.size |
|
3172 | iosize = r._io.size | |
3173 | buffer = util.buffer |
|
3173 | buffer = util.buffer | |
3174 |
|
3174 | |||
3175 | chunks = [] |
|
3175 | chunks = [] | |
3176 | ladd = chunks.append |
|
3176 | ladd = chunks.append | |
3177 | for idx, item in enumerate(chain): |
|
3177 | for idx, item in enumerate(chain): | |
3178 | offset = start(item[0]) |
|
3178 | offset = start(item[0]) | |
3179 | bits = data[idx] |
|
3179 | bits = data[idx] | |
3180 | for rev in item: |
|
3180 | for rev in item: | |
3181 | chunkstart = start(rev) |
|
3181 | chunkstart = start(rev) | |
3182 | if inline: |
|
3182 | if inline: | |
3183 | chunkstart += (rev + 1) * iosize |
|
3183 | chunkstart += (rev + 1) * iosize | |
3184 | chunklength = length(rev) |
|
3184 | chunklength = length(rev) | |
3185 | ladd(buffer(bits, chunkstart - offset, chunklength)) |
|
3185 | ladd(buffer(bits, chunkstart - offset, chunklength)) | |
3186 |
|
3186 | |||
3187 | return chunks |
|
3187 | return chunks | |
3188 |
|
3188 | |||
3189 | def dodeltachain(rev): |
|
3189 | def dodeltachain(rev): | |
3190 | if not cache: |
|
3190 | if not cache: | |
3191 | r.clearcaches() |
|
3191 | r.clearcaches() | |
3192 | r._deltachain(rev) |
|
3192 | r._deltachain(rev) | |
3193 |
|
3193 | |||
3194 | def doread(chain): |
|
3194 | def doread(chain): | |
3195 | if not cache: |
|
3195 | if not cache: | |
3196 | r.clearcaches() |
|
3196 | r.clearcaches() | |
3197 | for item in slicedchain: |
|
3197 | for item in slicedchain: | |
3198 | segmentforrevs(item[0], item[-1]) |
|
3198 | segmentforrevs(item[0], item[-1]) | |
3199 |
|
3199 | |||
3200 | def doslice(r, chain, size): |
|
3200 | def doslice(r, chain, size): | |
3201 | for s in slicechunk(r, chain, targetsize=size): |
|
3201 | for s in slicechunk(r, chain, targetsize=size): | |
3202 | pass |
|
3202 | pass | |
3203 |
|
3203 | |||
3204 | def dorawchunks(data, chain): |
|
3204 | def dorawchunks(data, chain): | |
3205 | if not cache: |
|
3205 | if not cache: | |
3206 | r.clearcaches() |
|
3206 | r.clearcaches() | |
3207 | getrawchunks(data, chain) |
|
3207 | getrawchunks(data, chain) | |
3208 |
|
3208 | |||
3209 | def dodecompress(chunks): |
|
3209 | def dodecompress(chunks): | |
3210 | decomp = r.decompress |
|
3210 | decomp = r.decompress | |
3211 | for chunk in chunks: |
|
3211 | for chunk in chunks: | |
3212 | decomp(chunk) |
|
3212 | decomp(chunk) | |
3213 |
|
3213 | |||
3214 | def dopatch(text, bins): |
|
3214 | def dopatch(text, bins): | |
3215 | if not cache: |
|
3215 | if not cache: | |
3216 | r.clearcaches() |
|
3216 | r.clearcaches() | |
3217 | mdiff.patches(text, bins) |
|
3217 | mdiff.patches(text, bins) | |
3218 |
|
3218 | |||
3219 | def dohash(text): |
|
3219 | def dohash(text): | |
3220 | if not cache: |
|
3220 | if not cache: | |
3221 | r.clearcaches() |
|
3221 | r.clearcaches() | |
3222 | r.checkhash(text, node, rev=rev) |
|
3222 | r.checkhash(text, node, rev=rev) | |
3223 |
|
3223 | |||
3224 | def dorevision(): |
|
3224 | def dorevision(): | |
3225 | if not cache: |
|
3225 | if not cache: | |
3226 | r.clearcaches() |
|
3226 | r.clearcaches() | |
3227 | r.revision(node) |
|
3227 | r.revision(node) | |
3228 |
|
3228 | |||
3229 | try: |
|
3229 | try: | |
3230 | from mercurial.revlogutils.deltas import slicechunk |
|
3230 | from mercurial.revlogutils.deltas import slicechunk | |
3231 | except ImportError: |
|
3231 | except ImportError: | |
3232 | slicechunk = getattr(revlog, '_slicechunk', None) |
|
3232 | slicechunk = getattr(revlog, '_slicechunk', None) | |
3233 |
|
3233 | |||
3234 | size = r.length(rev) |
|
3234 | size = r.length(rev) | |
3235 | chain = r._deltachain(rev)[0] |
|
3235 | chain = r._deltachain(rev)[0] | |
3236 | if not getattr(r, '_withsparseread', False): |
|
3236 | if not getattr(r, '_withsparseread', False): | |
3237 | slicedchain = (chain,) |
|
3237 | slicedchain = (chain,) | |
3238 | else: |
|
3238 | else: | |
3239 | slicedchain = tuple(slicechunk(r, chain, targetsize=size)) |
|
3239 | slicedchain = tuple(slicechunk(r, chain, targetsize=size)) | |
3240 | data = [segmentforrevs(seg[0], seg[-1])[1] for seg in slicedchain] |
|
3240 | data = [segmentforrevs(seg[0], seg[-1])[1] for seg in slicedchain] | |
3241 | rawchunks = getrawchunks(data, slicedchain) |
|
3241 | rawchunks = getrawchunks(data, slicedchain) | |
3242 | bins = r._chunks(chain) |
|
3242 | bins = r._chunks(chain) | |
3243 | text = bytes(bins[0]) |
|
3243 | text = bytes(bins[0]) | |
3244 | bins = bins[1:] |
|
3244 | bins = bins[1:] | |
3245 | text = mdiff.patches(text, bins) |
|
3245 | text = mdiff.patches(text, bins) | |
3246 |
|
3246 | |||
3247 | benches = [ |
|
3247 | benches = [ | |
3248 | (lambda: dorevision(), b'full'), |
|
3248 | (lambda: dorevision(), b'full'), | |
3249 | (lambda: dodeltachain(rev), b'deltachain'), |
|
3249 | (lambda: dodeltachain(rev), b'deltachain'), | |
3250 | (lambda: doread(chain), b'read'), |
|
3250 | (lambda: doread(chain), b'read'), | |
3251 | ] |
|
3251 | ] | |
3252 |
|
3252 | |||
3253 | if getattr(r, '_withsparseread', False): |
|
3253 | if getattr(r, '_withsparseread', False): | |
3254 | slicing = (lambda: doslice(r, chain, size), b'slice-sparse-chain') |
|
3254 | slicing = (lambda: doslice(r, chain, size), b'slice-sparse-chain') | |
3255 | benches.append(slicing) |
|
3255 | benches.append(slicing) | |
3256 |
|
3256 | |||
3257 | benches.extend( |
|
3257 | benches.extend( | |
3258 | [ |
|
3258 | [ | |
3259 | (lambda: dorawchunks(data, slicedchain), b'rawchunks'), |
|
3259 | (lambda: dorawchunks(data, slicedchain), b'rawchunks'), | |
3260 | (lambda: dodecompress(rawchunks), b'decompress'), |
|
3260 | (lambda: dodecompress(rawchunks), b'decompress'), | |
3261 | (lambda: dopatch(text, bins), b'patch'), |
|
3261 | (lambda: dopatch(text, bins), b'patch'), | |
3262 | (lambda: dohash(text), b'hash'), |
|
3262 | (lambda: dohash(text), b'hash'), | |
3263 | ] |
|
3263 | ] | |
3264 | ) |
|
3264 | ) | |
3265 |
|
3265 | |||
3266 | timer, fm = gettimer(ui, opts) |
|
3266 | timer, fm = gettimer(ui, opts) | |
3267 | for fn, title in benches: |
|
3267 | for fn, title in benches: | |
3268 | timer(fn, title=title) |
|
3268 | timer(fn, title=title) | |
3269 | fm.end() |
|
3269 | fm.end() | |
3270 |
|
3270 | |||
3271 |
|
3271 | |||
3272 | @command( |
|
3272 | @command( | |
3273 | b'perfrevset', |
|
3273 | b'perfrevset', | |
3274 | [ |
|
3274 | [ | |
3275 | (b'C', b'clear', False, b'clear volatile cache between each call.'), |
|
3275 | (b'C', b'clear', False, b'clear volatile cache between each call.'), | |
3276 | (b'', b'contexts', False, b'obtain changectx for each revision'), |
|
3276 | (b'', b'contexts', False, b'obtain changectx for each revision'), | |
3277 | ] |
|
3277 | ] | |
3278 | + formatteropts, |
|
3278 | + formatteropts, | |
3279 | b"REVSET", |
|
3279 | b"REVSET", | |
3280 | ) |
|
3280 | ) | |
3281 | def perfrevset(ui, repo, expr, clear=False, contexts=False, **opts): |
|
3281 | def perfrevset(ui, repo, expr, clear=False, contexts=False, **opts): | |
3282 | """benchmark the execution time of a revset |
|
3282 | """benchmark the execution time of a revset | |
3283 |
|
3283 | |||
3284 | Use the --clean option if need to evaluate the impact of build volatile |
|
3284 | Use the --clean option if need to evaluate the impact of build volatile | |
3285 | revisions set cache on the revset execution. Volatile cache hold filtered |
|
3285 | revisions set cache on the revset execution. Volatile cache hold filtered | |
3286 | and obsolete related cache.""" |
|
3286 | and obsolete related cache.""" | |
3287 | opts = _byteskwargs(opts) |
|
3287 | opts = _byteskwargs(opts) | |
3288 |
|
3288 | |||
3289 | timer, fm = gettimer(ui, opts) |
|
3289 | timer, fm = gettimer(ui, opts) | |
3290 |
|
3290 | |||
3291 | def d(): |
|
3291 | def d(): | |
3292 | if clear: |
|
3292 | if clear: | |
3293 | repo.invalidatevolatilesets() |
|
3293 | repo.invalidatevolatilesets() | |
3294 | if contexts: |
|
3294 | if contexts: | |
3295 | for ctx in repo.set(expr): |
|
3295 | for ctx in repo.set(expr): | |
3296 | pass |
|
3296 | pass | |
3297 | else: |
|
3297 | else: | |
3298 | for r in repo.revs(expr): |
|
3298 | for r in repo.revs(expr): | |
3299 | pass |
|
3299 | pass | |
3300 |
|
3300 | |||
3301 | timer(d) |
|
3301 | timer(d) | |
3302 | fm.end() |
|
3302 | fm.end() | |
3303 |
|
3303 | |||
3304 |
|
3304 | |||
3305 | @command( |
|
3305 | @command( | |
3306 | b'perfvolatilesets', |
|
3306 | b'perfvolatilesets', | |
3307 | [(b'', b'clear-obsstore', False, b'drop obsstore between each call.'),] |
|
3307 | [(b'', b'clear-obsstore', False, b'drop obsstore between each call.'),] | |
3308 | + formatteropts, |
|
3308 | + formatteropts, | |
3309 | ) |
|
3309 | ) | |
3310 | def perfvolatilesets(ui, repo, *names, **opts): |
|
3310 | def perfvolatilesets(ui, repo, *names, **opts): | |
3311 | """benchmark the computation of various volatile set |
|
3311 | """benchmark the computation of various volatile set | |
3312 |
|
3312 | |||
3313 | Volatile set computes element related to filtering and obsolescence.""" |
|
3313 | Volatile set computes element related to filtering and obsolescence.""" | |
3314 | opts = _byteskwargs(opts) |
|
3314 | opts = _byteskwargs(opts) | |
3315 | timer, fm = gettimer(ui, opts) |
|
3315 | timer, fm = gettimer(ui, opts) | |
3316 | repo = repo.unfiltered() |
|
3316 | repo = repo.unfiltered() | |
3317 |
|
3317 | |||
3318 | def getobs(name): |
|
3318 | def getobs(name): | |
3319 | def d(): |
|
3319 | def d(): | |
3320 | repo.invalidatevolatilesets() |
|
3320 | repo.invalidatevolatilesets() | |
3321 | if opts[b'clear_obsstore']: |
|
3321 | if opts[b'clear_obsstore']: | |
3322 | clearfilecache(repo, b'obsstore') |
|
3322 | clearfilecache(repo, b'obsstore') | |
3323 | obsolete.getrevs(repo, name) |
|
3323 | obsolete.getrevs(repo, name) | |
3324 |
|
3324 | |||
3325 | return d |
|
3325 | return d | |
3326 |
|
3326 | |||
3327 | allobs = sorted(obsolete.cachefuncs) |
|
3327 | allobs = sorted(obsolete.cachefuncs) | |
3328 | if names: |
|
3328 | if names: | |
3329 | allobs = [n for n in allobs if n in names] |
|
3329 | allobs = [n for n in allobs if n in names] | |
3330 |
|
3330 | |||
3331 | for name in allobs: |
|
3331 | for name in allobs: | |
3332 | timer(getobs(name), title=name) |
|
3332 | timer(getobs(name), title=name) | |
3333 |
|
3333 | |||
3334 | def getfiltered(name): |
|
3334 | def getfiltered(name): | |
3335 | def d(): |
|
3335 | def d(): | |
3336 | repo.invalidatevolatilesets() |
|
3336 | repo.invalidatevolatilesets() | |
3337 | if opts[b'clear_obsstore']: |
|
3337 | if opts[b'clear_obsstore']: | |
3338 | clearfilecache(repo, b'obsstore') |
|
3338 | clearfilecache(repo, b'obsstore') | |
3339 | repoview.filterrevs(repo, name) |
|
3339 | repoview.filterrevs(repo, name) | |
3340 |
|
3340 | |||
3341 | return d |
|
3341 | return d | |
3342 |
|
3342 | |||
3343 | allfilter = sorted(repoview.filtertable) |
|
3343 | allfilter = sorted(repoview.filtertable) | |
3344 | if names: |
|
3344 | if names: | |
3345 | allfilter = [n for n in allfilter if n in names] |
|
3345 | allfilter = [n for n in allfilter if n in names] | |
3346 |
|
3346 | |||
3347 | for name in allfilter: |
|
3347 | for name in allfilter: | |
3348 | timer(getfiltered(name), title=name) |
|
3348 | timer(getfiltered(name), title=name) | |
3349 | fm.end() |
|
3349 | fm.end() | |
3350 |
|
3350 | |||
3351 |
|
3351 | |||
3352 | @command( |
|
3352 | @command( | |
3353 | b'perfbranchmap', |
|
3353 | b'perfbranchmap', | |
3354 | [ |
|
3354 | [ | |
3355 | (b'f', b'full', False, b'Includes build time of subset'), |
|
3355 | (b'f', b'full', False, b'Includes build time of subset'), | |
3356 | ( |
|
3356 | ( | |
3357 | b'', |
|
3357 | b'', | |
3358 | b'clear-revbranch', |
|
3358 | b'clear-revbranch', | |
3359 | False, |
|
3359 | False, | |
3360 | b'purge the revbranch cache between computation', |
|
3360 | b'purge the revbranch cache between computation', | |
3361 | ), |
|
3361 | ), | |
3362 | ] |
|
3362 | ] | |
3363 | + formatteropts, |
|
3363 | + formatteropts, | |
3364 | ) |
|
3364 | ) | |
3365 | def perfbranchmap(ui, repo, *filternames, **opts): |
|
3365 | def perfbranchmap(ui, repo, *filternames, **opts): | |
3366 | """benchmark the update of a branchmap |
|
3366 | """benchmark the update of a branchmap | |
3367 |
|
3367 | |||
3368 | This benchmarks the full repo.branchmap() call with read and write disabled |
|
3368 | This benchmarks the full repo.branchmap() call with read and write disabled | |
3369 | """ |
|
3369 | """ | |
3370 | opts = _byteskwargs(opts) |
|
3370 | opts = _byteskwargs(opts) | |
3371 | full = opts.get(b"full", False) |
|
3371 | full = opts.get(b"full", False) | |
3372 | clear_revbranch = opts.get(b"clear_revbranch", False) |
|
3372 | clear_revbranch = opts.get(b"clear_revbranch", False) | |
3373 | timer, fm = gettimer(ui, opts) |
|
3373 | timer, fm = gettimer(ui, opts) | |
3374 |
|
3374 | |||
3375 | def getbranchmap(filtername): |
|
3375 | def getbranchmap(filtername): | |
3376 | """generate a benchmark function for the filtername""" |
|
3376 | """generate a benchmark function for the filtername""" | |
3377 | if filtername is None: |
|
3377 | if filtername is None: | |
3378 | view = repo |
|
3378 | view = repo | |
3379 | else: |
|
3379 | else: | |
3380 | view = repo.filtered(filtername) |
|
3380 | view = repo.filtered(filtername) | |
3381 | if util.safehasattr(view._branchcaches, '_per_filter'): |
|
3381 | if util.safehasattr(view._branchcaches, '_per_filter'): | |
3382 | filtered = view._branchcaches._per_filter |
|
3382 | filtered = view._branchcaches._per_filter | |
3383 | else: |
|
3383 | else: | |
3384 | # older versions |
|
3384 | # older versions | |
3385 | filtered = view._branchcaches |
|
3385 | filtered = view._branchcaches | |
3386 |
|
3386 | |||
3387 | def d(): |
|
3387 | def d(): | |
3388 | if clear_revbranch: |
|
3388 | if clear_revbranch: | |
3389 | repo.revbranchcache()._clear() |
|
3389 | repo.revbranchcache()._clear() | |
3390 | if full: |
|
3390 | if full: | |
3391 | view._branchcaches.clear() |
|
3391 | view._branchcaches.clear() | |
3392 | else: |
|
3392 | else: | |
3393 | filtered.pop(filtername, None) |
|
3393 | filtered.pop(filtername, None) | |
3394 | view.branchmap() |
|
3394 | view.branchmap() | |
3395 |
|
3395 | |||
3396 | return d |
|
3396 | return d | |
3397 |
|
3397 | |||
3398 | # add filter in smaller subset to bigger subset |
|
3398 | # add filter in smaller subset to bigger subset | |
3399 | possiblefilters = set(repoview.filtertable) |
|
3399 | possiblefilters = set(repoview.filtertable) | |
3400 | if filternames: |
|
3400 | if filternames: | |
3401 | possiblefilters &= set(filternames) |
|
3401 | possiblefilters &= set(filternames) | |
3402 | subsettable = getbranchmapsubsettable() |
|
3402 | subsettable = getbranchmapsubsettable() | |
3403 | allfilters = [] |
|
3403 | allfilters = [] | |
3404 | while possiblefilters: |
|
3404 | while possiblefilters: | |
3405 | for name in possiblefilters: |
|
3405 | for name in possiblefilters: | |
3406 | subset = subsettable.get(name) |
|
3406 | subset = subsettable.get(name) | |
3407 | if subset not in possiblefilters: |
|
3407 | if subset not in possiblefilters: | |
3408 | break |
|
3408 | break | |
3409 | else: |
|
3409 | else: | |
3410 | assert False, b'subset cycle %s!' % possiblefilters |
|
3410 | assert False, b'subset cycle %s!' % possiblefilters | |
3411 | allfilters.append(name) |
|
3411 | allfilters.append(name) | |
3412 | possiblefilters.remove(name) |
|
3412 | possiblefilters.remove(name) | |
3413 |
|
3413 | |||
3414 | # warm the cache |
|
3414 | # warm the cache | |
3415 | if not full: |
|
3415 | if not full: | |
3416 | for name in allfilters: |
|
3416 | for name in allfilters: | |
3417 | repo.filtered(name).branchmap() |
|
3417 | repo.filtered(name).branchmap() | |
3418 | if not filternames or b'unfiltered' in filternames: |
|
3418 | if not filternames or b'unfiltered' in filternames: | |
3419 | # add unfiltered |
|
3419 | # add unfiltered | |
3420 | allfilters.append(None) |
|
3420 | allfilters.append(None) | |
3421 |
|
3421 | |||
3422 | if util.safehasattr(branchmap.branchcache, 'fromfile'): |
|
3422 | if util.safehasattr(branchmap.branchcache, 'fromfile'): | |
3423 | branchcacheread = safeattrsetter(branchmap.branchcache, b'fromfile') |
|
3423 | branchcacheread = safeattrsetter(branchmap.branchcache, b'fromfile') | |
3424 | branchcacheread.set(classmethod(lambda *args: None)) |
|
3424 | branchcacheread.set(classmethod(lambda *args: None)) | |
3425 | else: |
|
3425 | else: | |
3426 | # older versions |
|
3426 | # older versions | |
3427 | branchcacheread = safeattrsetter(branchmap, b'read') |
|
3427 | branchcacheread = safeattrsetter(branchmap, b'read') | |
3428 | branchcacheread.set(lambda *args: None) |
|
3428 | branchcacheread.set(lambda *args: None) | |
3429 | branchcachewrite = safeattrsetter(branchmap.branchcache, b'write') |
|
3429 | branchcachewrite = safeattrsetter(branchmap.branchcache, b'write') | |
3430 | branchcachewrite.set(lambda *args: None) |
|
3430 | branchcachewrite.set(lambda *args: None) | |
3431 | try: |
|
3431 | try: | |
3432 | for name in allfilters: |
|
3432 | for name in allfilters: | |
3433 | printname = name |
|
3433 | printname = name | |
3434 | if name is None: |
|
3434 | if name is None: | |
3435 | printname = b'unfiltered' |
|
3435 | printname = b'unfiltered' | |
3436 | timer(getbranchmap(name), title=str(printname)) |
|
3436 | timer(getbranchmap(name), title=str(printname)) | |
3437 | finally: |
|
3437 | finally: | |
3438 | branchcacheread.restore() |
|
3438 | branchcacheread.restore() | |
3439 | branchcachewrite.restore() |
|
3439 | branchcachewrite.restore() | |
3440 | fm.end() |
|
3440 | fm.end() | |
3441 |
|
3441 | |||
3442 |
|
3442 | |||
3443 | @command( |
|
3443 | @command( | |
3444 | b'perfbranchmapupdate', |
|
3444 | b'perfbranchmapupdate', | |
3445 | [ |
|
3445 | [ | |
3446 | (b'', b'base', [], b'subset of revision to start from'), |
|
3446 | (b'', b'base', [], b'subset of revision to start from'), | |
3447 | (b'', b'target', [], b'subset of revision to end with'), |
|
3447 | (b'', b'target', [], b'subset of revision to end with'), | |
3448 | (b'', b'clear-caches', False, b'clear cache between each runs'), |
|
3448 | (b'', b'clear-caches', False, b'clear cache between each runs'), | |
3449 | ] |
|
3449 | ] | |
3450 | + formatteropts, |
|
3450 | + formatteropts, | |
3451 | ) |
|
3451 | ) | |
3452 | def perfbranchmapupdate(ui, repo, base=(), target=(), **opts): |
|
3452 | def perfbranchmapupdate(ui, repo, base=(), target=(), **opts): | |
3453 | """benchmark branchmap update from for <base> revs to <target> revs |
|
3453 | """benchmark branchmap update from for <base> revs to <target> revs | |
3454 |
|
3454 | |||
3455 | If `--clear-caches` is passed, the following items will be reset before |
|
3455 | If `--clear-caches` is passed, the following items will be reset before | |
3456 | each update: |
|
3456 | each update: | |
3457 | * the changelog instance and associated indexes |
|
3457 | * the changelog instance and associated indexes | |
3458 | * the rev-branch-cache instance |
|
3458 | * the rev-branch-cache instance | |
3459 |
|
3459 | |||
3460 | Examples: |
|
3460 | Examples: | |
3461 |
|
3461 | |||
3462 | # update for the one last revision |
|
3462 | # update for the one last revision | |
3463 | $ hg perfbranchmapupdate --base 'not tip' --target 'tip' |
|
3463 | $ hg perfbranchmapupdate --base 'not tip' --target 'tip' | |
3464 |
|
3464 | |||
3465 | $ update for change coming with a new branch |
|
3465 | $ update for change coming with a new branch | |
3466 | $ hg perfbranchmapupdate --base 'stable' --target 'default' |
|
3466 | $ hg perfbranchmapupdate --base 'stable' --target 'default' | |
3467 | """ |
|
3467 | """ | |
3468 | from mercurial import branchmap |
|
3468 | from mercurial import branchmap | |
3469 | from mercurial import repoview |
|
3469 | from mercurial import repoview | |
3470 |
|
3470 | |||
3471 | opts = _byteskwargs(opts) |
|
3471 | opts = _byteskwargs(opts) | |
3472 | timer, fm = gettimer(ui, opts) |
|
3472 | timer, fm = gettimer(ui, opts) | |
3473 | clearcaches = opts[b'clear_caches'] |
|
3473 | clearcaches = opts[b'clear_caches'] | |
3474 | unfi = repo.unfiltered() |
|
3474 | unfi = repo.unfiltered() | |
3475 | x = [None] # used to pass data between closure |
|
3475 | x = [None] # used to pass data between closure | |
3476 |
|
3476 | |||
3477 | # we use a `list` here to avoid possible side effect from smartset |
|
3477 | # we use a `list` here to avoid possible side effect from smartset | |
3478 | baserevs = list(scmutil.revrange(repo, base)) |
|
3478 | baserevs = list(scmutil.revrange(repo, base)) | |
3479 | targetrevs = list(scmutil.revrange(repo, target)) |
|
3479 | targetrevs = list(scmutil.revrange(repo, target)) | |
3480 | if not baserevs: |
|
3480 | if not baserevs: | |
3481 | raise error.Abort(b'no revisions selected for --base') |
|
3481 | raise error.Abort(b'no revisions selected for --base') | |
3482 | if not targetrevs: |
|
3482 | if not targetrevs: | |
3483 | raise error.Abort(b'no revisions selected for --target') |
|
3483 | raise error.Abort(b'no revisions selected for --target') | |
3484 |
|
3484 | |||
3485 | # make sure the target branchmap also contains the one in the base |
|
3485 | # make sure the target branchmap also contains the one in the base | |
3486 | targetrevs = list(set(baserevs) | set(targetrevs)) |
|
3486 | targetrevs = list(set(baserevs) | set(targetrevs)) | |
3487 | targetrevs.sort() |
|
3487 | targetrevs.sort() | |
3488 |
|
3488 | |||
3489 | cl = repo.changelog |
|
3489 | cl = repo.changelog | |
3490 | allbaserevs = list(cl.ancestors(baserevs, inclusive=True)) |
|
3490 | allbaserevs = list(cl.ancestors(baserevs, inclusive=True)) | |
3491 | allbaserevs.sort() |
|
3491 | allbaserevs.sort() | |
3492 | alltargetrevs = frozenset(cl.ancestors(targetrevs, inclusive=True)) |
|
3492 | alltargetrevs = frozenset(cl.ancestors(targetrevs, inclusive=True)) | |
3493 |
|
3493 | |||
3494 | newrevs = list(alltargetrevs.difference(allbaserevs)) |
|
3494 | newrevs = list(alltargetrevs.difference(allbaserevs)) | |
3495 | newrevs.sort() |
|
3495 | newrevs.sort() | |
3496 |
|
3496 | |||
3497 | allrevs = frozenset(unfi.changelog.revs()) |
|
3497 | allrevs = frozenset(unfi.changelog.revs()) | |
3498 | basefilterrevs = frozenset(allrevs.difference(allbaserevs)) |
|
3498 | basefilterrevs = frozenset(allrevs.difference(allbaserevs)) | |
3499 | targetfilterrevs = frozenset(allrevs.difference(alltargetrevs)) |
|
3499 | targetfilterrevs = frozenset(allrevs.difference(alltargetrevs)) | |
3500 |
|
3500 | |||
3501 | def basefilter(repo, visibilityexceptions=None): |
|
3501 | def basefilter(repo, visibilityexceptions=None): | |
3502 | return basefilterrevs |
|
3502 | return basefilterrevs | |
3503 |
|
3503 | |||
3504 | def targetfilter(repo, visibilityexceptions=None): |
|
3504 | def targetfilter(repo, visibilityexceptions=None): | |
3505 | return targetfilterrevs |
|
3505 | return targetfilterrevs | |
3506 |
|
3506 | |||
3507 | msg = b'benchmark of branchmap with %d revisions with %d new ones\n' |
|
3507 | msg = b'benchmark of branchmap with %d revisions with %d new ones\n' | |
3508 | ui.status(msg % (len(allbaserevs), len(newrevs))) |
|
3508 | ui.status(msg % (len(allbaserevs), len(newrevs))) | |
3509 | if targetfilterrevs: |
|
3509 | if targetfilterrevs: | |
3510 | msg = b'(%d revisions still filtered)\n' |
|
3510 | msg = b'(%d revisions still filtered)\n' | |
3511 | ui.status(msg % len(targetfilterrevs)) |
|
3511 | ui.status(msg % len(targetfilterrevs)) | |
3512 |
|
3512 | |||
3513 | try: |
|
3513 | try: | |
3514 | repoview.filtertable[b'__perf_branchmap_update_base'] = basefilter |
|
3514 | repoview.filtertable[b'__perf_branchmap_update_base'] = basefilter | |
3515 | repoview.filtertable[b'__perf_branchmap_update_target'] = targetfilter |
|
3515 | repoview.filtertable[b'__perf_branchmap_update_target'] = targetfilter | |
3516 |
|
3516 | |||
3517 | baserepo = repo.filtered(b'__perf_branchmap_update_base') |
|
3517 | baserepo = repo.filtered(b'__perf_branchmap_update_base') | |
3518 | targetrepo = repo.filtered(b'__perf_branchmap_update_target') |
|
3518 | targetrepo = repo.filtered(b'__perf_branchmap_update_target') | |
3519 |
|
3519 | |||
3520 | # try to find an existing branchmap to reuse |
|
3520 | # try to find an existing branchmap to reuse | |
3521 | subsettable = getbranchmapsubsettable() |
|
3521 | subsettable = getbranchmapsubsettable() | |
3522 | candidatefilter = subsettable.get(None) |
|
3522 | candidatefilter = subsettable.get(None) | |
3523 | while candidatefilter is not None: |
|
3523 | while candidatefilter is not None: | |
3524 | candidatebm = repo.filtered(candidatefilter).branchmap() |
|
3524 | candidatebm = repo.filtered(candidatefilter).branchmap() | |
3525 | if candidatebm.validfor(baserepo): |
|
3525 | if candidatebm.validfor(baserepo): | |
3526 | filtered = repoview.filterrevs(repo, candidatefilter) |
|
3526 | filtered = repoview.filterrevs(repo, candidatefilter) | |
3527 | missing = [r for r in allbaserevs if r in filtered] |
|
3527 | missing = [r for r in allbaserevs if r in filtered] | |
3528 | base = candidatebm.copy() |
|
3528 | base = candidatebm.copy() | |
3529 | base.update(baserepo, missing) |
|
3529 | base.update(baserepo, missing) | |
3530 | break |
|
3530 | break | |
3531 | candidatefilter = subsettable.get(candidatefilter) |
|
3531 | candidatefilter = subsettable.get(candidatefilter) | |
3532 | else: |
|
3532 | else: | |
3533 | # no suitable subset where found |
|
3533 | # no suitable subset where found | |
3534 | base = branchmap.branchcache() |
|
3534 | base = branchmap.branchcache() | |
3535 | base.update(baserepo, allbaserevs) |
|
3535 | base.update(baserepo, allbaserevs) | |
3536 |
|
3536 | |||
3537 | def setup(): |
|
3537 | def setup(): | |
3538 | x[0] = base.copy() |
|
3538 | x[0] = base.copy() | |
3539 | if clearcaches: |
|
3539 | if clearcaches: | |
3540 | unfi._revbranchcache = None |
|
3540 | unfi._revbranchcache = None | |
3541 | clearchangelog(repo) |
|
3541 | clearchangelog(repo) | |
3542 |
|
3542 | |||
3543 | def bench(): |
|
3543 | def bench(): | |
3544 | x[0].update(targetrepo, newrevs) |
|
3544 | x[0].update(targetrepo, newrevs) | |
3545 |
|
3545 | |||
3546 | timer(bench, setup=setup) |
|
3546 | timer(bench, setup=setup) | |
3547 | fm.end() |
|
3547 | fm.end() | |
3548 | finally: |
|
3548 | finally: | |
3549 | repoview.filtertable.pop(b'__perf_branchmap_update_base', None) |
|
3549 | repoview.filtertable.pop(b'__perf_branchmap_update_base', None) | |
3550 | repoview.filtertable.pop(b'__perf_branchmap_update_target', None) |
|
3550 | repoview.filtertable.pop(b'__perf_branchmap_update_target', None) | |
3551 |
|
3551 | |||
3552 |
|
3552 | |||
3553 | @command( |
|
3553 | @command( | |
3554 | b'perfbranchmapload', |
|
3554 | b'perfbranchmapload', | |
3555 | [ |
|
3555 | [ | |
3556 | (b'f', b'filter', b'', b'Specify repoview filter'), |
|
3556 | (b'f', b'filter', b'', b'Specify repoview filter'), | |
3557 | (b'', b'list', False, b'List brachmap filter caches'), |
|
3557 | (b'', b'list', False, b'List brachmap filter caches'), | |
3558 | (b'', b'clear-revlogs', False, b'refresh changelog and manifest'), |
|
3558 | (b'', b'clear-revlogs', False, b'refresh changelog and manifest'), | |
3559 | ] |
|
3559 | ] | |
3560 | + formatteropts, |
|
3560 | + formatteropts, | |
3561 | ) |
|
3561 | ) | |
3562 | def perfbranchmapload(ui, repo, filter=b'', list=False, **opts): |
|
3562 | def perfbranchmapload(ui, repo, filter=b'', list=False, **opts): | |
3563 | """benchmark reading the branchmap""" |
|
3563 | """benchmark reading the branchmap""" | |
3564 | opts = _byteskwargs(opts) |
|
3564 | opts = _byteskwargs(opts) | |
3565 | clearrevlogs = opts[b'clear_revlogs'] |
|
3565 | clearrevlogs = opts[b'clear_revlogs'] | |
3566 |
|
3566 | |||
3567 | if list: |
|
3567 | if list: | |
3568 | for name, kind, st in repo.cachevfs.readdir(stat=True): |
|
3568 | for name, kind, st in repo.cachevfs.readdir(stat=True): | |
3569 | if name.startswith(b'branch2'): |
|
3569 | if name.startswith(b'branch2'): | |
3570 | filtername = name.partition(b'-')[2] or b'unfiltered' |
|
3570 | filtername = name.partition(b'-')[2] or b'unfiltered' | |
3571 | ui.status( |
|
3571 | ui.status( | |
3572 | b'%s - %s\n' % (filtername, util.bytecount(st.st_size)) |
|
3572 | b'%s - %s\n' % (filtername, util.bytecount(st.st_size)) | |
3573 | ) |
|
3573 | ) | |
3574 | return |
|
3574 | return | |
3575 | if not filter: |
|
3575 | if not filter: | |
3576 | filter = None |
|
3576 | filter = None | |
3577 | subsettable = getbranchmapsubsettable() |
|
3577 | subsettable = getbranchmapsubsettable() | |
3578 | if filter is None: |
|
3578 | if filter is None: | |
3579 | repo = repo.unfiltered() |
|
3579 | repo = repo.unfiltered() | |
3580 | else: |
|
3580 | else: | |
3581 | repo = repoview.repoview(repo, filter) |
|
3581 | repo = repoview.repoview(repo, filter) | |
3582 |
|
3582 | |||
3583 | repo.branchmap() # make sure we have a relevant, up to date branchmap |
|
3583 | repo.branchmap() # make sure we have a relevant, up to date branchmap | |
3584 |
|
3584 | |||
3585 | try: |
|
3585 | try: | |
3586 | fromfile = branchmap.branchcache.fromfile |
|
3586 | fromfile = branchmap.branchcache.fromfile | |
3587 | except AttributeError: |
|
3587 | except AttributeError: | |
3588 | # older versions |
|
3588 | # older versions | |
3589 | fromfile = branchmap.read |
|
3589 | fromfile = branchmap.read | |
3590 |
|
3590 | |||
3591 | currentfilter = filter |
|
3591 | currentfilter = filter | |
3592 | # try once without timer, the filter may not be cached |
|
3592 | # try once without timer, the filter may not be cached | |
3593 | while fromfile(repo) is None: |
|
3593 | while fromfile(repo) is None: | |
3594 | currentfilter = subsettable.get(currentfilter) |
|
3594 | currentfilter = subsettable.get(currentfilter) | |
3595 | if currentfilter is None: |
|
3595 | if currentfilter is None: | |
3596 | raise error.Abort( |
|
3596 | raise error.Abort( | |
3597 | b'No branchmap cached for %s repo' % (filter or b'unfiltered') |
|
3597 | b'No branchmap cached for %s repo' % (filter or b'unfiltered') | |
3598 | ) |
|
3598 | ) | |
3599 | repo = repo.filtered(currentfilter) |
|
3599 | repo = repo.filtered(currentfilter) | |
3600 | timer, fm = gettimer(ui, opts) |
|
3600 | timer, fm = gettimer(ui, opts) | |
3601 |
|
3601 | |||
3602 | def setup(): |
|
3602 | def setup(): | |
3603 | if clearrevlogs: |
|
3603 | if clearrevlogs: | |
3604 | clearchangelog(repo) |
|
3604 | clearchangelog(repo) | |
3605 |
|
3605 | |||
3606 | def bench(): |
|
3606 | def bench(): | |
3607 | fromfile(repo) |
|
3607 | fromfile(repo) | |
3608 |
|
3608 | |||
3609 | timer(bench, setup=setup) |
|
3609 | timer(bench, setup=setup) | |
3610 | fm.end() |
|
3610 | fm.end() | |
3611 |
|
3611 | |||
3612 |
|
3612 | |||
3613 | @command(b'perfloadmarkers') |
|
3613 | @command(b'perfloadmarkers') | |
3614 | def perfloadmarkers(ui, repo): |
|
3614 | def perfloadmarkers(ui, repo): | |
3615 | """benchmark the time to parse the on-disk markers for a repo |
|
3615 | """benchmark the time to parse the on-disk markers for a repo | |
3616 |
|
3616 | |||
3617 | Result is the number of markers in the repo.""" |
|
3617 | Result is the number of markers in the repo.""" | |
3618 | timer, fm = gettimer(ui) |
|
3618 | timer, fm = gettimer(ui) | |
3619 | svfs = getsvfs(repo) |
|
3619 | svfs = getsvfs(repo) | |
3620 | timer(lambda: len(obsolete.obsstore(svfs))) |
|
3620 | timer(lambda: len(obsolete.obsstore(svfs))) | |
3621 | fm.end() |
|
3621 | fm.end() | |
3622 |
|
3622 | |||
3623 |
|
3623 | |||
3624 | @command( |
|
3624 | @command( | |
3625 | b'perflrucachedict', |
|
3625 | b'perflrucachedict', | |
3626 | formatteropts |
|
3626 | formatteropts | |
3627 | + [ |
|
3627 | + [ | |
3628 | (b'', b'costlimit', 0, b'maximum total cost of items in cache'), |
|
3628 | (b'', b'costlimit', 0, b'maximum total cost of items in cache'), | |
3629 | (b'', b'mincost', 0, b'smallest cost of items in cache'), |
|
3629 | (b'', b'mincost', 0, b'smallest cost of items in cache'), | |
3630 | (b'', b'maxcost', 100, b'maximum cost of items in cache'), |
|
3630 | (b'', b'maxcost', 100, b'maximum cost of items in cache'), | |
3631 | (b'', b'size', 4, b'size of cache'), |
|
3631 | (b'', b'size', 4, b'size of cache'), | |
3632 | (b'', b'gets', 10000, b'number of key lookups'), |
|
3632 | (b'', b'gets', 10000, b'number of key lookups'), | |
3633 | (b'', b'sets', 10000, b'number of key sets'), |
|
3633 | (b'', b'sets', 10000, b'number of key sets'), | |
3634 | (b'', b'mixed', 10000, b'number of mixed mode operations'), |
|
3634 | (b'', b'mixed', 10000, b'number of mixed mode operations'), | |
3635 | ( |
|
3635 | ( | |
3636 | b'', |
|
3636 | b'', | |
3637 | b'mixedgetfreq', |
|
3637 | b'mixedgetfreq', | |
3638 | 50, |
|
3638 | 50, | |
3639 | b'frequency of get vs set ops in mixed mode', |
|
3639 | b'frequency of get vs set ops in mixed mode', | |
3640 | ), |
|
3640 | ), | |
3641 | ], |
|
3641 | ], | |
3642 | norepo=True, |
|
3642 | norepo=True, | |
3643 | ) |
|
3643 | ) | |
3644 | def perflrucache( |
|
3644 | def perflrucache( | |
3645 | ui, |
|
3645 | ui, | |
3646 | mincost=0, |
|
3646 | mincost=0, | |
3647 | maxcost=100, |
|
3647 | maxcost=100, | |
3648 | costlimit=0, |
|
3648 | costlimit=0, | |
3649 | size=4, |
|
3649 | size=4, | |
3650 | gets=10000, |
|
3650 | gets=10000, | |
3651 | sets=10000, |
|
3651 | sets=10000, | |
3652 | mixed=10000, |
|
3652 | mixed=10000, | |
3653 | mixedgetfreq=50, |
|
3653 | mixedgetfreq=50, | |
3654 | **opts |
|
3654 | **opts | |
3655 | ): |
|
3655 | ): | |
3656 | opts = _byteskwargs(opts) |
|
3656 | opts = _byteskwargs(opts) | |
3657 |
|
3657 | |||
3658 | def doinit(): |
|
3658 | def doinit(): | |
3659 | for i in _xrange(10000): |
|
3659 | for i in _xrange(10000): | |
3660 | util.lrucachedict(size) |
|
3660 | util.lrucachedict(size) | |
3661 |
|
3661 | |||
3662 | costrange = list(range(mincost, maxcost + 1)) |
|
3662 | costrange = list(range(mincost, maxcost + 1)) | |
3663 |
|
3663 | |||
3664 | values = [] |
|
3664 | values = [] | |
3665 | for i in _xrange(size): |
|
3665 | for i in _xrange(size): | |
3666 | values.append(random.randint(0, _maxint)) |
|
3666 | values.append(random.randint(0, _maxint)) | |
3667 |
|
3667 | |||
3668 | # Get mode fills the cache and tests raw lookup performance with no |
|
3668 | # Get mode fills the cache and tests raw lookup performance with no | |
3669 | # eviction. |
|
3669 | # eviction. | |
3670 | getseq = [] |
|
3670 | getseq = [] | |
3671 | for i in _xrange(gets): |
|
3671 | for i in _xrange(gets): | |
3672 | getseq.append(random.choice(values)) |
|
3672 | getseq.append(random.choice(values)) | |
3673 |
|
3673 | |||
3674 | def dogets(): |
|
3674 | def dogets(): | |
3675 | d = util.lrucachedict(size) |
|
3675 | d = util.lrucachedict(size) | |
3676 | for v in values: |
|
3676 | for v in values: | |
3677 | d[v] = v |
|
3677 | d[v] = v | |
3678 | for key in getseq: |
|
3678 | for key in getseq: | |
3679 | value = d[key] |
|
3679 | value = d[key] | |
3680 | value # silence pyflakes warning |
|
3680 | value # silence pyflakes warning | |
3681 |
|
3681 | |||
3682 | def dogetscost(): |
|
3682 | def dogetscost(): | |
3683 | d = util.lrucachedict(size, maxcost=costlimit) |
|
3683 | d = util.lrucachedict(size, maxcost=costlimit) | |
3684 | for i, v in enumerate(values): |
|
3684 | for i, v in enumerate(values): | |
3685 | d.insert(v, v, cost=costs[i]) |
|
3685 | d.insert(v, v, cost=costs[i]) | |
3686 | for key in getseq: |
|
3686 | for key in getseq: | |
3687 | try: |
|
3687 | try: | |
3688 | value = d[key] |
|
3688 | value = d[key] | |
3689 | value # silence pyflakes warning |
|
3689 | value # silence pyflakes warning | |
3690 | except KeyError: |
|
3690 | except KeyError: | |
3691 | pass |
|
3691 | pass | |
3692 |
|
3692 | |||
3693 | # Set mode tests insertion speed with cache eviction. |
|
3693 | # Set mode tests insertion speed with cache eviction. | |
3694 | setseq = [] |
|
3694 | setseq = [] | |
3695 | costs = [] |
|
3695 | costs = [] | |
3696 | for i in _xrange(sets): |
|
3696 | for i in _xrange(sets): | |
3697 | setseq.append(random.randint(0, _maxint)) |
|
3697 | setseq.append(random.randint(0, _maxint)) | |
3698 | costs.append(random.choice(costrange)) |
|
3698 | costs.append(random.choice(costrange)) | |
3699 |
|
3699 | |||
3700 | def doinserts(): |
|
3700 | def doinserts(): | |
3701 | d = util.lrucachedict(size) |
|
3701 | d = util.lrucachedict(size) | |
3702 | for v in setseq: |
|
3702 | for v in setseq: | |
3703 | d.insert(v, v) |
|
3703 | d.insert(v, v) | |
3704 |
|
3704 | |||
3705 | def doinsertscost(): |
|
3705 | def doinsertscost(): | |
3706 | d = util.lrucachedict(size, maxcost=costlimit) |
|
3706 | d = util.lrucachedict(size, maxcost=costlimit) | |
3707 | for i, v in enumerate(setseq): |
|
3707 | for i, v in enumerate(setseq): | |
3708 | d.insert(v, v, cost=costs[i]) |
|
3708 | d.insert(v, v, cost=costs[i]) | |
3709 |
|
3709 | |||
3710 | def dosets(): |
|
3710 | def dosets(): | |
3711 | d = util.lrucachedict(size) |
|
3711 | d = util.lrucachedict(size) | |
3712 | for v in setseq: |
|
3712 | for v in setseq: | |
3713 | d[v] = v |
|
3713 | d[v] = v | |
3714 |
|
3714 | |||
3715 | # Mixed mode randomly performs gets and sets with eviction. |
|
3715 | # Mixed mode randomly performs gets and sets with eviction. | |
3716 | mixedops = [] |
|
3716 | mixedops = [] | |
3717 | for i in _xrange(mixed): |
|
3717 | for i in _xrange(mixed): | |
3718 | r = random.randint(0, 100) |
|
3718 | r = random.randint(0, 100) | |
3719 | if r < mixedgetfreq: |
|
3719 | if r < mixedgetfreq: | |
3720 | op = 0 |
|
3720 | op = 0 | |
3721 | else: |
|
3721 | else: | |
3722 | op = 1 |
|
3722 | op = 1 | |
3723 |
|
3723 | |||
3724 | mixedops.append( |
|
3724 | mixedops.append( | |
3725 | (op, random.randint(0, size * 2), random.choice(costrange)) |
|
3725 | (op, random.randint(0, size * 2), random.choice(costrange)) | |
3726 | ) |
|
3726 | ) | |
3727 |
|
3727 | |||
3728 | def domixed(): |
|
3728 | def domixed(): | |
3729 | d = util.lrucachedict(size) |
|
3729 | d = util.lrucachedict(size) | |
3730 |
|
3730 | |||
3731 | for op, v, cost in mixedops: |
|
3731 | for op, v, cost in mixedops: | |
3732 | if op == 0: |
|
3732 | if op == 0: | |
3733 | try: |
|
3733 | try: | |
3734 | d[v] |
|
3734 | d[v] | |
3735 | except KeyError: |
|
3735 | except KeyError: | |
3736 | pass |
|
3736 | pass | |
3737 | else: |
|
3737 | else: | |
3738 | d[v] = v |
|
3738 | d[v] = v | |
3739 |
|
3739 | |||
3740 | def domixedcost(): |
|
3740 | def domixedcost(): | |
3741 | d = util.lrucachedict(size, maxcost=costlimit) |
|
3741 | d = util.lrucachedict(size, maxcost=costlimit) | |
3742 |
|
3742 | |||
3743 | for op, v, cost in mixedops: |
|
3743 | for op, v, cost in mixedops: | |
3744 | if op == 0: |
|
3744 | if op == 0: | |
3745 | try: |
|
3745 | try: | |
3746 | d[v] |
|
3746 | d[v] | |
3747 | except KeyError: |
|
3747 | except KeyError: | |
3748 | pass |
|
3748 | pass | |
3749 | else: |
|
3749 | else: | |
3750 | d.insert(v, v, cost=cost) |
|
3750 | d.insert(v, v, cost=cost) | |
3751 |
|
3751 | |||
3752 | benches = [ |
|
3752 | benches = [ | |
3753 | (doinit, b'init'), |
|
3753 | (doinit, b'init'), | |
3754 | ] |
|
3754 | ] | |
3755 |
|
3755 | |||
3756 | if costlimit: |
|
3756 | if costlimit: | |
3757 | benches.extend( |
|
3757 | benches.extend( | |
3758 | [ |
|
3758 | [ | |
3759 | (dogetscost, b'gets w/ cost limit'), |
|
3759 | (dogetscost, b'gets w/ cost limit'), | |
3760 | (doinsertscost, b'inserts w/ cost limit'), |
|
3760 | (doinsertscost, b'inserts w/ cost limit'), | |
3761 | (domixedcost, b'mixed w/ cost limit'), |
|
3761 | (domixedcost, b'mixed w/ cost limit'), | |
3762 | ] |
|
3762 | ] | |
3763 | ) |
|
3763 | ) | |
3764 | else: |
|
3764 | else: | |
3765 | benches.extend( |
|
3765 | benches.extend( | |
3766 | [ |
|
3766 | [ | |
3767 | (dogets, b'gets'), |
|
3767 | (dogets, b'gets'), | |
3768 | (doinserts, b'inserts'), |
|
3768 | (doinserts, b'inserts'), | |
3769 | (dosets, b'sets'), |
|
3769 | (dosets, b'sets'), | |
3770 | (domixed, b'mixed'), |
|
3770 | (domixed, b'mixed'), | |
3771 | ] |
|
3771 | ] | |
3772 | ) |
|
3772 | ) | |
3773 |
|
3773 | |||
3774 | for fn, title in benches: |
|
3774 | for fn, title in benches: | |
3775 | timer, fm = gettimer(ui, opts) |
|
3775 | timer, fm = gettimer(ui, opts) | |
3776 | timer(fn, title=title) |
|
3776 | timer(fn, title=title) | |
3777 | fm.end() |
|
3777 | fm.end() | |
3778 |
|
3778 | |||
3779 |
|
3779 | |||
3780 | @command(b'perfwrite', formatteropts) |
|
3780 | @command(b'perfwrite', formatteropts) | |
3781 | def perfwrite(ui, repo, **opts): |
|
3781 | def perfwrite(ui, repo, **opts): | |
3782 | """microbenchmark ui.write |
|
3782 | """microbenchmark ui.write | |
3783 | """ |
|
3783 | """ | |
3784 | opts = _byteskwargs(opts) |
|
3784 | opts = _byteskwargs(opts) | |
3785 |
|
3785 | |||
3786 | timer, fm = gettimer(ui, opts) |
|
3786 | timer, fm = gettimer(ui, opts) | |
3787 |
|
3787 | |||
3788 | def write(): |
|
3788 | def write(): | |
3789 | for i in range(100000): |
|
3789 | for i in range(100000): | |
3790 | ui.writenoi18n(b'Testing write performance\n') |
|
3790 | ui.writenoi18n(b'Testing write performance\n') | |
3791 |
|
3791 | |||
3792 | timer(write) |
|
3792 | timer(write) | |
3793 | fm.end() |
|
3793 | fm.end() | |
3794 |
|
3794 | |||
3795 |
|
3795 | |||
3796 | def uisetup(ui): |
|
3796 | def uisetup(ui): | |
3797 | if util.safehasattr(cmdutil, b'openrevlog') and not util.safehasattr( |
|
3797 | if util.safehasattr(cmdutil, b'openrevlog') and not util.safehasattr( | |
3798 | commands, b'debugrevlogopts' |
|
3798 | commands, b'debugrevlogopts' | |
3799 | ): |
|
3799 | ): | |
3800 | # for "historical portability": |
|
3800 | # for "historical portability": | |
3801 | # In this case, Mercurial should be 1.9 (or a79fea6b3e77) - |
|
3801 | # In this case, Mercurial should be 1.9 (or a79fea6b3e77) - | |
3802 | # 3.7 (or 5606f7d0d063). Therefore, '--dir' option for |
|
3802 | # 3.7 (or 5606f7d0d063). Therefore, '--dir' option for | |
3803 | # openrevlog() should cause failure, because it has been |
|
3803 | # openrevlog() should cause failure, because it has been | |
3804 | # available since 3.5 (or 49c583ca48c4). |
|
3804 | # available since 3.5 (or 49c583ca48c4). | |
3805 | def openrevlog(orig, repo, cmd, file_, opts): |
|
3805 | def openrevlog(orig, repo, cmd, file_, opts): | |
3806 | if opts.get(b'dir') and not util.safehasattr(repo, b'dirlog'): |
|
3806 | if opts.get(b'dir') and not util.safehasattr(repo, b'dirlog'): | |
3807 | raise error.Abort( |
|
3807 | raise error.Abort( | |
3808 | b"This version doesn't support --dir option", |
|
3808 | b"This version doesn't support --dir option", | |
3809 | hint=b"use 3.5 or later", |
|
3809 | hint=b"use 3.5 or later", | |
3810 | ) |
|
3810 | ) | |
3811 | return orig(repo, cmd, file_, opts) |
|
3811 | return orig(repo, cmd, file_, opts) | |
3812 |
|
3812 | |||
3813 | extensions.wrapfunction(cmdutil, b'openrevlog', openrevlog) |
|
3813 | extensions.wrapfunction(cmdutil, b'openrevlog', openrevlog) | |
3814 |
|
3814 | |||
3815 |
|
3815 | |||
3816 | @command( |
|
3816 | @command( | |
3817 | b'perfprogress', |
|
3817 | b'perfprogress', | |
3818 | formatteropts |
|
3818 | formatteropts | |
3819 | + [ |
|
3819 | + [ | |
3820 | (b'', b'topic', b'topic', b'topic for progress messages'), |
|
3820 | (b'', b'topic', b'topic', b'topic for progress messages'), | |
3821 | (b'c', b'total', 1000000, b'total value we are progressing to'), |
|
3821 | (b'c', b'total', 1000000, b'total value we are progressing to'), | |
3822 | ], |
|
3822 | ], | |
3823 | norepo=True, |
|
3823 | norepo=True, | |
3824 | ) |
|
3824 | ) | |
3825 | def perfprogress(ui, topic=None, total=None, **opts): |
|
3825 | def perfprogress(ui, topic=None, total=None, **opts): | |
3826 | """printing of progress bars""" |
|
3826 | """printing of progress bars""" | |
3827 | opts = _byteskwargs(opts) |
|
3827 | opts = _byteskwargs(opts) | |
3828 |
|
3828 | |||
3829 | timer, fm = gettimer(ui, opts) |
|
3829 | timer, fm = gettimer(ui, opts) | |
3830 |
|
3830 | |||
3831 | def doprogress(): |
|
3831 | def doprogress(): | |
3832 | with ui.makeprogress(topic, total=total) as progress: |
|
3832 | with ui.makeprogress(topic, total=total) as progress: | |
3833 | for i in _xrange(total): |
|
3833 | for i in _xrange(total): | |
3834 | progress.increment() |
|
3834 | progress.increment() | |
3835 |
|
3835 | |||
3836 | timer(doprogress) |
|
3836 | timer(doprogress) | |
3837 | fm.end() |
|
3837 | fm.end() |
@@ -1,2947 +1,2947 b'' | |||||
1 | # revlog.py - storage back-end for mercurial |
|
1 | # revlog.py - storage back-end for mercurial | |
2 | # |
|
2 | # | |
3 | # Copyright 2005-2007 Matt Mackall <mpm@selenic.com> |
|
3 | # Copyright 2005-2007 Matt Mackall <mpm@selenic.com> | |
4 | # |
|
4 | # | |
5 | # This software may be used and distributed according to the terms of the |
|
5 | # This software may be used and distributed according to the terms of the | |
6 | # GNU General Public License version 2 or any later version. |
|
6 | # GNU General Public License version 2 or any later version. | |
7 |
|
7 | |||
8 | """Storage back-end for Mercurial. |
|
8 | """Storage back-end for Mercurial. | |
9 |
|
9 | |||
10 | This provides efficient delta storage with O(1) retrieve and append |
|
10 | This provides efficient delta storage with O(1) retrieve and append | |
11 | and O(changes) merge between branches. |
|
11 | and O(changes) merge between branches. | |
12 | """ |
|
12 | """ | |
13 |
|
13 | |||
14 | from __future__ import absolute_import |
|
14 | from __future__ import absolute_import | |
15 |
|
15 | |||
16 | import collections |
|
16 | import collections | |
17 | import contextlib |
|
17 | import contextlib | |
18 | import errno |
|
18 | import errno | |
19 | import io |
|
19 | import io | |
20 | import os |
|
20 | import os | |
21 | import struct |
|
21 | import struct | |
22 | import zlib |
|
22 | import zlib | |
23 |
|
23 | |||
24 | # import stuff from node for others to import from revlog |
|
24 | # import stuff from node for others to import from revlog | |
25 | from .node import ( |
|
25 | from .node import ( | |
26 | bin, |
|
26 | bin, | |
27 | hex, |
|
27 | hex, | |
28 | nullhex, |
|
28 | nullhex, | |
29 | nullid, |
|
29 | nullid, | |
30 | nullrev, |
|
30 | nullrev, | |
31 | short, |
|
31 | short, | |
32 | wdirfilenodeids, |
|
32 | wdirfilenodeids, | |
33 | wdirhex, |
|
33 | wdirhex, | |
34 | wdirid, |
|
34 | wdirid, | |
35 | wdirrev, |
|
35 | wdirrev, | |
36 | ) |
|
36 | ) | |
37 | from .i18n import _ |
|
37 | from .i18n import _ | |
38 | from .pycompat import getattr |
|
38 | from .pycompat import getattr | |
39 | from .revlogutils.constants import ( |
|
39 | from .revlogutils.constants import ( | |
40 | FLAG_GENERALDELTA, |
|
40 | FLAG_GENERALDELTA, | |
41 | FLAG_INLINE_DATA, |
|
41 | FLAG_INLINE_DATA, | |
42 | REVLOGV0, |
|
42 | REVLOGV0, | |
43 | REVLOGV1, |
|
43 | REVLOGV1, | |
44 | REVLOGV1_FLAGS, |
|
44 | REVLOGV1_FLAGS, | |
45 | REVLOGV2, |
|
45 | REVLOGV2, | |
46 | REVLOGV2_FLAGS, |
|
46 | REVLOGV2_FLAGS, | |
47 | REVLOG_DEFAULT_FLAGS, |
|
47 | REVLOG_DEFAULT_FLAGS, | |
48 | REVLOG_DEFAULT_FORMAT, |
|
48 | REVLOG_DEFAULT_FORMAT, | |
49 | REVLOG_DEFAULT_VERSION, |
|
49 | REVLOG_DEFAULT_VERSION, | |
50 | ) |
|
50 | ) | |
51 | from .revlogutils.flagutil import ( |
|
51 | from .revlogutils.flagutil import ( | |
52 | REVIDX_DEFAULT_FLAGS, |
|
52 | REVIDX_DEFAULT_FLAGS, | |
53 | REVIDX_ELLIPSIS, |
|
53 | REVIDX_ELLIPSIS, | |
54 | REVIDX_EXTSTORED, |
|
54 | REVIDX_EXTSTORED, | |
55 | REVIDX_FLAGS_ORDER, |
|
55 | REVIDX_FLAGS_ORDER, | |
56 | REVIDX_ISCENSORED, |
|
56 | REVIDX_ISCENSORED, | |
57 | REVIDX_RAWTEXT_CHANGING_FLAGS, |
|
57 | REVIDX_RAWTEXT_CHANGING_FLAGS, | |
58 | REVIDX_SIDEDATA, |
|
58 | REVIDX_SIDEDATA, | |
59 | ) |
|
59 | ) | |
60 | from .thirdparty import attr |
|
60 | from .thirdparty import attr | |
61 | from . import ( |
|
61 | from . import ( | |
62 | ancestor, |
|
62 | ancestor, | |
63 | dagop, |
|
63 | dagop, | |
64 | error, |
|
64 | error, | |
65 | mdiff, |
|
65 | mdiff, | |
66 | policy, |
|
66 | policy, | |
67 | pycompat, |
|
67 | pycompat, | |
68 | revlogutils, |
|
68 | revlogutils, | |
69 | templatefilters, |
|
69 | templatefilters, | |
70 | util, |
|
70 | util, | |
71 | ) |
|
71 | ) | |
72 | from .interfaces import ( |
|
72 | from .interfaces import ( | |
73 | repository, |
|
73 | repository, | |
74 | util as interfaceutil, |
|
74 | util as interfaceutil, | |
75 | ) |
|
75 | ) | |
76 | from .revlogutils import ( |
|
76 | from .revlogutils import ( | |
77 | deltas as deltautil, |
|
77 | deltas as deltautil, | |
78 | flagutil, |
|
78 | flagutil, | |
79 | sidedata as sidedatautil, |
|
79 | sidedata as sidedatautil, | |
80 | ) |
|
80 | ) | |
81 | from .utils import ( |
|
81 | from .utils import ( | |
82 | storageutil, |
|
82 | storageutil, | |
83 | stringutil, |
|
83 | stringutil, | |
84 | ) |
|
84 | ) | |
85 |
|
85 | |||
86 | # blanked usage of all the name to prevent pyflakes constraints |
|
86 | # blanked usage of all the name to prevent pyflakes constraints | |
87 | # We need these name available in the module for extensions. |
|
87 | # We need these name available in the module for extensions. | |
88 | REVLOGV0 |
|
88 | REVLOGV0 | |
89 | REVLOGV1 |
|
89 | REVLOGV1 | |
90 | REVLOGV2 |
|
90 | REVLOGV2 | |
91 | FLAG_INLINE_DATA |
|
91 | FLAG_INLINE_DATA | |
92 | FLAG_GENERALDELTA |
|
92 | FLAG_GENERALDELTA | |
93 | REVLOG_DEFAULT_FLAGS |
|
93 | REVLOG_DEFAULT_FLAGS | |
94 | REVLOG_DEFAULT_FORMAT |
|
94 | REVLOG_DEFAULT_FORMAT | |
95 | REVLOG_DEFAULT_VERSION |
|
95 | REVLOG_DEFAULT_VERSION | |
96 | REVLOGV1_FLAGS |
|
96 | REVLOGV1_FLAGS | |
97 | REVLOGV2_FLAGS |
|
97 | REVLOGV2_FLAGS | |
98 | REVIDX_ISCENSORED |
|
98 | REVIDX_ISCENSORED | |
99 | REVIDX_ELLIPSIS |
|
99 | REVIDX_ELLIPSIS | |
100 | REVIDX_SIDEDATA |
|
100 | REVIDX_SIDEDATA | |
101 | REVIDX_EXTSTORED |
|
101 | REVIDX_EXTSTORED | |
102 | REVIDX_DEFAULT_FLAGS |
|
102 | REVIDX_DEFAULT_FLAGS | |
103 | REVIDX_FLAGS_ORDER |
|
103 | REVIDX_FLAGS_ORDER | |
104 | REVIDX_RAWTEXT_CHANGING_FLAGS |
|
104 | REVIDX_RAWTEXT_CHANGING_FLAGS | |
105 |
|
105 | |||
106 | parsers = policy.importmod('parsers') |
|
106 | parsers = policy.importmod('parsers') | |
107 | rustancestor = policy.importrust('ancestor') |
|
107 | rustancestor = policy.importrust('ancestor') | |
108 | rustdagop = policy.importrust('dagop') |
|
108 | rustdagop = policy.importrust('dagop') | |
109 |
|
109 | |||
110 | # Aliased for performance. |
|
110 | # Aliased for performance. | |
111 | _zlibdecompress = zlib.decompress |
|
111 | _zlibdecompress = zlib.decompress | |
112 |
|
112 | |||
113 | # max size of revlog with inline data |
|
113 | # max size of revlog with inline data | |
114 | _maxinline = 131072 |
|
114 | _maxinline = 131072 | |
115 | _chunksize = 1048576 |
|
115 | _chunksize = 1048576 | |
116 |
|
116 | |||
117 | # Flag processors for REVIDX_ELLIPSIS. |
|
117 | # Flag processors for REVIDX_ELLIPSIS. | |
118 | def ellipsisreadprocessor(rl, text): |
|
118 | def ellipsisreadprocessor(rl, text): | |
119 | return text, False, {} |
|
119 | return text, False, {} | |
120 |
|
120 | |||
121 |
|
121 | |||
122 | def ellipsiswriteprocessor(rl, text, sidedata): |
|
122 | def ellipsiswriteprocessor(rl, text, sidedata): | |
123 | return text, False |
|
123 | return text, False | |
124 |
|
124 | |||
125 |
|
125 | |||
126 | def ellipsisrawprocessor(rl, text): |
|
126 | def ellipsisrawprocessor(rl, text): | |
127 | return False |
|
127 | return False | |
128 |
|
128 | |||
129 |
|
129 | |||
130 | ellipsisprocessor = ( |
|
130 | ellipsisprocessor = ( | |
131 | ellipsisreadprocessor, |
|
131 | ellipsisreadprocessor, | |
132 | ellipsiswriteprocessor, |
|
132 | ellipsiswriteprocessor, | |
133 | ellipsisrawprocessor, |
|
133 | ellipsisrawprocessor, | |
134 | ) |
|
134 | ) | |
135 |
|
135 | |||
136 |
|
136 | |||
137 | def getoffset(q): |
|
137 | def getoffset(q): | |
138 | return int(q >> 16) |
|
138 | return int(q >> 16) | |
139 |
|
139 | |||
140 |
|
140 | |||
141 | def gettype(q): |
|
141 | def gettype(q): | |
142 | return int(q & 0xFFFF) |
|
142 | return int(q & 0xFFFF) | |
143 |
|
143 | |||
144 |
|
144 | |||
145 | def offset_type(offset, type): |
|
145 | def offset_type(offset, type): | |
146 | if (type & ~flagutil.REVIDX_KNOWN_FLAGS) != 0: |
|
146 | if (type & ~flagutil.REVIDX_KNOWN_FLAGS) != 0: | |
147 | raise ValueError(b'unknown revlog index flags') |
|
147 | raise ValueError(b'unknown revlog index flags') | |
148 | return int(int(offset) << 16 | type) |
|
148 | return int(int(offset) << 16 | type) | |
149 |
|
149 | |||
150 |
|
150 | |||
151 | @attr.s(slots=True, frozen=True) |
|
151 | @attr.s(slots=True, frozen=True) | |
152 | class _revisioninfo(object): |
|
152 | class _revisioninfo(object): | |
153 | """Information about a revision that allows building its fulltext |
|
153 | """Information about a revision that allows building its fulltext | |
154 | node: expected hash of the revision |
|
154 | node: expected hash of the revision | |
155 | p1, p2: parent revs of the revision |
|
155 | p1, p2: parent revs of the revision | |
156 | btext: built text cache consisting of a one-element list |
|
156 | btext: built text cache consisting of a one-element list | |
157 | cachedelta: (baserev, uncompressed_delta) or None |
|
157 | cachedelta: (baserev, uncompressed_delta) or None | |
158 | flags: flags associated to the revision storage |
|
158 | flags: flags associated to the revision storage | |
159 |
|
159 | |||
160 | One of btext[0] or cachedelta must be set. |
|
160 | One of btext[0] or cachedelta must be set. | |
161 | """ |
|
161 | """ | |
162 |
|
162 | |||
163 | node = attr.ib() |
|
163 | node = attr.ib() | |
164 | p1 = attr.ib() |
|
164 | p1 = attr.ib() | |
165 | p2 = attr.ib() |
|
165 | p2 = attr.ib() | |
166 | btext = attr.ib() |
|
166 | btext = attr.ib() | |
167 | textlen = attr.ib() |
|
167 | textlen = attr.ib() | |
168 | cachedelta = attr.ib() |
|
168 | cachedelta = attr.ib() | |
169 | flags = attr.ib() |
|
169 | flags = attr.ib() | |
170 |
|
170 | |||
171 |
|
171 | |||
172 | @interfaceutil.implementer(repository.irevisiondelta) |
|
172 | @interfaceutil.implementer(repository.irevisiondelta) | |
173 | @attr.s(slots=True) |
|
173 | @attr.s(slots=True) | |
174 | class revlogrevisiondelta(object): |
|
174 | class revlogrevisiondelta(object): | |
175 | node = attr.ib() |
|
175 | node = attr.ib() | |
176 | p1node = attr.ib() |
|
176 | p1node = attr.ib() | |
177 | p2node = attr.ib() |
|
177 | p2node = attr.ib() | |
178 | basenode = attr.ib() |
|
178 | basenode = attr.ib() | |
179 | flags = attr.ib() |
|
179 | flags = attr.ib() | |
180 | baserevisionsize = attr.ib() |
|
180 | baserevisionsize = attr.ib() | |
181 | revision = attr.ib() |
|
181 | revision = attr.ib() | |
182 | delta = attr.ib() |
|
182 | delta = attr.ib() | |
183 | linknode = attr.ib(default=None) |
|
183 | linknode = attr.ib(default=None) | |
184 |
|
184 | |||
185 |
|
185 | |||
186 | @interfaceutil.implementer(repository.iverifyproblem) |
|
186 | @interfaceutil.implementer(repository.iverifyproblem) | |
187 | @attr.s(frozen=True) |
|
187 | @attr.s(frozen=True) | |
188 | class revlogproblem(object): |
|
188 | class revlogproblem(object): | |
189 | warning = attr.ib(default=None) |
|
189 | warning = attr.ib(default=None) | |
190 | error = attr.ib(default=None) |
|
190 | error = attr.ib(default=None) | |
191 | node = attr.ib(default=None) |
|
191 | node = attr.ib(default=None) | |
192 |
|
192 | |||
193 |
|
193 | |||
194 | # index v0: |
|
194 | # index v0: | |
195 | # 4 bytes: offset |
|
195 | # 4 bytes: offset | |
196 | # 4 bytes: compressed length |
|
196 | # 4 bytes: compressed length | |
197 | # 4 bytes: base rev |
|
197 | # 4 bytes: base rev | |
198 | # 4 bytes: link rev |
|
198 | # 4 bytes: link rev | |
199 | # 20 bytes: parent 1 nodeid |
|
199 | # 20 bytes: parent 1 nodeid | |
200 | # 20 bytes: parent 2 nodeid |
|
200 | # 20 bytes: parent 2 nodeid | |
201 | # 20 bytes: nodeid |
|
201 | # 20 bytes: nodeid | |
202 | indexformatv0 = struct.Struct(b">4l20s20s20s") |
|
202 | indexformatv0 = struct.Struct(b">4l20s20s20s") | |
203 | indexformatv0_pack = indexformatv0.pack |
|
203 | indexformatv0_pack = indexformatv0.pack | |
204 | indexformatv0_unpack = indexformatv0.unpack |
|
204 | indexformatv0_unpack = indexformatv0.unpack | |
205 |
|
205 | |||
206 |
|
206 | |||
207 | class revlogoldindex(list): |
|
207 | class revlogoldindex(list): | |
208 | @util.propertycache |
|
208 | @util.propertycache | |
209 | def nodemap(self): |
|
209 | def nodemap(self): | |
210 | nodemap = revlogutils.NodeMap({nullid: nullrev}) |
|
210 | nodemap = revlogutils.NodeMap({nullid: nullrev}) | |
211 | for r in range(0, len(self)): |
|
211 | for r in range(0, len(self)): | |
212 | n = self[r][7] |
|
212 | n = self[r][7] | |
213 | nodemap[n] = r |
|
213 | nodemap[n] = r | |
214 | return nodemap |
|
214 | return nodemap | |
215 |
|
215 | |||
216 | def clearcaches(self): |
|
216 | def clearcaches(self): | |
217 | self.__dict__.pop('nodemap', None) |
|
217 | self.__dict__.pop('nodemap', None) | |
218 |
|
218 | |||
219 | def __getitem__(self, i): |
|
219 | def __getitem__(self, i): | |
220 | if i == -1: |
|
220 | if i == -1: | |
221 | return (0, 0, 0, -1, -1, -1, -1, nullid) |
|
221 | return (0, 0, 0, -1, -1, -1, -1, nullid) | |
222 | return list.__getitem__(self, i) |
|
222 | return list.__getitem__(self, i) | |
223 |
|
223 | |||
224 |
|
224 | |||
225 | class revlogoldio(object): |
|
225 | class revlogoldio(object): | |
226 | def __init__(self): |
|
226 | def __init__(self): | |
227 | self.size = indexformatv0.size |
|
227 | self.size = indexformatv0.size | |
228 |
|
228 | |||
229 | def parseindex(self, data, inline): |
|
229 | def parseindex(self, data, inline): | |
230 | s = self.size |
|
230 | s = self.size | |
231 | index = [] |
|
231 | index = [] | |
232 | nodemap = revlogutils.NodeMap({nullid: nullrev}) |
|
232 | nodemap = revlogutils.NodeMap({nullid: nullrev}) | |
233 | n = off = 0 |
|
233 | n = off = 0 | |
234 | l = len(data) |
|
234 | l = len(data) | |
235 | while off + s <= l: |
|
235 | while off + s <= l: | |
236 | cur = data[off : off + s] |
|
236 | cur = data[off : off + s] | |
237 | off += s |
|
237 | off += s | |
238 | e = indexformatv0_unpack(cur) |
|
238 | e = indexformatv0_unpack(cur) | |
239 | # transform to revlogv1 format |
|
239 | # transform to revlogv1 format | |
240 | e2 = ( |
|
240 | e2 = ( | |
241 | offset_type(e[0], 0), |
|
241 | offset_type(e[0], 0), | |
242 | e[1], |
|
242 | e[1], | |
243 | -1, |
|
243 | -1, | |
244 | e[2], |
|
244 | e[2], | |
245 | e[3], |
|
245 | e[3], | |
246 | nodemap.get(e[4], nullrev), |
|
246 | nodemap.get(e[4], nullrev), | |
247 | nodemap.get(e[5], nullrev), |
|
247 | nodemap.get(e[5], nullrev), | |
248 | e[6], |
|
248 | e[6], | |
249 | ) |
|
249 | ) | |
250 | index.append(e2) |
|
250 | index.append(e2) | |
251 | nodemap[e[6]] = n |
|
251 | nodemap[e[6]] = n | |
252 | n += 1 |
|
252 | n += 1 | |
253 |
|
253 | |||
254 | index = revlogoldindex(index) |
|
254 | index = revlogoldindex(index) | |
255 |
return index, |
|
255 | return index, None | |
256 |
|
256 | |||
257 | def packentry(self, entry, node, version, rev): |
|
257 | def packentry(self, entry, node, version, rev): | |
258 | if gettype(entry[0]): |
|
258 | if gettype(entry[0]): | |
259 | raise error.RevlogError( |
|
259 | raise error.RevlogError( | |
260 | _(b'index entry flags need revlog version 1') |
|
260 | _(b'index entry flags need revlog version 1') | |
261 | ) |
|
261 | ) | |
262 | e2 = ( |
|
262 | e2 = ( | |
263 | getoffset(entry[0]), |
|
263 | getoffset(entry[0]), | |
264 | entry[1], |
|
264 | entry[1], | |
265 | entry[3], |
|
265 | entry[3], | |
266 | entry[4], |
|
266 | entry[4], | |
267 | node(entry[5]), |
|
267 | node(entry[5]), | |
268 | node(entry[6]), |
|
268 | node(entry[6]), | |
269 | entry[7], |
|
269 | entry[7], | |
270 | ) |
|
270 | ) | |
271 | return indexformatv0_pack(*e2) |
|
271 | return indexformatv0_pack(*e2) | |
272 |
|
272 | |||
273 |
|
273 | |||
274 | # index ng: |
|
274 | # index ng: | |
275 | # 6 bytes: offset |
|
275 | # 6 bytes: offset | |
276 | # 2 bytes: flags |
|
276 | # 2 bytes: flags | |
277 | # 4 bytes: compressed length |
|
277 | # 4 bytes: compressed length | |
278 | # 4 bytes: uncompressed length |
|
278 | # 4 bytes: uncompressed length | |
279 | # 4 bytes: base rev |
|
279 | # 4 bytes: base rev | |
280 | # 4 bytes: link rev |
|
280 | # 4 bytes: link rev | |
281 | # 4 bytes: parent 1 rev |
|
281 | # 4 bytes: parent 1 rev | |
282 | # 4 bytes: parent 2 rev |
|
282 | # 4 bytes: parent 2 rev | |
283 | # 32 bytes: nodeid |
|
283 | # 32 bytes: nodeid | |
284 | indexformatng = struct.Struct(b">Qiiiiii20s12x") |
|
284 | indexformatng = struct.Struct(b">Qiiiiii20s12x") | |
285 | indexformatng_pack = indexformatng.pack |
|
285 | indexformatng_pack = indexformatng.pack | |
286 | versionformat = struct.Struct(b">I") |
|
286 | versionformat = struct.Struct(b">I") | |
287 | versionformat_pack = versionformat.pack |
|
287 | versionformat_pack = versionformat.pack | |
288 | versionformat_unpack = versionformat.unpack |
|
288 | versionformat_unpack = versionformat.unpack | |
289 |
|
289 | |||
290 | # corresponds to uncompressed length of indexformatng (2 gigs, 4-byte |
|
290 | # corresponds to uncompressed length of indexformatng (2 gigs, 4-byte | |
291 | # signed integer) |
|
291 | # signed integer) | |
292 | _maxentrysize = 0x7FFFFFFF |
|
292 | _maxentrysize = 0x7FFFFFFF | |
293 |
|
293 | |||
294 |
|
294 | |||
295 | class revlogio(object): |
|
295 | class revlogio(object): | |
296 | def __init__(self): |
|
296 | def __init__(self): | |
297 | self.size = indexformatng.size |
|
297 | self.size = indexformatng.size | |
298 |
|
298 | |||
299 | def parseindex(self, data, inline): |
|
299 | def parseindex(self, data, inline): | |
300 | # call the C implementation to parse the index data |
|
300 | # call the C implementation to parse the index data | |
301 | index, cache = parsers.parse_index2(data, inline) |
|
301 | index, cache = parsers.parse_index2(data, inline) | |
302 |
return index, |
|
302 | return index, cache | |
303 |
|
303 | |||
304 | def packentry(self, entry, node, version, rev): |
|
304 | def packentry(self, entry, node, version, rev): | |
305 | p = indexformatng_pack(*entry) |
|
305 | p = indexformatng_pack(*entry) | |
306 | if rev == 0: |
|
306 | if rev == 0: | |
307 | p = versionformat_pack(version) + p[4:] |
|
307 | p = versionformat_pack(version) + p[4:] | |
308 | return p |
|
308 | return p | |
309 |
|
309 | |||
310 |
|
310 | |||
311 | class revlog(object): |
|
311 | class revlog(object): | |
312 | """ |
|
312 | """ | |
313 | the underlying revision storage object |
|
313 | the underlying revision storage object | |
314 |
|
314 | |||
315 | A revlog consists of two parts, an index and the revision data. |
|
315 | A revlog consists of two parts, an index and the revision data. | |
316 |
|
316 | |||
317 | The index is a file with a fixed record size containing |
|
317 | The index is a file with a fixed record size containing | |
318 | information on each revision, including its nodeid (hash), the |
|
318 | information on each revision, including its nodeid (hash), the | |
319 | nodeids of its parents, the position and offset of its data within |
|
319 | nodeids of its parents, the position and offset of its data within | |
320 | the data file, and the revision it's based on. Finally, each entry |
|
320 | the data file, and the revision it's based on. Finally, each entry | |
321 | contains a linkrev entry that can serve as a pointer to external |
|
321 | contains a linkrev entry that can serve as a pointer to external | |
322 | data. |
|
322 | data. | |
323 |
|
323 | |||
324 | The revision data itself is a linear collection of data chunks. |
|
324 | The revision data itself is a linear collection of data chunks. | |
325 | Each chunk represents a revision and is usually represented as a |
|
325 | Each chunk represents a revision and is usually represented as a | |
326 | delta against the previous chunk. To bound lookup time, runs of |
|
326 | delta against the previous chunk. To bound lookup time, runs of | |
327 | deltas are limited to about 2 times the length of the original |
|
327 | deltas are limited to about 2 times the length of the original | |
328 | version data. This makes retrieval of a version proportional to |
|
328 | version data. This makes retrieval of a version proportional to | |
329 | its size, or O(1) relative to the number of revisions. |
|
329 | its size, or O(1) relative to the number of revisions. | |
330 |
|
330 | |||
331 | Both pieces of the revlog are written to in an append-only |
|
331 | Both pieces of the revlog are written to in an append-only | |
332 | fashion, which means we never need to rewrite a file to insert or |
|
332 | fashion, which means we never need to rewrite a file to insert or | |
333 | remove data, and can use some simple techniques to avoid the need |
|
333 | remove data, and can use some simple techniques to avoid the need | |
334 | for locking while reading. |
|
334 | for locking while reading. | |
335 |
|
335 | |||
336 | If checkambig, indexfile is opened with checkambig=True at |
|
336 | If checkambig, indexfile is opened with checkambig=True at | |
337 | writing, to avoid file stat ambiguity. |
|
337 | writing, to avoid file stat ambiguity. | |
338 |
|
338 | |||
339 | If mmaplargeindex is True, and an mmapindexthreshold is set, the |
|
339 | If mmaplargeindex is True, and an mmapindexthreshold is set, the | |
340 | index will be mmapped rather than read if it is larger than the |
|
340 | index will be mmapped rather than read if it is larger than the | |
341 | configured threshold. |
|
341 | configured threshold. | |
342 |
|
342 | |||
343 | If censorable is True, the revlog can have censored revisions. |
|
343 | If censorable is True, the revlog can have censored revisions. | |
344 |
|
344 | |||
345 | If `upperboundcomp` is not None, this is the expected maximal gain from |
|
345 | If `upperboundcomp` is not None, this is the expected maximal gain from | |
346 | compression for the data content. |
|
346 | compression for the data content. | |
347 | """ |
|
347 | """ | |
348 |
|
348 | |||
349 | _flagserrorclass = error.RevlogError |
|
349 | _flagserrorclass = error.RevlogError | |
350 |
|
350 | |||
351 | def __init__( |
|
351 | def __init__( | |
352 | self, |
|
352 | self, | |
353 | opener, |
|
353 | opener, | |
354 | indexfile, |
|
354 | indexfile, | |
355 | datafile=None, |
|
355 | datafile=None, | |
356 | checkambig=False, |
|
356 | checkambig=False, | |
357 | mmaplargeindex=False, |
|
357 | mmaplargeindex=False, | |
358 | censorable=False, |
|
358 | censorable=False, | |
359 | upperboundcomp=None, |
|
359 | upperboundcomp=None, | |
360 | ): |
|
360 | ): | |
361 | """ |
|
361 | """ | |
362 | create a revlog object |
|
362 | create a revlog object | |
363 |
|
363 | |||
364 | opener is a function that abstracts the file opening operation |
|
364 | opener is a function that abstracts the file opening operation | |
365 | and can be used to implement COW semantics or the like. |
|
365 | and can be used to implement COW semantics or the like. | |
366 |
|
366 | |||
367 | """ |
|
367 | """ | |
368 | self.upperboundcomp = upperboundcomp |
|
368 | self.upperboundcomp = upperboundcomp | |
369 | self.indexfile = indexfile |
|
369 | self.indexfile = indexfile | |
370 | self.datafile = datafile or (indexfile[:-2] + b".d") |
|
370 | self.datafile = datafile or (indexfile[:-2] + b".d") | |
371 | self.opener = opener |
|
371 | self.opener = opener | |
372 | # When True, indexfile is opened with checkambig=True at writing, to |
|
372 | # When True, indexfile is opened with checkambig=True at writing, to | |
373 | # avoid file stat ambiguity. |
|
373 | # avoid file stat ambiguity. | |
374 | self._checkambig = checkambig |
|
374 | self._checkambig = checkambig | |
375 | self._mmaplargeindex = mmaplargeindex |
|
375 | self._mmaplargeindex = mmaplargeindex | |
376 | self._censorable = censorable |
|
376 | self._censorable = censorable | |
377 | # 3-tuple of (node, rev, text) for a raw revision. |
|
377 | # 3-tuple of (node, rev, text) for a raw revision. | |
378 | self._revisioncache = None |
|
378 | self._revisioncache = None | |
379 | # Maps rev to chain base rev. |
|
379 | # Maps rev to chain base rev. | |
380 | self._chainbasecache = util.lrucachedict(100) |
|
380 | self._chainbasecache = util.lrucachedict(100) | |
381 | # 2-tuple of (offset, data) of raw data from the revlog at an offset. |
|
381 | # 2-tuple of (offset, data) of raw data from the revlog at an offset. | |
382 | self._chunkcache = (0, b'') |
|
382 | self._chunkcache = (0, b'') | |
383 | # How much data to read and cache into the raw revlog data cache. |
|
383 | # How much data to read and cache into the raw revlog data cache. | |
384 | self._chunkcachesize = 65536 |
|
384 | self._chunkcachesize = 65536 | |
385 | self._maxchainlen = None |
|
385 | self._maxchainlen = None | |
386 | self._deltabothparents = True |
|
386 | self._deltabothparents = True | |
387 | self.index = None |
|
387 | self.index = None | |
388 | # Mapping of partial identifiers to full nodes. |
|
388 | # Mapping of partial identifiers to full nodes. | |
389 | self._pcache = {} |
|
389 | self._pcache = {} | |
390 | # Mapping of revision integer to full node. |
|
390 | # Mapping of revision integer to full node. | |
391 | self._nodecache = None |
|
391 | self._nodecache = None | |
392 | self._nodepos = None |
|
392 | self._nodepos = None | |
393 | self._compengine = b'zlib' |
|
393 | self._compengine = b'zlib' | |
394 | self._compengineopts = {} |
|
394 | self._compengineopts = {} | |
395 | self._maxdeltachainspan = -1 |
|
395 | self._maxdeltachainspan = -1 | |
396 | self._withsparseread = False |
|
396 | self._withsparseread = False | |
397 | self._sparserevlog = False |
|
397 | self._sparserevlog = False | |
398 | self._srdensitythreshold = 0.50 |
|
398 | self._srdensitythreshold = 0.50 | |
399 | self._srmingapsize = 262144 |
|
399 | self._srmingapsize = 262144 | |
400 |
|
400 | |||
401 | # Make copy of flag processors so each revlog instance can support |
|
401 | # Make copy of flag processors so each revlog instance can support | |
402 | # custom flags. |
|
402 | # custom flags. | |
403 | self._flagprocessors = dict(flagutil.flagprocessors) |
|
403 | self._flagprocessors = dict(flagutil.flagprocessors) | |
404 |
|
404 | |||
405 | # 2-tuple of file handles being used for active writing. |
|
405 | # 2-tuple of file handles being used for active writing. | |
406 | self._writinghandles = None |
|
406 | self._writinghandles = None | |
407 |
|
407 | |||
408 | self._loadindex() |
|
408 | self._loadindex() | |
409 |
|
409 | |||
410 | def _loadindex(self): |
|
410 | def _loadindex(self): | |
411 | mmapindexthreshold = None |
|
411 | mmapindexthreshold = None | |
412 | opts = self.opener.options |
|
412 | opts = self.opener.options | |
413 |
|
413 | |||
414 | if b'revlogv2' in opts: |
|
414 | if b'revlogv2' in opts: | |
415 | newversionflags = REVLOGV2 | FLAG_INLINE_DATA |
|
415 | newversionflags = REVLOGV2 | FLAG_INLINE_DATA | |
416 | elif b'revlogv1' in opts: |
|
416 | elif b'revlogv1' in opts: | |
417 | newversionflags = REVLOGV1 | FLAG_INLINE_DATA |
|
417 | newversionflags = REVLOGV1 | FLAG_INLINE_DATA | |
418 | if b'generaldelta' in opts: |
|
418 | if b'generaldelta' in opts: | |
419 | newversionflags |= FLAG_GENERALDELTA |
|
419 | newversionflags |= FLAG_GENERALDELTA | |
420 | elif b'revlogv0' in self.opener.options: |
|
420 | elif b'revlogv0' in self.opener.options: | |
421 | newversionflags = REVLOGV0 |
|
421 | newversionflags = REVLOGV0 | |
422 | else: |
|
422 | else: | |
423 | newversionflags = REVLOG_DEFAULT_VERSION |
|
423 | newversionflags = REVLOG_DEFAULT_VERSION | |
424 |
|
424 | |||
425 | if b'chunkcachesize' in opts: |
|
425 | if b'chunkcachesize' in opts: | |
426 | self._chunkcachesize = opts[b'chunkcachesize'] |
|
426 | self._chunkcachesize = opts[b'chunkcachesize'] | |
427 | if b'maxchainlen' in opts: |
|
427 | if b'maxchainlen' in opts: | |
428 | self._maxchainlen = opts[b'maxchainlen'] |
|
428 | self._maxchainlen = opts[b'maxchainlen'] | |
429 | if b'deltabothparents' in opts: |
|
429 | if b'deltabothparents' in opts: | |
430 | self._deltabothparents = opts[b'deltabothparents'] |
|
430 | self._deltabothparents = opts[b'deltabothparents'] | |
431 | self._lazydelta = bool(opts.get(b'lazydelta', True)) |
|
431 | self._lazydelta = bool(opts.get(b'lazydelta', True)) | |
432 | self._lazydeltabase = False |
|
432 | self._lazydeltabase = False | |
433 | if self._lazydelta: |
|
433 | if self._lazydelta: | |
434 | self._lazydeltabase = bool(opts.get(b'lazydeltabase', False)) |
|
434 | self._lazydeltabase = bool(opts.get(b'lazydeltabase', False)) | |
435 | if b'compengine' in opts: |
|
435 | if b'compengine' in opts: | |
436 | self._compengine = opts[b'compengine'] |
|
436 | self._compengine = opts[b'compengine'] | |
437 | if b'zlib.level' in opts: |
|
437 | if b'zlib.level' in opts: | |
438 | self._compengineopts[b'zlib.level'] = opts[b'zlib.level'] |
|
438 | self._compengineopts[b'zlib.level'] = opts[b'zlib.level'] | |
439 | if b'zstd.level' in opts: |
|
439 | if b'zstd.level' in opts: | |
440 | self._compengineopts[b'zstd.level'] = opts[b'zstd.level'] |
|
440 | self._compengineopts[b'zstd.level'] = opts[b'zstd.level'] | |
441 | if b'maxdeltachainspan' in opts: |
|
441 | if b'maxdeltachainspan' in opts: | |
442 | self._maxdeltachainspan = opts[b'maxdeltachainspan'] |
|
442 | self._maxdeltachainspan = opts[b'maxdeltachainspan'] | |
443 | if self._mmaplargeindex and b'mmapindexthreshold' in opts: |
|
443 | if self._mmaplargeindex and b'mmapindexthreshold' in opts: | |
444 | mmapindexthreshold = opts[b'mmapindexthreshold'] |
|
444 | mmapindexthreshold = opts[b'mmapindexthreshold'] | |
445 | self.hassidedata = bool(opts.get(b'side-data', False)) |
|
445 | self.hassidedata = bool(opts.get(b'side-data', False)) | |
446 | if self.hassidedata: |
|
446 | if self.hassidedata: | |
447 | self._flagprocessors[REVIDX_SIDEDATA] = sidedatautil.processors |
|
447 | self._flagprocessors[REVIDX_SIDEDATA] = sidedatautil.processors | |
448 | self._sparserevlog = bool(opts.get(b'sparse-revlog', False)) |
|
448 | self._sparserevlog = bool(opts.get(b'sparse-revlog', False)) | |
449 | withsparseread = bool(opts.get(b'with-sparse-read', False)) |
|
449 | withsparseread = bool(opts.get(b'with-sparse-read', False)) | |
450 | # sparse-revlog forces sparse-read |
|
450 | # sparse-revlog forces sparse-read | |
451 | self._withsparseread = self._sparserevlog or withsparseread |
|
451 | self._withsparseread = self._sparserevlog or withsparseread | |
452 | if b'sparse-read-density-threshold' in opts: |
|
452 | if b'sparse-read-density-threshold' in opts: | |
453 | self._srdensitythreshold = opts[b'sparse-read-density-threshold'] |
|
453 | self._srdensitythreshold = opts[b'sparse-read-density-threshold'] | |
454 | if b'sparse-read-min-gap-size' in opts: |
|
454 | if b'sparse-read-min-gap-size' in opts: | |
455 | self._srmingapsize = opts[b'sparse-read-min-gap-size'] |
|
455 | self._srmingapsize = opts[b'sparse-read-min-gap-size'] | |
456 | if opts.get(b'enableellipsis'): |
|
456 | if opts.get(b'enableellipsis'): | |
457 | self._flagprocessors[REVIDX_ELLIPSIS] = ellipsisprocessor |
|
457 | self._flagprocessors[REVIDX_ELLIPSIS] = ellipsisprocessor | |
458 |
|
458 | |||
459 | # revlog v0 doesn't have flag processors |
|
459 | # revlog v0 doesn't have flag processors | |
460 | for flag, processor in pycompat.iteritems( |
|
460 | for flag, processor in pycompat.iteritems( | |
461 | opts.get(b'flagprocessors', {}) |
|
461 | opts.get(b'flagprocessors', {}) | |
462 | ): |
|
462 | ): | |
463 | flagutil.insertflagprocessor(flag, processor, self._flagprocessors) |
|
463 | flagutil.insertflagprocessor(flag, processor, self._flagprocessors) | |
464 |
|
464 | |||
465 | if self._chunkcachesize <= 0: |
|
465 | if self._chunkcachesize <= 0: | |
466 | raise error.RevlogError( |
|
466 | raise error.RevlogError( | |
467 | _(b'revlog chunk cache size %r is not greater than 0') |
|
467 | _(b'revlog chunk cache size %r is not greater than 0') | |
468 | % self._chunkcachesize |
|
468 | % self._chunkcachesize | |
469 | ) |
|
469 | ) | |
470 | elif self._chunkcachesize & (self._chunkcachesize - 1): |
|
470 | elif self._chunkcachesize & (self._chunkcachesize - 1): | |
471 | raise error.RevlogError( |
|
471 | raise error.RevlogError( | |
472 | _(b'revlog chunk cache size %r is not a power of 2') |
|
472 | _(b'revlog chunk cache size %r is not a power of 2') | |
473 | % self._chunkcachesize |
|
473 | % self._chunkcachesize | |
474 | ) |
|
474 | ) | |
475 |
|
475 | |||
476 | indexdata = b'' |
|
476 | indexdata = b'' | |
477 | self._initempty = True |
|
477 | self._initempty = True | |
478 | try: |
|
478 | try: | |
479 | with self._indexfp() as f: |
|
479 | with self._indexfp() as f: | |
480 | if ( |
|
480 | if ( | |
481 | mmapindexthreshold is not None |
|
481 | mmapindexthreshold is not None | |
482 | and self.opener.fstat(f).st_size >= mmapindexthreshold |
|
482 | and self.opener.fstat(f).st_size >= mmapindexthreshold | |
483 | ): |
|
483 | ): | |
484 | # TODO: should .close() to release resources without |
|
484 | # TODO: should .close() to release resources without | |
485 | # relying on Python GC |
|
485 | # relying on Python GC | |
486 | indexdata = util.buffer(util.mmapread(f)) |
|
486 | indexdata = util.buffer(util.mmapread(f)) | |
487 | else: |
|
487 | else: | |
488 | indexdata = f.read() |
|
488 | indexdata = f.read() | |
489 | if len(indexdata) > 0: |
|
489 | if len(indexdata) > 0: | |
490 | versionflags = versionformat_unpack(indexdata[:4])[0] |
|
490 | versionflags = versionformat_unpack(indexdata[:4])[0] | |
491 | self._initempty = False |
|
491 | self._initempty = False | |
492 | else: |
|
492 | else: | |
493 | versionflags = newversionflags |
|
493 | versionflags = newversionflags | |
494 | except IOError as inst: |
|
494 | except IOError as inst: | |
495 | if inst.errno != errno.ENOENT: |
|
495 | if inst.errno != errno.ENOENT: | |
496 | raise |
|
496 | raise | |
497 |
|
497 | |||
498 | versionflags = newversionflags |
|
498 | versionflags = newversionflags | |
499 |
|
499 | |||
500 | self.version = versionflags |
|
500 | self.version = versionflags | |
501 |
|
501 | |||
502 | flags = versionflags & ~0xFFFF |
|
502 | flags = versionflags & ~0xFFFF | |
503 | fmt = versionflags & 0xFFFF |
|
503 | fmt = versionflags & 0xFFFF | |
504 |
|
504 | |||
505 | if fmt == REVLOGV0: |
|
505 | if fmt == REVLOGV0: | |
506 | if flags: |
|
506 | if flags: | |
507 | raise error.RevlogError( |
|
507 | raise error.RevlogError( | |
508 | _(b'unknown flags (%#04x) in version %d revlog %s') |
|
508 | _(b'unknown flags (%#04x) in version %d revlog %s') | |
509 | % (flags >> 16, fmt, self.indexfile) |
|
509 | % (flags >> 16, fmt, self.indexfile) | |
510 | ) |
|
510 | ) | |
511 |
|
511 | |||
512 | self._inline = False |
|
512 | self._inline = False | |
513 | self._generaldelta = False |
|
513 | self._generaldelta = False | |
514 |
|
514 | |||
515 | elif fmt == REVLOGV1: |
|
515 | elif fmt == REVLOGV1: | |
516 | if flags & ~REVLOGV1_FLAGS: |
|
516 | if flags & ~REVLOGV1_FLAGS: | |
517 | raise error.RevlogError( |
|
517 | raise error.RevlogError( | |
518 | _(b'unknown flags (%#04x) in version %d revlog %s') |
|
518 | _(b'unknown flags (%#04x) in version %d revlog %s') | |
519 | % (flags >> 16, fmt, self.indexfile) |
|
519 | % (flags >> 16, fmt, self.indexfile) | |
520 | ) |
|
520 | ) | |
521 |
|
521 | |||
522 | self._inline = versionflags & FLAG_INLINE_DATA |
|
522 | self._inline = versionflags & FLAG_INLINE_DATA | |
523 | self._generaldelta = versionflags & FLAG_GENERALDELTA |
|
523 | self._generaldelta = versionflags & FLAG_GENERALDELTA | |
524 |
|
524 | |||
525 | elif fmt == REVLOGV2: |
|
525 | elif fmt == REVLOGV2: | |
526 | if flags & ~REVLOGV2_FLAGS: |
|
526 | if flags & ~REVLOGV2_FLAGS: | |
527 | raise error.RevlogError( |
|
527 | raise error.RevlogError( | |
528 | _(b'unknown flags (%#04x) in version %d revlog %s') |
|
528 | _(b'unknown flags (%#04x) in version %d revlog %s') | |
529 | % (flags >> 16, fmt, self.indexfile) |
|
529 | % (flags >> 16, fmt, self.indexfile) | |
530 | ) |
|
530 | ) | |
531 |
|
531 | |||
532 | self._inline = versionflags & FLAG_INLINE_DATA |
|
532 | self._inline = versionflags & FLAG_INLINE_DATA | |
533 | # generaldelta implied by version 2 revlogs. |
|
533 | # generaldelta implied by version 2 revlogs. | |
534 | self._generaldelta = True |
|
534 | self._generaldelta = True | |
535 |
|
535 | |||
536 | else: |
|
536 | else: | |
537 | raise error.RevlogError( |
|
537 | raise error.RevlogError( | |
538 | _(b'unknown version (%d) in revlog %s') % (fmt, self.indexfile) |
|
538 | _(b'unknown version (%d) in revlog %s') % (fmt, self.indexfile) | |
539 | ) |
|
539 | ) | |
540 | # sparse-revlog can't be on without general-delta (issue6056) |
|
540 | # sparse-revlog can't be on without general-delta (issue6056) | |
541 | if not self._generaldelta: |
|
541 | if not self._generaldelta: | |
542 | self._sparserevlog = False |
|
542 | self._sparserevlog = False | |
543 |
|
543 | |||
544 | self._storedeltachains = True |
|
544 | self._storedeltachains = True | |
545 |
|
545 | |||
546 | self._io = revlogio() |
|
546 | self._io = revlogio() | |
547 | if self.version == REVLOGV0: |
|
547 | if self.version == REVLOGV0: | |
548 | self._io = revlogoldio() |
|
548 | self._io = revlogoldio() | |
549 | try: |
|
549 | try: | |
550 | d = self._io.parseindex(indexdata, self._inline) |
|
550 | d = self._io.parseindex(indexdata, self._inline) | |
551 | except (ValueError, IndexError): |
|
551 | except (ValueError, IndexError): | |
552 | raise error.RevlogError( |
|
552 | raise error.RevlogError( | |
553 | _(b"index %s is corrupted") % self.indexfile |
|
553 | _(b"index %s is corrupted") % self.indexfile | |
554 | ) |
|
554 | ) | |
555 |
self.index |
|
555 | self.index, self._chunkcache = d | |
556 | self.nodemap = self._nodecache = nodemap |
|
556 | self.nodemap = self._nodecache = self.index.nodemap | |
557 | if not self._chunkcache: |
|
557 | if not self._chunkcache: | |
558 | self._chunkclear() |
|
558 | self._chunkclear() | |
559 | # revnum -> (chain-length, sum-delta-length) |
|
559 | # revnum -> (chain-length, sum-delta-length) | |
560 | self._chaininfocache = {} |
|
560 | self._chaininfocache = {} | |
561 | # revlog header -> revlog compressor |
|
561 | # revlog header -> revlog compressor | |
562 | self._decompressors = {} |
|
562 | self._decompressors = {} | |
563 |
|
563 | |||
564 | @util.propertycache |
|
564 | @util.propertycache | |
565 | def _compressor(self): |
|
565 | def _compressor(self): | |
566 | engine = util.compengines[self._compengine] |
|
566 | engine = util.compengines[self._compengine] | |
567 | return engine.revlogcompressor(self._compengineopts) |
|
567 | return engine.revlogcompressor(self._compengineopts) | |
568 |
|
568 | |||
569 | def _indexfp(self, mode=b'r'): |
|
569 | def _indexfp(self, mode=b'r'): | |
570 | """file object for the revlog's index file""" |
|
570 | """file object for the revlog's index file""" | |
571 | args = {'mode': mode} |
|
571 | args = {'mode': mode} | |
572 | if mode != b'r': |
|
572 | if mode != b'r': | |
573 | args['checkambig'] = self._checkambig |
|
573 | args['checkambig'] = self._checkambig | |
574 | if mode == b'w': |
|
574 | if mode == b'w': | |
575 | args['atomictemp'] = True |
|
575 | args['atomictemp'] = True | |
576 | return self.opener(self.indexfile, **args) |
|
576 | return self.opener(self.indexfile, **args) | |
577 |
|
577 | |||
578 | def _datafp(self, mode=b'r'): |
|
578 | def _datafp(self, mode=b'r'): | |
579 | """file object for the revlog's data file""" |
|
579 | """file object for the revlog's data file""" | |
580 | return self.opener(self.datafile, mode=mode) |
|
580 | return self.opener(self.datafile, mode=mode) | |
581 |
|
581 | |||
582 | @contextlib.contextmanager |
|
582 | @contextlib.contextmanager | |
583 | def _datareadfp(self, existingfp=None): |
|
583 | def _datareadfp(self, existingfp=None): | |
584 | """file object suitable to read data""" |
|
584 | """file object suitable to read data""" | |
585 | # Use explicit file handle, if given. |
|
585 | # Use explicit file handle, if given. | |
586 | if existingfp is not None: |
|
586 | if existingfp is not None: | |
587 | yield existingfp |
|
587 | yield existingfp | |
588 |
|
588 | |||
589 | # Use a file handle being actively used for writes, if available. |
|
589 | # Use a file handle being actively used for writes, if available. | |
590 | # There is some danger to doing this because reads will seek the |
|
590 | # There is some danger to doing this because reads will seek the | |
591 | # file. However, _writeentry() performs a SEEK_END before all writes, |
|
591 | # file. However, _writeentry() performs a SEEK_END before all writes, | |
592 | # so we should be safe. |
|
592 | # so we should be safe. | |
593 | elif self._writinghandles: |
|
593 | elif self._writinghandles: | |
594 | if self._inline: |
|
594 | if self._inline: | |
595 | yield self._writinghandles[0] |
|
595 | yield self._writinghandles[0] | |
596 | else: |
|
596 | else: | |
597 | yield self._writinghandles[1] |
|
597 | yield self._writinghandles[1] | |
598 |
|
598 | |||
599 | # Otherwise open a new file handle. |
|
599 | # Otherwise open a new file handle. | |
600 | else: |
|
600 | else: | |
601 | if self._inline: |
|
601 | if self._inline: | |
602 | func = self._indexfp |
|
602 | func = self._indexfp | |
603 | else: |
|
603 | else: | |
604 | func = self._datafp |
|
604 | func = self._datafp | |
605 | with func() as fp: |
|
605 | with func() as fp: | |
606 | yield fp |
|
606 | yield fp | |
607 |
|
607 | |||
608 | def tiprev(self): |
|
608 | def tiprev(self): | |
609 | return len(self.index) - 1 |
|
609 | return len(self.index) - 1 | |
610 |
|
610 | |||
611 | def tip(self): |
|
611 | def tip(self): | |
612 | return self.node(self.tiprev()) |
|
612 | return self.node(self.tiprev()) | |
613 |
|
613 | |||
614 | def __contains__(self, rev): |
|
614 | def __contains__(self, rev): | |
615 | return 0 <= rev < len(self) |
|
615 | return 0 <= rev < len(self) | |
616 |
|
616 | |||
617 | def __len__(self): |
|
617 | def __len__(self): | |
618 | return len(self.index) |
|
618 | return len(self.index) | |
619 |
|
619 | |||
620 | def __iter__(self): |
|
620 | def __iter__(self): | |
621 | return iter(pycompat.xrange(len(self))) |
|
621 | return iter(pycompat.xrange(len(self))) | |
622 |
|
622 | |||
623 | def revs(self, start=0, stop=None): |
|
623 | def revs(self, start=0, stop=None): | |
624 | """iterate over all rev in this revlog (from start to stop)""" |
|
624 | """iterate over all rev in this revlog (from start to stop)""" | |
625 | return storageutil.iterrevs(len(self), start=start, stop=stop) |
|
625 | return storageutil.iterrevs(len(self), start=start, stop=stop) | |
626 |
|
626 | |||
627 | @util.propertycache |
|
627 | @util.propertycache | |
628 | def nodemap(self): |
|
628 | def nodemap(self): | |
629 | if self.index: |
|
629 | if self.index: | |
630 | # populate mapping down to the initial node |
|
630 | # populate mapping down to the initial node | |
631 | node0 = self.index[0][7] # get around changelog filtering |
|
631 | node0 = self.index[0][7] # get around changelog filtering | |
632 | self.rev(node0) |
|
632 | self.rev(node0) | |
633 | return self._nodecache |
|
633 | return self._nodecache | |
634 |
|
634 | |||
635 | def hasnode(self, node): |
|
635 | def hasnode(self, node): | |
636 | try: |
|
636 | try: | |
637 | self.rev(node) |
|
637 | self.rev(node) | |
638 | return True |
|
638 | return True | |
639 | except KeyError: |
|
639 | except KeyError: | |
640 | return False |
|
640 | return False | |
641 |
|
641 | |||
642 | def candelta(self, baserev, rev): |
|
642 | def candelta(self, baserev, rev): | |
643 | """whether two revisions (baserev, rev) can be delta-ed or not""" |
|
643 | """whether two revisions (baserev, rev) can be delta-ed or not""" | |
644 | # Disable delta if either rev requires a content-changing flag |
|
644 | # Disable delta if either rev requires a content-changing flag | |
645 | # processor (ex. LFS). This is because such flag processor can alter |
|
645 | # processor (ex. LFS). This is because such flag processor can alter | |
646 | # the rawtext content that the delta will be based on, and two clients |
|
646 | # the rawtext content that the delta will be based on, and two clients | |
647 | # could have a same revlog node with different flags (i.e. different |
|
647 | # could have a same revlog node with different flags (i.e. different | |
648 | # rawtext contents) and the delta could be incompatible. |
|
648 | # rawtext contents) and the delta could be incompatible. | |
649 | if (self.flags(baserev) & REVIDX_RAWTEXT_CHANGING_FLAGS) or ( |
|
649 | if (self.flags(baserev) & REVIDX_RAWTEXT_CHANGING_FLAGS) or ( | |
650 | self.flags(rev) & REVIDX_RAWTEXT_CHANGING_FLAGS |
|
650 | self.flags(rev) & REVIDX_RAWTEXT_CHANGING_FLAGS | |
651 | ): |
|
651 | ): | |
652 | return False |
|
652 | return False | |
653 | return True |
|
653 | return True | |
654 |
|
654 | |||
655 | def clearcaches(self): |
|
655 | def clearcaches(self): | |
656 | self._revisioncache = None |
|
656 | self._revisioncache = None | |
657 | self._chainbasecache.clear() |
|
657 | self._chainbasecache.clear() | |
658 | self._chunkcache = (0, b'') |
|
658 | self._chunkcache = (0, b'') | |
659 | self._pcache = {} |
|
659 | self._pcache = {} | |
660 | self.index.clearcaches() |
|
660 | self.index.clearcaches() | |
661 |
|
661 | |||
662 | def rev(self, node): |
|
662 | def rev(self, node): | |
663 | try: |
|
663 | try: | |
664 | return self._nodecache[node] |
|
664 | return self._nodecache[node] | |
665 | except TypeError: |
|
665 | except TypeError: | |
666 | raise |
|
666 | raise | |
667 | except error.RevlogError: |
|
667 | except error.RevlogError: | |
668 | # parsers.c radix tree lookup failed |
|
668 | # parsers.c radix tree lookup failed | |
669 | if node == wdirid or node in wdirfilenodeids: |
|
669 | if node == wdirid or node in wdirfilenodeids: | |
670 | raise error.WdirUnsupported |
|
670 | raise error.WdirUnsupported | |
671 | raise error.LookupError(node, self.indexfile, _(b'no node')) |
|
671 | raise error.LookupError(node, self.indexfile, _(b'no node')) | |
672 |
|
672 | |||
673 | # Accessors for index entries. |
|
673 | # Accessors for index entries. | |
674 |
|
674 | |||
675 | # First tuple entry is 8 bytes. First 6 bytes are offset. Last 2 bytes |
|
675 | # First tuple entry is 8 bytes. First 6 bytes are offset. Last 2 bytes | |
676 | # are flags. |
|
676 | # are flags. | |
677 | def start(self, rev): |
|
677 | def start(self, rev): | |
678 | return int(self.index[rev][0] >> 16) |
|
678 | return int(self.index[rev][0] >> 16) | |
679 |
|
679 | |||
680 | def flags(self, rev): |
|
680 | def flags(self, rev): | |
681 | return self.index[rev][0] & 0xFFFF |
|
681 | return self.index[rev][0] & 0xFFFF | |
682 |
|
682 | |||
683 | def length(self, rev): |
|
683 | def length(self, rev): | |
684 | return self.index[rev][1] |
|
684 | return self.index[rev][1] | |
685 |
|
685 | |||
686 | def rawsize(self, rev): |
|
686 | def rawsize(self, rev): | |
687 | """return the length of the uncompressed text for a given revision""" |
|
687 | """return the length of the uncompressed text for a given revision""" | |
688 | l = self.index[rev][2] |
|
688 | l = self.index[rev][2] | |
689 | if l >= 0: |
|
689 | if l >= 0: | |
690 | return l |
|
690 | return l | |
691 |
|
691 | |||
692 | t = self.rawdata(rev) |
|
692 | t = self.rawdata(rev) | |
693 | return len(t) |
|
693 | return len(t) | |
694 |
|
694 | |||
695 | def size(self, rev): |
|
695 | def size(self, rev): | |
696 | """length of non-raw text (processed by a "read" flag processor)""" |
|
696 | """length of non-raw text (processed by a "read" flag processor)""" | |
697 | # fast path: if no "read" flag processor could change the content, |
|
697 | # fast path: if no "read" flag processor could change the content, | |
698 | # size is rawsize. note: ELLIPSIS is known to not change the content. |
|
698 | # size is rawsize. note: ELLIPSIS is known to not change the content. | |
699 | flags = self.flags(rev) |
|
699 | flags = self.flags(rev) | |
700 | if flags & (flagutil.REVIDX_KNOWN_FLAGS ^ REVIDX_ELLIPSIS) == 0: |
|
700 | if flags & (flagutil.REVIDX_KNOWN_FLAGS ^ REVIDX_ELLIPSIS) == 0: | |
701 | return self.rawsize(rev) |
|
701 | return self.rawsize(rev) | |
702 |
|
702 | |||
703 | return len(self.revision(rev, raw=False)) |
|
703 | return len(self.revision(rev, raw=False)) | |
704 |
|
704 | |||
705 | def chainbase(self, rev): |
|
705 | def chainbase(self, rev): | |
706 | base = self._chainbasecache.get(rev) |
|
706 | base = self._chainbasecache.get(rev) | |
707 | if base is not None: |
|
707 | if base is not None: | |
708 | return base |
|
708 | return base | |
709 |
|
709 | |||
710 | index = self.index |
|
710 | index = self.index | |
711 | iterrev = rev |
|
711 | iterrev = rev | |
712 | base = index[iterrev][3] |
|
712 | base = index[iterrev][3] | |
713 | while base != iterrev: |
|
713 | while base != iterrev: | |
714 | iterrev = base |
|
714 | iterrev = base | |
715 | base = index[iterrev][3] |
|
715 | base = index[iterrev][3] | |
716 |
|
716 | |||
717 | self._chainbasecache[rev] = base |
|
717 | self._chainbasecache[rev] = base | |
718 | return base |
|
718 | return base | |
719 |
|
719 | |||
720 | def linkrev(self, rev): |
|
720 | def linkrev(self, rev): | |
721 | return self.index[rev][4] |
|
721 | return self.index[rev][4] | |
722 |
|
722 | |||
723 | def parentrevs(self, rev): |
|
723 | def parentrevs(self, rev): | |
724 | try: |
|
724 | try: | |
725 | entry = self.index[rev] |
|
725 | entry = self.index[rev] | |
726 | except IndexError: |
|
726 | except IndexError: | |
727 | if rev == wdirrev: |
|
727 | if rev == wdirrev: | |
728 | raise error.WdirUnsupported |
|
728 | raise error.WdirUnsupported | |
729 | raise |
|
729 | raise | |
730 |
|
730 | |||
731 | return entry[5], entry[6] |
|
731 | return entry[5], entry[6] | |
732 |
|
732 | |||
733 | # fast parentrevs(rev) where rev isn't filtered |
|
733 | # fast parentrevs(rev) where rev isn't filtered | |
734 | _uncheckedparentrevs = parentrevs |
|
734 | _uncheckedparentrevs = parentrevs | |
735 |
|
735 | |||
736 | def node(self, rev): |
|
736 | def node(self, rev): | |
737 | try: |
|
737 | try: | |
738 | return self.index[rev][7] |
|
738 | return self.index[rev][7] | |
739 | except IndexError: |
|
739 | except IndexError: | |
740 | if rev == wdirrev: |
|
740 | if rev == wdirrev: | |
741 | raise error.WdirUnsupported |
|
741 | raise error.WdirUnsupported | |
742 | raise |
|
742 | raise | |
743 |
|
743 | |||
744 | # Derived from index values. |
|
744 | # Derived from index values. | |
745 |
|
745 | |||
746 | def end(self, rev): |
|
746 | def end(self, rev): | |
747 | return self.start(rev) + self.length(rev) |
|
747 | return self.start(rev) + self.length(rev) | |
748 |
|
748 | |||
749 | def parents(self, node): |
|
749 | def parents(self, node): | |
750 | i = self.index |
|
750 | i = self.index | |
751 | d = i[self.rev(node)] |
|
751 | d = i[self.rev(node)] | |
752 | return i[d[5]][7], i[d[6]][7] # map revisions to nodes inline |
|
752 | return i[d[5]][7], i[d[6]][7] # map revisions to nodes inline | |
753 |
|
753 | |||
754 | def chainlen(self, rev): |
|
754 | def chainlen(self, rev): | |
755 | return self._chaininfo(rev)[0] |
|
755 | return self._chaininfo(rev)[0] | |
756 |
|
756 | |||
757 | def _chaininfo(self, rev): |
|
757 | def _chaininfo(self, rev): | |
758 | chaininfocache = self._chaininfocache |
|
758 | chaininfocache = self._chaininfocache | |
759 | if rev in chaininfocache: |
|
759 | if rev in chaininfocache: | |
760 | return chaininfocache[rev] |
|
760 | return chaininfocache[rev] | |
761 | index = self.index |
|
761 | index = self.index | |
762 | generaldelta = self._generaldelta |
|
762 | generaldelta = self._generaldelta | |
763 | iterrev = rev |
|
763 | iterrev = rev | |
764 | e = index[iterrev] |
|
764 | e = index[iterrev] | |
765 | clen = 0 |
|
765 | clen = 0 | |
766 | compresseddeltalen = 0 |
|
766 | compresseddeltalen = 0 | |
767 | while iterrev != e[3]: |
|
767 | while iterrev != e[3]: | |
768 | clen += 1 |
|
768 | clen += 1 | |
769 | compresseddeltalen += e[1] |
|
769 | compresseddeltalen += e[1] | |
770 | if generaldelta: |
|
770 | if generaldelta: | |
771 | iterrev = e[3] |
|
771 | iterrev = e[3] | |
772 | else: |
|
772 | else: | |
773 | iterrev -= 1 |
|
773 | iterrev -= 1 | |
774 | if iterrev in chaininfocache: |
|
774 | if iterrev in chaininfocache: | |
775 | t = chaininfocache[iterrev] |
|
775 | t = chaininfocache[iterrev] | |
776 | clen += t[0] |
|
776 | clen += t[0] | |
777 | compresseddeltalen += t[1] |
|
777 | compresseddeltalen += t[1] | |
778 | break |
|
778 | break | |
779 | e = index[iterrev] |
|
779 | e = index[iterrev] | |
780 | else: |
|
780 | else: | |
781 | # Add text length of base since decompressing that also takes |
|
781 | # Add text length of base since decompressing that also takes | |
782 | # work. For cache hits the length is already included. |
|
782 | # work. For cache hits the length is already included. | |
783 | compresseddeltalen += e[1] |
|
783 | compresseddeltalen += e[1] | |
784 | r = (clen, compresseddeltalen) |
|
784 | r = (clen, compresseddeltalen) | |
785 | chaininfocache[rev] = r |
|
785 | chaininfocache[rev] = r | |
786 | return r |
|
786 | return r | |
787 |
|
787 | |||
788 | def _deltachain(self, rev, stoprev=None): |
|
788 | def _deltachain(self, rev, stoprev=None): | |
789 | """Obtain the delta chain for a revision. |
|
789 | """Obtain the delta chain for a revision. | |
790 |
|
790 | |||
791 | ``stoprev`` specifies a revision to stop at. If not specified, we |
|
791 | ``stoprev`` specifies a revision to stop at. If not specified, we | |
792 | stop at the base of the chain. |
|
792 | stop at the base of the chain. | |
793 |
|
793 | |||
794 | Returns a 2-tuple of (chain, stopped) where ``chain`` is a list of |
|
794 | Returns a 2-tuple of (chain, stopped) where ``chain`` is a list of | |
795 | revs in ascending order and ``stopped`` is a bool indicating whether |
|
795 | revs in ascending order and ``stopped`` is a bool indicating whether | |
796 | ``stoprev`` was hit. |
|
796 | ``stoprev`` was hit. | |
797 | """ |
|
797 | """ | |
798 | # Try C implementation. |
|
798 | # Try C implementation. | |
799 | try: |
|
799 | try: | |
800 | return self.index.deltachain(rev, stoprev, self._generaldelta) |
|
800 | return self.index.deltachain(rev, stoprev, self._generaldelta) | |
801 | except AttributeError: |
|
801 | except AttributeError: | |
802 | pass |
|
802 | pass | |
803 |
|
803 | |||
804 | chain = [] |
|
804 | chain = [] | |
805 |
|
805 | |||
806 | # Alias to prevent attribute lookup in tight loop. |
|
806 | # Alias to prevent attribute lookup in tight loop. | |
807 | index = self.index |
|
807 | index = self.index | |
808 | generaldelta = self._generaldelta |
|
808 | generaldelta = self._generaldelta | |
809 |
|
809 | |||
810 | iterrev = rev |
|
810 | iterrev = rev | |
811 | e = index[iterrev] |
|
811 | e = index[iterrev] | |
812 | while iterrev != e[3] and iterrev != stoprev: |
|
812 | while iterrev != e[3] and iterrev != stoprev: | |
813 | chain.append(iterrev) |
|
813 | chain.append(iterrev) | |
814 | if generaldelta: |
|
814 | if generaldelta: | |
815 | iterrev = e[3] |
|
815 | iterrev = e[3] | |
816 | else: |
|
816 | else: | |
817 | iterrev -= 1 |
|
817 | iterrev -= 1 | |
818 | e = index[iterrev] |
|
818 | e = index[iterrev] | |
819 |
|
819 | |||
820 | if iterrev == stoprev: |
|
820 | if iterrev == stoprev: | |
821 | stopped = True |
|
821 | stopped = True | |
822 | else: |
|
822 | else: | |
823 | chain.append(iterrev) |
|
823 | chain.append(iterrev) | |
824 | stopped = False |
|
824 | stopped = False | |
825 |
|
825 | |||
826 | chain.reverse() |
|
826 | chain.reverse() | |
827 | return chain, stopped |
|
827 | return chain, stopped | |
828 |
|
828 | |||
829 | def ancestors(self, revs, stoprev=0, inclusive=False): |
|
829 | def ancestors(self, revs, stoprev=0, inclusive=False): | |
830 | """Generate the ancestors of 'revs' in reverse revision order. |
|
830 | """Generate the ancestors of 'revs' in reverse revision order. | |
831 | Does not generate revs lower than stoprev. |
|
831 | Does not generate revs lower than stoprev. | |
832 |
|
832 | |||
833 | See the documentation for ancestor.lazyancestors for more details.""" |
|
833 | See the documentation for ancestor.lazyancestors for more details.""" | |
834 |
|
834 | |||
835 | # first, make sure start revisions aren't filtered |
|
835 | # first, make sure start revisions aren't filtered | |
836 | revs = list(revs) |
|
836 | revs = list(revs) | |
837 | checkrev = self.node |
|
837 | checkrev = self.node | |
838 | for r in revs: |
|
838 | for r in revs: | |
839 | checkrev(r) |
|
839 | checkrev(r) | |
840 | # and we're sure ancestors aren't filtered as well |
|
840 | # and we're sure ancestors aren't filtered as well | |
841 |
|
841 | |||
842 | if rustancestor is not None: |
|
842 | if rustancestor is not None: | |
843 | lazyancestors = rustancestor.LazyAncestors |
|
843 | lazyancestors = rustancestor.LazyAncestors | |
844 | arg = self.index |
|
844 | arg = self.index | |
845 | elif util.safehasattr(parsers, b'rustlazyancestors'): |
|
845 | elif util.safehasattr(parsers, b'rustlazyancestors'): | |
846 | lazyancestors = ancestor.rustlazyancestors |
|
846 | lazyancestors = ancestor.rustlazyancestors | |
847 | arg = self.index |
|
847 | arg = self.index | |
848 | else: |
|
848 | else: | |
849 | lazyancestors = ancestor.lazyancestors |
|
849 | lazyancestors = ancestor.lazyancestors | |
850 | arg = self._uncheckedparentrevs |
|
850 | arg = self._uncheckedparentrevs | |
851 | return lazyancestors(arg, revs, stoprev=stoprev, inclusive=inclusive) |
|
851 | return lazyancestors(arg, revs, stoprev=stoprev, inclusive=inclusive) | |
852 |
|
852 | |||
853 | def descendants(self, revs): |
|
853 | def descendants(self, revs): | |
854 | return dagop.descendantrevs(revs, self.revs, self.parentrevs) |
|
854 | return dagop.descendantrevs(revs, self.revs, self.parentrevs) | |
855 |
|
855 | |||
856 | def findcommonmissing(self, common=None, heads=None): |
|
856 | def findcommonmissing(self, common=None, heads=None): | |
857 | """Return a tuple of the ancestors of common and the ancestors of heads |
|
857 | """Return a tuple of the ancestors of common and the ancestors of heads | |
858 | that are not ancestors of common. In revset terminology, we return the |
|
858 | that are not ancestors of common. In revset terminology, we return the | |
859 | tuple: |
|
859 | tuple: | |
860 |
|
860 | |||
861 | ::common, (::heads) - (::common) |
|
861 | ::common, (::heads) - (::common) | |
862 |
|
862 | |||
863 | The list is sorted by revision number, meaning it is |
|
863 | The list is sorted by revision number, meaning it is | |
864 | topologically sorted. |
|
864 | topologically sorted. | |
865 |
|
865 | |||
866 | 'heads' and 'common' are both lists of node IDs. If heads is |
|
866 | 'heads' and 'common' are both lists of node IDs. If heads is | |
867 | not supplied, uses all of the revlog's heads. If common is not |
|
867 | not supplied, uses all of the revlog's heads. If common is not | |
868 | supplied, uses nullid.""" |
|
868 | supplied, uses nullid.""" | |
869 | if common is None: |
|
869 | if common is None: | |
870 | common = [nullid] |
|
870 | common = [nullid] | |
871 | if heads is None: |
|
871 | if heads is None: | |
872 | heads = self.heads() |
|
872 | heads = self.heads() | |
873 |
|
873 | |||
874 | common = [self.rev(n) for n in common] |
|
874 | common = [self.rev(n) for n in common] | |
875 | heads = [self.rev(n) for n in heads] |
|
875 | heads = [self.rev(n) for n in heads] | |
876 |
|
876 | |||
877 | # we want the ancestors, but inclusive |
|
877 | # we want the ancestors, but inclusive | |
878 | class lazyset(object): |
|
878 | class lazyset(object): | |
879 | def __init__(self, lazyvalues): |
|
879 | def __init__(self, lazyvalues): | |
880 | self.addedvalues = set() |
|
880 | self.addedvalues = set() | |
881 | self.lazyvalues = lazyvalues |
|
881 | self.lazyvalues = lazyvalues | |
882 |
|
882 | |||
883 | def __contains__(self, value): |
|
883 | def __contains__(self, value): | |
884 | return value in self.addedvalues or value in self.lazyvalues |
|
884 | return value in self.addedvalues or value in self.lazyvalues | |
885 |
|
885 | |||
886 | def __iter__(self): |
|
886 | def __iter__(self): | |
887 | added = self.addedvalues |
|
887 | added = self.addedvalues | |
888 | for r in added: |
|
888 | for r in added: | |
889 | yield r |
|
889 | yield r | |
890 | for r in self.lazyvalues: |
|
890 | for r in self.lazyvalues: | |
891 | if not r in added: |
|
891 | if not r in added: | |
892 | yield r |
|
892 | yield r | |
893 |
|
893 | |||
894 | def add(self, value): |
|
894 | def add(self, value): | |
895 | self.addedvalues.add(value) |
|
895 | self.addedvalues.add(value) | |
896 |
|
896 | |||
897 | def update(self, values): |
|
897 | def update(self, values): | |
898 | self.addedvalues.update(values) |
|
898 | self.addedvalues.update(values) | |
899 |
|
899 | |||
900 | has = lazyset(self.ancestors(common)) |
|
900 | has = lazyset(self.ancestors(common)) | |
901 | has.add(nullrev) |
|
901 | has.add(nullrev) | |
902 | has.update(common) |
|
902 | has.update(common) | |
903 |
|
903 | |||
904 | # take all ancestors from heads that aren't in has |
|
904 | # take all ancestors from heads that aren't in has | |
905 | missing = set() |
|
905 | missing = set() | |
906 | visit = collections.deque(r for r in heads if r not in has) |
|
906 | visit = collections.deque(r for r in heads if r not in has) | |
907 | while visit: |
|
907 | while visit: | |
908 | r = visit.popleft() |
|
908 | r = visit.popleft() | |
909 | if r in missing: |
|
909 | if r in missing: | |
910 | continue |
|
910 | continue | |
911 | else: |
|
911 | else: | |
912 | missing.add(r) |
|
912 | missing.add(r) | |
913 | for p in self.parentrevs(r): |
|
913 | for p in self.parentrevs(r): | |
914 | if p not in has: |
|
914 | if p not in has: | |
915 | visit.append(p) |
|
915 | visit.append(p) | |
916 | missing = list(missing) |
|
916 | missing = list(missing) | |
917 | missing.sort() |
|
917 | missing.sort() | |
918 | return has, [self.node(miss) for miss in missing] |
|
918 | return has, [self.node(miss) for miss in missing] | |
919 |
|
919 | |||
920 | def incrementalmissingrevs(self, common=None): |
|
920 | def incrementalmissingrevs(self, common=None): | |
921 | """Return an object that can be used to incrementally compute the |
|
921 | """Return an object that can be used to incrementally compute the | |
922 | revision numbers of the ancestors of arbitrary sets that are not |
|
922 | revision numbers of the ancestors of arbitrary sets that are not | |
923 | ancestors of common. This is an ancestor.incrementalmissingancestors |
|
923 | ancestors of common. This is an ancestor.incrementalmissingancestors | |
924 | object. |
|
924 | object. | |
925 |
|
925 | |||
926 | 'common' is a list of revision numbers. If common is not supplied, uses |
|
926 | 'common' is a list of revision numbers. If common is not supplied, uses | |
927 | nullrev. |
|
927 | nullrev. | |
928 | """ |
|
928 | """ | |
929 | if common is None: |
|
929 | if common is None: | |
930 | common = [nullrev] |
|
930 | common = [nullrev] | |
931 |
|
931 | |||
932 | if rustancestor is not None: |
|
932 | if rustancestor is not None: | |
933 | return rustancestor.MissingAncestors(self.index, common) |
|
933 | return rustancestor.MissingAncestors(self.index, common) | |
934 | return ancestor.incrementalmissingancestors(self.parentrevs, common) |
|
934 | return ancestor.incrementalmissingancestors(self.parentrevs, common) | |
935 |
|
935 | |||
936 | def findmissingrevs(self, common=None, heads=None): |
|
936 | def findmissingrevs(self, common=None, heads=None): | |
937 | """Return the revision numbers of the ancestors of heads that |
|
937 | """Return the revision numbers of the ancestors of heads that | |
938 | are not ancestors of common. |
|
938 | are not ancestors of common. | |
939 |
|
939 | |||
940 | More specifically, return a list of revision numbers corresponding to |
|
940 | More specifically, return a list of revision numbers corresponding to | |
941 | nodes N such that every N satisfies the following constraints: |
|
941 | nodes N such that every N satisfies the following constraints: | |
942 |
|
942 | |||
943 | 1. N is an ancestor of some node in 'heads' |
|
943 | 1. N is an ancestor of some node in 'heads' | |
944 | 2. N is not an ancestor of any node in 'common' |
|
944 | 2. N is not an ancestor of any node in 'common' | |
945 |
|
945 | |||
946 | The list is sorted by revision number, meaning it is |
|
946 | The list is sorted by revision number, meaning it is | |
947 | topologically sorted. |
|
947 | topologically sorted. | |
948 |
|
948 | |||
949 | 'heads' and 'common' are both lists of revision numbers. If heads is |
|
949 | 'heads' and 'common' are both lists of revision numbers. If heads is | |
950 | not supplied, uses all of the revlog's heads. If common is not |
|
950 | not supplied, uses all of the revlog's heads. If common is not | |
951 | supplied, uses nullid.""" |
|
951 | supplied, uses nullid.""" | |
952 | if common is None: |
|
952 | if common is None: | |
953 | common = [nullrev] |
|
953 | common = [nullrev] | |
954 | if heads is None: |
|
954 | if heads is None: | |
955 | heads = self.headrevs() |
|
955 | heads = self.headrevs() | |
956 |
|
956 | |||
957 | inc = self.incrementalmissingrevs(common=common) |
|
957 | inc = self.incrementalmissingrevs(common=common) | |
958 | return inc.missingancestors(heads) |
|
958 | return inc.missingancestors(heads) | |
959 |
|
959 | |||
960 | def findmissing(self, common=None, heads=None): |
|
960 | def findmissing(self, common=None, heads=None): | |
961 | """Return the ancestors of heads that are not ancestors of common. |
|
961 | """Return the ancestors of heads that are not ancestors of common. | |
962 |
|
962 | |||
963 | More specifically, return a list of nodes N such that every N |
|
963 | More specifically, return a list of nodes N such that every N | |
964 | satisfies the following constraints: |
|
964 | satisfies the following constraints: | |
965 |
|
965 | |||
966 | 1. N is an ancestor of some node in 'heads' |
|
966 | 1. N is an ancestor of some node in 'heads' | |
967 | 2. N is not an ancestor of any node in 'common' |
|
967 | 2. N is not an ancestor of any node in 'common' | |
968 |
|
968 | |||
969 | The list is sorted by revision number, meaning it is |
|
969 | The list is sorted by revision number, meaning it is | |
970 | topologically sorted. |
|
970 | topologically sorted. | |
971 |
|
971 | |||
972 | 'heads' and 'common' are both lists of node IDs. If heads is |
|
972 | 'heads' and 'common' are both lists of node IDs. If heads is | |
973 | not supplied, uses all of the revlog's heads. If common is not |
|
973 | not supplied, uses all of the revlog's heads. If common is not | |
974 | supplied, uses nullid.""" |
|
974 | supplied, uses nullid.""" | |
975 | if common is None: |
|
975 | if common is None: | |
976 | common = [nullid] |
|
976 | common = [nullid] | |
977 | if heads is None: |
|
977 | if heads is None: | |
978 | heads = self.heads() |
|
978 | heads = self.heads() | |
979 |
|
979 | |||
980 | common = [self.rev(n) for n in common] |
|
980 | common = [self.rev(n) for n in common] | |
981 | heads = [self.rev(n) for n in heads] |
|
981 | heads = [self.rev(n) for n in heads] | |
982 |
|
982 | |||
983 | inc = self.incrementalmissingrevs(common=common) |
|
983 | inc = self.incrementalmissingrevs(common=common) | |
984 | return [self.node(r) for r in inc.missingancestors(heads)] |
|
984 | return [self.node(r) for r in inc.missingancestors(heads)] | |
985 |
|
985 | |||
986 | def nodesbetween(self, roots=None, heads=None): |
|
986 | def nodesbetween(self, roots=None, heads=None): | |
987 | """Return a topological path from 'roots' to 'heads'. |
|
987 | """Return a topological path from 'roots' to 'heads'. | |
988 |
|
988 | |||
989 | Return a tuple (nodes, outroots, outheads) where 'nodes' is a |
|
989 | Return a tuple (nodes, outroots, outheads) where 'nodes' is a | |
990 | topologically sorted list of all nodes N that satisfy both of |
|
990 | topologically sorted list of all nodes N that satisfy both of | |
991 | these constraints: |
|
991 | these constraints: | |
992 |
|
992 | |||
993 | 1. N is a descendant of some node in 'roots' |
|
993 | 1. N is a descendant of some node in 'roots' | |
994 | 2. N is an ancestor of some node in 'heads' |
|
994 | 2. N is an ancestor of some node in 'heads' | |
995 |
|
995 | |||
996 | Every node is considered to be both a descendant and an ancestor |
|
996 | Every node is considered to be both a descendant and an ancestor | |
997 | of itself, so every reachable node in 'roots' and 'heads' will be |
|
997 | of itself, so every reachable node in 'roots' and 'heads' will be | |
998 | included in 'nodes'. |
|
998 | included in 'nodes'. | |
999 |
|
999 | |||
1000 | 'outroots' is the list of reachable nodes in 'roots', i.e., the |
|
1000 | 'outroots' is the list of reachable nodes in 'roots', i.e., the | |
1001 | subset of 'roots' that is returned in 'nodes'. Likewise, |
|
1001 | subset of 'roots' that is returned in 'nodes'. Likewise, | |
1002 | 'outheads' is the subset of 'heads' that is also in 'nodes'. |
|
1002 | 'outheads' is the subset of 'heads' that is also in 'nodes'. | |
1003 |
|
1003 | |||
1004 | 'roots' and 'heads' are both lists of node IDs. If 'roots' is |
|
1004 | 'roots' and 'heads' are both lists of node IDs. If 'roots' is | |
1005 | unspecified, uses nullid as the only root. If 'heads' is |
|
1005 | unspecified, uses nullid as the only root. If 'heads' is | |
1006 | unspecified, uses list of all of the revlog's heads.""" |
|
1006 | unspecified, uses list of all of the revlog's heads.""" | |
1007 | nonodes = ([], [], []) |
|
1007 | nonodes = ([], [], []) | |
1008 | if roots is not None: |
|
1008 | if roots is not None: | |
1009 | roots = list(roots) |
|
1009 | roots = list(roots) | |
1010 | if not roots: |
|
1010 | if not roots: | |
1011 | return nonodes |
|
1011 | return nonodes | |
1012 | lowestrev = min([self.rev(n) for n in roots]) |
|
1012 | lowestrev = min([self.rev(n) for n in roots]) | |
1013 | else: |
|
1013 | else: | |
1014 | roots = [nullid] # Everybody's a descendant of nullid |
|
1014 | roots = [nullid] # Everybody's a descendant of nullid | |
1015 | lowestrev = nullrev |
|
1015 | lowestrev = nullrev | |
1016 | if (lowestrev == nullrev) and (heads is None): |
|
1016 | if (lowestrev == nullrev) and (heads is None): | |
1017 | # We want _all_ the nodes! |
|
1017 | # We want _all_ the nodes! | |
1018 | return ([self.node(r) for r in self], [nullid], list(self.heads())) |
|
1018 | return ([self.node(r) for r in self], [nullid], list(self.heads())) | |
1019 | if heads is None: |
|
1019 | if heads is None: | |
1020 | # All nodes are ancestors, so the latest ancestor is the last |
|
1020 | # All nodes are ancestors, so the latest ancestor is the last | |
1021 | # node. |
|
1021 | # node. | |
1022 | highestrev = len(self) - 1 |
|
1022 | highestrev = len(self) - 1 | |
1023 | # Set ancestors to None to signal that every node is an ancestor. |
|
1023 | # Set ancestors to None to signal that every node is an ancestor. | |
1024 | ancestors = None |
|
1024 | ancestors = None | |
1025 | # Set heads to an empty dictionary for later discovery of heads |
|
1025 | # Set heads to an empty dictionary for later discovery of heads | |
1026 | heads = {} |
|
1026 | heads = {} | |
1027 | else: |
|
1027 | else: | |
1028 | heads = list(heads) |
|
1028 | heads = list(heads) | |
1029 | if not heads: |
|
1029 | if not heads: | |
1030 | return nonodes |
|
1030 | return nonodes | |
1031 | ancestors = set() |
|
1031 | ancestors = set() | |
1032 | # Turn heads into a dictionary so we can remove 'fake' heads. |
|
1032 | # Turn heads into a dictionary so we can remove 'fake' heads. | |
1033 | # Also, later we will be using it to filter out the heads we can't |
|
1033 | # Also, later we will be using it to filter out the heads we can't | |
1034 | # find from roots. |
|
1034 | # find from roots. | |
1035 | heads = dict.fromkeys(heads, False) |
|
1035 | heads = dict.fromkeys(heads, False) | |
1036 | # Start at the top and keep marking parents until we're done. |
|
1036 | # Start at the top and keep marking parents until we're done. | |
1037 | nodestotag = set(heads) |
|
1037 | nodestotag = set(heads) | |
1038 | # Remember where the top was so we can use it as a limit later. |
|
1038 | # Remember where the top was so we can use it as a limit later. | |
1039 | highestrev = max([self.rev(n) for n in nodestotag]) |
|
1039 | highestrev = max([self.rev(n) for n in nodestotag]) | |
1040 | while nodestotag: |
|
1040 | while nodestotag: | |
1041 | # grab a node to tag |
|
1041 | # grab a node to tag | |
1042 | n = nodestotag.pop() |
|
1042 | n = nodestotag.pop() | |
1043 | # Never tag nullid |
|
1043 | # Never tag nullid | |
1044 | if n == nullid: |
|
1044 | if n == nullid: | |
1045 | continue |
|
1045 | continue | |
1046 | # A node's revision number represents its place in a |
|
1046 | # A node's revision number represents its place in a | |
1047 | # topologically sorted list of nodes. |
|
1047 | # topologically sorted list of nodes. | |
1048 | r = self.rev(n) |
|
1048 | r = self.rev(n) | |
1049 | if r >= lowestrev: |
|
1049 | if r >= lowestrev: | |
1050 | if n not in ancestors: |
|
1050 | if n not in ancestors: | |
1051 | # If we are possibly a descendant of one of the roots |
|
1051 | # If we are possibly a descendant of one of the roots | |
1052 | # and we haven't already been marked as an ancestor |
|
1052 | # and we haven't already been marked as an ancestor | |
1053 | ancestors.add(n) # Mark as ancestor |
|
1053 | ancestors.add(n) # Mark as ancestor | |
1054 | # Add non-nullid parents to list of nodes to tag. |
|
1054 | # Add non-nullid parents to list of nodes to tag. | |
1055 | nodestotag.update( |
|
1055 | nodestotag.update( | |
1056 | [p for p in self.parents(n) if p != nullid] |
|
1056 | [p for p in self.parents(n) if p != nullid] | |
1057 | ) |
|
1057 | ) | |
1058 | elif n in heads: # We've seen it before, is it a fake head? |
|
1058 | elif n in heads: # We've seen it before, is it a fake head? | |
1059 | # So it is, real heads should not be the ancestors of |
|
1059 | # So it is, real heads should not be the ancestors of | |
1060 | # any other heads. |
|
1060 | # any other heads. | |
1061 | heads.pop(n) |
|
1061 | heads.pop(n) | |
1062 | if not ancestors: |
|
1062 | if not ancestors: | |
1063 | return nonodes |
|
1063 | return nonodes | |
1064 | # Now that we have our set of ancestors, we want to remove any |
|
1064 | # Now that we have our set of ancestors, we want to remove any | |
1065 | # roots that are not ancestors. |
|
1065 | # roots that are not ancestors. | |
1066 |
|
1066 | |||
1067 | # If one of the roots was nullid, everything is included anyway. |
|
1067 | # If one of the roots was nullid, everything is included anyway. | |
1068 | if lowestrev > nullrev: |
|
1068 | if lowestrev > nullrev: | |
1069 | # But, since we weren't, let's recompute the lowest rev to not |
|
1069 | # But, since we weren't, let's recompute the lowest rev to not | |
1070 | # include roots that aren't ancestors. |
|
1070 | # include roots that aren't ancestors. | |
1071 |
|
1071 | |||
1072 | # Filter out roots that aren't ancestors of heads |
|
1072 | # Filter out roots that aren't ancestors of heads | |
1073 | roots = [root for root in roots if root in ancestors] |
|
1073 | roots = [root for root in roots if root in ancestors] | |
1074 | # Recompute the lowest revision |
|
1074 | # Recompute the lowest revision | |
1075 | if roots: |
|
1075 | if roots: | |
1076 | lowestrev = min([self.rev(root) for root in roots]) |
|
1076 | lowestrev = min([self.rev(root) for root in roots]) | |
1077 | else: |
|
1077 | else: | |
1078 | # No more roots? Return empty list |
|
1078 | # No more roots? Return empty list | |
1079 | return nonodes |
|
1079 | return nonodes | |
1080 | else: |
|
1080 | else: | |
1081 | # We are descending from nullid, and don't need to care about |
|
1081 | # We are descending from nullid, and don't need to care about | |
1082 | # any other roots. |
|
1082 | # any other roots. | |
1083 | lowestrev = nullrev |
|
1083 | lowestrev = nullrev | |
1084 | roots = [nullid] |
|
1084 | roots = [nullid] | |
1085 | # Transform our roots list into a set. |
|
1085 | # Transform our roots list into a set. | |
1086 | descendants = set(roots) |
|
1086 | descendants = set(roots) | |
1087 | # Also, keep the original roots so we can filter out roots that aren't |
|
1087 | # Also, keep the original roots so we can filter out roots that aren't | |
1088 | # 'real' roots (i.e. are descended from other roots). |
|
1088 | # 'real' roots (i.e. are descended from other roots). | |
1089 | roots = descendants.copy() |
|
1089 | roots = descendants.copy() | |
1090 | # Our topologically sorted list of output nodes. |
|
1090 | # Our topologically sorted list of output nodes. | |
1091 | orderedout = [] |
|
1091 | orderedout = [] | |
1092 | # Don't start at nullid since we don't want nullid in our output list, |
|
1092 | # Don't start at nullid since we don't want nullid in our output list, | |
1093 | # and if nullid shows up in descendants, empty parents will look like |
|
1093 | # and if nullid shows up in descendants, empty parents will look like | |
1094 | # they're descendants. |
|
1094 | # they're descendants. | |
1095 | for r in self.revs(start=max(lowestrev, 0), stop=highestrev + 1): |
|
1095 | for r in self.revs(start=max(lowestrev, 0), stop=highestrev + 1): | |
1096 | n = self.node(r) |
|
1096 | n = self.node(r) | |
1097 | isdescendant = False |
|
1097 | isdescendant = False | |
1098 | if lowestrev == nullrev: # Everybody is a descendant of nullid |
|
1098 | if lowestrev == nullrev: # Everybody is a descendant of nullid | |
1099 | isdescendant = True |
|
1099 | isdescendant = True | |
1100 | elif n in descendants: |
|
1100 | elif n in descendants: | |
1101 | # n is already a descendant |
|
1101 | # n is already a descendant | |
1102 | isdescendant = True |
|
1102 | isdescendant = True | |
1103 | # This check only needs to be done here because all the roots |
|
1103 | # This check only needs to be done here because all the roots | |
1104 | # will start being marked is descendants before the loop. |
|
1104 | # will start being marked is descendants before the loop. | |
1105 | if n in roots: |
|
1105 | if n in roots: | |
1106 | # If n was a root, check if it's a 'real' root. |
|
1106 | # If n was a root, check if it's a 'real' root. | |
1107 | p = tuple(self.parents(n)) |
|
1107 | p = tuple(self.parents(n)) | |
1108 | # If any of its parents are descendants, it's not a root. |
|
1108 | # If any of its parents are descendants, it's not a root. | |
1109 | if (p[0] in descendants) or (p[1] in descendants): |
|
1109 | if (p[0] in descendants) or (p[1] in descendants): | |
1110 | roots.remove(n) |
|
1110 | roots.remove(n) | |
1111 | else: |
|
1111 | else: | |
1112 | p = tuple(self.parents(n)) |
|
1112 | p = tuple(self.parents(n)) | |
1113 | # A node is a descendant if either of its parents are |
|
1113 | # A node is a descendant if either of its parents are | |
1114 | # descendants. (We seeded the dependents list with the roots |
|
1114 | # descendants. (We seeded the dependents list with the roots | |
1115 | # up there, remember?) |
|
1115 | # up there, remember?) | |
1116 | if (p[0] in descendants) or (p[1] in descendants): |
|
1116 | if (p[0] in descendants) or (p[1] in descendants): | |
1117 | descendants.add(n) |
|
1117 | descendants.add(n) | |
1118 | isdescendant = True |
|
1118 | isdescendant = True | |
1119 | if isdescendant and ((ancestors is None) or (n in ancestors)): |
|
1119 | if isdescendant and ((ancestors is None) or (n in ancestors)): | |
1120 | # Only include nodes that are both descendants and ancestors. |
|
1120 | # Only include nodes that are both descendants and ancestors. | |
1121 | orderedout.append(n) |
|
1121 | orderedout.append(n) | |
1122 | if (ancestors is not None) and (n in heads): |
|
1122 | if (ancestors is not None) and (n in heads): | |
1123 | # We're trying to figure out which heads are reachable |
|
1123 | # We're trying to figure out which heads are reachable | |
1124 | # from roots. |
|
1124 | # from roots. | |
1125 | # Mark this head as having been reached |
|
1125 | # Mark this head as having been reached | |
1126 | heads[n] = True |
|
1126 | heads[n] = True | |
1127 | elif ancestors is None: |
|
1127 | elif ancestors is None: | |
1128 | # Otherwise, we're trying to discover the heads. |
|
1128 | # Otherwise, we're trying to discover the heads. | |
1129 | # Assume this is a head because if it isn't, the next step |
|
1129 | # Assume this is a head because if it isn't, the next step | |
1130 | # will eventually remove it. |
|
1130 | # will eventually remove it. | |
1131 | heads[n] = True |
|
1131 | heads[n] = True | |
1132 | # But, obviously its parents aren't. |
|
1132 | # But, obviously its parents aren't. | |
1133 | for p in self.parents(n): |
|
1133 | for p in self.parents(n): | |
1134 | heads.pop(p, None) |
|
1134 | heads.pop(p, None) | |
1135 | heads = [head for head, flag in pycompat.iteritems(heads) if flag] |
|
1135 | heads = [head for head, flag in pycompat.iteritems(heads) if flag] | |
1136 | roots = list(roots) |
|
1136 | roots = list(roots) | |
1137 | assert orderedout |
|
1137 | assert orderedout | |
1138 | assert roots |
|
1138 | assert roots | |
1139 | assert heads |
|
1139 | assert heads | |
1140 | return (orderedout, roots, heads) |
|
1140 | return (orderedout, roots, heads) | |
1141 |
|
1141 | |||
1142 | def headrevs(self, revs=None): |
|
1142 | def headrevs(self, revs=None): | |
1143 | if revs is None: |
|
1143 | if revs is None: | |
1144 | try: |
|
1144 | try: | |
1145 | return self.index.headrevs() |
|
1145 | return self.index.headrevs() | |
1146 | except AttributeError: |
|
1146 | except AttributeError: | |
1147 | return self._headrevs() |
|
1147 | return self._headrevs() | |
1148 | if rustdagop is not None: |
|
1148 | if rustdagop is not None: | |
1149 | return rustdagop.headrevs(self.index, revs) |
|
1149 | return rustdagop.headrevs(self.index, revs) | |
1150 | return dagop.headrevs(revs, self._uncheckedparentrevs) |
|
1150 | return dagop.headrevs(revs, self._uncheckedparentrevs) | |
1151 |
|
1151 | |||
1152 | def computephases(self, roots): |
|
1152 | def computephases(self, roots): | |
1153 | return self.index.computephasesmapsets(roots) |
|
1153 | return self.index.computephasesmapsets(roots) | |
1154 |
|
1154 | |||
1155 | def _headrevs(self): |
|
1155 | def _headrevs(self): | |
1156 | count = len(self) |
|
1156 | count = len(self) | |
1157 | if not count: |
|
1157 | if not count: | |
1158 | return [nullrev] |
|
1158 | return [nullrev] | |
1159 | # we won't iter over filtered rev so nobody is a head at start |
|
1159 | # we won't iter over filtered rev so nobody is a head at start | |
1160 | ishead = [0] * (count + 1) |
|
1160 | ishead = [0] * (count + 1) | |
1161 | index = self.index |
|
1161 | index = self.index | |
1162 | for r in self: |
|
1162 | for r in self: | |
1163 | ishead[r] = 1 # I may be an head |
|
1163 | ishead[r] = 1 # I may be an head | |
1164 | e = index[r] |
|
1164 | e = index[r] | |
1165 | ishead[e[5]] = ishead[e[6]] = 0 # my parent are not |
|
1165 | ishead[e[5]] = ishead[e[6]] = 0 # my parent are not | |
1166 | return [r for r, val in enumerate(ishead) if val] |
|
1166 | return [r for r, val in enumerate(ishead) if val] | |
1167 |
|
1167 | |||
1168 | def heads(self, start=None, stop=None): |
|
1168 | def heads(self, start=None, stop=None): | |
1169 | """return the list of all nodes that have no children |
|
1169 | """return the list of all nodes that have no children | |
1170 |
|
1170 | |||
1171 | if start is specified, only heads that are descendants of |
|
1171 | if start is specified, only heads that are descendants of | |
1172 | start will be returned |
|
1172 | start will be returned | |
1173 | if stop is specified, it will consider all the revs from stop |
|
1173 | if stop is specified, it will consider all the revs from stop | |
1174 | as if they had no children |
|
1174 | as if they had no children | |
1175 | """ |
|
1175 | """ | |
1176 | if start is None and stop is None: |
|
1176 | if start is None and stop is None: | |
1177 | if not len(self): |
|
1177 | if not len(self): | |
1178 | return [nullid] |
|
1178 | return [nullid] | |
1179 | return [self.node(r) for r in self.headrevs()] |
|
1179 | return [self.node(r) for r in self.headrevs()] | |
1180 |
|
1180 | |||
1181 | if start is None: |
|
1181 | if start is None: | |
1182 | start = nullrev |
|
1182 | start = nullrev | |
1183 | else: |
|
1183 | else: | |
1184 | start = self.rev(start) |
|
1184 | start = self.rev(start) | |
1185 |
|
1185 | |||
1186 | stoprevs = set(self.rev(n) for n in stop or []) |
|
1186 | stoprevs = set(self.rev(n) for n in stop or []) | |
1187 |
|
1187 | |||
1188 | revs = dagop.headrevssubset( |
|
1188 | revs = dagop.headrevssubset( | |
1189 | self.revs, self.parentrevs, startrev=start, stoprevs=stoprevs |
|
1189 | self.revs, self.parentrevs, startrev=start, stoprevs=stoprevs | |
1190 | ) |
|
1190 | ) | |
1191 |
|
1191 | |||
1192 | return [self.node(rev) for rev in revs] |
|
1192 | return [self.node(rev) for rev in revs] | |
1193 |
|
1193 | |||
1194 | def children(self, node): |
|
1194 | def children(self, node): | |
1195 | """find the children of a given node""" |
|
1195 | """find the children of a given node""" | |
1196 | c = [] |
|
1196 | c = [] | |
1197 | p = self.rev(node) |
|
1197 | p = self.rev(node) | |
1198 | for r in self.revs(start=p + 1): |
|
1198 | for r in self.revs(start=p + 1): | |
1199 | prevs = [pr for pr in self.parentrevs(r) if pr != nullrev] |
|
1199 | prevs = [pr for pr in self.parentrevs(r) if pr != nullrev] | |
1200 | if prevs: |
|
1200 | if prevs: | |
1201 | for pr in prevs: |
|
1201 | for pr in prevs: | |
1202 | if pr == p: |
|
1202 | if pr == p: | |
1203 | c.append(self.node(r)) |
|
1203 | c.append(self.node(r)) | |
1204 | elif p == nullrev: |
|
1204 | elif p == nullrev: | |
1205 | c.append(self.node(r)) |
|
1205 | c.append(self.node(r)) | |
1206 | return c |
|
1206 | return c | |
1207 |
|
1207 | |||
1208 | def commonancestorsheads(self, a, b): |
|
1208 | def commonancestorsheads(self, a, b): | |
1209 | """calculate all the heads of the common ancestors of nodes a and b""" |
|
1209 | """calculate all the heads of the common ancestors of nodes a and b""" | |
1210 | a, b = self.rev(a), self.rev(b) |
|
1210 | a, b = self.rev(a), self.rev(b) | |
1211 | ancs = self._commonancestorsheads(a, b) |
|
1211 | ancs = self._commonancestorsheads(a, b) | |
1212 | return pycompat.maplist(self.node, ancs) |
|
1212 | return pycompat.maplist(self.node, ancs) | |
1213 |
|
1213 | |||
1214 | def _commonancestorsheads(self, *revs): |
|
1214 | def _commonancestorsheads(self, *revs): | |
1215 | """calculate all the heads of the common ancestors of revs""" |
|
1215 | """calculate all the heads of the common ancestors of revs""" | |
1216 | try: |
|
1216 | try: | |
1217 | ancs = self.index.commonancestorsheads(*revs) |
|
1217 | ancs = self.index.commonancestorsheads(*revs) | |
1218 | except (AttributeError, OverflowError): # C implementation failed |
|
1218 | except (AttributeError, OverflowError): # C implementation failed | |
1219 | ancs = ancestor.commonancestorsheads(self.parentrevs, *revs) |
|
1219 | ancs = ancestor.commonancestorsheads(self.parentrevs, *revs) | |
1220 | return ancs |
|
1220 | return ancs | |
1221 |
|
1221 | |||
1222 | def isancestor(self, a, b): |
|
1222 | def isancestor(self, a, b): | |
1223 | """return True if node a is an ancestor of node b |
|
1223 | """return True if node a is an ancestor of node b | |
1224 |
|
1224 | |||
1225 | A revision is considered an ancestor of itself.""" |
|
1225 | A revision is considered an ancestor of itself.""" | |
1226 | a, b = self.rev(a), self.rev(b) |
|
1226 | a, b = self.rev(a), self.rev(b) | |
1227 | return self.isancestorrev(a, b) |
|
1227 | return self.isancestorrev(a, b) | |
1228 |
|
1228 | |||
1229 | def isancestorrev(self, a, b): |
|
1229 | def isancestorrev(self, a, b): | |
1230 | """return True if revision a is an ancestor of revision b |
|
1230 | """return True if revision a is an ancestor of revision b | |
1231 |
|
1231 | |||
1232 | A revision is considered an ancestor of itself. |
|
1232 | A revision is considered an ancestor of itself. | |
1233 |
|
1233 | |||
1234 | The implementation of this is trivial but the use of |
|
1234 | The implementation of this is trivial but the use of | |
1235 | reachableroots is not.""" |
|
1235 | reachableroots is not.""" | |
1236 | if a == nullrev: |
|
1236 | if a == nullrev: | |
1237 | return True |
|
1237 | return True | |
1238 | elif a == b: |
|
1238 | elif a == b: | |
1239 | return True |
|
1239 | return True | |
1240 | elif a > b: |
|
1240 | elif a > b: | |
1241 | return False |
|
1241 | return False | |
1242 | return bool(self.reachableroots(a, [b], [a], includepath=False)) |
|
1242 | return bool(self.reachableroots(a, [b], [a], includepath=False)) | |
1243 |
|
1243 | |||
1244 | def reachableroots(self, minroot, heads, roots, includepath=False): |
|
1244 | def reachableroots(self, minroot, heads, roots, includepath=False): | |
1245 | """return (heads(::<roots> and <roots>::<heads>)) |
|
1245 | """return (heads(::<roots> and <roots>::<heads>)) | |
1246 |
|
1246 | |||
1247 | If includepath is True, return (<roots>::<heads>).""" |
|
1247 | If includepath is True, return (<roots>::<heads>).""" | |
1248 | try: |
|
1248 | try: | |
1249 | return self.index.reachableroots2( |
|
1249 | return self.index.reachableroots2( | |
1250 | minroot, heads, roots, includepath |
|
1250 | minroot, heads, roots, includepath | |
1251 | ) |
|
1251 | ) | |
1252 | except AttributeError: |
|
1252 | except AttributeError: | |
1253 | return dagop._reachablerootspure( |
|
1253 | return dagop._reachablerootspure( | |
1254 | self.parentrevs, minroot, roots, heads, includepath |
|
1254 | self.parentrevs, minroot, roots, heads, includepath | |
1255 | ) |
|
1255 | ) | |
1256 |
|
1256 | |||
1257 | def ancestor(self, a, b): |
|
1257 | def ancestor(self, a, b): | |
1258 | """calculate the "best" common ancestor of nodes a and b""" |
|
1258 | """calculate the "best" common ancestor of nodes a and b""" | |
1259 |
|
1259 | |||
1260 | a, b = self.rev(a), self.rev(b) |
|
1260 | a, b = self.rev(a), self.rev(b) | |
1261 | try: |
|
1261 | try: | |
1262 | ancs = self.index.ancestors(a, b) |
|
1262 | ancs = self.index.ancestors(a, b) | |
1263 | except (AttributeError, OverflowError): |
|
1263 | except (AttributeError, OverflowError): | |
1264 | ancs = ancestor.ancestors(self.parentrevs, a, b) |
|
1264 | ancs = ancestor.ancestors(self.parentrevs, a, b) | |
1265 | if ancs: |
|
1265 | if ancs: | |
1266 | # choose a consistent winner when there's a tie |
|
1266 | # choose a consistent winner when there's a tie | |
1267 | return min(map(self.node, ancs)) |
|
1267 | return min(map(self.node, ancs)) | |
1268 | return nullid |
|
1268 | return nullid | |
1269 |
|
1269 | |||
1270 | def _match(self, id): |
|
1270 | def _match(self, id): | |
1271 | if isinstance(id, int): |
|
1271 | if isinstance(id, int): | |
1272 | # rev |
|
1272 | # rev | |
1273 | return self.node(id) |
|
1273 | return self.node(id) | |
1274 | if len(id) == 20: |
|
1274 | if len(id) == 20: | |
1275 | # possibly a binary node |
|
1275 | # possibly a binary node | |
1276 | # odds of a binary node being all hex in ASCII are 1 in 10**25 |
|
1276 | # odds of a binary node being all hex in ASCII are 1 in 10**25 | |
1277 | try: |
|
1277 | try: | |
1278 | node = id |
|
1278 | node = id | |
1279 | self.rev(node) # quick search the index |
|
1279 | self.rev(node) # quick search the index | |
1280 | return node |
|
1280 | return node | |
1281 | except error.LookupError: |
|
1281 | except error.LookupError: | |
1282 | pass # may be partial hex id |
|
1282 | pass # may be partial hex id | |
1283 | try: |
|
1283 | try: | |
1284 | # str(rev) |
|
1284 | # str(rev) | |
1285 | rev = int(id) |
|
1285 | rev = int(id) | |
1286 | if b"%d" % rev != id: |
|
1286 | if b"%d" % rev != id: | |
1287 | raise ValueError |
|
1287 | raise ValueError | |
1288 | if rev < 0: |
|
1288 | if rev < 0: | |
1289 | rev = len(self) + rev |
|
1289 | rev = len(self) + rev | |
1290 | if rev < 0 or rev >= len(self): |
|
1290 | if rev < 0 or rev >= len(self): | |
1291 | raise ValueError |
|
1291 | raise ValueError | |
1292 | return self.node(rev) |
|
1292 | return self.node(rev) | |
1293 | except (ValueError, OverflowError): |
|
1293 | except (ValueError, OverflowError): | |
1294 | pass |
|
1294 | pass | |
1295 | if len(id) == 40: |
|
1295 | if len(id) == 40: | |
1296 | try: |
|
1296 | try: | |
1297 | # a full hex nodeid? |
|
1297 | # a full hex nodeid? | |
1298 | node = bin(id) |
|
1298 | node = bin(id) | |
1299 | self.rev(node) |
|
1299 | self.rev(node) | |
1300 | return node |
|
1300 | return node | |
1301 | except (TypeError, error.LookupError): |
|
1301 | except (TypeError, error.LookupError): | |
1302 | pass |
|
1302 | pass | |
1303 |
|
1303 | |||
1304 | def _partialmatch(self, id): |
|
1304 | def _partialmatch(self, id): | |
1305 | # we don't care wdirfilenodeids as they should be always full hash |
|
1305 | # we don't care wdirfilenodeids as they should be always full hash | |
1306 | maybewdir = wdirhex.startswith(id) |
|
1306 | maybewdir = wdirhex.startswith(id) | |
1307 | try: |
|
1307 | try: | |
1308 | partial = self.index.partialmatch(id) |
|
1308 | partial = self.index.partialmatch(id) | |
1309 | if partial and self.hasnode(partial): |
|
1309 | if partial and self.hasnode(partial): | |
1310 | if maybewdir: |
|
1310 | if maybewdir: | |
1311 | # single 'ff...' match in radix tree, ambiguous with wdir |
|
1311 | # single 'ff...' match in radix tree, ambiguous with wdir | |
1312 | raise error.RevlogError |
|
1312 | raise error.RevlogError | |
1313 | return partial |
|
1313 | return partial | |
1314 | if maybewdir: |
|
1314 | if maybewdir: | |
1315 | # no 'ff...' match in radix tree, wdir identified |
|
1315 | # no 'ff...' match in radix tree, wdir identified | |
1316 | raise error.WdirUnsupported |
|
1316 | raise error.WdirUnsupported | |
1317 | return None |
|
1317 | return None | |
1318 | except error.RevlogError: |
|
1318 | except error.RevlogError: | |
1319 | # parsers.c radix tree lookup gave multiple matches |
|
1319 | # parsers.c radix tree lookup gave multiple matches | |
1320 | # fast path: for unfiltered changelog, radix tree is accurate |
|
1320 | # fast path: for unfiltered changelog, radix tree is accurate | |
1321 | if not getattr(self, 'filteredrevs', None): |
|
1321 | if not getattr(self, 'filteredrevs', None): | |
1322 | raise error.AmbiguousPrefixLookupError( |
|
1322 | raise error.AmbiguousPrefixLookupError( | |
1323 | id, self.indexfile, _(b'ambiguous identifier') |
|
1323 | id, self.indexfile, _(b'ambiguous identifier') | |
1324 | ) |
|
1324 | ) | |
1325 | # fall through to slow path that filters hidden revisions |
|
1325 | # fall through to slow path that filters hidden revisions | |
1326 | except (AttributeError, ValueError): |
|
1326 | except (AttributeError, ValueError): | |
1327 | # we are pure python, or key was too short to search radix tree |
|
1327 | # we are pure python, or key was too short to search radix tree | |
1328 | pass |
|
1328 | pass | |
1329 |
|
1329 | |||
1330 | if id in self._pcache: |
|
1330 | if id in self._pcache: | |
1331 | return self._pcache[id] |
|
1331 | return self._pcache[id] | |
1332 |
|
1332 | |||
1333 | if len(id) <= 40: |
|
1333 | if len(id) <= 40: | |
1334 | try: |
|
1334 | try: | |
1335 | # hex(node)[:...] |
|
1335 | # hex(node)[:...] | |
1336 | l = len(id) // 2 # grab an even number of digits |
|
1336 | l = len(id) // 2 # grab an even number of digits | |
1337 | prefix = bin(id[: l * 2]) |
|
1337 | prefix = bin(id[: l * 2]) | |
1338 | nl = [e[7] for e in self.index if e[7].startswith(prefix)] |
|
1338 | nl = [e[7] for e in self.index if e[7].startswith(prefix)] | |
1339 | nl = [ |
|
1339 | nl = [ | |
1340 | n for n in nl if hex(n).startswith(id) and self.hasnode(n) |
|
1340 | n for n in nl if hex(n).startswith(id) and self.hasnode(n) | |
1341 | ] |
|
1341 | ] | |
1342 | if nullhex.startswith(id): |
|
1342 | if nullhex.startswith(id): | |
1343 | nl.append(nullid) |
|
1343 | nl.append(nullid) | |
1344 | if len(nl) > 0: |
|
1344 | if len(nl) > 0: | |
1345 | if len(nl) == 1 and not maybewdir: |
|
1345 | if len(nl) == 1 and not maybewdir: | |
1346 | self._pcache[id] = nl[0] |
|
1346 | self._pcache[id] = nl[0] | |
1347 | return nl[0] |
|
1347 | return nl[0] | |
1348 | raise error.AmbiguousPrefixLookupError( |
|
1348 | raise error.AmbiguousPrefixLookupError( | |
1349 | id, self.indexfile, _(b'ambiguous identifier') |
|
1349 | id, self.indexfile, _(b'ambiguous identifier') | |
1350 | ) |
|
1350 | ) | |
1351 | if maybewdir: |
|
1351 | if maybewdir: | |
1352 | raise error.WdirUnsupported |
|
1352 | raise error.WdirUnsupported | |
1353 | return None |
|
1353 | return None | |
1354 | except TypeError: |
|
1354 | except TypeError: | |
1355 | pass |
|
1355 | pass | |
1356 |
|
1356 | |||
1357 | def lookup(self, id): |
|
1357 | def lookup(self, id): | |
1358 | """locate a node based on: |
|
1358 | """locate a node based on: | |
1359 | - revision number or str(revision number) |
|
1359 | - revision number or str(revision number) | |
1360 | - nodeid or subset of hex nodeid |
|
1360 | - nodeid or subset of hex nodeid | |
1361 | """ |
|
1361 | """ | |
1362 | n = self._match(id) |
|
1362 | n = self._match(id) | |
1363 | if n is not None: |
|
1363 | if n is not None: | |
1364 | return n |
|
1364 | return n | |
1365 | n = self._partialmatch(id) |
|
1365 | n = self._partialmatch(id) | |
1366 | if n: |
|
1366 | if n: | |
1367 | return n |
|
1367 | return n | |
1368 |
|
1368 | |||
1369 | raise error.LookupError(id, self.indexfile, _(b'no match found')) |
|
1369 | raise error.LookupError(id, self.indexfile, _(b'no match found')) | |
1370 |
|
1370 | |||
1371 | def shortest(self, node, minlength=1): |
|
1371 | def shortest(self, node, minlength=1): | |
1372 | """Find the shortest unambiguous prefix that matches node.""" |
|
1372 | """Find the shortest unambiguous prefix that matches node.""" | |
1373 |
|
1373 | |||
1374 | def isvalid(prefix): |
|
1374 | def isvalid(prefix): | |
1375 | try: |
|
1375 | try: | |
1376 | matchednode = self._partialmatch(prefix) |
|
1376 | matchednode = self._partialmatch(prefix) | |
1377 | except error.AmbiguousPrefixLookupError: |
|
1377 | except error.AmbiguousPrefixLookupError: | |
1378 | return False |
|
1378 | return False | |
1379 | except error.WdirUnsupported: |
|
1379 | except error.WdirUnsupported: | |
1380 | # single 'ff...' match |
|
1380 | # single 'ff...' match | |
1381 | return True |
|
1381 | return True | |
1382 | if matchednode is None: |
|
1382 | if matchednode is None: | |
1383 | raise error.LookupError(node, self.indexfile, _(b'no node')) |
|
1383 | raise error.LookupError(node, self.indexfile, _(b'no node')) | |
1384 | return True |
|
1384 | return True | |
1385 |
|
1385 | |||
1386 | def maybewdir(prefix): |
|
1386 | def maybewdir(prefix): | |
1387 | return all(c == b'f' for c in pycompat.iterbytestr(prefix)) |
|
1387 | return all(c == b'f' for c in pycompat.iterbytestr(prefix)) | |
1388 |
|
1388 | |||
1389 | hexnode = hex(node) |
|
1389 | hexnode = hex(node) | |
1390 |
|
1390 | |||
1391 | def disambiguate(hexnode, minlength): |
|
1391 | def disambiguate(hexnode, minlength): | |
1392 | """Disambiguate against wdirid.""" |
|
1392 | """Disambiguate against wdirid.""" | |
1393 | for length in range(minlength, 41): |
|
1393 | for length in range(minlength, 41): | |
1394 | prefix = hexnode[:length] |
|
1394 | prefix = hexnode[:length] | |
1395 | if not maybewdir(prefix): |
|
1395 | if not maybewdir(prefix): | |
1396 | return prefix |
|
1396 | return prefix | |
1397 |
|
1397 | |||
1398 | if not getattr(self, 'filteredrevs', None): |
|
1398 | if not getattr(self, 'filteredrevs', None): | |
1399 | try: |
|
1399 | try: | |
1400 | length = max(self.index.shortest(node), minlength) |
|
1400 | length = max(self.index.shortest(node), minlength) | |
1401 | return disambiguate(hexnode, length) |
|
1401 | return disambiguate(hexnode, length) | |
1402 | except error.RevlogError: |
|
1402 | except error.RevlogError: | |
1403 | if node != wdirid: |
|
1403 | if node != wdirid: | |
1404 | raise error.LookupError(node, self.indexfile, _(b'no node')) |
|
1404 | raise error.LookupError(node, self.indexfile, _(b'no node')) | |
1405 | except AttributeError: |
|
1405 | except AttributeError: | |
1406 | # Fall through to pure code |
|
1406 | # Fall through to pure code | |
1407 | pass |
|
1407 | pass | |
1408 |
|
1408 | |||
1409 | if node == wdirid: |
|
1409 | if node == wdirid: | |
1410 | for length in range(minlength, 41): |
|
1410 | for length in range(minlength, 41): | |
1411 | prefix = hexnode[:length] |
|
1411 | prefix = hexnode[:length] | |
1412 | if isvalid(prefix): |
|
1412 | if isvalid(prefix): | |
1413 | return prefix |
|
1413 | return prefix | |
1414 |
|
1414 | |||
1415 | for length in range(minlength, 41): |
|
1415 | for length in range(minlength, 41): | |
1416 | prefix = hexnode[:length] |
|
1416 | prefix = hexnode[:length] | |
1417 | if isvalid(prefix): |
|
1417 | if isvalid(prefix): | |
1418 | return disambiguate(hexnode, length) |
|
1418 | return disambiguate(hexnode, length) | |
1419 |
|
1419 | |||
1420 | def cmp(self, node, text): |
|
1420 | def cmp(self, node, text): | |
1421 | """compare text with a given file revision |
|
1421 | """compare text with a given file revision | |
1422 |
|
1422 | |||
1423 | returns True if text is different than what is stored. |
|
1423 | returns True if text is different than what is stored. | |
1424 | """ |
|
1424 | """ | |
1425 | p1, p2 = self.parents(node) |
|
1425 | p1, p2 = self.parents(node) | |
1426 | return storageutil.hashrevisionsha1(text, p1, p2) != node |
|
1426 | return storageutil.hashrevisionsha1(text, p1, p2) != node | |
1427 |
|
1427 | |||
1428 | def _cachesegment(self, offset, data): |
|
1428 | def _cachesegment(self, offset, data): | |
1429 | """Add a segment to the revlog cache. |
|
1429 | """Add a segment to the revlog cache. | |
1430 |
|
1430 | |||
1431 | Accepts an absolute offset and the data that is at that location. |
|
1431 | Accepts an absolute offset and the data that is at that location. | |
1432 | """ |
|
1432 | """ | |
1433 | o, d = self._chunkcache |
|
1433 | o, d = self._chunkcache | |
1434 | # try to add to existing cache |
|
1434 | # try to add to existing cache | |
1435 | if o + len(d) == offset and len(d) + len(data) < _chunksize: |
|
1435 | if o + len(d) == offset and len(d) + len(data) < _chunksize: | |
1436 | self._chunkcache = o, d + data |
|
1436 | self._chunkcache = o, d + data | |
1437 | else: |
|
1437 | else: | |
1438 | self._chunkcache = offset, data |
|
1438 | self._chunkcache = offset, data | |
1439 |
|
1439 | |||
1440 | def _readsegment(self, offset, length, df=None): |
|
1440 | def _readsegment(self, offset, length, df=None): | |
1441 | """Load a segment of raw data from the revlog. |
|
1441 | """Load a segment of raw data from the revlog. | |
1442 |
|
1442 | |||
1443 | Accepts an absolute offset, length to read, and an optional existing |
|
1443 | Accepts an absolute offset, length to read, and an optional existing | |
1444 | file handle to read from. |
|
1444 | file handle to read from. | |
1445 |
|
1445 | |||
1446 | If an existing file handle is passed, it will be seeked and the |
|
1446 | If an existing file handle is passed, it will be seeked and the | |
1447 | original seek position will NOT be restored. |
|
1447 | original seek position will NOT be restored. | |
1448 |
|
1448 | |||
1449 | Returns a str or buffer of raw byte data. |
|
1449 | Returns a str or buffer of raw byte data. | |
1450 |
|
1450 | |||
1451 | Raises if the requested number of bytes could not be read. |
|
1451 | Raises if the requested number of bytes could not be read. | |
1452 | """ |
|
1452 | """ | |
1453 | # Cache data both forward and backward around the requested |
|
1453 | # Cache data both forward and backward around the requested | |
1454 | # data, in a fixed size window. This helps speed up operations |
|
1454 | # data, in a fixed size window. This helps speed up operations | |
1455 | # involving reading the revlog backwards. |
|
1455 | # involving reading the revlog backwards. | |
1456 | cachesize = self._chunkcachesize |
|
1456 | cachesize = self._chunkcachesize | |
1457 | realoffset = offset & ~(cachesize - 1) |
|
1457 | realoffset = offset & ~(cachesize - 1) | |
1458 | reallength = ( |
|
1458 | reallength = ( | |
1459 | (offset + length + cachesize) & ~(cachesize - 1) |
|
1459 | (offset + length + cachesize) & ~(cachesize - 1) | |
1460 | ) - realoffset |
|
1460 | ) - realoffset | |
1461 | with self._datareadfp(df) as df: |
|
1461 | with self._datareadfp(df) as df: | |
1462 | df.seek(realoffset) |
|
1462 | df.seek(realoffset) | |
1463 | d = df.read(reallength) |
|
1463 | d = df.read(reallength) | |
1464 |
|
1464 | |||
1465 | self._cachesegment(realoffset, d) |
|
1465 | self._cachesegment(realoffset, d) | |
1466 | if offset != realoffset or reallength != length: |
|
1466 | if offset != realoffset or reallength != length: | |
1467 | startoffset = offset - realoffset |
|
1467 | startoffset = offset - realoffset | |
1468 | if len(d) - startoffset < length: |
|
1468 | if len(d) - startoffset < length: | |
1469 | raise error.RevlogError( |
|
1469 | raise error.RevlogError( | |
1470 | _( |
|
1470 | _( | |
1471 | b'partial read of revlog %s; expected %d bytes from ' |
|
1471 | b'partial read of revlog %s; expected %d bytes from ' | |
1472 | b'offset %d, got %d' |
|
1472 | b'offset %d, got %d' | |
1473 | ) |
|
1473 | ) | |
1474 | % ( |
|
1474 | % ( | |
1475 | self.indexfile if self._inline else self.datafile, |
|
1475 | self.indexfile if self._inline else self.datafile, | |
1476 | length, |
|
1476 | length, | |
1477 | realoffset, |
|
1477 | realoffset, | |
1478 | len(d) - startoffset, |
|
1478 | len(d) - startoffset, | |
1479 | ) |
|
1479 | ) | |
1480 | ) |
|
1480 | ) | |
1481 |
|
1481 | |||
1482 | return util.buffer(d, startoffset, length) |
|
1482 | return util.buffer(d, startoffset, length) | |
1483 |
|
1483 | |||
1484 | if len(d) < length: |
|
1484 | if len(d) < length: | |
1485 | raise error.RevlogError( |
|
1485 | raise error.RevlogError( | |
1486 | _( |
|
1486 | _( | |
1487 | b'partial read of revlog %s; expected %d bytes from offset ' |
|
1487 | b'partial read of revlog %s; expected %d bytes from offset ' | |
1488 | b'%d, got %d' |
|
1488 | b'%d, got %d' | |
1489 | ) |
|
1489 | ) | |
1490 | % ( |
|
1490 | % ( | |
1491 | self.indexfile if self._inline else self.datafile, |
|
1491 | self.indexfile if self._inline else self.datafile, | |
1492 | length, |
|
1492 | length, | |
1493 | offset, |
|
1493 | offset, | |
1494 | len(d), |
|
1494 | len(d), | |
1495 | ) |
|
1495 | ) | |
1496 | ) |
|
1496 | ) | |
1497 |
|
1497 | |||
1498 | return d |
|
1498 | return d | |
1499 |
|
1499 | |||
1500 | def _getsegment(self, offset, length, df=None): |
|
1500 | def _getsegment(self, offset, length, df=None): | |
1501 | """Obtain a segment of raw data from the revlog. |
|
1501 | """Obtain a segment of raw data from the revlog. | |
1502 |
|
1502 | |||
1503 | Accepts an absolute offset, length of bytes to obtain, and an |
|
1503 | Accepts an absolute offset, length of bytes to obtain, and an | |
1504 | optional file handle to the already-opened revlog. If the file |
|
1504 | optional file handle to the already-opened revlog. If the file | |
1505 | handle is used, it's original seek position will not be preserved. |
|
1505 | handle is used, it's original seek position will not be preserved. | |
1506 |
|
1506 | |||
1507 | Requests for data may be returned from a cache. |
|
1507 | Requests for data may be returned from a cache. | |
1508 |
|
1508 | |||
1509 | Returns a str or a buffer instance of raw byte data. |
|
1509 | Returns a str or a buffer instance of raw byte data. | |
1510 | """ |
|
1510 | """ | |
1511 | o, d = self._chunkcache |
|
1511 | o, d = self._chunkcache | |
1512 | l = len(d) |
|
1512 | l = len(d) | |
1513 |
|
1513 | |||
1514 | # is it in the cache? |
|
1514 | # is it in the cache? | |
1515 | cachestart = offset - o |
|
1515 | cachestart = offset - o | |
1516 | cacheend = cachestart + length |
|
1516 | cacheend = cachestart + length | |
1517 | if cachestart >= 0 and cacheend <= l: |
|
1517 | if cachestart >= 0 and cacheend <= l: | |
1518 | if cachestart == 0 and cacheend == l: |
|
1518 | if cachestart == 0 and cacheend == l: | |
1519 | return d # avoid a copy |
|
1519 | return d # avoid a copy | |
1520 | return util.buffer(d, cachestart, cacheend - cachestart) |
|
1520 | return util.buffer(d, cachestart, cacheend - cachestart) | |
1521 |
|
1521 | |||
1522 | return self._readsegment(offset, length, df=df) |
|
1522 | return self._readsegment(offset, length, df=df) | |
1523 |
|
1523 | |||
1524 | def _getsegmentforrevs(self, startrev, endrev, df=None): |
|
1524 | def _getsegmentforrevs(self, startrev, endrev, df=None): | |
1525 | """Obtain a segment of raw data corresponding to a range of revisions. |
|
1525 | """Obtain a segment of raw data corresponding to a range of revisions. | |
1526 |
|
1526 | |||
1527 | Accepts the start and end revisions and an optional already-open |
|
1527 | Accepts the start and end revisions and an optional already-open | |
1528 | file handle to be used for reading. If the file handle is read, its |
|
1528 | file handle to be used for reading. If the file handle is read, its | |
1529 | seek position will not be preserved. |
|
1529 | seek position will not be preserved. | |
1530 |
|
1530 | |||
1531 | Requests for data may be satisfied by a cache. |
|
1531 | Requests for data may be satisfied by a cache. | |
1532 |
|
1532 | |||
1533 | Returns a 2-tuple of (offset, data) for the requested range of |
|
1533 | Returns a 2-tuple of (offset, data) for the requested range of | |
1534 | revisions. Offset is the integer offset from the beginning of the |
|
1534 | revisions. Offset is the integer offset from the beginning of the | |
1535 | revlog and data is a str or buffer of the raw byte data. |
|
1535 | revlog and data is a str or buffer of the raw byte data. | |
1536 |
|
1536 | |||
1537 | Callers will need to call ``self.start(rev)`` and ``self.length(rev)`` |
|
1537 | Callers will need to call ``self.start(rev)`` and ``self.length(rev)`` | |
1538 | to determine where each revision's data begins and ends. |
|
1538 | to determine where each revision's data begins and ends. | |
1539 | """ |
|
1539 | """ | |
1540 | # Inlined self.start(startrev) & self.end(endrev) for perf reasons |
|
1540 | # Inlined self.start(startrev) & self.end(endrev) for perf reasons | |
1541 | # (functions are expensive). |
|
1541 | # (functions are expensive). | |
1542 | index = self.index |
|
1542 | index = self.index | |
1543 | istart = index[startrev] |
|
1543 | istart = index[startrev] | |
1544 | start = int(istart[0] >> 16) |
|
1544 | start = int(istart[0] >> 16) | |
1545 | if startrev == endrev: |
|
1545 | if startrev == endrev: | |
1546 | end = start + istart[1] |
|
1546 | end = start + istart[1] | |
1547 | else: |
|
1547 | else: | |
1548 | iend = index[endrev] |
|
1548 | iend = index[endrev] | |
1549 | end = int(iend[0] >> 16) + iend[1] |
|
1549 | end = int(iend[0] >> 16) + iend[1] | |
1550 |
|
1550 | |||
1551 | if self._inline: |
|
1551 | if self._inline: | |
1552 | start += (startrev + 1) * self._io.size |
|
1552 | start += (startrev + 1) * self._io.size | |
1553 | end += (endrev + 1) * self._io.size |
|
1553 | end += (endrev + 1) * self._io.size | |
1554 | length = end - start |
|
1554 | length = end - start | |
1555 |
|
1555 | |||
1556 | return start, self._getsegment(start, length, df=df) |
|
1556 | return start, self._getsegment(start, length, df=df) | |
1557 |
|
1557 | |||
1558 | def _chunk(self, rev, df=None): |
|
1558 | def _chunk(self, rev, df=None): | |
1559 | """Obtain a single decompressed chunk for a revision. |
|
1559 | """Obtain a single decompressed chunk for a revision. | |
1560 |
|
1560 | |||
1561 | Accepts an integer revision and an optional already-open file handle |
|
1561 | Accepts an integer revision and an optional already-open file handle | |
1562 | to be used for reading. If used, the seek position of the file will not |
|
1562 | to be used for reading. If used, the seek position of the file will not | |
1563 | be preserved. |
|
1563 | be preserved. | |
1564 |
|
1564 | |||
1565 | Returns a str holding uncompressed data for the requested revision. |
|
1565 | Returns a str holding uncompressed data for the requested revision. | |
1566 | """ |
|
1566 | """ | |
1567 | return self.decompress(self._getsegmentforrevs(rev, rev, df=df)[1]) |
|
1567 | return self.decompress(self._getsegmentforrevs(rev, rev, df=df)[1]) | |
1568 |
|
1568 | |||
1569 | def _chunks(self, revs, df=None, targetsize=None): |
|
1569 | def _chunks(self, revs, df=None, targetsize=None): | |
1570 | """Obtain decompressed chunks for the specified revisions. |
|
1570 | """Obtain decompressed chunks for the specified revisions. | |
1571 |
|
1571 | |||
1572 | Accepts an iterable of numeric revisions that are assumed to be in |
|
1572 | Accepts an iterable of numeric revisions that are assumed to be in | |
1573 | ascending order. Also accepts an optional already-open file handle |
|
1573 | ascending order. Also accepts an optional already-open file handle | |
1574 | to be used for reading. If used, the seek position of the file will |
|
1574 | to be used for reading. If used, the seek position of the file will | |
1575 | not be preserved. |
|
1575 | not be preserved. | |
1576 |
|
1576 | |||
1577 | This function is similar to calling ``self._chunk()`` multiple times, |
|
1577 | This function is similar to calling ``self._chunk()`` multiple times, | |
1578 | but is faster. |
|
1578 | but is faster. | |
1579 |
|
1579 | |||
1580 | Returns a list with decompressed data for each requested revision. |
|
1580 | Returns a list with decompressed data for each requested revision. | |
1581 | """ |
|
1581 | """ | |
1582 | if not revs: |
|
1582 | if not revs: | |
1583 | return [] |
|
1583 | return [] | |
1584 | start = self.start |
|
1584 | start = self.start | |
1585 | length = self.length |
|
1585 | length = self.length | |
1586 | inline = self._inline |
|
1586 | inline = self._inline | |
1587 | iosize = self._io.size |
|
1587 | iosize = self._io.size | |
1588 | buffer = util.buffer |
|
1588 | buffer = util.buffer | |
1589 |
|
1589 | |||
1590 | l = [] |
|
1590 | l = [] | |
1591 | ladd = l.append |
|
1591 | ladd = l.append | |
1592 |
|
1592 | |||
1593 | if not self._withsparseread: |
|
1593 | if not self._withsparseread: | |
1594 | slicedchunks = (revs,) |
|
1594 | slicedchunks = (revs,) | |
1595 | else: |
|
1595 | else: | |
1596 | slicedchunks = deltautil.slicechunk( |
|
1596 | slicedchunks = deltautil.slicechunk( | |
1597 | self, revs, targetsize=targetsize |
|
1597 | self, revs, targetsize=targetsize | |
1598 | ) |
|
1598 | ) | |
1599 |
|
1599 | |||
1600 | for revschunk in slicedchunks: |
|
1600 | for revschunk in slicedchunks: | |
1601 | firstrev = revschunk[0] |
|
1601 | firstrev = revschunk[0] | |
1602 | # Skip trailing revisions with empty diff |
|
1602 | # Skip trailing revisions with empty diff | |
1603 | for lastrev in revschunk[::-1]: |
|
1603 | for lastrev in revschunk[::-1]: | |
1604 | if length(lastrev) != 0: |
|
1604 | if length(lastrev) != 0: | |
1605 | break |
|
1605 | break | |
1606 |
|
1606 | |||
1607 | try: |
|
1607 | try: | |
1608 | offset, data = self._getsegmentforrevs(firstrev, lastrev, df=df) |
|
1608 | offset, data = self._getsegmentforrevs(firstrev, lastrev, df=df) | |
1609 | except OverflowError: |
|
1609 | except OverflowError: | |
1610 | # issue4215 - we can't cache a run of chunks greater than |
|
1610 | # issue4215 - we can't cache a run of chunks greater than | |
1611 | # 2G on Windows |
|
1611 | # 2G on Windows | |
1612 | return [self._chunk(rev, df=df) for rev in revschunk] |
|
1612 | return [self._chunk(rev, df=df) for rev in revschunk] | |
1613 |
|
1613 | |||
1614 | decomp = self.decompress |
|
1614 | decomp = self.decompress | |
1615 | for rev in revschunk: |
|
1615 | for rev in revschunk: | |
1616 | chunkstart = start(rev) |
|
1616 | chunkstart = start(rev) | |
1617 | if inline: |
|
1617 | if inline: | |
1618 | chunkstart += (rev + 1) * iosize |
|
1618 | chunkstart += (rev + 1) * iosize | |
1619 | chunklength = length(rev) |
|
1619 | chunklength = length(rev) | |
1620 | ladd(decomp(buffer(data, chunkstart - offset, chunklength))) |
|
1620 | ladd(decomp(buffer(data, chunkstart - offset, chunklength))) | |
1621 |
|
1621 | |||
1622 | return l |
|
1622 | return l | |
1623 |
|
1623 | |||
1624 | def _chunkclear(self): |
|
1624 | def _chunkclear(self): | |
1625 | """Clear the raw chunk cache.""" |
|
1625 | """Clear the raw chunk cache.""" | |
1626 | self._chunkcache = (0, b'') |
|
1626 | self._chunkcache = (0, b'') | |
1627 |
|
1627 | |||
1628 | def deltaparent(self, rev): |
|
1628 | def deltaparent(self, rev): | |
1629 | """return deltaparent of the given revision""" |
|
1629 | """return deltaparent of the given revision""" | |
1630 | base = self.index[rev][3] |
|
1630 | base = self.index[rev][3] | |
1631 | if base == rev: |
|
1631 | if base == rev: | |
1632 | return nullrev |
|
1632 | return nullrev | |
1633 | elif self._generaldelta: |
|
1633 | elif self._generaldelta: | |
1634 | return base |
|
1634 | return base | |
1635 | else: |
|
1635 | else: | |
1636 | return rev - 1 |
|
1636 | return rev - 1 | |
1637 |
|
1637 | |||
1638 | def issnapshot(self, rev): |
|
1638 | def issnapshot(self, rev): | |
1639 | """tells whether rev is a snapshot |
|
1639 | """tells whether rev is a snapshot | |
1640 | """ |
|
1640 | """ | |
1641 | if not self._sparserevlog: |
|
1641 | if not self._sparserevlog: | |
1642 | return self.deltaparent(rev) == nullrev |
|
1642 | return self.deltaparent(rev) == nullrev | |
1643 | elif util.safehasattr(self.index, b'issnapshot'): |
|
1643 | elif util.safehasattr(self.index, b'issnapshot'): | |
1644 | # directly assign the method to cache the testing and access |
|
1644 | # directly assign the method to cache the testing and access | |
1645 | self.issnapshot = self.index.issnapshot |
|
1645 | self.issnapshot = self.index.issnapshot | |
1646 | return self.issnapshot(rev) |
|
1646 | return self.issnapshot(rev) | |
1647 | if rev == nullrev: |
|
1647 | if rev == nullrev: | |
1648 | return True |
|
1648 | return True | |
1649 | entry = self.index[rev] |
|
1649 | entry = self.index[rev] | |
1650 | base = entry[3] |
|
1650 | base = entry[3] | |
1651 | if base == rev: |
|
1651 | if base == rev: | |
1652 | return True |
|
1652 | return True | |
1653 | if base == nullrev: |
|
1653 | if base == nullrev: | |
1654 | return True |
|
1654 | return True | |
1655 | p1 = entry[5] |
|
1655 | p1 = entry[5] | |
1656 | p2 = entry[6] |
|
1656 | p2 = entry[6] | |
1657 | if base == p1 or base == p2: |
|
1657 | if base == p1 or base == p2: | |
1658 | return False |
|
1658 | return False | |
1659 | return self.issnapshot(base) |
|
1659 | return self.issnapshot(base) | |
1660 |
|
1660 | |||
1661 | def snapshotdepth(self, rev): |
|
1661 | def snapshotdepth(self, rev): | |
1662 | """number of snapshot in the chain before this one""" |
|
1662 | """number of snapshot in the chain before this one""" | |
1663 | if not self.issnapshot(rev): |
|
1663 | if not self.issnapshot(rev): | |
1664 | raise error.ProgrammingError(b'revision %d not a snapshot') |
|
1664 | raise error.ProgrammingError(b'revision %d not a snapshot') | |
1665 | return len(self._deltachain(rev)[0]) - 1 |
|
1665 | return len(self._deltachain(rev)[0]) - 1 | |
1666 |
|
1666 | |||
1667 | def revdiff(self, rev1, rev2): |
|
1667 | def revdiff(self, rev1, rev2): | |
1668 | """return or calculate a delta between two revisions |
|
1668 | """return or calculate a delta between two revisions | |
1669 |
|
1669 | |||
1670 | The delta calculated is in binary form and is intended to be written to |
|
1670 | The delta calculated is in binary form and is intended to be written to | |
1671 | revlog data directly. So this function needs raw revision data. |
|
1671 | revlog data directly. So this function needs raw revision data. | |
1672 | """ |
|
1672 | """ | |
1673 | if rev1 != nullrev and self.deltaparent(rev2) == rev1: |
|
1673 | if rev1 != nullrev and self.deltaparent(rev2) == rev1: | |
1674 | return bytes(self._chunk(rev2)) |
|
1674 | return bytes(self._chunk(rev2)) | |
1675 |
|
1675 | |||
1676 | return mdiff.textdiff(self.rawdata(rev1), self.rawdata(rev2)) |
|
1676 | return mdiff.textdiff(self.rawdata(rev1), self.rawdata(rev2)) | |
1677 |
|
1677 | |||
1678 | def _processflags(self, text, flags, operation, raw=False): |
|
1678 | def _processflags(self, text, flags, operation, raw=False): | |
1679 | """deprecated entry point to access flag processors""" |
|
1679 | """deprecated entry point to access flag processors""" | |
1680 | msg = b'_processflag(...) use the specialized variant' |
|
1680 | msg = b'_processflag(...) use the specialized variant' | |
1681 | util.nouideprecwarn(msg, b'5.2', stacklevel=2) |
|
1681 | util.nouideprecwarn(msg, b'5.2', stacklevel=2) | |
1682 | if raw: |
|
1682 | if raw: | |
1683 | return text, flagutil.processflagsraw(self, text, flags) |
|
1683 | return text, flagutil.processflagsraw(self, text, flags) | |
1684 | elif operation == b'read': |
|
1684 | elif operation == b'read': | |
1685 | return flagutil.processflagsread(self, text, flags) |
|
1685 | return flagutil.processflagsread(self, text, flags) | |
1686 | else: # write operation |
|
1686 | else: # write operation | |
1687 | return flagutil.processflagswrite(self, text, flags) |
|
1687 | return flagutil.processflagswrite(self, text, flags) | |
1688 |
|
1688 | |||
1689 | def revision(self, nodeorrev, _df=None, raw=False): |
|
1689 | def revision(self, nodeorrev, _df=None, raw=False): | |
1690 | """return an uncompressed revision of a given node or revision |
|
1690 | """return an uncompressed revision of a given node or revision | |
1691 | number. |
|
1691 | number. | |
1692 |
|
1692 | |||
1693 | _df - an existing file handle to read from. (internal-only) |
|
1693 | _df - an existing file handle to read from. (internal-only) | |
1694 | raw - an optional argument specifying if the revision data is to be |
|
1694 | raw - an optional argument specifying if the revision data is to be | |
1695 | treated as raw data when applying flag transforms. 'raw' should be set |
|
1695 | treated as raw data when applying flag transforms. 'raw' should be set | |
1696 | to True when generating changegroups or in debug commands. |
|
1696 | to True when generating changegroups or in debug commands. | |
1697 | """ |
|
1697 | """ | |
1698 | if raw: |
|
1698 | if raw: | |
1699 | msg = ( |
|
1699 | msg = ( | |
1700 | b'revlog.revision(..., raw=True) is deprecated, ' |
|
1700 | b'revlog.revision(..., raw=True) is deprecated, ' | |
1701 | b'use revlog.rawdata(...)' |
|
1701 | b'use revlog.rawdata(...)' | |
1702 | ) |
|
1702 | ) | |
1703 | util.nouideprecwarn(msg, b'5.2', stacklevel=2) |
|
1703 | util.nouideprecwarn(msg, b'5.2', stacklevel=2) | |
1704 | return self._revisiondata(nodeorrev, _df, raw=raw)[0] |
|
1704 | return self._revisiondata(nodeorrev, _df, raw=raw)[0] | |
1705 |
|
1705 | |||
1706 | def sidedata(self, nodeorrev, _df=None): |
|
1706 | def sidedata(self, nodeorrev, _df=None): | |
1707 | """a map of extra data related to the changeset but not part of the hash |
|
1707 | """a map of extra data related to the changeset but not part of the hash | |
1708 |
|
1708 | |||
1709 | This function currently return a dictionary. However, more advanced |
|
1709 | This function currently return a dictionary. However, more advanced | |
1710 | mapping object will likely be used in the future for a more |
|
1710 | mapping object will likely be used in the future for a more | |
1711 | efficient/lazy code. |
|
1711 | efficient/lazy code. | |
1712 | """ |
|
1712 | """ | |
1713 | return self._revisiondata(nodeorrev, _df)[1] |
|
1713 | return self._revisiondata(nodeorrev, _df)[1] | |
1714 |
|
1714 | |||
1715 | def _revisiondata(self, nodeorrev, _df=None, raw=False): |
|
1715 | def _revisiondata(self, nodeorrev, _df=None, raw=False): | |
1716 | # deal with <nodeorrev> argument type |
|
1716 | # deal with <nodeorrev> argument type | |
1717 | if isinstance(nodeorrev, int): |
|
1717 | if isinstance(nodeorrev, int): | |
1718 | rev = nodeorrev |
|
1718 | rev = nodeorrev | |
1719 | node = self.node(rev) |
|
1719 | node = self.node(rev) | |
1720 | else: |
|
1720 | else: | |
1721 | node = nodeorrev |
|
1721 | node = nodeorrev | |
1722 | rev = None |
|
1722 | rev = None | |
1723 |
|
1723 | |||
1724 | # fast path the special `nullid` rev |
|
1724 | # fast path the special `nullid` rev | |
1725 | if node == nullid: |
|
1725 | if node == nullid: | |
1726 | return b"", {} |
|
1726 | return b"", {} | |
1727 |
|
1727 | |||
1728 | # The text as stored inside the revlog. Might be the revision or might |
|
1728 | # The text as stored inside the revlog. Might be the revision or might | |
1729 | # need to be processed to retrieve the revision. |
|
1729 | # need to be processed to retrieve the revision. | |
1730 | rawtext = None |
|
1730 | rawtext = None | |
1731 |
|
1731 | |||
1732 | rev, rawtext, validated = self._rawtext(node, rev, _df=_df) |
|
1732 | rev, rawtext, validated = self._rawtext(node, rev, _df=_df) | |
1733 |
|
1733 | |||
1734 | if raw and validated: |
|
1734 | if raw and validated: | |
1735 | # if we don't want to process the raw text and that raw |
|
1735 | # if we don't want to process the raw text and that raw | |
1736 | # text is cached, we can exit early. |
|
1736 | # text is cached, we can exit early. | |
1737 | return rawtext, {} |
|
1737 | return rawtext, {} | |
1738 | if rev is None: |
|
1738 | if rev is None: | |
1739 | rev = self.rev(node) |
|
1739 | rev = self.rev(node) | |
1740 | # the revlog's flag for this revision |
|
1740 | # the revlog's flag for this revision | |
1741 | # (usually alter its state or content) |
|
1741 | # (usually alter its state or content) | |
1742 | flags = self.flags(rev) |
|
1742 | flags = self.flags(rev) | |
1743 |
|
1743 | |||
1744 | if validated and flags == REVIDX_DEFAULT_FLAGS: |
|
1744 | if validated and flags == REVIDX_DEFAULT_FLAGS: | |
1745 | # no extra flags set, no flag processor runs, text = rawtext |
|
1745 | # no extra flags set, no flag processor runs, text = rawtext | |
1746 | return rawtext, {} |
|
1746 | return rawtext, {} | |
1747 |
|
1747 | |||
1748 | sidedata = {} |
|
1748 | sidedata = {} | |
1749 | if raw: |
|
1749 | if raw: | |
1750 | validatehash = flagutil.processflagsraw(self, rawtext, flags) |
|
1750 | validatehash = flagutil.processflagsraw(self, rawtext, flags) | |
1751 | text = rawtext |
|
1751 | text = rawtext | |
1752 | else: |
|
1752 | else: | |
1753 | try: |
|
1753 | try: | |
1754 | r = flagutil.processflagsread(self, rawtext, flags) |
|
1754 | r = flagutil.processflagsread(self, rawtext, flags) | |
1755 | except error.SidedataHashError as exc: |
|
1755 | except error.SidedataHashError as exc: | |
1756 | msg = _(b"integrity check failed on %s:%s sidedata key %d") |
|
1756 | msg = _(b"integrity check failed on %s:%s sidedata key %d") | |
1757 | msg %= (self.indexfile, pycompat.bytestr(rev), exc.sidedatakey) |
|
1757 | msg %= (self.indexfile, pycompat.bytestr(rev), exc.sidedatakey) | |
1758 | raise error.RevlogError(msg) |
|
1758 | raise error.RevlogError(msg) | |
1759 | text, validatehash, sidedata = r |
|
1759 | text, validatehash, sidedata = r | |
1760 | if validatehash: |
|
1760 | if validatehash: | |
1761 | self.checkhash(text, node, rev=rev) |
|
1761 | self.checkhash(text, node, rev=rev) | |
1762 | if not validated: |
|
1762 | if not validated: | |
1763 | self._revisioncache = (node, rev, rawtext) |
|
1763 | self._revisioncache = (node, rev, rawtext) | |
1764 |
|
1764 | |||
1765 | return text, sidedata |
|
1765 | return text, sidedata | |
1766 |
|
1766 | |||
1767 | def _rawtext(self, node, rev, _df=None): |
|
1767 | def _rawtext(self, node, rev, _df=None): | |
1768 | """return the possibly unvalidated rawtext for a revision |
|
1768 | """return the possibly unvalidated rawtext for a revision | |
1769 |
|
1769 | |||
1770 | returns (rev, rawtext, validated) |
|
1770 | returns (rev, rawtext, validated) | |
1771 | """ |
|
1771 | """ | |
1772 |
|
1772 | |||
1773 | # revision in the cache (could be useful to apply delta) |
|
1773 | # revision in the cache (could be useful to apply delta) | |
1774 | cachedrev = None |
|
1774 | cachedrev = None | |
1775 | # An intermediate text to apply deltas to |
|
1775 | # An intermediate text to apply deltas to | |
1776 | basetext = None |
|
1776 | basetext = None | |
1777 |
|
1777 | |||
1778 | # Check if we have the entry in cache |
|
1778 | # Check if we have the entry in cache | |
1779 | # The cache entry looks like (node, rev, rawtext) |
|
1779 | # The cache entry looks like (node, rev, rawtext) | |
1780 | if self._revisioncache: |
|
1780 | if self._revisioncache: | |
1781 | if self._revisioncache[0] == node: |
|
1781 | if self._revisioncache[0] == node: | |
1782 | return (rev, self._revisioncache[2], True) |
|
1782 | return (rev, self._revisioncache[2], True) | |
1783 | cachedrev = self._revisioncache[1] |
|
1783 | cachedrev = self._revisioncache[1] | |
1784 |
|
1784 | |||
1785 | if rev is None: |
|
1785 | if rev is None: | |
1786 | rev = self.rev(node) |
|
1786 | rev = self.rev(node) | |
1787 |
|
1787 | |||
1788 | chain, stopped = self._deltachain(rev, stoprev=cachedrev) |
|
1788 | chain, stopped = self._deltachain(rev, stoprev=cachedrev) | |
1789 | if stopped: |
|
1789 | if stopped: | |
1790 | basetext = self._revisioncache[2] |
|
1790 | basetext = self._revisioncache[2] | |
1791 |
|
1791 | |||
1792 | # drop cache to save memory, the caller is expected to |
|
1792 | # drop cache to save memory, the caller is expected to | |
1793 | # update self._revisioncache after validating the text |
|
1793 | # update self._revisioncache after validating the text | |
1794 | self._revisioncache = None |
|
1794 | self._revisioncache = None | |
1795 |
|
1795 | |||
1796 | targetsize = None |
|
1796 | targetsize = None | |
1797 | rawsize = self.index[rev][2] |
|
1797 | rawsize = self.index[rev][2] | |
1798 | if 0 <= rawsize: |
|
1798 | if 0 <= rawsize: | |
1799 | targetsize = 4 * rawsize |
|
1799 | targetsize = 4 * rawsize | |
1800 |
|
1800 | |||
1801 | bins = self._chunks(chain, df=_df, targetsize=targetsize) |
|
1801 | bins = self._chunks(chain, df=_df, targetsize=targetsize) | |
1802 | if basetext is None: |
|
1802 | if basetext is None: | |
1803 | basetext = bytes(bins[0]) |
|
1803 | basetext = bytes(bins[0]) | |
1804 | bins = bins[1:] |
|
1804 | bins = bins[1:] | |
1805 |
|
1805 | |||
1806 | rawtext = mdiff.patches(basetext, bins) |
|
1806 | rawtext = mdiff.patches(basetext, bins) | |
1807 | del basetext # let us have a chance to free memory early |
|
1807 | del basetext # let us have a chance to free memory early | |
1808 | return (rev, rawtext, False) |
|
1808 | return (rev, rawtext, False) | |
1809 |
|
1809 | |||
1810 | def rawdata(self, nodeorrev, _df=None): |
|
1810 | def rawdata(self, nodeorrev, _df=None): | |
1811 | """return an uncompressed raw data of a given node or revision number. |
|
1811 | """return an uncompressed raw data of a given node or revision number. | |
1812 |
|
1812 | |||
1813 | _df - an existing file handle to read from. (internal-only) |
|
1813 | _df - an existing file handle to read from. (internal-only) | |
1814 | """ |
|
1814 | """ | |
1815 | return self._revisiondata(nodeorrev, _df, raw=True)[0] |
|
1815 | return self._revisiondata(nodeorrev, _df, raw=True)[0] | |
1816 |
|
1816 | |||
1817 | def hash(self, text, p1, p2): |
|
1817 | def hash(self, text, p1, p2): | |
1818 | """Compute a node hash. |
|
1818 | """Compute a node hash. | |
1819 |
|
1819 | |||
1820 | Available as a function so that subclasses can replace the hash |
|
1820 | Available as a function so that subclasses can replace the hash | |
1821 | as needed. |
|
1821 | as needed. | |
1822 | """ |
|
1822 | """ | |
1823 | return storageutil.hashrevisionsha1(text, p1, p2) |
|
1823 | return storageutil.hashrevisionsha1(text, p1, p2) | |
1824 |
|
1824 | |||
1825 | def checkhash(self, text, node, p1=None, p2=None, rev=None): |
|
1825 | def checkhash(self, text, node, p1=None, p2=None, rev=None): | |
1826 | """Check node hash integrity. |
|
1826 | """Check node hash integrity. | |
1827 |
|
1827 | |||
1828 | Available as a function so that subclasses can extend hash mismatch |
|
1828 | Available as a function so that subclasses can extend hash mismatch | |
1829 | behaviors as needed. |
|
1829 | behaviors as needed. | |
1830 | """ |
|
1830 | """ | |
1831 | try: |
|
1831 | try: | |
1832 | if p1 is None and p2 is None: |
|
1832 | if p1 is None and p2 is None: | |
1833 | p1, p2 = self.parents(node) |
|
1833 | p1, p2 = self.parents(node) | |
1834 | if node != self.hash(text, p1, p2): |
|
1834 | if node != self.hash(text, p1, p2): | |
1835 | # Clear the revision cache on hash failure. The revision cache |
|
1835 | # Clear the revision cache on hash failure. The revision cache | |
1836 | # only stores the raw revision and clearing the cache does have |
|
1836 | # only stores the raw revision and clearing the cache does have | |
1837 | # the side-effect that we won't have a cache hit when the raw |
|
1837 | # the side-effect that we won't have a cache hit when the raw | |
1838 | # revision data is accessed. But this case should be rare and |
|
1838 | # revision data is accessed. But this case should be rare and | |
1839 | # it is extra work to teach the cache about the hash |
|
1839 | # it is extra work to teach the cache about the hash | |
1840 | # verification state. |
|
1840 | # verification state. | |
1841 | if self._revisioncache and self._revisioncache[0] == node: |
|
1841 | if self._revisioncache and self._revisioncache[0] == node: | |
1842 | self._revisioncache = None |
|
1842 | self._revisioncache = None | |
1843 |
|
1843 | |||
1844 | revornode = rev |
|
1844 | revornode = rev | |
1845 | if revornode is None: |
|
1845 | if revornode is None: | |
1846 | revornode = templatefilters.short(hex(node)) |
|
1846 | revornode = templatefilters.short(hex(node)) | |
1847 | raise error.RevlogError( |
|
1847 | raise error.RevlogError( | |
1848 | _(b"integrity check failed on %s:%s") |
|
1848 | _(b"integrity check failed on %s:%s") | |
1849 | % (self.indexfile, pycompat.bytestr(revornode)) |
|
1849 | % (self.indexfile, pycompat.bytestr(revornode)) | |
1850 | ) |
|
1850 | ) | |
1851 | except error.RevlogError: |
|
1851 | except error.RevlogError: | |
1852 | if self._censorable and storageutil.iscensoredtext(text): |
|
1852 | if self._censorable and storageutil.iscensoredtext(text): | |
1853 | raise error.CensoredNodeError(self.indexfile, node, text) |
|
1853 | raise error.CensoredNodeError(self.indexfile, node, text) | |
1854 | raise |
|
1854 | raise | |
1855 |
|
1855 | |||
1856 | def _enforceinlinesize(self, tr, fp=None): |
|
1856 | def _enforceinlinesize(self, tr, fp=None): | |
1857 | """Check if the revlog is too big for inline and convert if so. |
|
1857 | """Check if the revlog is too big for inline and convert if so. | |
1858 |
|
1858 | |||
1859 | This should be called after revisions are added to the revlog. If the |
|
1859 | This should be called after revisions are added to the revlog. If the | |
1860 | revlog has grown too large to be an inline revlog, it will convert it |
|
1860 | revlog has grown too large to be an inline revlog, it will convert it | |
1861 | to use multiple index and data files. |
|
1861 | to use multiple index and data files. | |
1862 | """ |
|
1862 | """ | |
1863 | tiprev = len(self) - 1 |
|
1863 | tiprev = len(self) - 1 | |
1864 | if ( |
|
1864 | if ( | |
1865 | not self._inline |
|
1865 | not self._inline | |
1866 | or (self.start(tiprev) + self.length(tiprev)) < _maxinline |
|
1866 | or (self.start(tiprev) + self.length(tiprev)) < _maxinline | |
1867 | ): |
|
1867 | ): | |
1868 | return |
|
1868 | return | |
1869 |
|
1869 | |||
1870 | trinfo = tr.find(self.indexfile) |
|
1870 | trinfo = tr.find(self.indexfile) | |
1871 | if trinfo is None: |
|
1871 | if trinfo is None: | |
1872 | raise error.RevlogError( |
|
1872 | raise error.RevlogError( | |
1873 | _(b"%s not found in the transaction") % self.indexfile |
|
1873 | _(b"%s not found in the transaction") % self.indexfile | |
1874 | ) |
|
1874 | ) | |
1875 |
|
1875 | |||
1876 | trindex = trinfo[2] |
|
1876 | trindex = trinfo[2] | |
1877 | if trindex is not None: |
|
1877 | if trindex is not None: | |
1878 | dataoff = self.start(trindex) |
|
1878 | dataoff = self.start(trindex) | |
1879 | else: |
|
1879 | else: | |
1880 | # revlog was stripped at start of transaction, use all leftover data |
|
1880 | # revlog was stripped at start of transaction, use all leftover data | |
1881 | trindex = len(self) - 1 |
|
1881 | trindex = len(self) - 1 | |
1882 | dataoff = self.end(tiprev) |
|
1882 | dataoff = self.end(tiprev) | |
1883 |
|
1883 | |||
1884 | tr.add(self.datafile, dataoff) |
|
1884 | tr.add(self.datafile, dataoff) | |
1885 |
|
1885 | |||
1886 | if fp: |
|
1886 | if fp: | |
1887 | fp.flush() |
|
1887 | fp.flush() | |
1888 | fp.close() |
|
1888 | fp.close() | |
1889 | # We can't use the cached file handle after close(). So prevent |
|
1889 | # We can't use the cached file handle after close(). So prevent | |
1890 | # its usage. |
|
1890 | # its usage. | |
1891 | self._writinghandles = None |
|
1891 | self._writinghandles = None | |
1892 |
|
1892 | |||
1893 | with self._indexfp(b'r') as ifh, self._datafp(b'w') as dfh: |
|
1893 | with self._indexfp(b'r') as ifh, self._datafp(b'w') as dfh: | |
1894 | for r in self: |
|
1894 | for r in self: | |
1895 | dfh.write(self._getsegmentforrevs(r, r, df=ifh)[1]) |
|
1895 | dfh.write(self._getsegmentforrevs(r, r, df=ifh)[1]) | |
1896 |
|
1896 | |||
1897 | with self._indexfp(b'w') as fp: |
|
1897 | with self._indexfp(b'w') as fp: | |
1898 | self.version &= ~FLAG_INLINE_DATA |
|
1898 | self.version &= ~FLAG_INLINE_DATA | |
1899 | self._inline = False |
|
1899 | self._inline = False | |
1900 | io = self._io |
|
1900 | io = self._io | |
1901 | for i in self: |
|
1901 | for i in self: | |
1902 | e = io.packentry(self.index[i], self.node, self.version, i) |
|
1902 | e = io.packentry(self.index[i], self.node, self.version, i) | |
1903 | fp.write(e) |
|
1903 | fp.write(e) | |
1904 |
|
1904 | |||
1905 | # the temp file replace the real index when we exit the context |
|
1905 | # the temp file replace the real index when we exit the context | |
1906 | # manager |
|
1906 | # manager | |
1907 |
|
1907 | |||
1908 | tr.replace(self.indexfile, trindex * self._io.size) |
|
1908 | tr.replace(self.indexfile, trindex * self._io.size) | |
1909 | self._chunkclear() |
|
1909 | self._chunkclear() | |
1910 |
|
1910 | |||
1911 | def _nodeduplicatecallback(self, transaction, node): |
|
1911 | def _nodeduplicatecallback(self, transaction, node): | |
1912 | """called when trying to add a node already stored. |
|
1912 | """called when trying to add a node already stored. | |
1913 | """ |
|
1913 | """ | |
1914 |
|
1914 | |||
1915 | def addrevision( |
|
1915 | def addrevision( | |
1916 | self, |
|
1916 | self, | |
1917 | text, |
|
1917 | text, | |
1918 | transaction, |
|
1918 | transaction, | |
1919 | link, |
|
1919 | link, | |
1920 | p1, |
|
1920 | p1, | |
1921 | p2, |
|
1921 | p2, | |
1922 | cachedelta=None, |
|
1922 | cachedelta=None, | |
1923 | node=None, |
|
1923 | node=None, | |
1924 | flags=REVIDX_DEFAULT_FLAGS, |
|
1924 | flags=REVIDX_DEFAULT_FLAGS, | |
1925 | deltacomputer=None, |
|
1925 | deltacomputer=None, | |
1926 | sidedata=None, |
|
1926 | sidedata=None, | |
1927 | ): |
|
1927 | ): | |
1928 | """add a revision to the log |
|
1928 | """add a revision to the log | |
1929 |
|
1929 | |||
1930 | text - the revision data to add |
|
1930 | text - the revision data to add | |
1931 | transaction - the transaction object used for rollback |
|
1931 | transaction - the transaction object used for rollback | |
1932 | link - the linkrev data to add |
|
1932 | link - the linkrev data to add | |
1933 | p1, p2 - the parent nodeids of the revision |
|
1933 | p1, p2 - the parent nodeids of the revision | |
1934 | cachedelta - an optional precomputed delta |
|
1934 | cachedelta - an optional precomputed delta | |
1935 | node - nodeid of revision; typically node is not specified, and it is |
|
1935 | node - nodeid of revision; typically node is not specified, and it is | |
1936 | computed by default as hash(text, p1, p2), however subclasses might |
|
1936 | computed by default as hash(text, p1, p2), however subclasses might | |
1937 | use different hashing method (and override checkhash() in such case) |
|
1937 | use different hashing method (and override checkhash() in such case) | |
1938 | flags - the known flags to set on the revision |
|
1938 | flags - the known flags to set on the revision | |
1939 | deltacomputer - an optional deltacomputer instance shared between |
|
1939 | deltacomputer - an optional deltacomputer instance shared between | |
1940 | multiple calls |
|
1940 | multiple calls | |
1941 | """ |
|
1941 | """ | |
1942 | if link == nullrev: |
|
1942 | if link == nullrev: | |
1943 | raise error.RevlogError( |
|
1943 | raise error.RevlogError( | |
1944 | _(b"attempted to add linkrev -1 to %s") % self.indexfile |
|
1944 | _(b"attempted to add linkrev -1 to %s") % self.indexfile | |
1945 | ) |
|
1945 | ) | |
1946 |
|
1946 | |||
1947 | if sidedata is None: |
|
1947 | if sidedata is None: | |
1948 | sidedata = {} |
|
1948 | sidedata = {} | |
1949 | flags = flags & ~REVIDX_SIDEDATA |
|
1949 | flags = flags & ~REVIDX_SIDEDATA | |
1950 | elif not self.hassidedata: |
|
1950 | elif not self.hassidedata: | |
1951 | raise error.ProgrammingError( |
|
1951 | raise error.ProgrammingError( | |
1952 | _(b"trying to add sidedata to a revlog who don't support them") |
|
1952 | _(b"trying to add sidedata to a revlog who don't support them") | |
1953 | ) |
|
1953 | ) | |
1954 | else: |
|
1954 | else: | |
1955 | flags |= REVIDX_SIDEDATA |
|
1955 | flags |= REVIDX_SIDEDATA | |
1956 |
|
1956 | |||
1957 | if flags: |
|
1957 | if flags: | |
1958 | node = node or self.hash(text, p1, p2) |
|
1958 | node = node or self.hash(text, p1, p2) | |
1959 |
|
1959 | |||
1960 | rawtext, validatehash = flagutil.processflagswrite( |
|
1960 | rawtext, validatehash = flagutil.processflagswrite( | |
1961 | self, text, flags, sidedata=sidedata |
|
1961 | self, text, flags, sidedata=sidedata | |
1962 | ) |
|
1962 | ) | |
1963 |
|
1963 | |||
1964 | # If the flag processor modifies the revision data, ignore any provided |
|
1964 | # If the flag processor modifies the revision data, ignore any provided | |
1965 | # cachedelta. |
|
1965 | # cachedelta. | |
1966 | if rawtext != text: |
|
1966 | if rawtext != text: | |
1967 | cachedelta = None |
|
1967 | cachedelta = None | |
1968 |
|
1968 | |||
1969 | if len(rawtext) > _maxentrysize: |
|
1969 | if len(rawtext) > _maxentrysize: | |
1970 | raise error.RevlogError( |
|
1970 | raise error.RevlogError( | |
1971 | _( |
|
1971 | _( | |
1972 | b"%s: size of %d bytes exceeds maximum revlog storage of 2GiB" |
|
1972 | b"%s: size of %d bytes exceeds maximum revlog storage of 2GiB" | |
1973 | ) |
|
1973 | ) | |
1974 | % (self.indexfile, len(rawtext)) |
|
1974 | % (self.indexfile, len(rawtext)) | |
1975 | ) |
|
1975 | ) | |
1976 |
|
1976 | |||
1977 | node = node or self.hash(rawtext, p1, p2) |
|
1977 | node = node or self.hash(rawtext, p1, p2) | |
1978 | if node in self.nodemap: |
|
1978 | if node in self.nodemap: | |
1979 | return node |
|
1979 | return node | |
1980 |
|
1980 | |||
1981 | if validatehash: |
|
1981 | if validatehash: | |
1982 | self.checkhash(rawtext, node, p1=p1, p2=p2) |
|
1982 | self.checkhash(rawtext, node, p1=p1, p2=p2) | |
1983 |
|
1983 | |||
1984 | return self.addrawrevision( |
|
1984 | return self.addrawrevision( | |
1985 | rawtext, |
|
1985 | rawtext, | |
1986 | transaction, |
|
1986 | transaction, | |
1987 | link, |
|
1987 | link, | |
1988 | p1, |
|
1988 | p1, | |
1989 | p2, |
|
1989 | p2, | |
1990 | node, |
|
1990 | node, | |
1991 | flags, |
|
1991 | flags, | |
1992 | cachedelta=cachedelta, |
|
1992 | cachedelta=cachedelta, | |
1993 | deltacomputer=deltacomputer, |
|
1993 | deltacomputer=deltacomputer, | |
1994 | ) |
|
1994 | ) | |
1995 |
|
1995 | |||
1996 | def addrawrevision( |
|
1996 | def addrawrevision( | |
1997 | self, |
|
1997 | self, | |
1998 | rawtext, |
|
1998 | rawtext, | |
1999 | transaction, |
|
1999 | transaction, | |
2000 | link, |
|
2000 | link, | |
2001 | p1, |
|
2001 | p1, | |
2002 | p2, |
|
2002 | p2, | |
2003 | node, |
|
2003 | node, | |
2004 | flags, |
|
2004 | flags, | |
2005 | cachedelta=None, |
|
2005 | cachedelta=None, | |
2006 | deltacomputer=None, |
|
2006 | deltacomputer=None, | |
2007 | ): |
|
2007 | ): | |
2008 | """add a raw revision with known flags, node and parents |
|
2008 | """add a raw revision with known flags, node and parents | |
2009 | useful when reusing a revision not stored in this revlog (ex: received |
|
2009 | useful when reusing a revision not stored in this revlog (ex: received | |
2010 | over wire, or read from an external bundle). |
|
2010 | over wire, or read from an external bundle). | |
2011 | """ |
|
2011 | """ | |
2012 | dfh = None |
|
2012 | dfh = None | |
2013 | if not self._inline: |
|
2013 | if not self._inline: | |
2014 | dfh = self._datafp(b"a+") |
|
2014 | dfh = self._datafp(b"a+") | |
2015 | ifh = self._indexfp(b"a+") |
|
2015 | ifh = self._indexfp(b"a+") | |
2016 | try: |
|
2016 | try: | |
2017 | return self._addrevision( |
|
2017 | return self._addrevision( | |
2018 | node, |
|
2018 | node, | |
2019 | rawtext, |
|
2019 | rawtext, | |
2020 | transaction, |
|
2020 | transaction, | |
2021 | link, |
|
2021 | link, | |
2022 | p1, |
|
2022 | p1, | |
2023 | p2, |
|
2023 | p2, | |
2024 | flags, |
|
2024 | flags, | |
2025 | cachedelta, |
|
2025 | cachedelta, | |
2026 | ifh, |
|
2026 | ifh, | |
2027 | dfh, |
|
2027 | dfh, | |
2028 | deltacomputer=deltacomputer, |
|
2028 | deltacomputer=deltacomputer, | |
2029 | ) |
|
2029 | ) | |
2030 | finally: |
|
2030 | finally: | |
2031 | if dfh: |
|
2031 | if dfh: | |
2032 | dfh.close() |
|
2032 | dfh.close() | |
2033 | ifh.close() |
|
2033 | ifh.close() | |
2034 |
|
2034 | |||
2035 | def compress(self, data): |
|
2035 | def compress(self, data): | |
2036 | """Generate a possibly-compressed representation of data.""" |
|
2036 | """Generate a possibly-compressed representation of data.""" | |
2037 | if not data: |
|
2037 | if not data: | |
2038 | return b'', data |
|
2038 | return b'', data | |
2039 |
|
2039 | |||
2040 | compressed = self._compressor.compress(data) |
|
2040 | compressed = self._compressor.compress(data) | |
2041 |
|
2041 | |||
2042 | if compressed: |
|
2042 | if compressed: | |
2043 | # The revlog compressor added the header in the returned data. |
|
2043 | # The revlog compressor added the header in the returned data. | |
2044 | return b'', compressed |
|
2044 | return b'', compressed | |
2045 |
|
2045 | |||
2046 | if data[0:1] == b'\0': |
|
2046 | if data[0:1] == b'\0': | |
2047 | return b'', data |
|
2047 | return b'', data | |
2048 | return b'u', data |
|
2048 | return b'u', data | |
2049 |
|
2049 | |||
2050 | def decompress(self, data): |
|
2050 | def decompress(self, data): | |
2051 | """Decompress a revlog chunk. |
|
2051 | """Decompress a revlog chunk. | |
2052 |
|
2052 | |||
2053 | The chunk is expected to begin with a header identifying the |
|
2053 | The chunk is expected to begin with a header identifying the | |
2054 | format type so it can be routed to an appropriate decompressor. |
|
2054 | format type so it can be routed to an appropriate decompressor. | |
2055 | """ |
|
2055 | """ | |
2056 | if not data: |
|
2056 | if not data: | |
2057 | return data |
|
2057 | return data | |
2058 |
|
2058 | |||
2059 | # Revlogs are read much more frequently than they are written and many |
|
2059 | # Revlogs are read much more frequently than they are written and many | |
2060 | # chunks only take microseconds to decompress, so performance is |
|
2060 | # chunks only take microseconds to decompress, so performance is | |
2061 | # important here. |
|
2061 | # important here. | |
2062 | # |
|
2062 | # | |
2063 | # We can make a few assumptions about revlogs: |
|
2063 | # We can make a few assumptions about revlogs: | |
2064 | # |
|
2064 | # | |
2065 | # 1) the majority of chunks will be compressed (as opposed to inline |
|
2065 | # 1) the majority of chunks will be compressed (as opposed to inline | |
2066 | # raw data). |
|
2066 | # raw data). | |
2067 | # 2) decompressing *any* data will likely by at least 10x slower than |
|
2067 | # 2) decompressing *any* data will likely by at least 10x slower than | |
2068 | # returning raw inline data. |
|
2068 | # returning raw inline data. | |
2069 | # 3) we want to prioritize common and officially supported compression |
|
2069 | # 3) we want to prioritize common and officially supported compression | |
2070 | # engines |
|
2070 | # engines | |
2071 | # |
|
2071 | # | |
2072 | # It follows that we want to optimize for "decompress compressed data |
|
2072 | # It follows that we want to optimize for "decompress compressed data | |
2073 | # when encoded with common and officially supported compression engines" |
|
2073 | # when encoded with common and officially supported compression engines" | |
2074 | # case over "raw data" and "data encoded by less common or non-official |
|
2074 | # case over "raw data" and "data encoded by less common or non-official | |
2075 | # compression engines." That is why we have the inline lookup first |
|
2075 | # compression engines." That is why we have the inline lookup first | |
2076 | # followed by the compengines lookup. |
|
2076 | # followed by the compengines lookup. | |
2077 | # |
|
2077 | # | |
2078 | # According to `hg perfrevlogchunks`, this is ~0.5% faster for zlib |
|
2078 | # According to `hg perfrevlogchunks`, this is ~0.5% faster for zlib | |
2079 | # compressed chunks. And this matters for changelog and manifest reads. |
|
2079 | # compressed chunks. And this matters for changelog and manifest reads. | |
2080 | t = data[0:1] |
|
2080 | t = data[0:1] | |
2081 |
|
2081 | |||
2082 | if t == b'x': |
|
2082 | if t == b'x': | |
2083 | try: |
|
2083 | try: | |
2084 | return _zlibdecompress(data) |
|
2084 | return _zlibdecompress(data) | |
2085 | except zlib.error as e: |
|
2085 | except zlib.error as e: | |
2086 | raise error.RevlogError( |
|
2086 | raise error.RevlogError( | |
2087 | _(b'revlog decompress error: %s') |
|
2087 | _(b'revlog decompress error: %s') | |
2088 | % stringutil.forcebytestr(e) |
|
2088 | % stringutil.forcebytestr(e) | |
2089 | ) |
|
2089 | ) | |
2090 | # '\0' is more common than 'u' so it goes first. |
|
2090 | # '\0' is more common than 'u' so it goes first. | |
2091 | elif t == b'\0': |
|
2091 | elif t == b'\0': | |
2092 | return data |
|
2092 | return data | |
2093 | elif t == b'u': |
|
2093 | elif t == b'u': | |
2094 | return util.buffer(data, 1) |
|
2094 | return util.buffer(data, 1) | |
2095 |
|
2095 | |||
2096 | try: |
|
2096 | try: | |
2097 | compressor = self._decompressors[t] |
|
2097 | compressor = self._decompressors[t] | |
2098 | except KeyError: |
|
2098 | except KeyError: | |
2099 | try: |
|
2099 | try: | |
2100 | engine = util.compengines.forrevlogheader(t) |
|
2100 | engine = util.compengines.forrevlogheader(t) | |
2101 | compressor = engine.revlogcompressor(self._compengineopts) |
|
2101 | compressor = engine.revlogcompressor(self._compengineopts) | |
2102 | self._decompressors[t] = compressor |
|
2102 | self._decompressors[t] = compressor | |
2103 | except KeyError: |
|
2103 | except KeyError: | |
2104 | raise error.RevlogError(_(b'unknown compression type %r') % t) |
|
2104 | raise error.RevlogError(_(b'unknown compression type %r') % t) | |
2105 |
|
2105 | |||
2106 | return compressor.decompress(data) |
|
2106 | return compressor.decompress(data) | |
2107 |
|
2107 | |||
2108 | def _addrevision( |
|
2108 | def _addrevision( | |
2109 | self, |
|
2109 | self, | |
2110 | node, |
|
2110 | node, | |
2111 | rawtext, |
|
2111 | rawtext, | |
2112 | transaction, |
|
2112 | transaction, | |
2113 | link, |
|
2113 | link, | |
2114 | p1, |
|
2114 | p1, | |
2115 | p2, |
|
2115 | p2, | |
2116 | flags, |
|
2116 | flags, | |
2117 | cachedelta, |
|
2117 | cachedelta, | |
2118 | ifh, |
|
2118 | ifh, | |
2119 | dfh, |
|
2119 | dfh, | |
2120 | alwayscache=False, |
|
2120 | alwayscache=False, | |
2121 | deltacomputer=None, |
|
2121 | deltacomputer=None, | |
2122 | ): |
|
2122 | ): | |
2123 | """internal function to add revisions to the log |
|
2123 | """internal function to add revisions to the log | |
2124 |
|
2124 | |||
2125 | see addrevision for argument descriptions. |
|
2125 | see addrevision for argument descriptions. | |
2126 |
|
2126 | |||
2127 | note: "addrevision" takes non-raw text, "_addrevision" takes raw text. |
|
2127 | note: "addrevision" takes non-raw text, "_addrevision" takes raw text. | |
2128 |
|
2128 | |||
2129 | if "deltacomputer" is not provided or None, a defaultdeltacomputer will |
|
2129 | if "deltacomputer" is not provided or None, a defaultdeltacomputer will | |
2130 | be used. |
|
2130 | be used. | |
2131 |
|
2131 | |||
2132 | invariants: |
|
2132 | invariants: | |
2133 | - rawtext is optional (can be None); if not set, cachedelta must be set. |
|
2133 | - rawtext is optional (can be None); if not set, cachedelta must be set. | |
2134 | if both are set, they must correspond to each other. |
|
2134 | if both are set, they must correspond to each other. | |
2135 | """ |
|
2135 | """ | |
2136 | if node == nullid: |
|
2136 | if node == nullid: | |
2137 | raise error.RevlogError( |
|
2137 | raise error.RevlogError( | |
2138 | _(b"%s: attempt to add null revision") % self.indexfile |
|
2138 | _(b"%s: attempt to add null revision") % self.indexfile | |
2139 | ) |
|
2139 | ) | |
2140 | if node == wdirid or node in wdirfilenodeids: |
|
2140 | if node == wdirid or node in wdirfilenodeids: | |
2141 | raise error.RevlogError( |
|
2141 | raise error.RevlogError( | |
2142 | _(b"%s: attempt to add wdir revision") % self.indexfile |
|
2142 | _(b"%s: attempt to add wdir revision") % self.indexfile | |
2143 | ) |
|
2143 | ) | |
2144 |
|
2144 | |||
2145 | if self._inline: |
|
2145 | if self._inline: | |
2146 | fh = ifh |
|
2146 | fh = ifh | |
2147 | else: |
|
2147 | else: | |
2148 | fh = dfh |
|
2148 | fh = dfh | |
2149 |
|
2149 | |||
2150 | btext = [rawtext] |
|
2150 | btext = [rawtext] | |
2151 |
|
2151 | |||
2152 | curr = len(self) |
|
2152 | curr = len(self) | |
2153 | prev = curr - 1 |
|
2153 | prev = curr - 1 | |
2154 | offset = self.end(prev) |
|
2154 | offset = self.end(prev) | |
2155 | p1r, p2r = self.rev(p1), self.rev(p2) |
|
2155 | p1r, p2r = self.rev(p1), self.rev(p2) | |
2156 |
|
2156 | |||
2157 | # full versions are inserted when the needed deltas |
|
2157 | # full versions are inserted when the needed deltas | |
2158 | # become comparable to the uncompressed text |
|
2158 | # become comparable to the uncompressed text | |
2159 | if rawtext is None: |
|
2159 | if rawtext is None: | |
2160 | # need rawtext size, before changed by flag processors, which is |
|
2160 | # need rawtext size, before changed by flag processors, which is | |
2161 | # the non-raw size. use revlog explicitly to avoid filelog's extra |
|
2161 | # the non-raw size. use revlog explicitly to avoid filelog's extra | |
2162 | # logic that might remove metadata size. |
|
2162 | # logic that might remove metadata size. | |
2163 | textlen = mdiff.patchedsize( |
|
2163 | textlen = mdiff.patchedsize( | |
2164 | revlog.size(self, cachedelta[0]), cachedelta[1] |
|
2164 | revlog.size(self, cachedelta[0]), cachedelta[1] | |
2165 | ) |
|
2165 | ) | |
2166 | else: |
|
2166 | else: | |
2167 | textlen = len(rawtext) |
|
2167 | textlen = len(rawtext) | |
2168 |
|
2168 | |||
2169 | if deltacomputer is None: |
|
2169 | if deltacomputer is None: | |
2170 | deltacomputer = deltautil.deltacomputer(self) |
|
2170 | deltacomputer = deltautil.deltacomputer(self) | |
2171 |
|
2171 | |||
2172 | revinfo = _revisioninfo(node, p1, p2, btext, textlen, cachedelta, flags) |
|
2172 | revinfo = _revisioninfo(node, p1, p2, btext, textlen, cachedelta, flags) | |
2173 |
|
2173 | |||
2174 | deltainfo = deltacomputer.finddeltainfo(revinfo, fh) |
|
2174 | deltainfo = deltacomputer.finddeltainfo(revinfo, fh) | |
2175 |
|
2175 | |||
2176 | e = ( |
|
2176 | e = ( | |
2177 | offset_type(offset, flags), |
|
2177 | offset_type(offset, flags), | |
2178 | deltainfo.deltalen, |
|
2178 | deltainfo.deltalen, | |
2179 | textlen, |
|
2179 | textlen, | |
2180 | deltainfo.base, |
|
2180 | deltainfo.base, | |
2181 | link, |
|
2181 | link, | |
2182 | p1r, |
|
2182 | p1r, | |
2183 | p2r, |
|
2183 | p2r, | |
2184 | node, |
|
2184 | node, | |
2185 | ) |
|
2185 | ) | |
2186 | self.index.append(e) |
|
2186 | self.index.append(e) | |
2187 | self.nodemap[node] = curr |
|
2187 | self.nodemap[node] = curr | |
2188 |
|
2188 | |||
2189 | # Reset the pure node cache start lookup offset to account for new |
|
2189 | # Reset the pure node cache start lookup offset to account for new | |
2190 | # revision. |
|
2190 | # revision. | |
2191 | if self._nodepos is not None: |
|
2191 | if self._nodepos is not None: | |
2192 | self._nodepos = curr |
|
2192 | self._nodepos = curr | |
2193 |
|
2193 | |||
2194 | entry = self._io.packentry(e, self.node, self.version, curr) |
|
2194 | entry = self._io.packentry(e, self.node, self.version, curr) | |
2195 | self._writeentry( |
|
2195 | self._writeentry( | |
2196 | transaction, ifh, dfh, entry, deltainfo.data, link, offset |
|
2196 | transaction, ifh, dfh, entry, deltainfo.data, link, offset | |
2197 | ) |
|
2197 | ) | |
2198 |
|
2198 | |||
2199 | rawtext = btext[0] |
|
2199 | rawtext = btext[0] | |
2200 |
|
2200 | |||
2201 | if alwayscache and rawtext is None: |
|
2201 | if alwayscache and rawtext is None: | |
2202 | rawtext = deltacomputer.buildtext(revinfo, fh) |
|
2202 | rawtext = deltacomputer.buildtext(revinfo, fh) | |
2203 |
|
2203 | |||
2204 | if type(rawtext) == bytes: # only accept immutable objects |
|
2204 | if type(rawtext) == bytes: # only accept immutable objects | |
2205 | self._revisioncache = (node, curr, rawtext) |
|
2205 | self._revisioncache = (node, curr, rawtext) | |
2206 | self._chainbasecache[curr] = deltainfo.chainbase |
|
2206 | self._chainbasecache[curr] = deltainfo.chainbase | |
2207 | return node |
|
2207 | return node | |
2208 |
|
2208 | |||
2209 | def _writeentry(self, transaction, ifh, dfh, entry, data, link, offset): |
|
2209 | def _writeentry(self, transaction, ifh, dfh, entry, data, link, offset): | |
2210 | # Files opened in a+ mode have inconsistent behavior on various |
|
2210 | # Files opened in a+ mode have inconsistent behavior on various | |
2211 | # platforms. Windows requires that a file positioning call be made |
|
2211 | # platforms. Windows requires that a file positioning call be made | |
2212 | # when the file handle transitions between reads and writes. See |
|
2212 | # when the file handle transitions between reads and writes. See | |
2213 | # 3686fa2b8eee and the mixedfilemodewrapper in windows.py. On other |
|
2213 | # 3686fa2b8eee and the mixedfilemodewrapper in windows.py. On other | |
2214 | # platforms, Python or the platform itself can be buggy. Some versions |
|
2214 | # platforms, Python or the platform itself can be buggy. Some versions | |
2215 | # of Solaris have been observed to not append at the end of the file |
|
2215 | # of Solaris have been observed to not append at the end of the file | |
2216 | # if the file was seeked to before the end. See issue4943 for more. |
|
2216 | # if the file was seeked to before the end. See issue4943 for more. | |
2217 | # |
|
2217 | # | |
2218 | # We work around this issue by inserting a seek() before writing. |
|
2218 | # We work around this issue by inserting a seek() before writing. | |
2219 | # Note: This is likely not necessary on Python 3. However, because |
|
2219 | # Note: This is likely not necessary on Python 3. However, because | |
2220 | # the file handle is reused for reads and may be seeked there, we need |
|
2220 | # the file handle is reused for reads and may be seeked there, we need | |
2221 | # to be careful before changing this. |
|
2221 | # to be careful before changing this. | |
2222 | ifh.seek(0, os.SEEK_END) |
|
2222 | ifh.seek(0, os.SEEK_END) | |
2223 | if dfh: |
|
2223 | if dfh: | |
2224 | dfh.seek(0, os.SEEK_END) |
|
2224 | dfh.seek(0, os.SEEK_END) | |
2225 |
|
2225 | |||
2226 | curr = len(self) - 1 |
|
2226 | curr = len(self) - 1 | |
2227 | if not self._inline: |
|
2227 | if not self._inline: | |
2228 | transaction.add(self.datafile, offset) |
|
2228 | transaction.add(self.datafile, offset) | |
2229 | transaction.add(self.indexfile, curr * len(entry)) |
|
2229 | transaction.add(self.indexfile, curr * len(entry)) | |
2230 | if data[0]: |
|
2230 | if data[0]: | |
2231 | dfh.write(data[0]) |
|
2231 | dfh.write(data[0]) | |
2232 | dfh.write(data[1]) |
|
2232 | dfh.write(data[1]) | |
2233 | ifh.write(entry) |
|
2233 | ifh.write(entry) | |
2234 | else: |
|
2234 | else: | |
2235 | offset += curr * self._io.size |
|
2235 | offset += curr * self._io.size | |
2236 | transaction.add(self.indexfile, offset, curr) |
|
2236 | transaction.add(self.indexfile, offset, curr) | |
2237 | ifh.write(entry) |
|
2237 | ifh.write(entry) | |
2238 | ifh.write(data[0]) |
|
2238 | ifh.write(data[0]) | |
2239 | ifh.write(data[1]) |
|
2239 | ifh.write(data[1]) | |
2240 | self._enforceinlinesize(transaction, ifh) |
|
2240 | self._enforceinlinesize(transaction, ifh) | |
2241 |
|
2241 | |||
2242 | def addgroup(self, deltas, linkmapper, transaction, addrevisioncb=None): |
|
2242 | def addgroup(self, deltas, linkmapper, transaction, addrevisioncb=None): | |
2243 | """ |
|
2243 | """ | |
2244 | add a delta group |
|
2244 | add a delta group | |
2245 |
|
2245 | |||
2246 | given a set of deltas, add them to the revision log. the |
|
2246 | given a set of deltas, add them to the revision log. the | |
2247 | first delta is against its parent, which should be in our |
|
2247 | first delta is against its parent, which should be in our | |
2248 | log, the rest are against the previous delta. |
|
2248 | log, the rest are against the previous delta. | |
2249 |
|
2249 | |||
2250 | If ``addrevisioncb`` is defined, it will be called with arguments of |
|
2250 | If ``addrevisioncb`` is defined, it will be called with arguments of | |
2251 | this revlog and the node that was added. |
|
2251 | this revlog and the node that was added. | |
2252 | """ |
|
2252 | """ | |
2253 |
|
2253 | |||
2254 | if self._writinghandles: |
|
2254 | if self._writinghandles: | |
2255 | raise error.ProgrammingError(b'cannot nest addgroup() calls') |
|
2255 | raise error.ProgrammingError(b'cannot nest addgroup() calls') | |
2256 |
|
2256 | |||
2257 | nodes = [] |
|
2257 | nodes = [] | |
2258 |
|
2258 | |||
2259 | r = len(self) |
|
2259 | r = len(self) | |
2260 | end = 0 |
|
2260 | end = 0 | |
2261 | if r: |
|
2261 | if r: | |
2262 | end = self.end(r - 1) |
|
2262 | end = self.end(r - 1) | |
2263 | ifh = self._indexfp(b"a+") |
|
2263 | ifh = self._indexfp(b"a+") | |
2264 | isize = r * self._io.size |
|
2264 | isize = r * self._io.size | |
2265 | if self._inline: |
|
2265 | if self._inline: | |
2266 | transaction.add(self.indexfile, end + isize, r) |
|
2266 | transaction.add(self.indexfile, end + isize, r) | |
2267 | dfh = None |
|
2267 | dfh = None | |
2268 | else: |
|
2268 | else: | |
2269 | transaction.add(self.indexfile, isize, r) |
|
2269 | transaction.add(self.indexfile, isize, r) | |
2270 | transaction.add(self.datafile, end) |
|
2270 | transaction.add(self.datafile, end) | |
2271 | dfh = self._datafp(b"a+") |
|
2271 | dfh = self._datafp(b"a+") | |
2272 |
|
2272 | |||
2273 | def flush(): |
|
2273 | def flush(): | |
2274 | if dfh: |
|
2274 | if dfh: | |
2275 | dfh.flush() |
|
2275 | dfh.flush() | |
2276 | ifh.flush() |
|
2276 | ifh.flush() | |
2277 |
|
2277 | |||
2278 | self._writinghandles = (ifh, dfh) |
|
2278 | self._writinghandles = (ifh, dfh) | |
2279 |
|
2279 | |||
2280 | try: |
|
2280 | try: | |
2281 | deltacomputer = deltautil.deltacomputer(self) |
|
2281 | deltacomputer = deltautil.deltacomputer(self) | |
2282 | # loop through our set of deltas |
|
2282 | # loop through our set of deltas | |
2283 | for data in deltas: |
|
2283 | for data in deltas: | |
2284 | node, p1, p2, linknode, deltabase, delta, flags = data |
|
2284 | node, p1, p2, linknode, deltabase, delta, flags = data | |
2285 | link = linkmapper(linknode) |
|
2285 | link = linkmapper(linknode) | |
2286 | flags = flags or REVIDX_DEFAULT_FLAGS |
|
2286 | flags = flags or REVIDX_DEFAULT_FLAGS | |
2287 |
|
2287 | |||
2288 | nodes.append(node) |
|
2288 | nodes.append(node) | |
2289 |
|
2289 | |||
2290 | if node in self.nodemap: |
|
2290 | if node in self.nodemap: | |
2291 | self._nodeduplicatecallback(transaction, node) |
|
2291 | self._nodeduplicatecallback(transaction, node) | |
2292 | # this can happen if two branches make the same change |
|
2292 | # this can happen if two branches make the same change | |
2293 | continue |
|
2293 | continue | |
2294 |
|
2294 | |||
2295 | for p in (p1, p2): |
|
2295 | for p in (p1, p2): | |
2296 | if p not in self.nodemap: |
|
2296 | if p not in self.nodemap: | |
2297 | raise error.LookupError( |
|
2297 | raise error.LookupError( | |
2298 | p, self.indexfile, _(b'unknown parent') |
|
2298 | p, self.indexfile, _(b'unknown parent') | |
2299 | ) |
|
2299 | ) | |
2300 |
|
2300 | |||
2301 | if deltabase not in self.nodemap: |
|
2301 | if deltabase not in self.nodemap: | |
2302 | raise error.LookupError( |
|
2302 | raise error.LookupError( | |
2303 | deltabase, self.indexfile, _(b'unknown delta base') |
|
2303 | deltabase, self.indexfile, _(b'unknown delta base') | |
2304 | ) |
|
2304 | ) | |
2305 |
|
2305 | |||
2306 | baserev = self.rev(deltabase) |
|
2306 | baserev = self.rev(deltabase) | |
2307 |
|
2307 | |||
2308 | if baserev != nullrev and self.iscensored(baserev): |
|
2308 | if baserev != nullrev and self.iscensored(baserev): | |
2309 | # if base is censored, delta must be full replacement in a |
|
2309 | # if base is censored, delta must be full replacement in a | |
2310 | # single patch operation |
|
2310 | # single patch operation | |
2311 | hlen = struct.calcsize(b">lll") |
|
2311 | hlen = struct.calcsize(b">lll") | |
2312 | oldlen = self.rawsize(baserev) |
|
2312 | oldlen = self.rawsize(baserev) | |
2313 | newlen = len(delta) - hlen |
|
2313 | newlen = len(delta) - hlen | |
2314 | if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen): |
|
2314 | if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen): | |
2315 | raise error.CensoredBaseError( |
|
2315 | raise error.CensoredBaseError( | |
2316 | self.indexfile, self.node(baserev) |
|
2316 | self.indexfile, self.node(baserev) | |
2317 | ) |
|
2317 | ) | |
2318 |
|
2318 | |||
2319 | if not flags and self._peek_iscensored(baserev, delta, flush): |
|
2319 | if not flags and self._peek_iscensored(baserev, delta, flush): | |
2320 | flags |= REVIDX_ISCENSORED |
|
2320 | flags |= REVIDX_ISCENSORED | |
2321 |
|
2321 | |||
2322 | # We assume consumers of addrevisioncb will want to retrieve |
|
2322 | # We assume consumers of addrevisioncb will want to retrieve | |
2323 | # the added revision, which will require a call to |
|
2323 | # the added revision, which will require a call to | |
2324 | # revision(). revision() will fast path if there is a cache |
|
2324 | # revision(). revision() will fast path if there is a cache | |
2325 | # hit. So, we tell _addrevision() to always cache in this case. |
|
2325 | # hit. So, we tell _addrevision() to always cache in this case. | |
2326 | # We're only using addgroup() in the context of changegroup |
|
2326 | # We're only using addgroup() in the context of changegroup | |
2327 | # generation so the revision data can always be handled as raw |
|
2327 | # generation so the revision data can always be handled as raw | |
2328 | # by the flagprocessor. |
|
2328 | # by the flagprocessor. | |
2329 | self._addrevision( |
|
2329 | self._addrevision( | |
2330 | node, |
|
2330 | node, | |
2331 | None, |
|
2331 | None, | |
2332 | transaction, |
|
2332 | transaction, | |
2333 | link, |
|
2333 | link, | |
2334 | p1, |
|
2334 | p1, | |
2335 | p2, |
|
2335 | p2, | |
2336 | flags, |
|
2336 | flags, | |
2337 | (baserev, delta), |
|
2337 | (baserev, delta), | |
2338 | ifh, |
|
2338 | ifh, | |
2339 | dfh, |
|
2339 | dfh, | |
2340 | alwayscache=bool(addrevisioncb), |
|
2340 | alwayscache=bool(addrevisioncb), | |
2341 | deltacomputer=deltacomputer, |
|
2341 | deltacomputer=deltacomputer, | |
2342 | ) |
|
2342 | ) | |
2343 |
|
2343 | |||
2344 | if addrevisioncb: |
|
2344 | if addrevisioncb: | |
2345 | addrevisioncb(self, node) |
|
2345 | addrevisioncb(self, node) | |
2346 |
|
2346 | |||
2347 | if not dfh and not self._inline: |
|
2347 | if not dfh and not self._inline: | |
2348 | # addrevision switched from inline to conventional |
|
2348 | # addrevision switched from inline to conventional | |
2349 | # reopen the index |
|
2349 | # reopen the index | |
2350 | ifh.close() |
|
2350 | ifh.close() | |
2351 | dfh = self._datafp(b"a+") |
|
2351 | dfh = self._datafp(b"a+") | |
2352 | ifh = self._indexfp(b"a+") |
|
2352 | ifh = self._indexfp(b"a+") | |
2353 | self._writinghandles = (ifh, dfh) |
|
2353 | self._writinghandles = (ifh, dfh) | |
2354 | finally: |
|
2354 | finally: | |
2355 | self._writinghandles = None |
|
2355 | self._writinghandles = None | |
2356 |
|
2356 | |||
2357 | if dfh: |
|
2357 | if dfh: | |
2358 | dfh.close() |
|
2358 | dfh.close() | |
2359 | ifh.close() |
|
2359 | ifh.close() | |
2360 |
|
2360 | |||
2361 | return nodes |
|
2361 | return nodes | |
2362 |
|
2362 | |||
2363 | def iscensored(self, rev): |
|
2363 | def iscensored(self, rev): | |
2364 | """Check if a file revision is censored.""" |
|
2364 | """Check if a file revision is censored.""" | |
2365 | if not self._censorable: |
|
2365 | if not self._censorable: | |
2366 | return False |
|
2366 | return False | |
2367 |
|
2367 | |||
2368 | return self.flags(rev) & REVIDX_ISCENSORED |
|
2368 | return self.flags(rev) & REVIDX_ISCENSORED | |
2369 |
|
2369 | |||
2370 | def _peek_iscensored(self, baserev, delta, flush): |
|
2370 | def _peek_iscensored(self, baserev, delta, flush): | |
2371 | """Quickly check if a delta produces a censored revision.""" |
|
2371 | """Quickly check if a delta produces a censored revision.""" | |
2372 | if not self._censorable: |
|
2372 | if not self._censorable: | |
2373 | return False |
|
2373 | return False | |
2374 |
|
2374 | |||
2375 | return storageutil.deltaiscensored(delta, baserev, self.rawsize) |
|
2375 | return storageutil.deltaiscensored(delta, baserev, self.rawsize) | |
2376 |
|
2376 | |||
2377 | def getstrippoint(self, minlink): |
|
2377 | def getstrippoint(self, minlink): | |
2378 | """find the minimum rev that must be stripped to strip the linkrev |
|
2378 | """find the minimum rev that must be stripped to strip the linkrev | |
2379 |
|
2379 | |||
2380 | Returns a tuple containing the minimum rev and a set of all revs that |
|
2380 | Returns a tuple containing the minimum rev and a set of all revs that | |
2381 | have linkrevs that will be broken by this strip. |
|
2381 | have linkrevs that will be broken by this strip. | |
2382 | """ |
|
2382 | """ | |
2383 | return storageutil.resolvestripinfo( |
|
2383 | return storageutil.resolvestripinfo( | |
2384 | minlink, |
|
2384 | minlink, | |
2385 | len(self) - 1, |
|
2385 | len(self) - 1, | |
2386 | self.headrevs(), |
|
2386 | self.headrevs(), | |
2387 | self.linkrev, |
|
2387 | self.linkrev, | |
2388 | self.parentrevs, |
|
2388 | self.parentrevs, | |
2389 | ) |
|
2389 | ) | |
2390 |
|
2390 | |||
2391 | def strip(self, minlink, transaction): |
|
2391 | def strip(self, minlink, transaction): | |
2392 | """truncate the revlog on the first revision with a linkrev >= minlink |
|
2392 | """truncate the revlog on the first revision with a linkrev >= minlink | |
2393 |
|
2393 | |||
2394 | This function is called when we're stripping revision minlink and |
|
2394 | This function is called when we're stripping revision minlink and | |
2395 | its descendants from the repository. |
|
2395 | its descendants from the repository. | |
2396 |
|
2396 | |||
2397 | We have to remove all revisions with linkrev >= minlink, because |
|
2397 | We have to remove all revisions with linkrev >= minlink, because | |
2398 | the equivalent changelog revisions will be renumbered after the |
|
2398 | the equivalent changelog revisions will be renumbered after the | |
2399 | strip. |
|
2399 | strip. | |
2400 |
|
2400 | |||
2401 | So we truncate the revlog on the first of these revisions, and |
|
2401 | So we truncate the revlog on the first of these revisions, and | |
2402 | trust that the caller has saved the revisions that shouldn't be |
|
2402 | trust that the caller has saved the revisions that shouldn't be | |
2403 | removed and that it'll re-add them after this truncation. |
|
2403 | removed and that it'll re-add them after this truncation. | |
2404 | """ |
|
2404 | """ | |
2405 | if len(self) == 0: |
|
2405 | if len(self) == 0: | |
2406 | return |
|
2406 | return | |
2407 |
|
2407 | |||
2408 | rev, _ = self.getstrippoint(minlink) |
|
2408 | rev, _ = self.getstrippoint(minlink) | |
2409 | if rev == len(self): |
|
2409 | if rev == len(self): | |
2410 | return |
|
2410 | return | |
2411 |
|
2411 | |||
2412 | # first truncate the files on disk |
|
2412 | # first truncate the files on disk | |
2413 | end = self.start(rev) |
|
2413 | end = self.start(rev) | |
2414 | if not self._inline: |
|
2414 | if not self._inline: | |
2415 | transaction.add(self.datafile, end) |
|
2415 | transaction.add(self.datafile, end) | |
2416 | end = rev * self._io.size |
|
2416 | end = rev * self._io.size | |
2417 | else: |
|
2417 | else: | |
2418 | end += rev * self._io.size |
|
2418 | end += rev * self._io.size | |
2419 |
|
2419 | |||
2420 | transaction.add(self.indexfile, end) |
|
2420 | transaction.add(self.indexfile, end) | |
2421 |
|
2421 | |||
2422 | # then reset internal state in memory to forget those revisions |
|
2422 | # then reset internal state in memory to forget those revisions | |
2423 | self._revisioncache = None |
|
2423 | self._revisioncache = None | |
2424 | self._chaininfocache = {} |
|
2424 | self._chaininfocache = {} | |
2425 | self._chunkclear() |
|
2425 | self._chunkclear() | |
2426 | for x in pycompat.xrange(rev, len(self)): |
|
2426 | for x in pycompat.xrange(rev, len(self)): | |
2427 | del self.nodemap[self.node(x)] |
|
2427 | del self.nodemap[self.node(x)] | |
2428 |
|
2428 | |||
2429 | del self.index[rev:-1] |
|
2429 | del self.index[rev:-1] | |
2430 | self._nodepos = None |
|
2430 | self._nodepos = None | |
2431 |
|
2431 | |||
2432 | def checksize(self): |
|
2432 | def checksize(self): | |
2433 | """Check size of index and data files |
|
2433 | """Check size of index and data files | |
2434 |
|
2434 | |||
2435 | return a (dd, di) tuple. |
|
2435 | return a (dd, di) tuple. | |
2436 | - dd: extra bytes for the "data" file |
|
2436 | - dd: extra bytes for the "data" file | |
2437 | - di: extra bytes for the "index" file |
|
2437 | - di: extra bytes for the "index" file | |
2438 |
|
2438 | |||
2439 | A healthy revlog will return (0, 0). |
|
2439 | A healthy revlog will return (0, 0). | |
2440 | """ |
|
2440 | """ | |
2441 | expected = 0 |
|
2441 | expected = 0 | |
2442 | if len(self): |
|
2442 | if len(self): | |
2443 | expected = max(0, self.end(len(self) - 1)) |
|
2443 | expected = max(0, self.end(len(self) - 1)) | |
2444 |
|
2444 | |||
2445 | try: |
|
2445 | try: | |
2446 | with self._datafp() as f: |
|
2446 | with self._datafp() as f: | |
2447 | f.seek(0, io.SEEK_END) |
|
2447 | f.seek(0, io.SEEK_END) | |
2448 | actual = f.tell() |
|
2448 | actual = f.tell() | |
2449 | dd = actual - expected |
|
2449 | dd = actual - expected | |
2450 | except IOError as inst: |
|
2450 | except IOError as inst: | |
2451 | if inst.errno != errno.ENOENT: |
|
2451 | if inst.errno != errno.ENOENT: | |
2452 | raise |
|
2452 | raise | |
2453 | dd = 0 |
|
2453 | dd = 0 | |
2454 |
|
2454 | |||
2455 | try: |
|
2455 | try: | |
2456 | f = self.opener(self.indexfile) |
|
2456 | f = self.opener(self.indexfile) | |
2457 | f.seek(0, io.SEEK_END) |
|
2457 | f.seek(0, io.SEEK_END) | |
2458 | actual = f.tell() |
|
2458 | actual = f.tell() | |
2459 | f.close() |
|
2459 | f.close() | |
2460 | s = self._io.size |
|
2460 | s = self._io.size | |
2461 | i = max(0, actual // s) |
|
2461 | i = max(0, actual // s) | |
2462 | di = actual - (i * s) |
|
2462 | di = actual - (i * s) | |
2463 | if self._inline: |
|
2463 | if self._inline: | |
2464 | databytes = 0 |
|
2464 | databytes = 0 | |
2465 | for r in self: |
|
2465 | for r in self: | |
2466 | databytes += max(0, self.length(r)) |
|
2466 | databytes += max(0, self.length(r)) | |
2467 | dd = 0 |
|
2467 | dd = 0 | |
2468 | di = actual - len(self) * s - databytes |
|
2468 | di = actual - len(self) * s - databytes | |
2469 | except IOError as inst: |
|
2469 | except IOError as inst: | |
2470 | if inst.errno != errno.ENOENT: |
|
2470 | if inst.errno != errno.ENOENT: | |
2471 | raise |
|
2471 | raise | |
2472 | di = 0 |
|
2472 | di = 0 | |
2473 |
|
2473 | |||
2474 | return (dd, di) |
|
2474 | return (dd, di) | |
2475 |
|
2475 | |||
2476 | def files(self): |
|
2476 | def files(self): | |
2477 | res = [self.indexfile] |
|
2477 | res = [self.indexfile] | |
2478 | if not self._inline: |
|
2478 | if not self._inline: | |
2479 | res.append(self.datafile) |
|
2479 | res.append(self.datafile) | |
2480 | return res |
|
2480 | return res | |
2481 |
|
2481 | |||
2482 | def emitrevisions( |
|
2482 | def emitrevisions( | |
2483 | self, |
|
2483 | self, | |
2484 | nodes, |
|
2484 | nodes, | |
2485 | nodesorder=None, |
|
2485 | nodesorder=None, | |
2486 | revisiondata=False, |
|
2486 | revisiondata=False, | |
2487 | assumehaveparentrevisions=False, |
|
2487 | assumehaveparentrevisions=False, | |
2488 | deltamode=repository.CG_DELTAMODE_STD, |
|
2488 | deltamode=repository.CG_DELTAMODE_STD, | |
2489 | ): |
|
2489 | ): | |
2490 | if nodesorder not in (b'nodes', b'storage', b'linear', None): |
|
2490 | if nodesorder not in (b'nodes', b'storage', b'linear', None): | |
2491 | raise error.ProgrammingError( |
|
2491 | raise error.ProgrammingError( | |
2492 | b'unhandled value for nodesorder: %s' % nodesorder |
|
2492 | b'unhandled value for nodesorder: %s' % nodesorder | |
2493 | ) |
|
2493 | ) | |
2494 |
|
2494 | |||
2495 | if nodesorder is None and not self._generaldelta: |
|
2495 | if nodesorder is None and not self._generaldelta: | |
2496 | nodesorder = b'storage' |
|
2496 | nodesorder = b'storage' | |
2497 |
|
2497 | |||
2498 | if ( |
|
2498 | if ( | |
2499 | not self._storedeltachains |
|
2499 | not self._storedeltachains | |
2500 | and deltamode != repository.CG_DELTAMODE_PREV |
|
2500 | and deltamode != repository.CG_DELTAMODE_PREV | |
2501 | ): |
|
2501 | ): | |
2502 | deltamode = repository.CG_DELTAMODE_FULL |
|
2502 | deltamode = repository.CG_DELTAMODE_FULL | |
2503 |
|
2503 | |||
2504 | return storageutil.emitrevisions( |
|
2504 | return storageutil.emitrevisions( | |
2505 | self, |
|
2505 | self, | |
2506 | nodes, |
|
2506 | nodes, | |
2507 | nodesorder, |
|
2507 | nodesorder, | |
2508 | revlogrevisiondelta, |
|
2508 | revlogrevisiondelta, | |
2509 | deltaparentfn=self.deltaparent, |
|
2509 | deltaparentfn=self.deltaparent, | |
2510 | candeltafn=self.candelta, |
|
2510 | candeltafn=self.candelta, | |
2511 | rawsizefn=self.rawsize, |
|
2511 | rawsizefn=self.rawsize, | |
2512 | revdifffn=self.revdiff, |
|
2512 | revdifffn=self.revdiff, | |
2513 | flagsfn=self.flags, |
|
2513 | flagsfn=self.flags, | |
2514 | deltamode=deltamode, |
|
2514 | deltamode=deltamode, | |
2515 | revisiondata=revisiondata, |
|
2515 | revisiondata=revisiondata, | |
2516 | assumehaveparentrevisions=assumehaveparentrevisions, |
|
2516 | assumehaveparentrevisions=assumehaveparentrevisions, | |
2517 | ) |
|
2517 | ) | |
2518 |
|
2518 | |||
2519 | DELTAREUSEALWAYS = b'always' |
|
2519 | DELTAREUSEALWAYS = b'always' | |
2520 | DELTAREUSESAMEREVS = b'samerevs' |
|
2520 | DELTAREUSESAMEREVS = b'samerevs' | |
2521 | DELTAREUSENEVER = b'never' |
|
2521 | DELTAREUSENEVER = b'never' | |
2522 |
|
2522 | |||
2523 | DELTAREUSEFULLADD = b'fulladd' |
|
2523 | DELTAREUSEFULLADD = b'fulladd' | |
2524 |
|
2524 | |||
2525 | DELTAREUSEALL = {b'always', b'samerevs', b'never', b'fulladd'} |
|
2525 | DELTAREUSEALL = {b'always', b'samerevs', b'never', b'fulladd'} | |
2526 |
|
2526 | |||
2527 | def clone( |
|
2527 | def clone( | |
2528 | self, |
|
2528 | self, | |
2529 | tr, |
|
2529 | tr, | |
2530 | destrevlog, |
|
2530 | destrevlog, | |
2531 | addrevisioncb=None, |
|
2531 | addrevisioncb=None, | |
2532 | deltareuse=DELTAREUSESAMEREVS, |
|
2532 | deltareuse=DELTAREUSESAMEREVS, | |
2533 | forcedeltabothparents=None, |
|
2533 | forcedeltabothparents=None, | |
2534 | sidedatacompanion=None, |
|
2534 | sidedatacompanion=None, | |
2535 | ): |
|
2535 | ): | |
2536 | """Copy this revlog to another, possibly with format changes. |
|
2536 | """Copy this revlog to another, possibly with format changes. | |
2537 |
|
2537 | |||
2538 | The destination revlog will contain the same revisions and nodes. |
|
2538 | The destination revlog will contain the same revisions and nodes. | |
2539 | However, it may not be bit-for-bit identical due to e.g. delta encoding |
|
2539 | However, it may not be bit-for-bit identical due to e.g. delta encoding | |
2540 | differences. |
|
2540 | differences. | |
2541 |
|
2541 | |||
2542 | The ``deltareuse`` argument control how deltas from the existing revlog |
|
2542 | The ``deltareuse`` argument control how deltas from the existing revlog | |
2543 | are preserved in the destination revlog. The argument can have the |
|
2543 | are preserved in the destination revlog. The argument can have the | |
2544 | following values: |
|
2544 | following values: | |
2545 |
|
2545 | |||
2546 | DELTAREUSEALWAYS |
|
2546 | DELTAREUSEALWAYS | |
2547 | Deltas will always be reused (if possible), even if the destination |
|
2547 | Deltas will always be reused (if possible), even if the destination | |
2548 | revlog would not select the same revisions for the delta. This is the |
|
2548 | revlog would not select the same revisions for the delta. This is the | |
2549 | fastest mode of operation. |
|
2549 | fastest mode of operation. | |
2550 | DELTAREUSESAMEREVS |
|
2550 | DELTAREUSESAMEREVS | |
2551 | Deltas will be reused if the destination revlog would pick the same |
|
2551 | Deltas will be reused if the destination revlog would pick the same | |
2552 | revisions for the delta. This mode strikes a balance between speed |
|
2552 | revisions for the delta. This mode strikes a balance between speed | |
2553 | and optimization. |
|
2553 | and optimization. | |
2554 | DELTAREUSENEVER |
|
2554 | DELTAREUSENEVER | |
2555 | Deltas will never be reused. This is the slowest mode of execution. |
|
2555 | Deltas will never be reused. This is the slowest mode of execution. | |
2556 | This mode can be used to recompute deltas (e.g. if the diff/delta |
|
2556 | This mode can be used to recompute deltas (e.g. if the diff/delta | |
2557 | algorithm changes). |
|
2557 | algorithm changes). | |
2558 | DELTAREUSEFULLADD |
|
2558 | DELTAREUSEFULLADD | |
2559 | Revision will be re-added as if their were new content. This is |
|
2559 | Revision will be re-added as if their were new content. This is | |
2560 | slower than DELTAREUSEALWAYS but allow more mechanism to kicks in. |
|
2560 | slower than DELTAREUSEALWAYS but allow more mechanism to kicks in. | |
2561 | eg: large file detection and handling. |
|
2561 | eg: large file detection and handling. | |
2562 |
|
2562 | |||
2563 | Delta computation can be slow, so the choice of delta reuse policy can |
|
2563 | Delta computation can be slow, so the choice of delta reuse policy can | |
2564 | significantly affect run time. |
|
2564 | significantly affect run time. | |
2565 |
|
2565 | |||
2566 | The default policy (``DELTAREUSESAMEREVS``) strikes a balance between |
|
2566 | The default policy (``DELTAREUSESAMEREVS``) strikes a balance between | |
2567 | two extremes. Deltas will be reused if they are appropriate. But if the |
|
2567 | two extremes. Deltas will be reused if they are appropriate. But if the | |
2568 | delta could choose a better revision, it will do so. This means if you |
|
2568 | delta could choose a better revision, it will do so. This means if you | |
2569 | are converting a non-generaldelta revlog to a generaldelta revlog, |
|
2569 | are converting a non-generaldelta revlog to a generaldelta revlog, | |
2570 | deltas will be recomputed if the delta's parent isn't a parent of the |
|
2570 | deltas will be recomputed if the delta's parent isn't a parent of the | |
2571 | revision. |
|
2571 | revision. | |
2572 |
|
2572 | |||
2573 | In addition to the delta policy, the ``forcedeltabothparents`` |
|
2573 | In addition to the delta policy, the ``forcedeltabothparents`` | |
2574 | argument controls whether to force compute deltas against both parents |
|
2574 | argument controls whether to force compute deltas against both parents | |
2575 | for merges. By default, the current default is used. |
|
2575 | for merges. By default, the current default is used. | |
2576 |
|
2576 | |||
2577 | If not None, the `sidedatacompanion` is callable that accept two |
|
2577 | If not None, the `sidedatacompanion` is callable that accept two | |
2578 | arguments: |
|
2578 | arguments: | |
2579 |
|
2579 | |||
2580 | (srcrevlog, rev) |
|
2580 | (srcrevlog, rev) | |
2581 |
|
2581 | |||
2582 | and return a triplet that control changes to sidedata content from the |
|
2582 | and return a triplet that control changes to sidedata content from the | |
2583 | old revision to the new clone result: |
|
2583 | old revision to the new clone result: | |
2584 |
|
2584 | |||
2585 | (dropall, filterout, update) |
|
2585 | (dropall, filterout, update) | |
2586 |
|
2586 | |||
2587 | * if `dropall` is True, all sidedata should be dropped |
|
2587 | * if `dropall` is True, all sidedata should be dropped | |
2588 | * `filterout` is a set of sidedata keys that should be dropped |
|
2588 | * `filterout` is a set of sidedata keys that should be dropped | |
2589 | * `update` is a mapping of additionnal/new key -> value |
|
2589 | * `update` is a mapping of additionnal/new key -> value | |
2590 | """ |
|
2590 | """ | |
2591 | if deltareuse not in self.DELTAREUSEALL: |
|
2591 | if deltareuse not in self.DELTAREUSEALL: | |
2592 | raise ValueError( |
|
2592 | raise ValueError( | |
2593 | _(b'value for deltareuse invalid: %s') % deltareuse |
|
2593 | _(b'value for deltareuse invalid: %s') % deltareuse | |
2594 | ) |
|
2594 | ) | |
2595 |
|
2595 | |||
2596 | if len(destrevlog): |
|
2596 | if len(destrevlog): | |
2597 | raise ValueError(_(b'destination revlog is not empty')) |
|
2597 | raise ValueError(_(b'destination revlog is not empty')) | |
2598 |
|
2598 | |||
2599 | if getattr(self, 'filteredrevs', None): |
|
2599 | if getattr(self, 'filteredrevs', None): | |
2600 | raise ValueError(_(b'source revlog has filtered revisions')) |
|
2600 | raise ValueError(_(b'source revlog has filtered revisions')) | |
2601 | if getattr(destrevlog, 'filteredrevs', None): |
|
2601 | if getattr(destrevlog, 'filteredrevs', None): | |
2602 | raise ValueError(_(b'destination revlog has filtered revisions')) |
|
2602 | raise ValueError(_(b'destination revlog has filtered revisions')) | |
2603 |
|
2603 | |||
2604 | # lazydelta and lazydeltabase controls whether to reuse a cached delta, |
|
2604 | # lazydelta and lazydeltabase controls whether to reuse a cached delta, | |
2605 | # if possible. |
|
2605 | # if possible. | |
2606 | oldlazydelta = destrevlog._lazydelta |
|
2606 | oldlazydelta = destrevlog._lazydelta | |
2607 | oldlazydeltabase = destrevlog._lazydeltabase |
|
2607 | oldlazydeltabase = destrevlog._lazydeltabase | |
2608 | oldamd = destrevlog._deltabothparents |
|
2608 | oldamd = destrevlog._deltabothparents | |
2609 |
|
2609 | |||
2610 | try: |
|
2610 | try: | |
2611 | if deltareuse == self.DELTAREUSEALWAYS: |
|
2611 | if deltareuse == self.DELTAREUSEALWAYS: | |
2612 | destrevlog._lazydeltabase = True |
|
2612 | destrevlog._lazydeltabase = True | |
2613 | destrevlog._lazydelta = True |
|
2613 | destrevlog._lazydelta = True | |
2614 | elif deltareuse == self.DELTAREUSESAMEREVS: |
|
2614 | elif deltareuse == self.DELTAREUSESAMEREVS: | |
2615 | destrevlog._lazydeltabase = False |
|
2615 | destrevlog._lazydeltabase = False | |
2616 | destrevlog._lazydelta = True |
|
2616 | destrevlog._lazydelta = True | |
2617 | elif deltareuse == self.DELTAREUSENEVER: |
|
2617 | elif deltareuse == self.DELTAREUSENEVER: | |
2618 | destrevlog._lazydeltabase = False |
|
2618 | destrevlog._lazydeltabase = False | |
2619 | destrevlog._lazydelta = False |
|
2619 | destrevlog._lazydelta = False | |
2620 |
|
2620 | |||
2621 | destrevlog._deltabothparents = forcedeltabothparents or oldamd |
|
2621 | destrevlog._deltabothparents = forcedeltabothparents or oldamd | |
2622 |
|
2622 | |||
2623 | self._clone( |
|
2623 | self._clone( | |
2624 | tr, |
|
2624 | tr, | |
2625 | destrevlog, |
|
2625 | destrevlog, | |
2626 | addrevisioncb, |
|
2626 | addrevisioncb, | |
2627 | deltareuse, |
|
2627 | deltareuse, | |
2628 | forcedeltabothparents, |
|
2628 | forcedeltabothparents, | |
2629 | sidedatacompanion, |
|
2629 | sidedatacompanion, | |
2630 | ) |
|
2630 | ) | |
2631 |
|
2631 | |||
2632 | finally: |
|
2632 | finally: | |
2633 | destrevlog._lazydelta = oldlazydelta |
|
2633 | destrevlog._lazydelta = oldlazydelta | |
2634 | destrevlog._lazydeltabase = oldlazydeltabase |
|
2634 | destrevlog._lazydeltabase = oldlazydeltabase | |
2635 | destrevlog._deltabothparents = oldamd |
|
2635 | destrevlog._deltabothparents = oldamd | |
2636 |
|
2636 | |||
2637 | def _clone( |
|
2637 | def _clone( | |
2638 | self, |
|
2638 | self, | |
2639 | tr, |
|
2639 | tr, | |
2640 | destrevlog, |
|
2640 | destrevlog, | |
2641 | addrevisioncb, |
|
2641 | addrevisioncb, | |
2642 | deltareuse, |
|
2642 | deltareuse, | |
2643 | forcedeltabothparents, |
|
2643 | forcedeltabothparents, | |
2644 | sidedatacompanion, |
|
2644 | sidedatacompanion, | |
2645 | ): |
|
2645 | ): | |
2646 | """perform the core duty of `revlog.clone` after parameter processing""" |
|
2646 | """perform the core duty of `revlog.clone` after parameter processing""" | |
2647 | deltacomputer = deltautil.deltacomputer(destrevlog) |
|
2647 | deltacomputer = deltautil.deltacomputer(destrevlog) | |
2648 | index = self.index |
|
2648 | index = self.index | |
2649 | for rev in self: |
|
2649 | for rev in self: | |
2650 | entry = index[rev] |
|
2650 | entry = index[rev] | |
2651 |
|
2651 | |||
2652 | # Some classes override linkrev to take filtered revs into |
|
2652 | # Some classes override linkrev to take filtered revs into | |
2653 | # account. Use raw entry from index. |
|
2653 | # account. Use raw entry from index. | |
2654 | flags = entry[0] & 0xFFFF |
|
2654 | flags = entry[0] & 0xFFFF | |
2655 | linkrev = entry[4] |
|
2655 | linkrev = entry[4] | |
2656 | p1 = index[entry[5]][7] |
|
2656 | p1 = index[entry[5]][7] | |
2657 | p2 = index[entry[6]][7] |
|
2657 | p2 = index[entry[6]][7] | |
2658 | node = entry[7] |
|
2658 | node = entry[7] | |
2659 |
|
2659 | |||
2660 | sidedataactions = (False, [], {}) |
|
2660 | sidedataactions = (False, [], {}) | |
2661 | if sidedatacompanion is not None: |
|
2661 | if sidedatacompanion is not None: | |
2662 | sidedataactions = sidedatacompanion(self, rev) |
|
2662 | sidedataactions = sidedatacompanion(self, rev) | |
2663 |
|
2663 | |||
2664 | # (Possibly) reuse the delta from the revlog if allowed and |
|
2664 | # (Possibly) reuse the delta from the revlog if allowed and | |
2665 | # the revlog chunk is a delta. |
|
2665 | # the revlog chunk is a delta. | |
2666 | cachedelta = None |
|
2666 | cachedelta = None | |
2667 | rawtext = None |
|
2667 | rawtext = None | |
2668 | if any(sidedataactions) or deltareuse == self.DELTAREUSEFULLADD: |
|
2668 | if any(sidedataactions) or deltareuse == self.DELTAREUSEFULLADD: | |
2669 | dropall, filterout, update = sidedataactions |
|
2669 | dropall, filterout, update = sidedataactions | |
2670 | text, sidedata = self._revisiondata(rev) |
|
2670 | text, sidedata = self._revisiondata(rev) | |
2671 | if dropall: |
|
2671 | if dropall: | |
2672 | sidedata = {} |
|
2672 | sidedata = {} | |
2673 | for key in filterout: |
|
2673 | for key in filterout: | |
2674 | sidedata.pop(key, None) |
|
2674 | sidedata.pop(key, None) | |
2675 | sidedata.update(update) |
|
2675 | sidedata.update(update) | |
2676 | if not sidedata: |
|
2676 | if not sidedata: | |
2677 | sidedata = None |
|
2677 | sidedata = None | |
2678 | destrevlog.addrevision( |
|
2678 | destrevlog.addrevision( | |
2679 | text, |
|
2679 | text, | |
2680 | tr, |
|
2680 | tr, | |
2681 | linkrev, |
|
2681 | linkrev, | |
2682 | p1, |
|
2682 | p1, | |
2683 | p2, |
|
2683 | p2, | |
2684 | cachedelta=cachedelta, |
|
2684 | cachedelta=cachedelta, | |
2685 | node=node, |
|
2685 | node=node, | |
2686 | flags=flags, |
|
2686 | flags=flags, | |
2687 | deltacomputer=deltacomputer, |
|
2687 | deltacomputer=deltacomputer, | |
2688 | sidedata=sidedata, |
|
2688 | sidedata=sidedata, | |
2689 | ) |
|
2689 | ) | |
2690 | else: |
|
2690 | else: | |
2691 | if destrevlog._lazydelta: |
|
2691 | if destrevlog._lazydelta: | |
2692 | dp = self.deltaparent(rev) |
|
2692 | dp = self.deltaparent(rev) | |
2693 | if dp != nullrev: |
|
2693 | if dp != nullrev: | |
2694 | cachedelta = (dp, bytes(self._chunk(rev))) |
|
2694 | cachedelta = (dp, bytes(self._chunk(rev))) | |
2695 |
|
2695 | |||
2696 | if not cachedelta: |
|
2696 | if not cachedelta: | |
2697 | rawtext = self.rawdata(rev) |
|
2697 | rawtext = self.rawdata(rev) | |
2698 |
|
2698 | |||
2699 | ifh = destrevlog.opener( |
|
2699 | ifh = destrevlog.opener( | |
2700 | destrevlog.indexfile, b'a+', checkambig=False |
|
2700 | destrevlog.indexfile, b'a+', checkambig=False | |
2701 | ) |
|
2701 | ) | |
2702 | dfh = None |
|
2702 | dfh = None | |
2703 | if not destrevlog._inline: |
|
2703 | if not destrevlog._inline: | |
2704 | dfh = destrevlog.opener(destrevlog.datafile, b'a+') |
|
2704 | dfh = destrevlog.opener(destrevlog.datafile, b'a+') | |
2705 | try: |
|
2705 | try: | |
2706 | destrevlog._addrevision( |
|
2706 | destrevlog._addrevision( | |
2707 | node, |
|
2707 | node, | |
2708 | rawtext, |
|
2708 | rawtext, | |
2709 | tr, |
|
2709 | tr, | |
2710 | linkrev, |
|
2710 | linkrev, | |
2711 | p1, |
|
2711 | p1, | |
2712 | p2, |
|
2712 | p2, | |
2713 | flags, |
|
2713 | flags, | |
2714 | cachedelta, |
|
2714 | cachedelta, | |
2715 | ifh, |
|
2715 | ifh, | |
2716 | dfh, |
|
2716 | dfh, | |
2717 | deltacomputer=deltacomputer, |
|
2717 | deltacomputer=deltacomputer, | |
2718 | ) |
|
2718 | ) | |
2719 | finally: |
|
2719 | finally: | |
2720 | if dfh: |
|
2720 | if dfh: | |
2721 | dfh.close() |
|
2721 | dfh.close() | |
2722 | ifh.close() |
|
2722 | ifh.close() | |
2723 |
|
2723 | |||
2724 | if addrevisioncb: |
|
2724 | if addrevisioncb: | |
2725 | addrevisioncb(self, rev, node) |
|
2725 | addrevisioncb(self, rev, node) | |
2726 |
|
2726 | |||
2727 | def censorrevision(self, tr, censornode, tombstone=b''): |
|
2727 | def censorrevision(self, tr, censornode, tombstone=b''): | |
2728 | if (self.version & 0xFFFF) == REVLOGV0: |
|
2728 | if (self.version & 0xFFFF) == REVLOGV0: | |
2729 | raise error.RevlogError( |
|
2729 | raise error.RevlogError( | |
2730 | _(b'cannot censor with version %d revlogs') % self.version |
|
2730 | _(b'cannot censor with version %d revlogs') % self.version | |
2731 | ) |
|
2731 | ) | |
2732 |
|
2732 | |||
2733 | censorrev = self.rev(censornode) |
|
2733 | censorrev = self.rev(censornode) | |
2734 | tombstone = storageutil.packmeta({b'censored': tombstone}, b'') |
|
2734 | tombstone = storageutil.packmeta({b'censored': tombstone}, b'') | |
2735 |
|
2735 | |||
2736 | if len(tombstone) > self.rawsize(censorrev): |
|
2736 | if len(tombstone) > self.rawsize(censorrev): | |
2737 | raise error.Abort( |
|
2737 | raise error.Abort( | |
2738 | _(b'censor tombstone must be no longer than censored data') |
|
2738 | _(b'censor tombstone must be no longer than censored data') | |
2739 | ) |
|
2739 | ) | |
2740 |
|
2740 | |||
2741 | # Rewriting the revlog in place is hard. Our strategy for censoring is |
|
2741 | # Rewriting the revlog in place is hard. Our strategy for censoring is | |
2742 | # to create a new revlog, copy all revisions to it, then replace the |
|
2742 | # to create a new revlog, copy all revisions to it, then replace the | |
2743 | # revlogs on transaction close. |
|
2743 | # revlogs on transaction close. | |
2744 |
|
2744 | |||
2745 | newindexfile = self.indexfile + b'.tmpcensored' |
|
2745 | newindexfile = self.indexfile + b'.tmpcensored' | |
2746 | newdatafile = self.datafile + b'.tmpcensored' |
|
2746 | newdatafile = self.datafile + b'.tmpcensored' | |
2747 |
|
2747 | |||
2748 | # This is a bit dangerous. We could easily have a mismatch of state. |
|
2748 | # This is a bit dangerous. We could easily have a mismatch of state. | |
2749 | newrl = revlog(self.opener, newindexfile, newdatafile, censorable=True) |
|
2749 | newrl = revlog(self.opener, newindexfile, newdatafile, censorable=True) | |
2750 | newrl.version = self.version |
|
2750 | newrl.version = self.version | |
2751 | newrl._generaldelta = self._generaldelta |
|
2751 | newrl._generaldelta = self._generaldelta | |
2752 | newrl._io = self._io |
|
2752 | newrl._io = self._io | |
2753 |
|
2753 | |||
2754 | for rev in self.revs(): |
|
2754 | for rev in self.revs(): | |
2755 | node = self.node(rev) |
|
2755 | node = self.node(rev) | |
2756 | p1, p2 = self.parents(node) |
|
2756 | p1, p2 = self.parents(node) | |
2757 |
|
2757 | |||
2758 | if rev == censorrev: |
|
2758 | if rev == censorrev: | |
2759 | newrl.addrawrevision( |
|
2759 | newrl.addrawrevision( | |
2760 | tombstone, |
|
2760 | tombstone, | |
2761 | tr, |
|
2761 | tr, | |
2762 | self.linkrev(censorrev), |
|
2762 | self.linkrev(censorrev), | |
2763 | p1, |
|
2763 | p1, | |
2764 | p2, |
|
2764 | p2, | |
2765 | censornode, |
|
2765 | censornode, | |
2766 | REVIDX_ISCENSORED, |
|
2766 | REVIDX_ISCENSORED, | |
2767 | ) |
|
2767 | ) | |
2768 |
|
2768 | |||
2769 | if newrl.deltaparent(rev) != nullrev: |
|
2769 | if newrl.deltaparent(rev) != nullrev: | |
2770 | raise error.Abort( |
|
2770 | raise error.Abort( | |
2771 | _( |
|
2771 | _( | |
2772 | b'censored revision stored as delta; ' |
|
2772 | b'censored revision stored as delta; ' | |
2773 | b'cannot censor' |
|
2773 | b'cannot censor' | |
2774 | ), |
|
2774 | ), | |
2775 | hint=_( |
|
2775 | hint=_( | |
2776 | b'censoring of revlogs is not ' |
|
2776 | b'censoring of revlogs is not ' | |
2777 | b'fully implemented; please report ' |
|
2777 | b'fully implemented; please report ' | |
2778 | b'this bug' |
|
2778 | b'this bug' | |
2779 | ), |
|
2779 | ), | |
2780 | ) |
|
2780 | ) | |
2781 | continue |
|
2781 | continue | |
2782 |
|
2782 | |||
2783 | if self.iscensored(rev): |
|
2783 | if self.iscensored(rev): | |
2784 | if self.deltaparent(rev) != nullrev: |
|
2784 | if self.deltaparent(rev) != nullrev: | |
2785 | raise error.Abort( |
|
2785 | raise error.Abort( | |
2786 | _( |
|
2786 | _( | |
2787 | b'cannot censor due to censored ' |
|
2787 | b'cannot censor due to censored ' | |
2788 | b'revision having delta stored' |
|
2788 | b'revision having delta stored' | |
2789 | ) |
|
2789 | ) | |
2790 | ) |
|
2790 | ) | |
2791 | rawtext = self._chunk(rev) |
|
2791 | rawtext = self._chunk(rev) | |
2792 | else: |
|
2792 | else: | |
2793 | rawtext = self.rawdata(rev) |
|
2793 | rawtext = self.rawdata(rev) | |
2794 |
|
2794 | |||
2795 | newrl.addrawrevision( |
|
2795 | newrl.addrawrevision( | |
2796 | rawtext, tr, self.linkrev(rev), p1, p2, node, self.flags(rev) |
|
2796 | rawtext, tr, self.linkrev(rev), p1, p2, node, self.flags(rev) | |
2797 | ) |
|
2797 | ) | |
2798 |
|
2798 | |||
2799 | tr.addbackup(self.indexfile, location=b'store') |
|
2799 | tr.addbackup(self.indexfile, location=b'store') | |
2800 | if not self._inline: |
|
2800 | if not self._inline: | |
2801 | tr.addbackup(self.datafile, location=b'store') |
|
2801 | tr.addbackup(self.datafile, location=b'store') | |
2802 |
|
2802 | |||
2803 | self.opener.rename(newrl.indexfile, self.indexfile) |
|
2803 | self.opener.rename(newrl.indexfile, self.indexfile) | |
2804 | if not self._inline: |
|
2804 | if not self._inline: | |
2805 | self.opener.rename(newrl.datafile, self.datafile) |
|
2805 | self.opener.rename(newrl.datafile, self.datafile) | |
2806 |
|
2806 | |||
2807 | self.clearcaches() |
|
2807 | self.clearcaches() | |
2808 | self._loadindex() |
|
2808 | self._loadindex() | |
2809 |
|
2809 | |||
2810 | def verifyintegrity(self, state): |
|
2810 | def verifyintegrity(self, state): | |
2811 | """Verifies the integrity of the revlog. |
|
2811 | """Verifies the integrity of the revlog. | |
2812 |
|
2812 | |||
2813 | Yields ``revlogproblem`` instances describing problems that are |
|
2813 | Yields ``revlogproblem`` instances describing problems that are | |
2814 | found. |
|
2814 | found. | |
2815 | """ |
|
2815 | """ | |
2816 | dd, di = self.checksize() |
|
2816 | dd, di = self.checksize() | |
2817 | if dd: |
|
2817 | if dd: | |
2818 | yield revlogproblem(error=_(b'data length off by %d bytes') % dd) |
|
2818 | yield revlogproblem(error=_(b'data length off by %d bytes') % dd) | |
2819 | if di: |
|
2819 | if di: | |
2820 | yield revlogproblem(error=_(b'index contains %d extra bytes') % di) |
|
2820 | yield revlogproblem(error=_(b'index contains %d extra bytes') % di) | |
2821 |
|
2821 | |||
2822 | version = self.version & 0xFFFF |
|
2822 | version = self.version & 0xFFFF | |
2823 |
|
2823 | |||
2824 | # The verifier tells us what version revlog we should be. |
|
2824 | # The verifier tells us what version revlog we should be. | |
2825 | if version != state[b'expectedversion']: |
|
2825 | if version != state[b'expectedversion']: | |
2826 | yield revlogproblem( |
|
2826 | yield revlogproblem( | |
2827 | warning=_(b"warning: '%s' uses revlog format %d; expected %d") |
|
2827 | warning=_(b"warning: '%s' uses revlog format %d; expected %d") | |
2828 | % (self.indexfile, version, state[b'expectedversion']) |
|
2828 | % (self.indexfile, version, state[b'expectedversion']) | |
2829 | ) |
|
2829 | ) | |
2830 |
|
2830 | |||
2831 | state[b'skipread'] = set() |
|
2831 | state[b'skipread'] = set() | |
2832 |
|
2832 | |||
2833 | for rev in self: |
|
2833 | for rev in self: | |
2834 | node = self.node(rev) |
|
2834 | node = self.node(rev) | |
2835 |
|
2835 | |||
2836 | # Verify contents. 4 cases to care about: |
|
2836 | # Verify contents. 4 cases to care about: | |
2837 | # |
|
2837 | # | |
2838 | # common: the most common case |
|
2838 | # common: the most common case | |
2839 | # rename: with a rename |
|
2839 | # rename: with a rename | |
2840 | # meta: file content starts with b'\1\n', the metadata |
|
2840 | # meta: file content starts with b'\1\n', the metadata | |
2841 | # header defined in filelog.py, but without a rename |
|
2841 | # header defined in filelog.py, but without a rename | |
2842 | # ext: content stored externally |
|
2842 | # ext: content stored externally | |
2843 | # |
|
2843 | # | |
2844 | # More formally, their differences are shown below: |
|
2844 | # More formally, their differences are shown below: | |
2845 | # |
|
2845 | # | |
2846 | # | common | rename | meta | ext |
|
2846 | # | common | rename | meta | ext | |
2847 | # ------------------------------------------------------- |
|
2847 | # ------------------------------------------------------- | |
2848 | # flags() | 0 | 0 | 0 | not 0 |
|
2848 | # flags() | 0 | 0 | 0 | not 0 | |
2849 | # renamed() | False | True | False | ? |
|
2849 | # renamed() | False | True | False | ? | |
2850 | # rawtext[0:2]=='\1\n'| False | True | True | ? |
|
2850 | # rawtext[0:2]=='\1\n'| False | True | True | ? | |
2851 | # |
|
2851 | # | |
2852 | # "rawtext" means the raw text stored in revlog data, which |
|
2852 | # "rawtext" means the raw text stored in revlog data, which | |
2853 | # could be retrieved by "rawdata(rev)". "text" |
|
2853 | # could be retrieved by "rawdata(rev)". "text" | |
2854 | # mentioned below is "revision(rev)". |
|
2854 | # mentioned below is "revision(rev)". | |
2855 | # |
|
2855 | # | |
2856 | # There are 3 different lengths stored physically: |
|
2856 | # There are 3 different lengths stored physically: | |
2857 | # 1. L1: rawsize, stored in revlog index |
|
2857 | # 1. L1: rawsize, stored in revlog index | |
2858 | # 2. L2: len(rawtext), stored in revlog data |
|
2858 | # 2. L2: len(rawtext), stored in revlog data | |
2859 | # 3. L3: len(text), stored in revlog data if flags==0, or |
|
2859 | # 3. L3: len(text), stored in revlog data if flags==0, or | |
2860 | # possibly somewhere else if flags!=0 |
|
2860 | # possibly somewhere else if flags!=0 | |
2861 | # |
|
2861 | # | |
2862 | # L1 should be equal to L2. L3 could be different from them. |
|
2862 | # L1 should be equal to L2. L3 could be different from them. | |
2863 | # "text" may or may not affect commit hash depending on flag |
|
2863 | # "text" may or may not affect commit hash depending on flag | |
2864 | # processors (see flagutil.addflagprocessor). |
|
2864 | # processors (see flagutil.addflagprocessor). | |
2865 | # |
|
2865 | # | |
2866 | # | common | rename | meta | ext |
|
2866 | # | common | rename | meta | ext | |
2867 | # ------------------------------------------------- |
|
2867 | # ------------------------------------------------- | |
2868 | # rawsize() | L1 | L1 | L1 | L1 |
|
2868 | # rawsize() | L1 | L1 | L1 | L1 | |
2869 | # size() | L1 | L2-LM | L1(*) | L1 (?) |
|
2869 | # size() | L1 | L2-LM | L1(*) | L1 (?) | |
2870 | # len(rawtext) | L2 | L2 | L2 | L2 |
|
2870 | # len(rawtext) | L2 | L2 | L2 | L2 | |
2871 | # len(text) | L2 | L2 | L2 | L3 |
|
2871 | # len(text) | L2 | L2 | L2 | L3 | |
2872 | # len(read()) | L2 | L2-LM | L2-LM | L3 (?) |
|
2872 | # len(read()) | L2 | L2-LM | L2-LM | L3 (?) | |
2873 | # |
|
2873 | # | |
2874 | # LM: length of metadata, depending on rawtext |
|
2874 | # LM: length of metadata, depending on rawtext | |
2875 | # (*): not ideal, see comment in filelog.size |
|
2875 | # (*): not ideal, see comment in filelog.size | |
2876 | # (?): could be "- len(meta)" if the resolved content has |
|
2876 | # (?): could be "- len(meta)" if the resolved content has | |
2877 | # rename metadata |
|
2877 | # rename metadata | |
2878 | # |
|
2878 | # | |
2879 | # Checks needed to be done: |
|
2879 | # Checks needed to be done: | |
2880 | # 1. length check: L1 == L2, in all cases. |
|
2880 | # 1. length check: L1 == L2, in all cases. | |
2881 | # 2. hash check: depending on flag processor, we may need to |
|
2881 | # 2. hash check: depending on flag processor, we may need to | |
2882 | # use either "text" (external), or "rawtext" (in revlog). |
|
2882 | # use either "text" (external), or "rawtext" (in revlog). | |
2883 |
|
2883 | |||
2884 | try: |
|
2884 | try: | |
2885 | skipflags = state.get(b'skipflags', 0) |
|
2885 | skipflags = state.get(b'skipflags', 0) | |
2886 | if skipflags: |
|
2886 | if skipflags: | |
2887 | skipflags &= self.flags(rev) |
|
2887 | skipflags &= self.flags(rev) | |
2888 |
|
2888 | |||
2889 | if skipflags: |
|
2889 | if skipflags: | |
2890 | state[b'skipread'].add(node) |
|
2890 | state[b'skipread'].add(node) | |
2891 | else: |
|
2891 | else: | |
2892 | # Side-effect: read content and verify hash. |
|
2892 | # Side-effect: read content and verify hash. | |
2893 | self.revision(node) |
|
2893 | self.revision(node) | |
2894 |
|
2894 | |||
2895 | l1 = self.rawsize(rev) |
|
2895 | l1 = self.rawsize(rev) | |
2896 | l2 = len(self.rawdata(node)) |
|
2896 | l2 = len(self.rawdata(node)) | |
2897 |
|
2897 | |||
2898 | if l1 != l2: |
|
2898 | if l1 != l2: | |
2899 | yield revlogproblem( |
|
2899 | yield revlogproblem( | |
2900 | error=_(b'unpacked size is %d, %d expected') % (l2, l1), |
|
2900 | error=_(b'unpacked size is %d, %d expected') % (l2, l1), | |
2901 | node=node, |
|
2901 | node=node, | |
2902 | ) |
|
2902 | ) | |
2903 |
|
2903 | |||
2904 | except error.CensoredNodeError: |
|
2904 | except error.CensoredNodeError: | |
2905 | if state[b'erroroncensored']: |
|
2905 | if state[b'erroroncensored']: | |
2906 | yield revlogproblem( |
|
2906 | yield revlogproblem( | |
2907 | error=_(b'censored file data'), node=node |
|
2907 | error=_(b'censored file data'), node=node | |
2908 | ) |
|
2908 | ) | |
2909 | state[b'skipread'].add(node) |
|
2909 | state[b'skipread'].add(node) | |
2910 | except Exception as e: |
|
2910 | except Exception as e: | |
2911 | yield revlogproblem( |
|
2911 | yield revlogproblem( | |
2912 | error=_(b'unpacking %s: %s') |
|
2912 | error=_(b'unpacking %s: %s') | |
2913 | % (short(node), stringutil.forcebytestr(e)), |
|
2913 | % (short(node), stringutil.forcebytestr(e)), | |
2914 | node=node, |
|
2914 | node=node, | |
2915 | ) |
|
2915 | ) | |
2916 | state[b'skipread'].add(node) |
|
2916 | state[b'skipread'].add(node) | |
2917 |
|
2917 | |||
2918 | def storageinfo( |
|
2918 | def storageinfo( | |
2919 | self, |
|
2919 | self, | |
2920 | exclusivefiles=False, |
|
2920 | exclusivefiles=False, | |
2921 | sharedfiles=False, |
|
2921 | sharedfiles=False, | |
2922 | revisionscount=False, |
|
2922 | revisionscount=False, | |
2923 | trackedsize=False, |
|
2923 | trackedsize=False, | |
2924 | storedsize=False, |
|
2924 | storedsize=False, | |
2925 | ): |
|
2925 | ): | |
2926 | d = {} |
|
2926 | d = {} | |
2927 |
|
2927 | |||
2928 | if exclusivefiles: |
|
2928 | if exclusivefiles: | |
2929 | d[b'exclusivefiles'] = [(self.opener, self.indexfile)] |
|
2929 | d[b'exclusivefiles'] = [(self.opener, self.indexfile)] | |
2930 | if not self._inline: |
|
2930 | if not self._inline: | |
2931 | d[b'exclusivefiles'].append((self.opener, self.datafile)) |
|
2931 | d[b'exclusivefiles'].append((self.opener, self.datafile)) | |
2932 |
|
2932 | |||
2933 | if sharedfiles: |
|
2933 | if sharedfiles: | |
2934 | d[b'sharedfiles'] = [] |
|
2934 | d[b'sharedfiles'] = [] | |
2935 |
|
2935 | |||
2936 | if revisionscount: |
|
2936 | if revisionscount: | |
2937 | d[b'revisionscount'] = len(self) |
|
2937 | d[b'revisionscount'] = len(self) | |
2938 |
|
2938 | |||
2939 | if trackedsize: |
|
2939 | if trackedsize: | |
2940 | d[b'trackedsize'] = sum(map(self.rawsize, iter(self))) |
|
2940 | d[b'trackedsize'] = sum(map(self.rawsize, iter(self))) | |
2941 |
|
2941 | |||
2942 | if storedsize: |
|
2942 | if storedsize: | |
2943 | d[b'storedsize'] = sum( |
|
2943 | d[b'storedsize'] = sum( | |
2944 | self.opener.stat(path).st_size for path in self.files() |
|
2944 | self.opener.stat(path).st_size for path in self.files() | |
2945 | ) |
|
2945 | ) | |
2946 |
|
2946 | |||
2947 | return d |
|
2947 | return d |
General Comments 0
You need to be logged in to leave comments.
Login now