Show More
@@ -1,4635 +1,4637 b'' | |||
|
1 | 1 | # perf.py - performance test routines |
|
2 | 2 | '''helper extension to measure performance |
|
3 | 3 | |
|
4 | 4 | Configurations |
|
5 | 5 | ============== |
|
6 | 6 | |
|
7 | 7 | ``perf`` |
|
8 | 8 | -------- |
|
9 | 9 | |
|
10 | 10 | ``all-timing`` |
|
11 | 11 | When set, additional statistics will be reported for each benchmark: best, |
|
12 | 12 | worst, median average. If not set only the best timing is reported |
|
13 | 13 | (default: off). |
|
14 | 14 | |
|
15 | 15 | ``presleep`` |
|
16 | 16 | number of second to wait before any group of runs (default: 1) |
|
17 | 17 | |
|
18 | 18 | ``pre-run`` |
|
19 | 19 | number of run to perform before starting measurement. |
|
20 | 20 | |
|
21 | 21 | ``profile-benchmark`` |
|
22 | 22 | Enable profiling for the benchmarked section. |
|
23 | 23 | (The first iteration is benchmarked) |
|
24 | 24 | |
|
25 | 25 | ``run-limits`` |
|
26 | 26 | Control the number of runs each benchmark will perform. The option value |
|
27 | 27 | should be a list of `<time>-<numberofrun>` pairs. After each run the |
|
28 | 28 | conditions are considered in order with the following logic: |
|
29 | 29 | |
|
30 | 30 | If benchmark has been running for <time> seconds, and we have performed |
|
31 | 31 | <numberofrun> iterations, stop the benchmark, |
|
32 | 32 | |
|
33 | 33 | The default value is: `3.0-100, 10.0-3` |
|
34 | 34 | |
|
35 | 35 | ``stub`` |
|
36 | 36 | When set, benchmarks will only be run once, useful for testing |
|
37 | 37 | (default: off) |
|
38 | 38 | ''' |
|
39 | 39 | |
|
40 | 40 | # "historical portability" policy of perf.py: |
|
41 | 41 | # |
|
42 | 42 | # We have to do: |
|
43 | 43 | # - make perf.py "loadable" with as wide Mercurial version as possible |
|
44 | 44 | # This doesn't mean that perf commands work correctly with that Mercurial. |
|
45 | 45 | # BTW, perf.py itself has been available since 1.1 (or eb240755386d). |
|
46 | 46 | # - make historical perf command work correctly with as wide Mercurial |
|
47 | 47 | # version as possible |
|
48 | 48 | # |
|
49 | 49 | # We have to do, if possible with reasonable cost: |
|
50 | 50 | # - make recent perf command for historical feature work correctly |
|
51 | 51 | # with early Mercurial |
|
52 | 52 | # |
|
53 | 53 | # We don't have to do: |
|
54 | 54 | # - make perf command for recent feature work correctly with early |
|
55 | 55 | # Mercurial |
|
56 | 56 | |
|
57 | 57 | import contextlib |
|
58 | 58 | import functools |
|
59 | 59 | import gc |
|
60 | 60 | import os |
|
61 | 61 | import random |
|
62 | 62 | import shutil |
|
63 | 63 | import struct |
|
64 | 64 | import sys |
|
65 | 65 | import tempfile |
|
66 | 66 | import threading |
|
67 | 67 | import time |
|
68 | 68 | |
|
69 | 69 | import mercurial.revlog |
|
70 | 70 | from mercurial import ( |
|
71 | 71 | changegroup, |
|
72 | 72 | cmdutil, |
|
73 | 73 | commands, |
|
74 | 74 | copies, |
|
75 | 75 | error, |
|
76 | 76 | extensions, |
|
77 | 77 | hg, |
|
78 | 78 | mdiff, |
|
79 | 79 | merge, |
|
80 | 80 | util, |
|
81 | 81 | ) |
|
82 | 82 | |
|
83 | 83 | # for "historical portability": |
|
84 | 84 | # try to import modules separately (in dict order), and ignore |
|
85 | 85 | # failure, because these aren't available with early Mercurial |
|
86 | 86 | try: |
|
87 | 87 | from mercurial import branchmap # since 2.5 (or bcee63733aad) |
|
88 | 88 | except ImportError: |
|
89 | 89 | pass |
|
90 | 90 | try: |
|
91 | 91 | from mercurial import obsolete # since 2.3 (or ad0d6c2b3279) |
|
92 | 92 | except ImportError: |
|
93 | 93 | pass |
|
94 | 94 | try: |
|
95 | 95 | from mercurial import registrar # since 3.7 (or 37d50250b696) |
|
96 | 96 | |
|
97 | 97 | dir(registrar) # forcibly load it |
|
98 | 98 | except ImportError: |
|
99 | 99 | registrar = None |
|
100 | 100 | try: |
|
101 | 101 | from mercurial import repoview # since 2.5 (or 3a6ddacb7198) |
|
102 | 102 | except ImportError: |
|
103 | 103 | pass |
|
104 | 104 | try: |
|
105 | 105 | from mercurial.utils import repoviewutil # since 5.0 |
|
106 | 106 | except ImportError: |
|
107 | 107 | repoviewutil = None |
|
108 | 108 | try: |
|
109 | 109 | from mercurial import scmutil # since 1.9 (or 8b252e826c68) |
|
110 | 110 | except ImportError: |
|
111 | 111 | pass |
|
112 | 112 | try: |
|
113 | 113 | from mercurial import setdiscovery # since 1.9 (or cb98fed52495) |
|
114 | 114 | except ImportError: |
|
115 | 115 | pass |
|
116 | 116 | |
|
117 | 117 | try: |
|
118 | 118 | from mercurial import profiling |
|
119 | 119 | except ImportError: |
|
120 | 120 | profiling = None |
|
121 | 121 | |
|
122 | 122 | try: |
|
123 | 123 | from mercurial.revlogutils import constants as revlog_constants |
|
124 | 124 | |
|
125 | 125 | perf_rl_kind = (revlog_constants.KIND_OTHER, b'created-by-perf') |
|
126 | 126 | |
|
127 | 127 | def revlog(opener, *args, **kwargs): |
|
128 | 128 | return mercurial.revlog.revlog(opener, perf_rl_kind, *args, **kwargs) |
|
129 | 129 | |
|
130 | 130 | |
|
131 | 131 | except (ImportError, AttributeError): |
|
132 | 132 | perf_rl_kind = None |
|
133 | 133 | |
|
134 | 134 | def revlog(opener, *args, **kwargs): |
|
135 | 135 | return mercurial.revlog.revlog(opener, *args, **kwargs) |
|
136 | 136 | |
|
137 | 137 | |
|
138 | 138 | def identity(a): |
|
139 | 139 | return a |
|
140 | 140 | |
|
141 | 141 | |
|
142 | 142 | try: |
|
143 | 143 | from mercurial import pycompat |
|
144 | 144 | |
|
145 | 145 | getargspec = pycompat.getargspec # added to module after 4.5 |
|
146 | 146 | _byteskwargs = pycompat.byteskwargs # since 4.1 (or fbc3f73dc802) |
|
147 | 147 | _sysstr = pycompat.sysstr # since 4.0 (or 2219f4f82ede) |
|
148 | 148 | _bytestr = pycompat.bytestr # since 4.2 (or b70407bd84d5) |
|
149 | 149 | _xrange = pycompat.xrange # since 4.8 (or 7eba8f83129b) |
|
150 | 150 | fsencode = pycompat.fsencode # since 3.9 (or f4a5e0e86a7e) |
|
151 | 151 | if pycompat.ispy3: |
|
152 | 152 | _maxint = sys.maxsize # per py3 docs for replacing maxint |
|
153 | 153 | else: |
|
154 | 154 | _maxint = sys.maxint |
|
155 | 155 | except (NameError, ImportError, AttributeError): |
|
156 | 156 | import inspect |
|
157 | 157 | |
|
158 | 158 | getargspec = inspect.getargspec |
|
159 | 159 | _byteskwargs = identity |
|
160 | 160 | _bytestr = str |
|
161 | 161 | fsencode = identity # no py3 support |
|
162 | 162 | _maxint = sys.maxint # no py3 support |
|
163 | 163 | _sysstr = lambda x: x # no py3 support |
|
164 | 164 | _xrange = xrange |
|
165 | 165 | |
|
166 | 166 | try: |
|
167 | 167 | # 4.7+ |
|
168 | 168 | queue = pycompat.queue.Queue |
|
169 | 169 | except (NameError, AttributeError, ImportError): |
|
170 | 170 | # <4.7. |
|
171 | 171 | try: |
|
172 | 172 | queue = pycompat.queue |
|
173 | 173 | except (NameError, AttributeError, ImportError): |
|
174 | 174 | import Queue as queue |
|
175 | 175 | |
|
176 | 176 | try: |
|
177 | 177 | from mercurial import logcmdutil |
|
178 | 178 | |
|
179 | 179 | makelogtemplater = logcmdutil.maketemplater |
|
180 | 180 | except (AttributeError, ImportError): |
|
181 | 181 | try: |
|
182 | 182 | makelogtemplater = cmdutil.makelogtemplater |
|
183 | 183 | except (AttributeError, ImportError): |
|
184 | 184 | makelogtemplater = None |
|
185 | 185 | |
|
186 | 186 | # for "historical portability": |
|
187 | 187 | # define util.safehasattr forcibly, because util.safehasattr has been |
|
188 | 188 | # available since 1.9.3 (or 94b200a11cf7) |
|
189 | 189 | _undefined = object() |
|
190 | 190 | |
|
191 | 191 | |
|
192 | 192 | def safehasattr(thing, attr): |
|
193 | 193 | return getattr(thing, _sysstr(attr), _undefined) is not _undefined |
|
194 | 194 | |
|
195 | 195 | |
|
196 | 196 | setattr(util, 'safehasattr', safehasattr) |
|
197 | 197 | |
|
198 | 198 | # for "historical portability": |
|
199 | 199 | # define util.timer forcibly, because util.timer has been available |
|
200 | 200 | # since ae5d60bb70c9 |
|
201 | 201 | if safehasattr(time, 'perf_counter'): |
|
202 | 202 | util.timer = time.perf_counter |
|
203 | 203 | elif os.name == b'nt': |
|
204 | 204 | util.timer = time.clock |
|
205 | 205 | else: |
|
206 | 206 | util.timer = time.time |
|
207 | 207 | |
|
208 | 208 | # for "historical portability": |
|
209 | 209 | # use locally defined empty option list, if formatteropts isn't |
|
210 | 210 | # available, because commands.formatteropts has been available since |
|
211 | 211 | # 3.2 (or 7a7eed5176a4), even though formatting itself has been |
|
212 | 212 | # available since 2.2 (or ae5f92e154d3) |
|
213 | 213 | formatteropts = getattr( |
|
214 | 214 | cmdutil, "formatteropts", getattr(commands, "formatteropts", []) |
|
215 | 215 | ) |
|
216 | 216 | |
|
217 | 217 | # for "historical portability": |
|
218 | 218 | # use locally defined option list, if debugrevlogopts isn't available, |
|
219 | 219 | # because commands.debugrevlogopts has been available since 3.7 (or |
|
220 | 220 | # 5606f7d0d063), even though cmdutil.openrevlog() has been available |
|
221 | 221 | # since 1.9 (or a79fea6b3e77). |
|
222 | 222 | revlogopts = getattr( |
|
223 | 223 | cmdutil, |
|
224 | 224 | "debugrevlogopts", |
|
225 | 225 | getattr( |
|
226 | 226 | commands, |
|
227 | 227 | "debugrevlogopts", |
|
228 | 228 | [ |
|
229 | 229 | (b'c', b'changelog', False, b'open changelog'), |
|
230 | 230 | (b'm', b'manifest', False, b'open manifest'), |
|
231 | 231 | (b'', b'dir', False, b'open directory manifest'), |
|
232 | 232 | ], |
|
233 | 233 | ), |
|
234 | 234 | ) |
|
235 | 235 | |
|
236 | 236 | cmdtable = {} |
|
237 | 237 | |
|
238 | 238 | |
|
239 | 239 | # for "historical portability": |
|
240 | 240 | # define parsealiases locally, because cmdutil.parsealiases has been |
|
241 | 241 | # available since 1.5 (or 6252852b4332) |
|
242 | 242 | def parsealiases(cmd): |
|
243 | 243 | return cmd.split(b"|") |
|
244 | 244 | |
|
245 | 245 | |
|
246 | 246 | if safehasattr(registrar, 'command'): |
|
247 | 247 | command = registrar.command(cmdtable) |
|
248 | 248 | elif safehasattr(cmdutil, 'command'): |
|
249 | 249 | command = cmdutil.command(cmdtable) |
|
250 | 250 | if 'norepo' not in getargspec(command).args: |
|
251 | 251 | # for "historical portability": |
|
252 | 252 | # wrap original cmdutil.command, because "norepo" option has |
|
253 | 253 | # been available since 3.1 (or 75a96326cecb) |
|
254 | 254 | _command = command |
|
255 | 255 | |
|
256 | 256 | def command(name, options=(), synopsis=None, norepo=False): |
|
257 | 257 | if norepo: |
|
258 | 258 | commands.norepo += b' %s' % b' '.join(parsealiases(name)) |
|
259 | 259 | return _command(name, list(options), synopsis) |
|
260 | 260 | |
|
261 | 261 | |
|
262 | 262 | else: |
|
263 | 263 | # for "historical portability": |
|
264 | 264 | # define "@command" annotation locally, because cmdutil.command |
|
265 | 265 | # has been available since 1.9 (or 2daa5179e73f) |
|
266 | 266 | def command(name, options=(), synopsis=None, norepo=False): |
|
267 | 267 | def decorator(func): |
|
268 | 268 | if synopsis: |
|
269 | 269 | cmdtable[name] = func, list(options), synopsis |
|
270 | 270 | else: |
|
271 | 271 | cmdtable[name] = func, list(options) |
|
272 | 272 | if norepo: |
|
273 | 273 | commands.norepo += b' %s' % b' '.join(parsealiases(name)) |
|
274 | 274 | return func |
|
275 | 275 | |
|
276 | 276 | return decorator |
|
277 | 277 | |
|
278 | 278 | |
|
279 | 279 | try: |
|
280 | 280 | import mercurial.registrar |
|
281 | 281 | import mercurial.configitems |
|
282 | 282 | |
|
283 | 283 | configtable = {} |
|
284 | 284 | configitem = mercurial.registrar.configitem(configtable) |
|
285 | 285 | configitem( |
|
286 | 286 | b'perf', |
|
287 | 287 | b'presleep', |
|
288 | 288 | default=mercurial.configitems.dynamicdefault, |
|
289 | 289 | experimental=True, |
|
290 | 290 | ) |
|
291 | 291 | configitem( |
|
292 | 292 | b'perf', |
|
293 | 293 | b'stub', |
|
294 | 294 | default=mercurial.configitems.dynamicdefault, |
|
295 | 295 | experimental=True, |
|
296 | 296 | ) |
|
297 | 297 | configitem( |
|
298 | 298 | b'perf', |
|
299 | 299 | b'parentscount', |
|
300 | 300 | default=mercurial.configitems.dynamicdefault, |
|
301 | 301 | experimental=True, |
|
302 | 302 | ) |
|
303 | 303 | configitem( |
|
304 | 304 | b'perf', |
|
305 | 305 | b'all-timing', |
|
306 | 306 | default=mercurial.configitems.dynamicdefault, |
|
307 | 307 | experimental=True, |
|
308 | 308 | ) |
|
309 | 309 | configitem( |
|
310 | 310 | b'perf', |
|
311 | 311 | b'pre-run', |
|
312 | 312 | default=mercurial.configitems.dynamicdefault, |
|
313 | 313 | ) |
|
314 | 314 | configitem( |
|
315 | 315 | b'perf', |
|
316 | 316 | b'profile-benchmark', |
|
317 | 317 | default=mercurial.configitems.dynamicdefault, |
|
318 | 318 | ) |
|
319 | 319 | configitem( |
|
320 | 320 | b'perf', |
|
321 | 321 | b'run-limits', |
|
322 | 322 | default=mercurial.configitems.dynamicdefault, |
|
323 | 323 | experimental=True, |
|
324 | 324 | ) |
|
325 | 325 | except (ImportError, AttributeError): |
|
326 | 326 | pass |
|
327 | 327 | except TypeError: |
|
328 | 328 | # compatibility fix for a11fd395e83f |
|
329 | 329 | # hg version: 5.2 |
|
330 | 330 | configitem( |
|
331 | 331 | b'perf', |
|
332 | 332 | b'presleep', |
|
333 | 333 | default=mercurial.configitems.dynamicdefault, |
|
334 | 334 | ) |
|
335 | 335 | configitem( |
|
336 | 336 | b'perf', |
|
337 | 337 | b'stub', |
|
338 | 338 | default=mercurial.configitems.dynamicdefault, |
|
339 | 339 | ) |
|
340 | 340 | configitem( |
|
341 | 341 | b'perf', |
|
342 | 342 | b'parentscount', |
|
343 | 343 | default=mercurial.configitems.dynamicdefault, |
|
344 | 344 | ) |
|
345 | 345 | configitem( |
|
346 | 346 | b'perf', |
|
347 | 347 | b'all-timing', |
|
348 | 348 | default=mercurial.configitems.dynamicdefault, |
|
349 | 349 | ) |
|
350 | 350 | configitem( |
|
351 | 351 | b'perf', |
|
352 | 352 | b'pre-run', |
|
353 | 353 | default=mercurial.configitems.dynamicdefault, |
|
354 | 354 | ) |
|
355 | 355 | configitem( |
|
356 | 356 | b'perf', |
|
357 | 357 | b'profile-benchmark', |
|
358 | 358 | default=mercurial.configitems.dynamicdefault, |
|
359 | 359 | ) |
|
360 | 360 | configitem( |
|
361 | 361 | b'perf', |
|
362 | 362 | b'run-limits', |
|
363 | 363 | default=mercurial.configitems.dynamicdefault, |
|
364 | 364 | ) |
|
365 | 365 | |
|
366 | 366 | |
|
367 | 367 | def getlen(ui): |
|
368 | 368 | if ui.configbool(b"perf", b"stub", False): |
|
369 | 369 | return lambda x: 1 |
|
370 | 370 | return len |
|
371 | 371 | |
|
372 | 372 | |
|
373 | 373 | class noop: |
|
374 | 374 | """dummy context manager""" |
|
375 | 375 | |
|
376 | 376 | def __enter__(self): |
|
377 | 377 | pass |
|
378 | 378 | |
|
379 | 379 | def __exit__(self, *args): |
|
380 | 380 | pass |
|
381 | 381 | |
|
382 | 382 | |
|
383 | 383 | NOOPCTX = noop() |
|
384 | 384 | |
|
385 | 385 | |
|
386 | 386 | def gettimer(ui, opts=None): |
|
387 | 387 | """return a timer function and formatter: (timer, formatter) |
|
388 | 388 | |
|
389 | 389 | This function exists to gather the creation of formatter in a single |
|
390 | 390 | place instead of duplicating it in all performance commands.""" |
|
391 | 391 | |
|
392 | 392 | # enforce an idle period before execution to counteract power management |
|
393 | 393 | # experimental config: perf.presleep |
|
394 | 394 | time.sleep(getint(ui, b"perf", b"presleep", 1)) |
|
395 | 395 | |
|
396 | 396 | if opts is None: |
|
397 | 397 | opts = {} |
|
398 | 398 | # redirect all to stderr unless buffer api is in use |
|
399 | 399 | if not ui._buffers: |
|
400 | 400 | ui = ui.copy() |
|
401 | 401 | uifout = safeattrsetter(ui, b'fout', ignoremissing=True) |
|
402 | 402 | if uifout: |
|
403 | 403 | # for "historical portability": |
|
404 | 404 | # ui.fout/ferr have been available since 1.9 (or 4e1ccd4c2b6d) |
|
405 | 405 | uifout.set(ui.ferr) |
|
406 | 406 | |
|
407 | 407 | # get a formatter |
|
408 | 408 | uiformatter = getattr(ui, 'formatter', None) |
|
409 | 409 | if uiformatter: |
|
410 | 410 | fm = uiformatter(b'perf', opts) |
|
411 | 411 | else: |
|
412 | 412 | # for "historical portability": |
|
413 | 413 | # define formatter locally, because ui.formatter has been |
|
414 | 414 | # available since 2.2 (or ae5f92e154d3) |
|
415 | 415 | from mercurial import node |
|
416 | 416 | |
|
417 | 417 | class defaultformatter: |
|
418 | 418 | """Minimized composition of baseformatter and plainformatter""" |
|
419 | 419 | |
|
420 | 420 | def __init__(self, ui, topic, opts): |
|
421 | 421 | self._ui = ui |
|
422 | 422 | if ui.debugflag: |
|
423 | 423 | self.hexfunc = node.hex |
|
424 | 424 | else: |
|
425 | 425 | self.hexfunc = node.short |
|
426 | 426 | |
|
427 | 427 | def __nonzero__(self): |
|
428 | 428 | return False |
|
429 | 429 | |
|
430 | 430 | __bool__ = __nonzero__ |
|
431 | 431 | |
|
432 | 432 | def startitem(self): |
|
433 | 433 | pass |
|
434 | 434 | |
|
435 | 435 | def data(self, **data): |
|
436 | 436 | pass |
|
437 | 437 | |
|
438 | 438 | def write(self, fields, deftext, *fielddata, **opts): |
|
439 | 439 | self._ui.write(deftext % fielddata, **opts) |
|
440 | 440 | |
|
441 | 441 | def condwrite(self, cond, fields, deftext, *fielddata, **opts): |
|
442 | 442 | if cond: |
|
443 | 443 | self._ui.write(deftext % fielddata, **opts) |
|
444 | 444 | |
|
445 | 445 | def plain(self, text, **opts): |
|
446 | 446 | self._ui.write(text, **opts) |
|
447 | 447 | |
|
448 | 448 | def end(self): |
|
449 | 449 | pass |
|
450 | 450 | |
|
451 | 451 | fm = defaultformatter(ui, b'perf', opts) |
|
452 | 452 | |
|
453 | 453 | # stub function, runs code only once instead of in a loop |
|
454 | 454 | # experimental config: perf.stub |
|
455 | 455 | if ui.configbool(b"perf", b"stub", False): |
|
456 | 456 | return functools.partial(stub_timer, fm), fm |
|
457 | 457 | |
|
458 | 458 | # experimental config: perf.all-timing |
|
459 | 459 | displayall = ui.configbool(b"perf", b"all-timing", True) |
|
460 | 460 | |
|
461 | 461 | # experimental config: perf.run-limits |
|
462 | 462 | limitspec = ui.configlist(b"perf", b"run-limits", []) |
|
463 | 463 | limits = [] |
|
464 | 464 | for item in limitspec: |
|
465 | 465 | parts = item.split(b'-', 1) |
|
466 | 466 | if len(parts) < 2: |
|
467 | 467 | ui.warn((b'malformatted run limit entry, missing "-": %s\n' % item)) |
|
468 | 468 | continue |
|
469 | 469 | try: |
|
470 | 470 | time_limit = float(_sysstr(parts[0])) |
|
471 | 471 | except ValueError as e: |
|
472 | 472 | ui.warn( |
|
473 | 473 | ( |
|
474 | 474 | b'malformatted run limit entry, %s: %s\n' |
|
475 | 475 | % (_bytestr(e), item) |
|
476 | 476 | ) |
|
477 | 477 | ) |
|
478 | 478 | continue |
|
479 | 479 | try: |
|
480 | 480 | run_limit = int(_sysstr(parts[1])) |
|
481 | 481 | except ValueError as e: |
|
482 | 482 | ui.warn( |
|
483 | 483 | ( |
|
484 | 484 | b'malformatted run limit entry, %s: %s\n' |
|
485 | 485 | % (_bytestr(e), item) |
|
486 | 486 | ) |
|
487 | 487 | ) |
|
488 | 488 | continue |
|
489 | 489 | limits.append((time_limit, run_limit)) |
|
490 | 490 | if not limits: |
|
491 | 491 | limits = DEFAULTLIMITS |
|
492 | 492 | |
|
493 | 493 | profiler = None |
|
494 | 494 | if profiling is not None: |
|
495 | 495 | if ui.configbool(b"perf", b"profile-benchmark", False): |
|
496 | 496 | profiler = profiling.profile(ui) |
|
497 | 497 | |
|
498 | 498 | prerun = getint(ui, b"perf", b"pre-run", 0) |
|
499 | 499 | t = functools.partial( |
|
500 | 500 | _timer, |
|
501 | 501 | fm, |
|
502 | 502 | displayall=displayall, |
|
503 | 503 | limits=limits, |
|
504 | 504 | prerun=prerun, |
|
505 | 505 | profiler=profiler, |
|
506 | 506 | ) |
|
507 | 507 | return t, fm |
|
508 | 508 | |
|
509 | 509 | |
|
510 | 510 | def stub_timer(fm, func, setup=None, title=None): |
|
511 | 511 | if setup is not None: |
|
512 | 512 | setup() |
|
513 | 513 | func() |
|
514 | 514 | |
|
515 | 515 | |
|
516 | 516 | @contextlib.contextmanager |
|
517 | 517 | def timeone(): |
|
518 | 518 | r = [] |
|
519 | 519 | ostart = os.times() |
|
520 | 520 | cstart = util.timer() |
|
521 | 521 | yield r |
|
522 | 522 | cstop = util.timer() |
|
523 | 523 | ostop = os.times() |
|
524 | 524 | a, b = ostart, ostop |
|
525 | 525 | r.append((cstop - cstart, b[0] - a[0], b[1] - a[1])) |
|
526 | 526 | |
|
527 | 527 | |
|
528 | 528 | # list of stop condition (elapsed time, minimal run count) |
|
529 | 529 | DEFAULTLIMITS = ( |
|
530 | 530 | (3.0, 100), |
|
531 | 531 | (10.0, 3), |
|
532 | 532 | ) |
|
533 | 533 | |
|
534 | 534 | |
|
535 | 535 | @contextlib.contextmanager |
|
536 | 536 | def noop_context(): |
|
537 | 537 | yield |
|
538 | 538 | |
|
539 | 539 | |
|
540 | 540 | def _timer( |
|
541 | 541 | fm, |
|
542 | 542 | func, |
|
543 | 543 | setup=None, |
|
544 | 544 | context=noop_context, |
|
545 | 545 | title=None, |
|
546 | 546 | displayall=False, |
|
547 | 547 | limits=DEFAULTLIMITS, |
|
548 | 548 | prerun=0, |
|
549 | 549 | profiler=None, |
|
550 | 550 | ): |
|
551 | 551 | gc.collect() |
|
552 | 552 | results = [] |
|
553 | 553 | begin = util.timer() |
|
554 | 554 | count = 0 |
|
555 | 555 | if profiler is None: |
|
556 | 556 | profiler = NOOPCTX |
|
557 | 557 | for i in range(prerun): |
|
558 | 558 | if setup is not None: |
|
559 | 559 | setup() |
|
560 | 560 | with context(): |
|
561 | 561 | func() |
|
562 | 562 | keepgoing = True |
|
563 | 563 | while keepgoing: |
|
564 | 564 | if setup is not None: |
|
565 | 565 | setup() |
|
566 | 566 | with context(): |
|
567 | 567 | with profiler: |
|
568 | 568 | with timeone() as item: |
|
569 | 569 | r = func() |
|
570 | 570 | profiler = NOOPCTX |
|
571 | 571 | count += 1 |
|
572 | 572 | results.append(item[0]) |
|
573 | 573 | cstop = util.timer() |
|
574 | 574 | # Look for a stop condition. |
|
575 | 575 | elapsed = cstop - begin |
|
576 | 576 | for t, mincount in limits: |
|
577 | 577 | if elapsed >= t and count >= mincount: |
|
578 | 578 | keepgoing = False |
|
579 | 579 | break |
|
580 | 580 | |
|
581 | 581 | formatone(fm, results, title=title, result=r, displayall=displayall) |
|
582 | 582 | |
|
583 | 583 | |
|
584 | 584 | def formatone(fm, timings, title=None, result=None, displayall=False): |
|
585 | 585 | count = len(timings) |
|
586 | 586 | |
|
587 | 587 | fm.startitem() |
|
588 | 588 | |
|
589 | 589 | if title: |
|
590 | 590 | fm.write(b'title', b'! %s\n', title) |
|
591 | 591 | if result: |
|
592 | 592 | fm.write(b'result', b'! result: %s\n', result) |
|
593 | 593 | |
|
594 | 594 | def display(role, entry): |
|
595 | 595 | prefix = b'' |
|
596 | 596 | if role != b'best': |
|
597 | 597 | prefix = b'%s.' % role |
|
598 | 598 | fm.plain(b'!') |
|
599 | 599 | fm.write(prefix + b'wall', b' wall %f', entry[0]) |
|
600 | 600 | fm.write(prefix + b'comb', b' comb %f', entry[1] + entry[2]) |
|
601 | 601 | fm.write(prefix + b'user', b' user %f', entry[1]) |
|
602 | 602 | fm.write(prefix + b'sys', b' sys %f', entry[2]) |
|
603 | 603 | fm.write(prefix + b'count', b' (%s of %%d)' % role, count) |
|
604 | 604 | fm.plain(b'\n') |
|
605 | 605 | |
|
606 | 606 | timings.sort() |
|
607 | 607 | min_val = timings[0] |
|
608 | 608 | display(b'best', min_val) |
|
609 | 609 | if displayall: |
|
610 | 610 | max_val = timings[-1] |
|
611 | 611 | display(b'max', max_val) |
|
612 | 612 | avg = tuple([sum(x) / count for x in zip(*timings)]) |
|
613 | 613 | display(b'avg', avg) |
|
614 | 614 | median = timings[len(timings) // 2] |
|
615 | 615 | display(b'median', median) |
|
616 | 616 | |
|
617 | 617 | |
|
618 | 618 | # utilities for historical portability |
|
619 | 619 | |
|
620 | 620 | |
|
621 | 621 | def getint(ui, section, name, default): |
|
622 | 622 | # for "historical portability": |
|
623 | 623 | # ui.configint has been available since 1.9 (or fa2b596db182) |
|
624 | 624 | v = ui.config(section, name, None) |
|
625 | 625 | if v is None: |
|
626 | 626 | return default |
|
627 | 627 | try: |
|
628 | 628 | return int(v) |
|
629 | 629 | except ValueError: |
|
630 | 630 | raise error.ConfigError( |
|
631 | 631 | b"%s.%s is not an integer ('%s')" % (section, name, v) |
|
632 | 632 | ) |
|
633 | 633 | |
|
634 | 634 | |
|
635 | 635 | def safeattrsetter(obj, name, ignoremissing=False): |
|
636 | 636 | """Ensure that 'obj' has 'name' attribute before subsequent setattr |
|
637 | 637 | |
|
638 | 638 | This function is aborted, if 'obj' doesn't have 'name' attribute |
|
639 | 639 | at runtime. This avoids overlooking removal of an attribute, which |
|
640 | 640 | breaks assumption of performance measurement, in the future. |
|
641 | 641 | |
|
642 | 642 | This function returns the object to (1) assign a new value, and |
|
643 | 643 | (2) restore an original value to the attribute. |
|
644 | 644 | |
|
645 | 645 | If 'ignoremissing' is true, missing 'name' attribute doesn't cause |
|
646 | 646 | abortion, and this function returns None. This is useful to |
|
647 | 647 | examine an attribute, which isn't ensured in all Mercurial |
|
648 | 648 | versions. |
|
649 | 649 | """ |
|
650 | 650 | if not util.safehasattr(obj, name): |
|
651 | 651 | if ignoremissing: |
|
652 | 652 | return None |
|
653 | 653 | raise error.Abort( |
|
654 | 654 | ( |
|
655 | 655 | b"missing attribute %s of %s might break assumption" |
|
656 | 656 | b" of performance measurement" |
|
657 | 657 | ) |
|
658 | 658 | % (name, obj) |
|
659 | 659 | ) |
|
660 | 660 | |
|
661 | 661 | origvalue = getattr(obj, _sysstr(name)) |
|
662 | 662 | |
|
663 | 663 | class attrutil: |
|
664 | 664 | def set(self, newvalue): |
|
665 | 665 | setattr(obj, _sysstr(name), newvalue) |
|
666 | 666 | |
|
667 | 667 | def restore(self): |
|
668 | 668 | setattr(obj, _sysstr(name), origvalue) |
|
669 | 669 | |
|
670 | 670 | return attrutil() |
|
671 | 671 | |
|
672 | 672 | |
|
673 | 673 | # utilities to examine each internal API changes |
|
674 | 674 | |
|
675 | 675 | |
|
676 | 676 | def getbranchmapsubsettable(): |
|
677 | 677 | # for "historical portability": |
|
678 | 678 | # subsettable is defined in: |
|
679 | 679 | # - branchmap since 2.9 (or 175c6fd8cacc) |
|
680 | 680 | # - repoview since 2.5 (or 59a9f18d4587) |
|
681 | 681 | # - repoviewutil since 5.0 |
|
682 | 682 | for mod in (branchmap, repoview, repoviewutil): |
|
683 | 683 | subsettable = getattr(mod, 'subsettable', None) |
|
684 | 684 | if subsettable: |
|
685 | 685 | return subsettable |
|
686 | 686 | |
|
687 | 687 | # bisecting in bcee63733aad::59a9f18d4587 can reach here (both |
|
688 | 688 | # branchmap and repoview modules exist, but subsettable attribute |
|
689 | 689 | # doesn't) |
|
690 | 690 | raise error.Abort( |
|
691 | 691 | b"perfbranchmap not available with this Mercurial", |
|
692 | 692 | hint=b"use 2.5 or later", |
|
693 | 693 | ) |
|
694 | 694 | |
|
695 | 695 | |
|
696 | 696 | def getsvfs(repo): |
|
697 | 697 | """Return appropriate object to access files under .hg/store""" |
|
698 | 698 | # for "historical portability": |
|
699 | 699 | # repo.svfs has been available since 2.3 (or 7034365089bf) |
|
700 | 700 | svfs = getattr(repo, 'svfs', None) |
|
701 | 701 | if svfs: |
|
702 | 702 | return svfs |
|
703 | 703 | else: |
|
704 | 704 | return getattr(repo, 'sopener') |
|
705 | 705 | |
|
706 | 706 | |
|
707 | 707 | def getvfs(repo): |
|
708 | 708 | """Return appropriate object to access files under .hg""" |
|
709 | 709 | # for "historical portability": |
|
710 | 710 | # repo.vfs has been available since 2.3 (or 7034365089bf) |
|
711 | 711 | vfs = getattr(repo, 'vfs', None) |
|
712 | 712 | if vfs: |
|
713 | 713 | return vfs |
|
714 | 714 | else: |
|
715 | 715 | return getattr(repo, 'opener') |
|
716 | 716 | |
|
717 | 717 | |
|
718 | 718 | def repocleartagscachefunc(repo): |
|
719 | 719 | """Return the function to clear tags cache according to repo internal API""" |
|
720 | 720 | if util.safehasattr(repo, b'_tagscache'): # since 2.0 (or 9dca7653b525) |
|
721 | 721 | # in this case, setattr(repo, '_tagscache', None) or so isn't |
|
722 | 722 | # correct way to clear tags cache, because existing code paths |
|
723 | 723 | # expect _tagscache to be a structured object. |
|
724 | 724 | def clearcache(): |
|
725 | 725 | # _tagscache has been filteredpropertycache since 2.5 (or |
|
726 | 726 | # 98c867ac1330), and delattr() can't work in such case |
|
727 | 727 | if '_tagscache' in vars(repo): |
|
728 | 728 | del repo.__dict__['_tagscache'] |
|
729 | 729 | |
|
730 | 730 | return clearcache |
|
731 | 731 | |
|
732 | 732 | repotags = safeattrsetter(repo, b'_tags', ignoremissing=True) |
|
733 | 733 | if repotags: # since 1.4 (or 5614a628d173) |
|
734 | 734 | return lambda: repotags.set(None) |
|
735 | 735 | |
|
736 | 736 | repotagscache = safeattrsetter(repo, b'tagscache', ignoremissing=True) |
|
737 | 737 | if repotagscache: # since 0.6 (or d7df759d0e97) |
|
738 | 738 | return lambda: repotagscache.set(None) |
|
739 | 739 | |
|
740 | 740 | # Mercurial earlier than 0.6 (or d7df759d0e97) logically reaches |
|
741 | 741 | # this point, but it isn't so problematic, because: |
|
742 | 742 | # - repo.tags of such Mercurial isn't "callable", and repo.tags() |
|
743 | 743 | # in perftags() causes failure soon |
|
744 | 744 | # - perf.py itself has been available since 1.1 (or eb240755386d) |
|
745 | 745 | raise error.Abort(b"tags API of this hg command is unknown") |
|
746 | 746 | |
|
747 | 747 | |
|
748 | 748 | # utilities to clear cache |
|
749 | 749 | |
|
750 | 750 | |
|
751 | 751 | def clearfilecache(obj, attrname): |
|
752 | 752 | unfiltered = getattr(obj, 'unfiltered', None) |
|
753 | 753 | if unfiltered is not None: |
|
754 | 754 | obj = obj.unfiltered() |
|
755 | 755 | if attrname in vars(obj): |
|
756 | 756 | delattr(obj, attrname) |
|
757 | 757 | obj._filecache.pop(attrname, None) |
|
758 | 758 | |
|
759 | 759 | |
|
760 | 760 | def clearchangelog(repo): |
|
761 | 761 | if repo is not repo.unfiltered(): |
|
762 | 762 | object.__setattr__(repo, '_clcachekey', None) |
|
763 | 763 | object.__setattr__(repo, '_clcache', None) |
|
764 | 764 | clearfilecache(repo.unfiltered(), 'changelog') |
|
765 | 765 | |
|
766 | 766 | |
|
767 | 767 | # perf commands |
|
768 | 768 | |
|
769 | 769 | |
|
770 | 770 | @command(b'perf::walk|perfwalk', formatteropts) |
|
771 | 771 | def perfwalk(ui, repo, *pats, **opts): |
|
772 | 772 | opts = _byteskwargs(opts) |
|
773 | 773 | timer, fm = gettimer(ui, opts) |
|
774 | 774 | m = scmutil.match(repo[None], pats, {}) |
|
775 | 775 | timer( |
|
776 | 776 | lambda: len( |
|
777 | 777 | list( |
|
778 | 778 | repo.dirstate.walk(m, subrepos=[], unknown=True, ignored=False) |
|
779 | 779 | ) |
|
780 | 780 | ) |
|
781 | 781 | ) |
|
782 | 782 | fm.end() |
|
783 | 783 | |
|
784 | 784 | |
|
785 | 785 | @command(b'perf::annotate|perfannotate', formatteropts) |
|
786 | 786 | def perfannotate(ui, repo, f, **opts): |
|
787 | 787 | opts = _byteskwargs(opts) |
|
788 | 788 | timer, fm = gettimer(ui, opts) |
|
789 | 789 | fc = repo[b'.'][f] |
|
790 | 790 | timer(lambda: len(fc.annotate(True))) |
|
791 | 791 | fm.end() |
|
792 | 792 | |
|
793 | 793 | |
|
794 | 794 | @command( |
|
795 | 795 | b'perf::status|perfstatus', |
|
796 | 796 | [ |
|
797 | 797 | (b'u', b'unknown', False, b'ask status to look for unknown files'), |
|
798 | 798 | (b'', b'dirstate', False, b'benchmark the internal dirstate call'), |
|
799 | 799 | ] |
|
800 | 800 | + formatteropts, |
|
801 | 801 | ) |
|
802 | 802 | def perfstatus(ui, repo, **opts): |
|
803 | 803 | """benchmark the performance of a single status call |
|
804 | 804 | |
|
805 | 805 | The repository data are preserved between each call. |
|
806 | 806 | |
|
807 | 807 | By default, only the status of the tracked file are requested. If |
|
808 | 808 | `--unknown` is passed, the "unknown" files are also tracked. |
|
809 | 809 | """ |
|
810 | 810 | opts = _byteskwargs(opts) |
|
811 | 811 | # m = match.always(repo.root, repo.getcwd()) |
|
812 | 812 | # timer(lambda: sum(map(len, repo.dirstate.status(m, [], False, False, |
|
813 | 813 | # False)))) |
|
814 | 814 | timer, fm = gettimer(ui, opts) |
|
815 | 815 | if opts[b'dirstate']: |
|
816 | 816 | dirstate = repo.dirstate |
|
817 | 817 | m = scmutil.matchall(repo) |
|
818 | 818 | unknown = opts[b'unknown'] |
|
819 | 819 | |
|
820 | 820 | def status_dirstate(): |
|
821 | 821 | s = dirstate.status( |
|
822 | 822 | m, subrepos=[], ignored=False, clean=False, unknown=unknown |
|
823 | 823 | ) |
|
824 | 824 | sum(map(bool, s)) |
|
825 | 825 | |
|
826 | 826 | if util.safehasattr(dirstate, 'running_status'): |
|
827 | 827 | with dirstate.running_status(repo): |
|
828 | 828 | timer(status_dirstate) |
|
829 | 829 | dirstate.invalidate() |
|
830 | 830 | else: |
|
831 | 831 | timer(status_dirstate) |
|
832 | 832 | else: |
|
833 | 833 | timer(lambda: sum(map(len, repo.status(unknown=opts[b'unknown'])))) |
|
834 | 834 | fm.end() |
|
835 | 835 | |
|
836 | 836 | |
|
837 | 837 | @command(b'perf::addremove|perfaddremove', formatteropts) |
|
838 | 838 | def perfaddremove(ui, repo, **opts): |
|
839 | 839 | opts = _byteskwargs(opts) |
|
840 | 840 | timer, fm = gettimer(ui, opts) |
|
841 | 841 | try: |
|
842 | 842 | oldquiet = repo.ui.quiet |
|
843 | 843 | repo.ui.quiet = True |
|
844 | 844 | matcher = scmutil.match(repo[None]) |
|
845 | 845 | opts[b'dry_run'] = True |
|
846 | 846 | if 'uipathfn' in getargspec(scmutil.addremove).args: |
|
847 | 847 | uipathfn = scmutil.getuipathfn(repo) |
|
848 | 848 | timer(lambda: scmutil.addremove(repo, matcher, b"", uipathfn, opts)) |
|
849 | 849 | else: |
|
850 | 850 | timer(lambda: scmutil.addremove(repo, matcher, b"", opts)) |
|
851 | 851 | finally: |
|
852 | 852 | repo.ui.quiet = oldquiet |
|
853 | 853 | fm.end() |
|
854 | 854 | |
|
855 | 855 | |
|
856 | 856 | def clearcaches(cl): |
|
857 | 857 | # behave somewhat consistently across internal API changes |
|
858 | 858 | if util.safehasattr(cl, b'clearcaches'): |
|
859 | 859 | cl.clearcaches() |
|
860 | 860 | elif util.safehasattr(cl, b'_nodecache'): |
|
861 | 861 | # <= hg-5.2 |
|
862 | 862 | from mercurial.node import nullid, nullrev |
|
863 | 863 | |
|
864 | 864 | cl._nodecache = {nullid: nullrev} |
|
865 | 865 | cl._nodepos = None |
|
866 | 866 | |
|
867 | 867 | |
|
868 | 868 | @command(b'perf::heads|perfheads', formatteropts) |
|
869 | 869 | def perfheads(ui, repo, **opts): |
|
870 | 870 | """benchmark the computation of a changelog heads""" |
|
871 | 871 | opts = _byteskwargs(opts) |
|
872 | 872 | timer, fm = gettimer(ui, opts) |
|
873 | 873 | cl = repo.changelog |
|
874 | 874 | |
|
875 | 875 | def s(): |
|
876 | 876 | clearcaches(cl) |
|
877 | 877 | |
|
878 | 878 | def d(): |
|
879 | 879 | len(cl.headrevs()) |
|
880 | 880 | |
|
881 | 881 | timer(d, setup=s) |
|
882 | 882 | fm.end() |
|
883 | 883 | |
|
884 | 884 | |
|
885 | 885 | def _default_clear_on_disk_tags_cache(repo): |
|
886 | 886 | from mercurial import tags |
|
887 | 887 | |
|
888 | 888 | repo.cachevfs.tryunlink(tags._filename(repo)) |
|
889 | 889 | |
|
890 | 890 | |
|
891 | 891 | def _default_clear_on_disk_tags_fnodes_cache(repo): |
|
892 | 892 | from mercurial import tags |
|
893 | 893 | |
|
894 | 894 | repo.cachevfs.tryunlink(tags._fnodescachefile) |
|
895 | 895 | |
|
896 | 896 | |
|
897 | 897 | def _default_forget_fnodes(repo, revs): |
|
898 | 898 | """function used by the perf extension to prune some entries from the |
|
899 | 899 | fnodes cache""" |
|
900 | 900 | from mercurial import tags |
|
901 | 901 | |
|
902 | 902 | missing_1 = b'\xff' * 4 |
|
903 | 903 | missing_2 = b'\xff' * 20 |
|
904 | 904 | cache = tags.hgtagsfnodescache(repo.unfiltered()) |
|
905 | 905 | for r in revs: |
|
906 | 906 | cache._writeentry(r * tags._fnodesrecsize, missing_1, missing_2) |
|
907 | 907 | cache.write() |
|
908 | 908 | |
|
909 | 909 | |
|
910 | 910 | @command( |
|
911 | 911 | b'perf::tags|perftags', |
|
912 | 912 | formatteropts |
|
913 | 913 | + [ |
|
914 | 914 | (b'', b'clear-revlogs', False, b'refresh changelog and manifest'), |
|
915 | 915 | ( |
|
916 | 916 | b'', |
|
917 | 917 | b'clear-on-disk-cache', |
|
918 | 918 | False, |
|
919 | 919 | b'clear on disk tags cache (DESTRUCTIVE)', |
|
920 | 920 | ), |
|
921 | 921 | ( |
|
922 | 922 | b'', |
|
923 | 923 | b'clear-fnode-cache-all', |
|
924 | 924 | False, |
|
925 | 925 | b'clear on disk file node cache (DESTRUCTIVE),', |
|
926 | 926 | ), |
|
927 | 927 | ( |
|
928 | 928 | b'', |
|
929 | 929 | b'clear-fnode-cache-rev', |
|
930 | 930 | [], |
|
931 | 931 | b'clear on disk file node cache (DESTRUCTIVE),', |
|
932 | 932 | b'REVS', |
|
933 | 933 | ), |
|
934 | 934 | ( |
|
935 | 935 | b'', |
|
936 | 936 | b'update-last', |
|
937 | 937 | b'', |
|
938 | 938 | b'simulate an update over the last N revisions (DESTRUCTIVE),', |
|
939 | 939 | b'N', |
|
940 | 940 | ), |
|
941 | 941 | ], |
|
942 | 942 | ) |
|
943 | 943 | def perftags(ui, repo, **opts): |
|
944 | 944 | """Benchmark tags retrieval in various situation |
|
945 | 945 | |
|
946 | 946 | The option marked as (DESTRUCTIVE) will alter the on-disk cache, possibly |
|
947 | 947 | altering performance after the command was run. However, it does not |
|
948 | 948 | destroy any stored data. |
|
949 | 949 | """ |
|
950 | 950 | from mercurial import tags |
|
951 | 951 | |
|
952 | 952 | opts = _byteskwargs(opts) |
|
953 | 953 | timer, fm = gettimer(ui, opts) |
|
954 | 954 | repocleartagscache = repocleartagscachefunc(repo) |
|
955 | 955 | clearrevlogs = opts[b'clear_revlogs'] |
|
956 | 956 | clear_disk = opts[b'clear_on_disk_cache'] |
|
957 | 957 | clear_fnode = opts[b'clear_fnode_cache_all'] |
|
958 | 958 | |
|
959 | 959 | clear_fnode_revs = opts[b'clear_fnode_cache_rev'] |
|
960 | 960 | update_last_str = opts[b'update_last'] |
|
961 | 961 | update_last = None |
|
962 | 962 | if update_last_str: |
|
963 | 963 | try: |
|
964 | 964 | update_last = int(update_last_str) |
|
965 | 965 | except ValueError: |
|
966 | 966 | msg = b'could not parse value for update-last: "%s"' |
|
967 | 967 | msg %= update_last_str |
|
968 | 968 | hint = b'value should be an integer' |
|
969 | 969 | raise error.Abort(msg, hint=hint) |
|
970 | 970 | |
|
971 | 971 | clear_disk_fn = getattr( |
|
972 | 972 | tags, |
|
973 | 973 | "clear_cache_on_disk", |
|
974 | 974 | _default_clear_on_disk_tags_cache, |
|
975 | 975 | ) |
|
976 | 976 | clear_fnodes_fn = getattr( |
|
977 | 977 | tags, |
|
978 | 978 | "clear_cache_fnodes", |
|
979 | 979 | _default_clear_on_disk_tags_fnodes_cache, |
|
980 | 980 | ) |
|
981 | 981 | clear_fnodes_rev_fn = getattr( |
|
982 | 982 | tags, |
|
983 | 983 | "forget_fnodes", |
|
984 | 984 | _default_forget_fnodes, |
|
985 | 985 | ) |
|
986 | 986 | |
|
987 | 987 | clear_revs = [] |
|
988 | 988 | if clear_fnode_revs: |
|
989 | 989 | clear_revs.extends(scmutil.revrange(repo, clear_fnode_revs)) |
|
990 | 990 | |
|
991 | 991 | if update_last: |
|
992 | 992 | revset = b'last(all(), %d)' % update_last |
|
993 | 993 | last_revs = repo.unfiltered().revs(revset) |
|
994 | 994 | clear_revs.extend(last_revs) |
|
995 | 995 | |
|
996 | 996 | from mercurial import repoview |
|
997 | 997 | |
|
998 | 998 | rev_filter = {(b'experimental', b'extra-filter-revs'): revset} |
|
999 | 999 | with repo.ui.configoverride(rev_filter, source=b"perf"): |
|
1000 | 1000 | filter_id = repoview.extrafilter(repo.ui) |
|
1001 | 1001 | |
|
1002 | 1002 | filter_name = b'%s%%%s' % (repo.filtername, filter_id) |
|
1003 | 1003 | pre_repo = repo.filtered(filter_name) |
|
1004 | 1004 | pre_repo.tags() # warm the cache |
|
1005 | 1005 | old_tags_path = repo.cachevfs.join(tags._filename(pre_repo)) |
|
1006 | 1006 | new_tags_path = repo.cachevfs.join(tags._filename(repo)) |
|
1007 | 1007 | |
|
1008 | 1008 | clear_revs = sorted(set(clear_revs)) |
|
1009 | 1009 | |
|
1010 | 1010 | def s(): |
|
1011 | 1011 | if update_last: |
|
1012 | 1012 | util.copyfile(old_tags_path, new_tags_path) |
|
1013 | 1013 | if clearrevlogs: |
|
1014 | 1014 | clearchangelog(repo) |
|
1015 | 1015 | clearfilecache(repo.unfiltered(), 'manifest') |
|
1016 | 1016 | if clear_disk: |
|
1017 | 1017 | clear_disk_fn(repo) |
|
1018 | 1018 | if clear_fnode: |
|
1019 | 1019 | clear_fnodes_fn(repo) |
|
1020 | 1020 | elif clear_revs: |
|
1021 | 1021 | clear_fnodes_rev_fn(repo, clear_revs) |
|
1022 | 1022 | repocleartagscache() |
|
1023 | 1023 | |
|
1024 | 1024 | def t(): |
|
1025 | 1025 | len(repo.tags()) |
|
1026 | 1026 | |
|
1027 | 1027 | timer(t, setup=s) |
|
1028 | 1028 | fm.end() |
|
1029 | 1029 | |
|
1030 | 1030 | |
|
1031 | 1031 | @command(b'perf::ancestors|perfancestors', formatteropts) |
|
1032 | 1032 | def perfancestors(ui, repo, **opts): |
|
1033 | 1033 | opts = _byteskwargs(opts) |
|
1034 | 1034 | timer, fm = gettimer(ui, opts) |
|
1035 | 1035 | heads = repo.changelog.headrevs() |
|
1036 | 1036 | |
|
1037 | 1037 | def d(): |
|
1038 | 1038 | for a in repo.changelog.ancestors(heads): |
|
1039 | 1039 | pass |
|
1040 | 1040 | |
|
1041 | 1041 | timer(d) |
|
1042 | 1042 | fm.end() |
|
1043 | 1043 | |
|
1044 | 1044 | |
|
1045 | 1045 | @command(b'perf::ancestorset|perfancestorset', formatteropts) |
|
1046 | 1046 | def perfancestorset(ui, repo, revset, **opts): |
|
1047 | 1047 | opts = _byteskwargs(opts) |
|
1048 | 1048 | timer, fm = gettimer(ui, opts) |
|
1049 | 1049 | revs = repo.revs(revset) |
|
1050 | 1050 | heads = repo.changelog.headrevs() |
|
1051 | 1051 | |
|
1052 | 1052 | def d(): |
|
1053 | 1053 | s = repo.changelog.ancestors(heads) |
|
1054 | 1054 | for rev in revs: |
|
1055 | 1055 | rev in s |
|
1056 | 1056 | |
|
1057 | 1057 | timer(d) |
|
1058 | 1058 | fm.end() |
|
1059 | 1059 | |
|
1060 | 1060 | |
|
1061 | 1061 | @command( |
|
1062 | 1062 | b'perf::delta-find', |
|
1063 | 1063 | revlogopts + formatteropts, |
|
1064 | 1064 | b'-c|-m|FILE REV', |
|
1065 | 1065 | ) |
|
1066 | 1066 | def perf_delta_find(ui, repo, arg_1, arg_2=None, **opts): |
|
1067 | 1067 | """benchmark the process of finding a valid delta for a revlog revision |
|
1068 | 1068 | |
|
1069 | 1069 | When a revlog receives a new revision (e.g. from a commit, or from an |
|
1070 | 1070 | incoming bundle), it searches for a suitable delta-base to produce a delta. |
|
1071 | 1071 | This perf command measures how much time we spend in this process. It |
|
1072 | 1072 | operates on an already stored revision. |
|
1073 | 1073 | |
|
1074 | 1074 | See `hg help debug-delta-find` for another related command. |
|
1075 | 1075 | """ |
|
1076 | 1076 | from mercurial import revlogutils |
|
1077 | 1077 | import mercurial.revlogutils.deltas as deltautil |
|
1078 | 1078 | |
|
1079 | 1079 | opts = _byteskwargs(opts) |
|
1080 | 1080 | if arg_2 is None: |
|
1081 | 1081 | file_ = None |
|
1082 | 1082 | rev = arg_1 |
|
1083 | 1083 | else: |
|
1084 | 1084 | file_ = arg_1 |
|
1085 | 1085 | rev = arg_2 |
|
1086 | 1086 | |
|
1087 | 1087 | repo = repo.unfiltered() |
|
1088 | 1088 | |
|
1089 | 1089 | timer, fm = gettimer(ui, opts) |
|
1090 | 1090 | |
|
1091 | 1091 | rev = int(rev) |
|
1092 | 1092 | |
|
1093 | 1093 | revlog = cmdutil.openrevlog(repo, b'perf::delta-find', file_, opts) |
|
1094 | 1094 | |
|
1095 | 1095 | deltacomputer = deltautil.deltacomputer(revlog) |
|
1096 | 1096 | |
|
1097 | 1097 | node = revlog.node(rev) |
|
1098 | 1098 | p1r, p2r = revlog.parentrevs(rev) |
|
1099 | 1099 | p1 = revlog.node(p1r) |
|
1100 | 1100 | p2 = revlog.node(p2r) |
|
1101 | 1101 | full_text = revlog.revision(rev) |
|
1102 | 1102 | textlen = len(full_text) |
|
1103 | 1103 | cachedelta = None |
|
1104 | 1104 | flags = revlog.flags(rev) |
|
1105 | 1105 | |
|
1106 | 1106 | revinfo = revlogutils.revisioninfo( |
|
1107 | 1107 | node, |
|
1108 | 1108 | p1, |
|
1109 | 1109 | p2, |
|
1110 | 1110 | [full_text], # btext |
|
1111 | 1111 | textlen, |
|
1112 | 1112 | cachedelta, |
|
1113 | 1113 | flags, |
|
1114 | 1114 | ) |
|
1115 | 1115 | |
|
1116 | 1116 | # Note: we should probably purge the potential caches (like the full |
|
1117 | 1117 | # manifest cache) between runs. |
|
1118 | 1118 | def find_one(): |
|
1119 | 1119 | with revlog._datafp() as fh: |
|
1120 | 1120 | deltacomputer.finddeltainfo(revinfo, fh, target_rev=rev) |
|
1121 | 1121 | |
|
1122 | 1122 | timer(find_one) |
|
1123 | 1123 | fm.end() |
|
1124 | 1124 | |
|
1125 | 1125 | |
|
1126 | 1126 | @command(b'perf::discovery|perfdiscovery', formatteropts, b'PATH') |
|
1127 | 1127 | def perfdiscovery(ui, repo, path, **opts): |
|
1128 | 1128 | """benchmark discovery between local repo and the peer at given path""" |
|
1129 | 1129 | repos = [repo, None] |
|
1130 | 1130 | timer, fm = gettimer(ui, opts) |
|
1131 | 1131 | |
|
1132 | 1132 | try: |
|
1133 | 1133 | from mercurial.utils.urlutil import get_unique_pull_path_obj |
|
1134 | 1134 | |
|
1135 | 1135 | path = get_unique_pull_path_obj(b'perfdiscovery', ui, path) |
|
1136 | 1136 | except ImportError: |
|
1137 | 1137 | try: |
|
1138 | 1138 | from mercurial.utils.urlutil import get_unique_pull_path |
|
1139 | 1139 | |
|
1140 | 1140 | path = get_unique_pull_path(b'perfdiscovery', repo, ui, path)[0] |
|
1141 | 1141 | except ImportError: |
|
1142 | 1142 | path = ui.expandpath(path) |
|
1143 | 1143 | |
|
1144 | 1144 | def s(): |
|
1145 | 1145 | repos[1] = hg.peer(ui, opts, path) |
|
1146 | 1146 | |
|
1147 | 1147 | def d(): |
|
1148 | 1148 | setdiscovery.findcommonheads(ui, *repos) |
|
1149 | 1149 | |
|
1150 | 1150 | timer(d, setup=s) |
|
1151 | 1151 | fm.end() |
|
1152 | 1152 | |
|
1153 | 1153 | |
|
1154 | 1154 | @command( |
|
1155 | 1155 | b'perf::bookmarks|perfbookmarks', |
|
1156 | 1156 | formatteropts |
|
1157 | 1157 | + [ |
|
1158 | 1158 | (b'', b'clear-revlogs', False, b'refresh changelog and manifest'), |
|
1159 | 1159 | ], |
|
1160 | 1160 | ) |
|
1161 | 1161 | def perfbookmarks(ui, repo, **opts): |
|
1162 | 1162 | """benchmark parsing bookmarks from disk to memory""" |
|
1163 | 1163 | opts = _byteskwargs(opts) |
|
1164 | 1164 | timer, fm = gettimer(ui, opts) |
|
1165 | 1165 | |
|
1166 | 1166 | clearrevlogs = opts[b'clear_revlogs'] |
|
1167 | 1167 | |
|
1168 | 1168 | def s(): |
|
1169 | 1169 | if clearrevlogs: |
|
1170 | 1170 | clearchangelog(repo) |
|
1171 | 1171 | clearfilecache(repo, b'_bookmarks') |
|
1172 | 1172 | |
|
1173 | 1173 | def d(): |
|
1174 | 1174 | repo._bookmarks |
|
1175 | 1175 | |
|
1176 | 1176 | timer(d, setup=s) |
|
1177 | 1177 | fm.end() |
|
1178 | 1178 | |
|
1179 | 1179 | |
|
1180 | 1180 | @command( |
|
1181 | 1181 | b'perf::bundle', |
|
1182 | 1182 | [ |
|
1183 | 1183 | ( |
|
1184 | 1184 | b'r', |
|
1185 | 1185 | b'rev', |
|
1186 | 1186 | [], |
|
1187 | 1187 | b'changesets to bundle', |
|
1188 | 1188 | b'REV', |
|
1189 | 1189 | ), |
|
1190 | 1190 | ( |
|
1191 | 1191 | b't', |
|
1192 | 1192 | b'type', |
|
1193 | 1193 | b'none', |
|
1194 | 1194 | b'bundlespec to use (see `hg help bundlespec`)', |
|
1195 | 1195 | b'TYPE', |
|
1196 | 1196 | ), |
|
1197 | 1197 | ] |
|
1198 | 1198 | + formatteropts, |
|
1199 | 1199 | b'REVS', |
|
1200 | 1200 | ) |
|
1201 | 1201 | def perfbundle(ui, repo, *revs, **opts): |
|
1202 | 1202 | """benchmark the creation of a bundle from a repository |
|
1203 | 1203 | |
|
1204 | 1204 | For now, this only supports "none" compression. |
|
1205 | 1205 | """ |
|
1206 | 1206 | try: |
|
1207 | 1207 | from mercurial import bundlecaches |
|
1208 | 1208 | |
|
1209 | 1209 | parsebundlespec = bundlecaches.parsebundlespec |
|
1210 | 1210 | except ImportError: |
|
1211 | 1211 | from mercurial import exchange |
|
1212 | 1212 | |
|
1213 | 1213 | parsebundlespec = exchange.parsebundlespec |
|
1214 | 1214 | |
|
1215 | 1215 | from mercurial import discovery |
|
1216 | 1216 | from mercurial import bundle2 |
|
1217 | 1217 | |
|
1218 | 1218 | opts = _byteskwargs(opts) |
|
1219 | 1219 | timer, fm = gettimer(ui, opts) |
|
1220 | 1220 | |
|
1221 | 1221 | cl = repo.changelog |
|
1222 | 1222 | revs = list(revs) |
|
1223 | 1223 | revs.extend(opts.get(b'rev', ())) |
|
1224 | 1224 | revs = scmutil.revrange(repo, revs) |
|
1225 | 1225 | if not revs: |
|
1226 | 1226 | raise error.Abort(b"not revision specified") |
|
1227 | 1227 | # make it a consistent set (ie: without topological gaps) |
|
1228 | 1228 | old_len = len(revs) |
|
1229 | 1229 | revs = list(repo.revs(b"%ld::%ld", revs, revs)) |
|
1230 | 1230 | if old_len != len(revs): |
|
1231 | 1231 | new_count = len(revs) - old_len |
|
1232 | 1232 | msg = b"add %d new revisions to make it a consistent set\n" |
|
1233 | 1233 | ui.write_err(msg % new_count) |
|
1234 | 1234 | |
|
1235 | 1235 | targets = [cl.node(r) for r in repo.revs(b"heads(::%ld)", revs)] |
|
1236 | 1236 | bases = [cl.node(r) for r in repo.revs(b"heads(::%ld - %ld)", revs, revs)] |
|
1237 | 1237 | outgoing = discovery.outgoing(repo, bases, targets) |
|
1238 | 1238 | |
|
1239 | 1239 | bundle_spec = opts.get(b'type') |
|
1240 | 1240 | |
|
1241 | 1241 | bundle_spec = parsebundlespec(repo, bundle_spec, strict=False) |
|
1242 | 1242 | |
|
1243 | 1243 | cgversion = bundle_spec.params.get(b"cg.version") |
|
1244 | 1244 | if cgversion is None: |
|
1245 | 1245 | if bundle_spec.version == b'v1': |
|
1246 | 1246 | cgversion = b'01' |
|
1247 | 1247 | if bundle_spec.version == b'v2': |
|
1248 | 1248 | cgversion = b'02' |
|
1249 | 1249 | if cgversion not in changegroup.supportedoutgoingversions(repo): |
|
1250 | 1250 | err = b"repository does not support bundle version %s" |
|
1251 | 1251 | raise error.Abort(err % cgversion) |
|
1252 | 1252 | |
|
1253 | 1253 | if cgversion == b'01': # bundle1 |
|
1254 | 1254 | bversion = b'HG10' + bundle_spec.wirecompression |
|
1255 | 1255 | bcompression = None |
|
1256 | 1256 | elif cgversion in (b'02', b'03'): |
|
1257 | 1257 | bversion = b'HG20' |
|
1258 | 1258 | bcompression = bundle_spec.wirecompression |
|
1259 | 1259 | else: |
|
1260 | 1260 | err = b'perf::bundle: unexpected changegroup version %s' |
|
1261 | 1261 | raise error.ProgrammingError(err % cgversion) |
|
1262 | 1262 | |
|
1263 | 1263 | if bcompression is None: |
|
1264 | 1264 | bcompression = b'UN' |
|
1265 | 1265 | |
|
1266 | 1266 | if bcompression != b'UN': |
|
1267 | 1267 | err = b'perf::bundle: compression currently unsupported: %s' |
|
1268 | 1268 | raise error.ProgrammingError(err % bcompression) |
|
1269 | 1269 | |
|
1270 | 1270 | def do_bundle(): |
|
1271 | 1271 | bundle2.writenewbundle( |
|
1272 | 1272 | ui, |
|
1273 | 1273 | repo, |
|
1274 | 1274 | b'perf::bundle', |
|
1275 | 1275 | os.devnull, |
|
1276 | 1276 | bversion, |
|
1277 | 1277 | outgoing, |
|
1278 | 1278 | bundle_spec.params, |
|
1279 | 1279 | ) |
|
1280 | 1280 | |
|
1281 | 1281 | timer(do_bundle) |
|
1282 | 1282 | fm.end() |
|
1283 | 1283 | |
|
1284 | 1284 | |
|
1285 | 1285 | @command(b'perf::bundleread|perfbundleread', formatteropts, b'BUNDLE') |
|
1286 | 1286 | def perfbundleread(ui, repo, bundlepath, **opts): |
|
1287 | 1287 | """Benchmark reading of bundle files. |
|
1288 | 1288 | |
|
1289 | 1289 | This command is meant to isolate the I/O part of bundle reading as |
|
1290 | 1290 | much as possible. |
|
1291 | 1291 | """ |
|
1292 | 1292 | from mercurial import ( |
|
1293 | 1293 | bundle2, |
|
1294 | 1294 | exchange, |
|
1295 | 1295 | streamclone, |
|
1296 | 1296 | ) |
|
1297 | 1297 | |
|
1298 | 1298 | opts = _byteskwargs(opts) |
|
1299 | 1299 | |
|
1300 | 1300 | def makebench(fn): |
|
1301 | 1301 | def run(): |
|
1302 | 1302 | with open(bundlepath, b'rb') as fh: |
|
1303 | 1303 | bundle = exchange.readbundle(ui, fh, bundlepath) |
|
1304 | 1304 | fn(bundle) |
|
1305 | 1305 | |
|
1306 | 1306 | return run |
|
1307 | 1307 | |
|
1308 | 1308 | def makereadnbytes(size): |
|
1309 | 1309 | def run(): |
|
1310 | 1310 | with open(bundlepath, b'rb') as fh: |
|
1311 | 1311 | bundle = exchange.readbundle(ui, fh, bundlepath) |
|
1312 | 1312 | while bundle.read(size): |
|
1313 | 1313 | pass |
|
1314 | 1314 | |
|
1315 | 1315 | return run |
|
1316 | 1316 | |
|
1317 | 1317 | def makestdioread(size): |
|
1318 | 1318 | def run(): |
|
1319 | 1319 | with open(bundlepath, b'rb') as fh: |
|
1320 | 1320 | while fh.read(size): |
|
1321 | 1321 | pass |
|
1322 | 1322 | |
|
1323 | 1323 | return run |
|
1324 | 1324 | |
|
1325 | 1325 | # bundle1 |
|
1326 | 1326 | |
|
1327 | 1327 | def deltaiter(bundle): |
|
1328 | 1328 | for delta in bundle.deltaiter(): |
|
1329 | 1329 | pass |
|
1330 | 1330 | |
|
1331 | 1331 | def iterchunks(bundle): |
|
1332 | 1332 | for chunk in bundle.getchunks(): |
|
1333 | 1333 | pass |
|
1334 | 1334 | |
|
1335 | 1335 | # bundle2 |
|
1336 | 1336 | |
|
1337 | 1337 | def forwardchunks(bundle): |
|
1338 | 1338 | for chunk in bundle._forwardchunks(): |
|
1339 | 1339 | pass |
|
1340 | 1340 | |
|
1341 | 1341 | def iterparts(bundle): |
|
1342 | 1342 | for part in bundle.iterparts(): |
|
1343 | 1343 | pass |
|
1344 | 1344 | |
|
1345 | 1345 | def iterpartsseekable(bundle): |
|
1346 | 1346 | for part in bundle.iterparts(seekable=True): |
|
1347 | 1347 | pass |
|
1348 | 1348 | |
|
1349 | 1349 | def seek(bundle): |
|
1350 | 1350 | for part in bundle.iterparts(seekable=True): |
|
1351 | 1351 | part.seek(0, os.SEEK_END) |
|
1352 | 1352 | |
|
1353 | 1353 | def makepartreadnbytes(size): |
|
1354 | 1354 | def run(): |
|
1355 | 1355 | with open(bundlepath, b'rb') as fh: |
|
1356 | 1356 | bundle = exchange.readbundle(ui, fh, bundlepath) |
|
1357 | 1357 | for part in bundle.iterparts(): |
|
1358 | 1358 | while part.read(size): |
|
1359 | 1359 | pass |
|
1360 | 1360 | |
|
1361 | 1361 | return run |
|
1362 | 1362 | |
|
1363 | 1363 | benches = [ |
|
1364 | 1364 | (makestdioread(8192), b'read(8k)'), |
|
1365 | 1365 | (makestdioread(16384), b'read(16k)'), |
|
1366 | 1366 | (makestdioread(32768), b'read(32k)'), |
|
1367 | 1367 | (makestdioread(131072), b'read(128k)'), |
|
1368 | 1368 | ] |
|
1369 | 1369 | |
|
1370 | 1370 | with open(bundlepath, b'rb') as fh: |
|
1371 | 1371 | bundle = exchange.readbundle(ui, fh, bundlepath) |
|
1372 | 1372 | |
|
1373 | 1373 | if isinstance(bundle, changegroup.cg1unpacker): |
|
1374 | 1374 | benches.extend( |
|
1375 | 1375 | [ |
|
1376 | 1376 | (makebench(deltaiter), b'cg1 deltaiter()'), |
|
1377 | 1377 | (makebench(iterchunks), b'cg1 getchunks()'), |
|
1378 | 1378 | (makereadnbytes(8192), b'cg1 read(8k)'), |
|
1379 | 1379 | (makereadnbytes(16384), b'cg1 read(16k)'), |
|
1380 | 1380 | (makereadnbytes(32768), b'cg1 read(32k)'), |
|
1381 | 1381 | (makereadnbytes(131072), b'cg1 read(128k)'), |
|
1382 | 1382 | ] |
|
1383 | 1383 | ) |
|
1384 | 1384 | elif isinstance(bundle, bundle2.unbundle20): |
|
1385 | 1385 | benches.extend( |
|
1386 | 1386 | [ |
|
1387 | 1387 | (makebench(forwardchunks), b'bundle2 forwardchunks()'), |
|
1388 | 1388 | (makebench(iterparts), b'bundle2 iterparts()'), |
|
1389 | 1389 | ( |
|
1390 | 1390 | makebench(iterpartsseekable), |
|
1391 | 1391 | b'bundle2 iterparts() seekable', |
|
1392 | 1392 | ), |
|
1393 | 1393 | (makebench(seek), b'bundle2 part seek()'), |
|
1394 | 1394 | (makepartreadnbytes(8192), b'bundle2 part read(8k)'), |
|
1395 | 1395 | (makepartreadnbytes(16384), b'bundle2 part read(16k)'), |
|
1396 | 1396 | (makepartreadnbytes(32768), b'bundle2 part read(32k)'), |
|
1397 | 1397 | (makepartreadnbytes(131072), b'bundle2 part read(128k)'), |
|
1398 | 1398 | ] |
|
1399 | 1399 | ) |
|
1400 | 1400 | elif isinstance(bundle, streamclone.streamcloneapplier): |
|
1401 | 1401 | raise error.Abort(b'stream clone bundles not supported') |
|
1402 | 1402 | else: |
|
1403 | 1403 | raise error.Abort(b'unhandled bundle type: %s' % type(bundle)) |
|
1404 | 1404 | |
|
1405 | 1405 | for fn, title in benches: |
|
1406 | 1406 | timer, fm = gettimer(ui, opts) |
|
1407 | 1407 | timer(fn, title=title) |
|
1408 | 1408 | fm.end() |
|
1409 | 1409 | |
|
1410 | 1410 | |
|
1411 | 1411 | @command( |
|
1412 | 1412 | b'perf::changegroupchangelog|perfchangegroupchangelog', |
|
1413 | 1413 | formatteropts |
|
1414 | 1414 | + [ |
|
1415 | 1415 | (b'', b'cgversion', b'02', b'changegroup version'), |
|
1416 | 1416 | (b'r', b'rev', b'', b'revisions to add to changegroup'), |
|
1417 | 1417 | ], |
|
1418 | 1418 | ) |
|
1419 | 1419 | def perfchangegroupchangelog(ui, repo, cgversion=b'02', rev=None, **opts): |
|
1420 | 1420 | """Benchmark producing a changelog group for a changegroup. |
|
1421 | 1421 | |
|
1422 | 1422 | This measures the time spent processing the changelog during a |
|
1423 | 1423 | bundle operation. This occurs during `hg bundle` and on a server |
|
1424 | 1424 | processing a `getbundle` wire protocol request (handles clones |
|
1425 | 1425 | and pull requests). |
|
1426 | 1426 | |
|
1427 | 1427 | By default, all revisions are added to the changegroup. |
|
1428 | 1428 | """ |
|
1429 | 1429 | opts = _byteskwargs(opts) |
|
1430 | 1430 | cl = repo.changelog |
|
1431 | 1431 | nodes = [cl.lookup(r) for r in repo.revs(rev or b'all()')] |
|
1432 | 1432 | bundler = changegroup.getbundler(cgversion, repo) |
|
1433 | 1433 | |
|
1434 | 1434 | def d(): |
|
1435 | 1435 | state, chunks = bundler._generatechangelog(cl, nodes) |
|
1436 | 1436 | for chunk in chunks: |
|
1437 | 1437 | pass |
|
1438 | 1438 | |
|
1439 | 1439 | timer, fm = gettimer(ui, opts) |
|
1440 | 1440 | |
|
1441 | 1441 | # Terminal printing can interfere with timing. So disable it. |
|
1442 | 1442 | with ui.configoverride({(b'progress', b'disable'): True}): |
|
1443 | 1443 | timer(d) |
|
1444 | 1444 | |
|
1445 | 1445 | fm.end() |
|
1446 | 1446 | |
|
1447 | 1447 | |
|
1448 | 1448 | @command(b'perf::dirs|perfdirs', formatteropts) |
|
1449 | 1449 | def perfdirs(ui, repo, **opts): |
|
1450 | 1450 | opts = _byteskwargs(opts) |
|
1451 | 1451 | timer, fm = gettimer(ui, opts) |
|
1452 | 1452 | dirstate = repo.dirstate |
|
1453 | 1453 | b'a' in dirstate |
|
1454 | 1454 | |
|
1455 | 1455 | def d(): |
|
1456 | 1456 | dirstate.hasdir(b'a') |
|
1457 | 1457 | try: |
|
1458 | 1458 | del dirstate._map._dirs |
|
1459 | 1459 | except AttributeError: |
|
1460 | 1460 | pass |
|
1461 | 1461 | |
|
1462 | 1462 | timer(d) |
|
1463 | 1463 | fm.end() |
|
1464 | 1464 | |
|
1465 | 1465 | |
|
1466 | 1466 | @command( |
|
1467 | 1467 | b'perf::dirstate|perfdirstate', |
|
1468 | 1468 | [ |
|
1469 | 1469 | ( |
|
1470 | 1470 | b'', |
|
1471 | 1471 | b'iteration', |
|
1472 | 1472 | None, |
|
1473 | 1473 | b'benchmark a full iteration for the dirstate', |
|
1474 | 1474 | ), |
|
1475 | 1475 | ( |
|
1476 | 1476 | b'', |
|
1477 | 1477 | b'contains', |
|
1478 | 1478 | None, |
|
1479 | 1479 | b'benchmark a large amount of `nf in dirstate` calls', |
|
1480 | 1480 | ), |
|
1481 | 1481 | ] |
|
1482 | 1482 | + formatteropts, |
|
1483 | 1483 | ) |
|
1484 | 1484 | def perfdirstate(ui, repo, **opts): |
|
1485 | 1485 | """benchmap the time of various distate operations |
|
1486 | 1486 | |
|
1487 | 1487 | By default benchmark the time necessary to load a dirstate from scratch. |
|
1488 | 1488 | The dirstate is loaded to the point were a "contains" request can be |
|
1489 | 1489 | answered. |
|
1490 | 1490 | """ |
|
1491 | 1491 | opts = _byteskwargs(opts) |
|
1492 | 1492 | timer, fm = gettimer(ui, opts) |
|
1493 | 1493 | b"a" in repo.dirstate |
|
1494 | 1494 | |
|
1495 | 1495 | if opts[b'iteration'] and opts[b'contains']: |
|
1496 | 1496 | msg = b'only specify one of --iteration or --contains' |
|
1497 | 1497 | raise error.Abort(msg) |
|
1498 | 1498 | |
|
1499 | 1499 | if opts[b'iteration']: |
|
1500 | 1500 | setup = None |
|
1501 | 1501 | dirstate = repo.dirstate |
|
1502 | 1502 | |
|
1503 | 1503 | def d(): |
|
1504 | 1504 | for f in dirstate: |
|
1505 | 1505 | pass |
|
1506 | 1506 | |
|
1507 | 1507 | elif opts[b'contains']: |
|
1508 | 1508 | setup = None |
|
1509 | 1509 | dirstate = repo.dirstate |
|
1510 | 1510 | allfiles = list(dirstate) |
|
1511 | 1511 | # also add file path that will be "missing" from the dirstate |
|
1512 | 1512 | allfiles.extend([f[::-1] for f in allfiles]) |
|
1513 | 1513 | |
|
1514 | 1514 | def d(): |
|
1515 | 1515 | for f in allfiles: |
|
1516 | 1516 | f in dirstate |
|
1517 | 1517 | |
|
1518 | 1518 | else: |
|
1519 | 1519 | |
|
1520 | 1520 | def setup(): |
|
1521 | 1521 | repo.dirstate.invalidate() |
|
1522 | 1522 | |
|
1523 | 1523 | def d(): |
|
1524 | 1524 | b"a" in repo.dirstate |
|
1525 | 1525 | |
|
1526 | 1526 | timer(d, setup=setup) |
|
1527 | 1527 | fm.end() |
|
1528 | 1528 | |
|
1529 | 1529 | |
|
1530 | 1530 | @command(b'perf::dirstatedirs|perfdirstatedirs', formatteropts) |
|
1531 | 1531 | def perfdirstatedirs(ui, repo, **opts): |
|
1532 | 1532 | """benchmap a 'dirstate.hasdir' call from an empty `dirs` cache""" |
|
1533 | 1533 | opts = _byteskwargs(opts) |
|
1534 | 1534 | timer, fm = gettimer(ui, opts) |
|
1535 | 1535 | repo.dirstate.hasdir(b"a") |
|
1536 | 1536 | |
|
1537 | 1537 | def setup(): |
|
1538 | 1538 | try: |
|
1539 | 1539 | del repo.dirstate._map._dirs |
|
1540 | 1540 | except AttributeError: |
|
1541 | 1541 | pass |
|
1542 | 1542 | |
|
1543 | 1543 | def d(): |
|
1544 | 1544 | repo.dirstate.hasdir(b"a") |
|
1545 | 1545 | |
|
1546 | 1546 | timer(d, setup=setup) |
|
1547 | 1547 | fm.end() |
|
1548 | 1548 | |
|
1549 | 1549 | |
|
1550 | 1550 | @command(b'perf::dirstatefoldmap|perfdirstatefoldmap', formatteropts) |
|
1551 | 1551 | def perfdirstatefoldmap(ui, repo, **opts): |
|
1552 | 1552 | """benchmap a `dirstate._map.filefoldmap.get()` request |
|
1553 | 1553 | |
|
1554 | 1554 | The dirstate filefoldmap cache is dropped between every request. |
|
1555 | 1555 | """ |
|
1556 | 1556 | opts = _byteskwargs(opts) |
|
1557 | 1557 | timer, fm = gettimer(ui, opts) |
|
1558 | 1558 | dirstate = repo.dirstate |
|
1559 | 1559 | dirstate._map.filefoldmap.get(b'a') |
|
1560 | 1560 | |
|
1561 | 1561 | def setup(): |
|
1562 | 1562 | del dirstate._map.filefoldmap |
|
1563 | 1563 | |
|
1564 | 1564 | def d(): |
|
1565 | 1565 | dirstate._map.filefoldmap.get(b'a') |
|
1566 | 1566 | |
|
1567 | 1567 | timer(d, setup=setup) |
|
1568 | 1568 | fm.end() |
|
1569 | 1569 | |
|
1570 | 1570 | |
|
1571 | 1571 | @command(b'perf::dirfoldmap|perfdirfoldmap', formatteropts) |
|
1572 | 1572 | def perfdirfoldmap(ui, repo, **opts): |
|
1573 | 1573 | """benchmap a `dirstate._map.dirfoldmap.get()` request |
|
1574 | 1574 | |
|
1575 | 1575 | The dirstate dirfoldmap cache is dropped between every request. |
|
1576 | 1576 | """ |
|
1577 | 1577 | opts = _byteskwargs(opts) |
|
1578 | 1578 | timer, fm = gettimer(ui, opts) |
|
1579 | 1579 | dirstate = repo.dirstate |
|
1580 | 1580 | dirstate._map.dirfoldmap.get(b'a') |
|
1581 | 1581 | |
|
1582 | 1582 | def setup(): |
|
1583 | 1583 | del dirstate._map.dirfoldmap |
|
1584 | 1584 | try: |
|
1585 | 1585 | del dirstate._map._dirs |
|
1586 | 1586 | except AttributeError: |
|
1587 | 1587 | pass |
|
1588 | 1588 | |
|
1589 | 1589 | def d(): |
|
1590 | 1590 | dirstate._map.dirfoldmap.get(b'a') |
|
1591 | 1591 | |
|
1592 | 1592 | timer(d, setup=setup) |
|
1593 | 1593 | fm.end() |
|
1594 | 1594 | |
|
1595 | 1595 | |
|
1596 | 1596 | @command(b'perf::dirstatewrite|perfdirstatewrite', formatteropts) |
|
1597 | 1597 | def perfdirstatewrite(ui, repo, **opts): |
|
1598 | 1598 | """benchmap the time it take to write a dirstate on disk""" |
|
1599 | 1599 | opts = _byteskwargs(opts) |
|
1600 | 1600 | timer, fm = gettimer(ui, opts) |
|
1601 | 1601 | ds = repo.dirstate |
|
1602 | 1602 | b"a" in ds |
|
1603 | 1603 | |
|
1604 | 1604 | def setup(): |
|
1605 | 1605 | ds._dirty = True |
|
1606 | 1606 | |
|
1607 | 1607 | def d(): |
|
1608 | 1608 | ds.write(repo.currenttransaction()) |
|
1609 | 1609 | |
|
1610 | 1610 | with repo.wlock(): |
|
1611 | 1611 | timer(d, setup=setup) |
|
1612 | 1612 | fm.end() |
|
1613 | 1613 | |
|
1614 | 1614 | |
|
1615 | 1615 | def _getmergerevs(repo, opts): |
|
1616 | 1616 | """parse command argument to return rev involved in merge |
|
1617 | 1617 | |
|
1618 | 1618 | input: options dictionnary with `rev`, `from` and `bse` |
|
1619 | 1619 | output: (localctx, otherctx, basectx) |
|
1620 | 1620 | """ |
|
1621 | 1621 | if opts[b'from']: |
|
1622 | 1622 | fromrev = scmutil.revsingle(repo, opts[b'from']) |
|
1623 | 1623 | wctx = repo[fromrev] |
|
1624 | 1624 | else: |
|
1625 | 1625 | wctx = repo[None] |
|
1626 | 1626 | # we don't want working dir files to be stat'd in the benchmark, so |
|
1627 | 1627 | # prime that cache |
|
1628 | 1628 | wctx.dirty() |
|
1629 | 1629 | rctx = scmutil.revsingle(repo, opts[b'rev'], opts[b'rev']) |
|
1630 | 1630 | if opts[b'base']: |
|
1631 | 1631 | fromrev = scmutil.revsingle(repo, opts[b'base']) |
|
1632 | 1632 | ancestor = repo[fromrev] |
|
1633 | 1633 | else: |
|
1634 | 1634 | ancestor = wctx.ancestor(rctx) |
|
1635 | 1635 | return (wctx, rctx, ancestor) |
|
1636 | 1636 | |
|
1637 | 1637 | |
|
1638 | 1638 | @command( |
|
1639 | 1639 | b'perf::mergecalculate|perfmergecalculate', |
|
1640 | 1640 | [ |
|
1641 | 1641 | (b'r', b'rev', b'.', b'rev to merge against'), |
|
1642 | 1642 | (b'', b'from', b'', b'rev to merge from'), |
|
1643 | 1643 | (b'', b'base', b'', b'the revision to use as base'), |
|
1644 | 1644 | ] |
|
1645 | 1645 | + formatteropts, |
|
1646 | 1646 | ) |
|
1647 | 1647 | def perfmergecalculate(ui, repo, **opts): |
|
1648 | 1648 | opts = _byteskwargs(opts) |
|
1649 | 1649 | timer, fm = gettimer(ui, opts) |
|
1650 | 1650 | |
|
1651 | 1651 | wctx, rctx, ancestor = _getmergerevs(repo, opts) |
|
1652 | 1652 | |
|
1653 | 1653 | def d(): |
|
1654 | 1654 | # acceptremote is True because we don't want prompts in the middle of |
|
1655 | 1655 | # our benchmark |
|
1656 | 1656 | merge.calculateupdates( |
|
1657 | 1657 | repo, |
|
1658 | 1658 | wctx, |
|
1659 | 1659 | rctx, |
|
1660 | 1660 | [ancestor], |
|
1661 | 1661 | branchmerge=False, |
|
1662 | 1662 | force=False, |
|
1663 | 1663 | acceptremote=True, |
|
1664 | 1664 | followcopies=True, |
|
1665 | 1665 | ) |
|
1666 | 1666 | |
|
1667 | 1667 | timer(d) |
|
1668 | 1668 | fm.end() |
|
1669 | 1669 | |
|
1670 | 1670 | |
|
1671 | 1671 | @command( |
|
1672 | 1672 | b'perf::mergecopies|perfmergecopies', |
|
1673 | 1673 | [ |
|
1674 | 1674 | (b'r', b'rev', b'.', b'rev to merge against'), |
|
1675 | 1675 | (b'', b'from', b'', b'rev to merge from'), |
|
1676 | 1676 | (b'', b'base', b'', b'the revision to use as base'), |
|
1677 | 1677 | ] |
|
1678 | 1678 | + formatteropts, |
|
1679 | 1679 | ) |
|
1680 | 1680 | def perfmergecopies(ui, repo, **opts): |
|
1681 | 1681 | """measure runtime of `copies.mergecopies`""" |
|
1682 | 1682 | opts = _byteskwargs(opts) |
|
1683 | 1683 | timer, fm = gettimer(ui, opts) |
|
1684 | 1684 | wctx, rctx, ancestor = _getmergerevs(repo, opts) |
|
1685 | 1685 | |
|
1686 | 1686 | def d(): |
|
1687 | 1687 | # acceptremote is True because we don't want prompts in the middle of |
|
1688 | 1688 | # our benchmark |
|
1689 | 1689 | copies.mergecopies(repo, wctx, rctx, ancestor) |
|
1690 | 1690 | |
|
1691 | 1691 | timer(d) |
|
1692 | 1692 | fm.end() |
|
1693 | 1693 | |
|
1694 | 1694 | |
|
1695 | 1695 | @command(b'perf::pathcopies|perfpathcopies', [], b"REV REV") |
|
1696 | 1696 | def perfpathcopies(ui, repo, rev1, rev2, **opts): |
|
1697 | 1697 | """benchmark the copy tracing logic""" |
|
1698 | 1698 | opts = _byteskwargs(opts) |
|
1699 | 1699 | timer, fm = gettimer(ui, opts) |
|
1700 | 1700 | ctx1 = scmutil.revsingle(repo, rev1, rev1) |
|
1701 | 1701 | ctx2 = scmutil.revsingle(repo, rev2, rev2) |
|
1702 | 1702 | |
|
1703 | 1703 | def d(): |
|
1704 | 1704 | copies.pathcopies(ctx1, ctx2) |
|
1705 | 1705 | |
|
1706 | 1706 | timer(d) |
|
1707 | 1707 | fm.end() |
|
1708 | 1708 | |
|
1709 | 1709 | |
|
1710 | 1710 | @command( |
|
1711 | 1711 | b'perf::phases|perfphases', |
|
1712 | 1712 | [ |
|
1713 | 1713 | (b'', b'full', False, b'include file reading time too'), |
|
1714 | 1714 | ], |
|
1715 | 1715 | b"", |
|
1716 | 1716 | ) |
|
1717 | 1717 | def perfphases(ui, repo, **opts): |
|
1718 | 1718 | """benchmark phasesets computation""" |
|
1719 | 1719 | opts = _byteskwargs(opts) |
|
1720 | 1720 | timer, fm = gettimer(ui, opts) |
|
1721 | 1721 | _phases = repo._phasecache |
|
1722 | 1722 | full = opts.get(b'full') |
|
1723 | 1723 | |
|
1724 | 1724 | def d(): |
|
1725 | 1725 | phases = _phases |
|
1726 | 1726 | if full: |
|
1727 | 1727 | clearfilecache(repo, b'_phasecache') |
|
1728 | 1728 | phases = repo._phasecache |
|
1729 | 1729 | phases.invalidate() |
|
1730 | 1730 | phases.loadphaserevs(repo) |
|
1731 | 1731 | |
|
1732 | 1732 | timer(d) |
|
1733 | 1733 | fm.end() |
|
1734 | 1734 | |
|
1735 | 1735 | |
|
1736 | 1736 | @command(b'perf::phasesremote|perfphasesremote', [], b"[DEST]") |
|
1737 | 1737 | def perfphasesremote(ui, repo, dest=None, **opts): |
|
1738 | 1738 | """benchmark time needed to analyse phases of the remote server""" |
|
1739 | 1739 | from mercurial.node import bin |
|
1740 | 1740 | from mercurial import ( |
|
1741 | 1741 | exchange, |
|
1742 | 1742 | hg, |
|
1743 | 1743 | phases, |
|
1744 | 1744 | ) |
|
1745 | 1745 | |
|
1746 | 1746 | opts = _byteskwargs(opts) |
|
1747 | 1747 | timer, fm = gettimer(ui, opts) |
|
1748 | 1748 | |
|
1749 | 1749 | path = ui.getpath(dest, default=(b'default-push', b'default')) |
|
1750 | 1750 | if not path: |
|
1751 | 1751 | raise error.Abort( |
|
1752 | 1752 | b'default repository not configured!', |
|
1753 | 1753 | hint=b"see 'hg help config.paths'", |
|
1754 | 1754 | ) |
|
1755 | 1755 | if util.safehasattr(path, 'main_path'): |
|
1756 | 1756 | path = path.get_push_variant() |
|
1757 | 1757 | dest = path.loc |
|
1758 | 1758 | else: |
|
1759 | 1759 | dest = path.pushloc or path.loc |
|
1760 | 1760 | ui.statusnoi18n(b'analysing phase of %s\n' % util.hidepassword(dest)) |
|
1761 | 1761 | other = hg.peer(repo, opts, dest) |
|
1762 | 1762 | |
|
1763 | 1763 | # easier to perform discovery through the operation |
|
1764 | 1764 | op = exchange.pushoperation(repo, other) |
|
1765 | 1765 | exchange._pushdiscoverychangeset(op) |
|
1766 | 1766 | |
|
1767 | 1767 | remotesubset = op.fallbackheads |
|
1768 | 1768 | |
|
1769 | 1769 | with other.commandexecutor() as e: |
|
1770 | 1770 | remotephases = e.callcommand( |
|
1771 | 1771 | b'listkeys', {b'namespace': b'phases'} |
|
1772 | 1772 | ).result() |
|
1773 | 1773 | del other |
|
1774 | 1774 | publishing = remotephases.get(b'publishing', False) |
|
1775 | 1775 | if publishing: |
|
1776 | 1776 | ui.statusnoi18n(b'publishing: yes\n') |
|
1777 | 1777 | else: |
|
1778 | 1778 | ui.statusnoi18n(b'publishing: no\n') |
|
1779 | 1779 | |
|
1780 | 1780 | has_node = getattr(repo.changelog.index, 'has_node', None) |
|
1781 | 1781 | if has_node is None: |
|
1782 | 1782 | has_node = repo.changelog.nodemap.__contains__ |
|
1783 | 1783 | nonpublishroots = 0 |
|
1784 | 1784 | for nhex, phase in remotephases.iteritems(): |
|
1785 | 1785 | if nhex == b'publishing': # ignore data related to publish option |
|
1786 | 1786 | continue |
|
1787 | 1787 | node = bin(nhex) |
|
1788 | 1788 | if has_node(node) and int(phase): |
|
1789 | 1789 | nonpublishroots += 1 |
|
1790 | 1790 | ui.statusnoi18n(b'number of roots: %d\n' % len(remotephases)) |
|
1791 | 1791 | ui.statusnoi18n(b'number of known non public roots: %d\n' % nonpublishroots) |
|
1792 | 1792 | |
|
1793 | 1793 | def d(): |
|
1794 | 1794 | phases.remotephasessummary(repo, remotesubset, remotephases) |
|
1795 | 1795 | |
|
1796 | 1796 | timer(d) |
|
1797 | 1797 | fm.end() |
|
1798 | 1798 | |
|
1799 | 1799 | |
|
1800 | 1800 | @command( |
|
1801 | 1801 | b'perf::manifest|perfmanifest', |
|
1802 | 1802 | [ |
|
1803 | 1803 | (b'm', b'manifest-rev', False, b'Look up a manifest node revision'), |
|
1804 | 1804 | (b'', b'clear-disk', False, b'clear on-disk caches too'), |
|
1805 | 1805 | ] |
|
1806 | 1806 | + formatteropts, |
|
1807 | 1807 | b'REV|NODE', |
|
1808 | 1808 | ) |
|
1809 | 1809 | def perfmanifest(ui, repo, rev, manifest_rev=False, clear_disk=False, **opts): |
|
1810 | 1810 | """benchmark the time to read a manifest from disk and return a usable |
|
1811 | 1811 | dict-like object |
|
1812 | 1812 | |
|
1813 | 1813 | Manifest caches are cleared before retrieval.""" |
|
1814 | 1814 | opts = _byteskwargs(opts) |
|
1815 | 1815 | timer, fm = gettimer(ui, opts) |
|
1816 | 1816 | if not manifest_rev: |
|
1817 | 1817 | ctx = scmutil.revsingle(repo, rev, rev) |
|
1818 | 1818 | t = ctx.manifestnode() |
|
1819 | 1819 | else: |
|
1820 | 1820 | from mercurial.node import bin |
|
1821 | 1821 | |
|
1822 | 1822 | if len(rev) == 40: |
|
1823 | 1823 | t = bin(rev) |
|
1824 | 1824 | else: |
|
1825 | 1825 | try: |
|
1826 | 1826 | rev = int(rev) |
|
1827 | 1827 | |
|
1828 | 1828 | if util.safehasattr(repo.manifestlog, b'getstorage'): |
|
1829 | 1829 | t = repo.manifestlog.getstorage(b'').node(rev) |
|
1830 | 1830 | else: |
|
1831 | 1831 | t = repo.manifestlog._revlog.lookup(rev) |
|
1832 | 1832 | except ValueError: |
|
1833 | 1833 | raise error.Abort( |
|
1834 | 1834 | b'manifest revision must be integer or full node' |
|
1835 | 1835 | ) |
|
1836 | 1836 | |
|
1837 | 1837 | def d(): |
|
1838 | 1838 | repo.manifestlog.clearcaches(clear_persisted_data=clear_disk) |
|
1839 | 1839 | repo.manifestlog[t].read() |
|
1840 | 1840 | |
|
1841 | 1841 | timer(d) |
|
1842 | 1842 | fm.end() |
|
1843 | 1843 | |
|
1844 | 1844 | |
|
1845 | 1845 | @command(b'perf::changeset|perfchangeset', formatteropts) |
|
1846 | 1846 | def perfchangeset(ui, repo, rev, **opts): |
|
1847 | 1847 | opts = _byteskwargs(opts) |
|
1848 | 1848 | timer, fm = gettimer(ui, opts) |
|
1849 | 1849 | n = scmutil.revsingle(repo, rev).node() |
|
1850 | 1850 | |
|
1851 | 1851 | def d(): |
|
1852 | 1852 | repo.changelog.read(n) |
|
1853 | 1853 | # repo.changelog._cache = None |
|
1854 | 1854 | |
|
1855 | 1855 | timer(d) |
|
1856 | 1856 | fm.end() |
|
1857 | 1857 | |
|
1858 | 1858 | |
|
1859 | 1859 | @command(b'perf::ignore|perfignore', formatteropts) |
|
1860 | 1860 | def perfignore(ui, repo, **opts): |
|
1861 | 1861 | """benchmark operation related to computing ignore""" |
|
1862 | 1862 | opts = _byteskwargs(opts) |
|
1863 | 1863 | timer, fm = gettimer(ui, opts) |
|
1864 | 1864 | dirstate = repo.dirstate |
|
1865 | 1865 | |
|
1866 | 1866 | def setupone(): |
|
1867 | 1867 | dirstate.invalidate() |
|
1868 | 1868 | clearfilecache(dirstate, b'_ignore') |
|
1869 | 1869 | |
|
1870 | 1870 | def runone(): |
|
1871 | 1871 | dirstate._ignore |
|
1872 | 1872 | |
|
1873 | 1873 | timer(runone, setup=setupone, title=b"load") |
|
1874 | 1874 | fm.end() |
|
1875 | 1875 | |
|
1876 | 1876 | |
|
1877 | 1877 | @command( |
|
1878 | 1878 | b'perf::index|perfindex', |
|
1879 | 1879 | [ |
|
1880 | 1880 | (b'', b'rev', [], b'revision to be looked up (default tip)'), |
|
1881 | 1881 | (b'', b'no-lookup', None, b'do not revision lookup post creation'), |
|
1882 | 1882 | ] |
|
1883 | 1883 | + formatteropts, |
|
1884 | 1884 | ) |
|
1885 | 1885 | def perfindex(ui, repo, **opts): |
|
1886 | 1886 | """benchmark index creation time followed by a lookup |
|
1887 | 1887 | |
|
1888 | 1888 | The default is to look `tip` up. Depending on the index implementation, |
|
1889 | 1889 | the revision looked up can matters. For example, an implementation |
|
1890 | 1890 | scanning the index will have a faster lookup time for `--rev tip` than for |
|
1891 | 1891 | `--rev 0`. The number of looked up revisions and their order can also |
|
1892 | 1892 | matters. |
|
1893 | 1893 | |
|
1894 | 1894 | Example of useful set to test: |
|
1895 | 1895 | |
|
1896 | 1896 | * tip |
|
1897 | 1897 | * 0 |
|
1898 | 1898 | * -10: |
|
1899 | 1899 | * :10 |
|
1900 | 1900 | * -10: + :10 |
|
1901 | 1901 | * :10: + -10: |
|
1902 | 1902 | * -10000: |
|
1903 | 1903 | * -10000: + 0 |
|
1904 | 1904 | |
|
1905 | 1905 | It is not currently possible to check for lookup of a missing node. For |
|
1906 | 1906 | deeper lookup benchmarking, checkout the `perfnodemap` command.""" |
|
1907 | 1907 | import mercurial.revlog |
|
1908 | 1908 | |
|
1909 | 1909 | opts = _byteskwargs(opts) |
|
1910 | 1910 | timer, fm = gettimer(ui, opts) |
|
1911 | 1911 | mercurial.revlog._prereadsize = 2 ** 24 # disable lazy parser in old hg |
|
1912 | 1912 | if opts[b'no_lookup']: |
|
1913 | 1913 | if opts['rev']: |
|
1914 | 1914 | raise error.Abort('--no-lookup and --rev are mutually exclusive') |
|
1915 | 1915 | nodes = [] |
|
1916 | 1916 | elif not opts[b'rev']: |
|
1917 | 1917 | nodes = [repo[b"tip"].node()] |
|
1918 | 1918 | else: |
|
1919 | 1919 | revs = scmutil.revrange(repo, opts[b'rev']) |
|
1920 | 1920 | cl = repo.changelog |
|
1921 | 1921 | nodes = [cl.node(r) for r in revs] |
|
1922 | 1922 | |
|
1923 | 1923 | unfi = repo.unfiltered() |
|
1924 | 1924 | # find the filecache func directly |
|
1925 | 1925 | # This avoid polluting the benchmark with the filecache logic |
|
1926 | 1926 | makecl = unfi.__class__.changelog.func |
|
1927 | 1927 | |
|
1928 | 1928 | def setup(): |
|
1929 | 1929 | # probably not necessary, but for good measure |
|
1930 | 1930 | clearchangelog(unfi) |
|
1931 | 1931 | |
|
1932 | 1932 | def d(): |
|
1933 | 1933 | cl = makecl(unfi) |
|
1934 | 1934 | for n in nodes: |
|
1935 | 1935 | cl.rev(n) |
|
1936 | 1936 | |
|
1937 | 1937 | timer(d, setup=setup) |
|
1938 | 1938 | fm.end() |
|
1939 | 1939 | |
|
1940 | 1940 | |
|
1941 | 1941 | @command( |
|
1942 | 1942 | b'perf::nodemap|perfnodemap', |
|
1943 | 1943 | [ |
|
1944 | 1944 | (b'', b'rev', [], b'revision to be looked up (default tip)'), |
|
1945 | 1945 | (b'', b'clear-caches', True, b'clear revlog cache between calls'), |
|
1946 | 1946 | ] |
|
1947 | 1947 | + formatteropts, |
|
1948 | 1948 | ) |
|
1949 | 1949 | def perfnodemap(ui, repo, **opts): |
|
1950 | 1950 | """benchmark the time necessary to look up revision from a cold nodemap |
|
1951 | 1951 | |
|
1952 | 1952 | Depending on the implementation, the amount and order of revision we look |
|
1953 | 1953 | up can varies. Example of useful set to test: |
|
1954 | 1954 | * tip |
|
1955 | 1955 | * 0 |
|
1956 | 1956 | * -10: |
|
1957 | 1957 | * :10 |
|
1958 | 1958 | * -10: + :10 |
|
1959 | 1959 | * :10: + -10: |
|
1960 | 1960 | * -10000: |
|
1961 | 1961 | * -10000: + 0 |
|
1962 | 1962 | |
|
1963 | 1963 | The command currently focus on valid binary lookup. Benchmarking for |
|
1964 | 1964 | hexlookup, prefix lookup and missing lookup would also be valuable. |
|
1965 | 1965 | """ |
|
1966 | 1966 | import mercurial.revlog |
|
1967 | 1967 | |
|
1968 | 1968 | opts = _byteskwargs(opts) |
|
1969 | 1969 | timer, fm = gettimer(ui, opts) |
|
1970 | 1970 | mercurial.revlog._prereadsize = 2 ** 24 # disable lazy parser in old hg |
|
1971 | 1971 | |
|
1972 | 1972 | unfi = repo.unfiltered() |
|
1973 | 1973 | clearcaches = opts[b'clear_caches'] |
|
1974 | 1974 | # find the filecache func directly |
|
1975 | 1975 | # This avoid polluting the benchmark with the filecache logic |
|
1976 | 1976 | makecl = unfi.__class__.changelog.func |
|
1977 | 1977 | if not opts[b'rev']: |
|
1978 | 1978 | raise error.Abort(b'use --rev to specify revisions to look up') |
|
1979 | 1979 | revs = scmutil.revrange(repo, opts[b'rev']) |
|
1980 | 1980 | cl = repo.changelog |
|
1981 | 1981 | nodes = [cl.node(r) for r in revs] |
|
1982 | 1982 | |
|
1983 | 1983 | # use a list to pass reference to a nodemap from one closure to the next |
|
1984 | 1984 | nodeget = [None] |
|
1985 | 1985 | |
|
1986 | 1986 | def setnodeget(): |
|
1987 | 1987 | # probably not necessary, but for good measure |
|
1988 | 1988 | clearchangelog(unfi) |
|
1989 | 1989 | cl = makecl(unfi) |
|
1990 | 1990 | if util.safehasattr(cl.index, 'get_rev'): |
|
1991 | 1991 | nodeget[0] = cl.index.get_rev |
|
1992 | 1992 | else: |
|
1993 | 1993 | nodeget[0] = cl.nodemap.get |
|
1994 | 1994 | |
|
1995 | 1995 | def d(): |
|
1996 | 1996 | get = nodeget[0] |
|
1997 | 1997 | for n in nodes: |
|
1998 | 1998 | get(n) |
|
1999 | 1999 | |
|
2000 | 2000 | setup = None |
|
2001 | 2001 | if clearcaches: |
|
2002 | 2002 | |
|
2003 | 2003 | def setup(): |
|
2004 | 2004 | setnodeget() |
|
2005 | 2005 | |
|
2006 | 2006 | else: |
|
2007 | 2007 | setnodeget() |
|
2008 | 2008 | d() # prewarm the data structure |
|
2009 | 2009 | timer(d, setup=setup) |
|
2010 | 2010 | fm.end() |
|
2011 | 2011 | |
|
2012 | 2012 | |
|
2013 | 2013 | @command(b'perf::startup|perfstartup', formatteropts) |
|
2014 | 2014 | def perfstartup(ui, repo, **opts): |
|
2015 | 2015 | opts = _byteskwargs(opts) |
|
2016 | 2016 | timer, fm = gettimer(ui, opts) |
|
2017 | 2017 | |
|
2018 | 2018 | def d(): |
|
2019 | 2019 | if os.name != 'nt': |
|
2020 | 2020 | os.system( |
|
2021 | 2021 | b"HGRCPATH= %s version -q > /dev/null" % fsencode(sys.argv[0]) |
|
2022 | 2022 | ) |
|
2023 | 2023 | else: |
|
2024 | 2024 | os.environ['HGRCPATH'] = r' ' |
|
2025 | 2025 | os.system("%s version -q > NUL" % sys.argv[0]) |
|
2026 | 2026 | |
|
2027 | 2027 | timer(d) |
|
2028 | 2028 | fm.end() |
|
2029 | 2029 | |
|
2030 | 2030 | |
|
2031 | 2031 | def _find_stream_generator(version): |
|
2032 | 2032 | """find the proper generator function for this stream version""" |
|
2033 | 2033 | import mercurial.streamclone |
|
2034 | 2034 | |
|
2035 | 2035 | available = {} |
|
2036 | 2036 | |
|
2037 | 2037 | # try to fetch a v1 generator |
|
2038 | 2038 | generatev1 = getattr(mercurial.streamclone, "generatev1", None) |
|
2039 | 2039 | if generatev1 is not None: |
|
2040 | 2040 | |
|
2041 | 2041 | def generate(repo): |
|
2042 | 2042 | entries, bytes, data = generatev2(repo, None, None, True) |
|
2043 | 2043 | return data |
|
2044 | 2044 | |
|
2045 | 2045 | available[b'v1'] = generatev1 |
|
2046 | 2046 | # try to fetch a v2 generator |
|
2047 | 2047 | generatev2 = getattr(mercurial.streamclone, "generatev2", None) |
|
2048 | 2048 | if generatev2 is not None: |
|
2049 | 2049 | |
|
2050 | 2050 | def generate(repo): |
|
2051 | 2051 | entries, bytes, data = generatev2(repo, None, None, True) |
|
2052 | 2052 | return data |
|
2053 | 2053 | |
|
2054 | 2054 | available[b'v2'] = generate |
|
2055 | 2055 | # try to fetch a v3 generator |
|
2056 | 2056 | generatev3 = getattr(mercurial.streamclone, "generatev3", None) |
|
2057 | 2057 | if generatev3 is not None: |
|
2058 | 2058 | |
|
2059 | 2059 | def generate(repo): |
|
2060 | 2060 | entries, bytes, data = generatev3(repo, None, None, True) |
|
2061 | 2061 | return data |
|
2062 | 2062 | |
|
2063 | 2063 | available[b'v3-exp'] = generate |
|
2064 | 2064 | |
|
2065 | 2065 | # resolve the request |
|
2066 | 2066 | if version == b"latest": |
|
2067 | 2067 | # latest is the highest non experimental version |
|
2068 | 2068 | latest_key = max(v for v in available if b'-exp' not in v) |
|
2069 | 2069 | return available[latest_key] |
|
2070 | 2070 | elif version in available: |
|
2071 | 2071 | return available[version] |
|
2072 | 2072 | else: |
|
2073 | 2073 | msg = b"unkown or unavailable version: %s" |
|
2074 | 2074 | msg %= version |
|
2075 | 2075 | hint = b"available versions: %s" |
|
2076 | 2076 | hint %= b', '.join(sorted(available)) |
|
2077 | 2077 | raise error.Abort(msg, hint=hint) |
|
2078 | 2078 | |
|
2079 | 2079 | |
|
2080 | 2080 | @command( |
|
2081 | 2081 | b'perf::stream-locked-section', |
|
2082 | 2082 | [ |
|
2083 | 2083 | ( |
|
2084 | 2084 | b'', |
|
2085 | 2085 | b'stream-version', |
|
2086 | 2086 | b'latest', |
|
2087 | 2087 | b'stream version to use ("v1", "v2", "v3" or "latest", (the default))', |
|
2088 | 2088 | ), |
|
2089 | 2089 | ] |
|
2090 | 2090 | + formatteropts, |
|
2091 | 2091 | ) |
|
2092 | 2092 | def perf_stream_clone_scan(ui, repo, stream_version, **opts): |
|
2093 | 2093 | """benchmark the initial, repo-locked, section of a stream-clone""" |
|
2094 | 2094 | |
|
2095 | 2095 | opts = _byteskwargs(opts) |
|
2096 | 2096 | timer, fm = gettimer(ui, opts) |
|
2097 | 2097 | |
|
2098 | 2098 | # deletion of the generator may trigger some cleanup that we do not want to |
|
2099 | 2099 | # measure |
|
2100 | 2100 | result_holder = [None] |
|
2101 | 2101 | |
|
2102 | 2102 | def setupone(): |
|
2103 | 2103 | result_holder[0] = None |
|
2104 | 2104 | |
|
2105 | 2105 | generate = _find_stream_generator(stream_version) |
|
2106 | 2106 | |
|
2107 | 2107 | def runone(): |
|
2108 | 2108 | # the lock is held for the duration the initialisation |
|
2109 | 2109 | result_holder[0] = generate(repo) |
|
2110 | 2110 | |
|
2111 | 2111 | timer(runone, setup=setupone, title=b"load") |
|
2112 | 2112 | fm.end() |
|
2113 | 2113 | |
|
2114 | 2114 | |
|
2115 | 2115 | @command( |
|
2116 | 2116 | b'perf::stream-generate', |
|
2117 | 2117 | [ |
|
2118 | 2118 | ( |
|
2119 | 2119 | b'', |
|
2120 | 2120 | b'stream-version', |
|
2121 | 2121 | b'latest', |
|
2122 | 2122 | b'stream version to us ("v1", "v2" or "latest", (the default))', |
|
2123 | 2123 | ), |
|
2124 | 2124 | ] |
|
2125 | 2125 | + formatteropts, |
|
2126 | 2126 | ) |
|
2127 | 2127 | def perf_stream_clone_generate(ui, repo, stream_version, **opts): |
|
2128 | 2128 | """benchmark the full generation of a stream clone""" |
|
2129 | 2129 | |
|
2130 | 2130 | opts = _byteskwargs(opts) |
|
2131 | 2131 | timer, fm = gettimer(ui, opts) |
|
2132 | 2132 | |
|
2133 | 2133 | # deletion of the generator may trigger some cleanup that we do not want to |
|
2134 | 2134 | # measure |
|
2135 | 2135 | |
|
2136 | 2136 | generate = _find_stream_generator(stream_version) |
|
2137 | 2137 | |
|
2138 | 2138 | def runone(): |
|
2139 | 2139 | # the lock is held for the duration the initialisation |
|
2140 | 2140 | for chunk in generate(repo): |
|
2141 | 2141 | pass |
|
2142 | 2142 | |
|
2143 | 2143 | timer(runone, title=b"generate") |
|
2144 | 2144 | fm.end() |
|
2145 | 2145 | |
|
2146 | 2146 | |
|
2147 | 2147 | @command( |
|
2148 | 2148 | b'perf::stream-consume', |
|
2149 | 2149 | formatteropts, |
|
2150 | 2150 | ) |
|
2151 | 2151 | def perf_stream_clone_consume(ui, repo, filename, **opts): |
|
2152 | 2152 | """benchmark the full application of a stream clone |
|
2153 | 2153 | |
|
2154 | 2154 | This include the creation of the repository |
|
2155 | 2155 | """ |
|
2156 | 2156 | # try except to appease check code |
|
2157 | 2157 | msg = b"mercurial too old, missing necessary module: %s" |
|
2158 | 2158 | try: |
|
2159 | 2159 | from mercurial import bundle2 |
|
2160 | 2160 | except ImportError as exc: |
|
2161 | 2161 | msg %= _bytestr(exc) |
|
2162 | 2162 | raise error.Abort(msg) |
|
2163 | 2163 | try: |
|
2164 | 2164 | from mercurial import exchange |
|
2165 | 2165 | except ImportError as exc: |
|
2166 | 2166 | msg %= _bytestr(exc) |
|
2167 | 2167 | raise error.Abort(msg) |
|
2168 | 2168 | try: |
|
2169 | 2169 | from mercurial import hg |
|
2170 | 2170 | except ImportError as exc: |
|
2171 | 2171 | msg %= _bytestr(exc) |
|
2172 | 2172 | raise error.Abort(msg) |
|
2173 | 2173 | try: |
|
2174 | 2174 | from mercurial import localrepo |
|
2175 | 2175 | except ImportError as exc: |
|
2176 | 2176 | msg %= _bytestr(exc) |
|
2177 | 2177 | raise error.Abort(msg) |
|
2178 | 2178 | |
|
2179 | 2179 | opts = _byteskwargs(opts) |
|
2180 | 2180 | timer, fm = gettimer(ui, opts) |
|
2181 | 2181 | |
|
2182 | 2182 | # deletion of the generator may trigger some cleanup that we do not want to |
|
2183 | 2183 | # measure |
|
2184 | 2184 | if not (os.path.isfile(filename) and os.access(filename, os.R_OK)): |
|
2185 | 2185 | raise error.Abort("not a readable file: %s" % filename) |
|
2186 | 2186 | |
|
2187 | 2187 | run_variables = [None, None] |
|
2188 | 2188 | |
|
2189 | 2189 | @contextlib.contextmanager |
|
2190 | 2190 | def context(): |
|
2191 | 2191 | with open(filename, mode='rb') as bundle: |
|
2192 | 2192 | with tempfile.TemporaryDirectory() as tmp_dir: |
|
2193 | 2193 | tmp_dir = fsencode(tmp_dir) |
|
2194 | 2194 | run_variables[0] = bundle |
|
2195 | 2195 | run_variables[1] = tmp_dir |
|
2196 | 2196 | yield |
|
2197 | 2197 | run_variables[0] = None |
|
2198 | 2198 | run_variables[1] = None |
|
2199 | 2199 | |
|
2200 | 2200 | def runone(): |
|
2201 | 2201 | bundle = run_variables[0] |
|
2202 | 2202 | tmp_dir = run_variables[1] |
|
2203 | 2203 | # only pass ui when no srcrepo |
|
2204 | 2204 | localrepo.createrepository( |
|
2205 | 2205 | repo.ui, tmp_dir, requirements=repo.requirements |
|
2206 | 2206 | ) |
|
2207 | 2207 | target = hg.repository(repo.ui, tmp_dir) |
|
2208 | 2208 | gen = exchange.readbundle(target.ui, bundle, bundle.name) |
|
2209 | 2209 | # stream v1 |
|
2210 | 2210 | if util.safehasattr(gen, 'apply'): |
|
2211 | 2211 | gen.apply(target) |
|
2212 | 2212 | else: |
|
2213 | 2213 | with target.transaction(b"perf::stream-consume") as tr: |
|
2214 | 2214 | bundle2.applybundle( |
|
2215 | 2215 | target, |
|
2216 | 2216 | gen, |
|
2217 | 2217 | tr, |
|
2218 | 2218 | source=b'unbundle', |
|
2219 | 2219 | url=filename, |
|
2220 | 2220 | ) |
|
2221 | 2221 | |
|
2222 | 2222 | timer(runone, context=context, title=b"consume") |
|
2223 | 2223 | fm.end() |
|
2224 | 2224 | |
|
2225 | 2225 | |
|
2226 | 2226 | @command(b'perf::parents|perfparents', formatteropts) |
|
2227 | 2227 | def perfparents(ui, repo, **opts): |
|
2228 | 2228 | """benchmark the time necessary to fetch one changeset's parents. |
|
2229 | 2229 | |
|
2230 | 2230 | The fetch is done using the `node identifier`, traversing all object layers |
|
2231 | 2231 | from the repository object. The first N revisions will be used for this |
|
2232 | 2232 | benchmark. N is controlled by the ``perf.parentscount`` config option |
|
2233 | 2233 | (default: 1000). |
|
2234 | 2234 | """ |
|
2235 | 2235 | opts = _byteskwargs(opts) |
|
2236 | 2236 | timer, fm = gettimer(ui, opts) |
|
2237 | 2237 | # control the number of commits perfparents iterates over |
|
2238 | 2238 | # experimental config: perf.parentscount |
|
2239 | 2239 | count = getint(ui, b"perf", b"parentscount", 1000) |
|
2240 | 2240 | if len(repo.changelog) < count: |
|
2241 | 2241 | raise error.Abort(b"repo needs %d commits for this test" % count) |
|
2242 | 2242 | repo = repo.unfiltered() |
|
2243 | 2243 | nl = [repo.changelog.node(i) for i in _xrange(count)] |
|
2244 | 2244 | |
|
2245 | 2245 | def d(): |
|
2246 | 2246 | for n in nl: |
|
2247 | 2247 | repo.changelog.parents(n) |
|
2248 | 2248 | |
|
2249 | 2249 | timer(d) |
|
2250 | 2250 | fm.end() |
|
2251 | 2251 | |
|
2252 | 2252 | |
|
2253 | 2253 | @command(b'perf::ctxfiles|perfctxfiles', formatteropts) |
|
2254 | 2254 | def perfctxfiles(ui, repo, x, **opts): |
|
2255 | 2255 | opts = _byteskwargs(opts) |
|
2256 | 2256 | x = int(x) |
|
2257 | 2257 | timer, fm = gettimer(ui, opts) |
|
2258 | 2258 | |
|
2259 | 2259 | def d(): |
|
2260 | 2260 | len(repo[x].files()) |
|
2261 | 2261 | |
|
2262 | 2262 | timer(d) |
|
2263 | 2263 | fm.end() |
|
2264 | 2264 | |
|
2265 | 2265 | |
|
2266 | 2266 | @command(b'perf::rawfiles|perfrawfiles', formatteropts) |
|
2267 | 2267 | def perfrawfiles(ui, repo, x, **opts): |
|
2268 | 2268 | opts = _byteskwargs(opts) |
|
2269 | 2269 | x = int(x) |
|
2270 | 2270 | timer, fm = gettimer(ui, opts) |
|
2271 | 2271 | cl = repo.changelog |
|
2272 | 2272 | |
|
2273 | 2273 | def d(): |
|
2274 | 2274 | len(cl.read(x)[3]) |
|
2275 | 2275 | |
|
2276 | 2276 | timer(d) |
|
2277 | 2277 | fm.end() |
|
2278 | 2278 | |
|
2279 | 2279 | |
|
2280 | 2280 | @command(b'perf::lookup|perflookup', formatteropts) |
|
2281 | 2281 | def perflookup(ui, repo, rev, **opts): |
|
2282 | 2282 | opts = _byteskwargs(opts) |
|
2283 | 2283 | timer, fm = gettimer(ui, opts) |
|
2284 | 2284 | timer(lambda: len(repo.lookup(rev))) |
|
2285 | 2285 | fm.end() |
|
2286 | 2286 | |
|
2287 | 2287 | |
|
2288 | 2288 | @command( |
|
2289 | 2289 | b'perf::linelogedits|perflinelogedits', |
|
2290 | 2290 | [ |
|
2291 | 2291 | (b'n', b'edits', 10000, b'number of edits'), |
|
2292 | 2292 | (b'', b'max-hunk-lines', 10, b'max lines in a hunk'), |
|
2293 | 2293 | ], |
|
2294 | 2294 | norepo=True, |
|
2295 | 2295 | ) |
|
2296 | 2296 | def perflinelogedits(ui, **opts): |
|
2297 | 2297 | from mercurial import linelog |
|
2298 | 2298 | |
|
2299 | 2299 | opts = _byteskwargs(opts) |
|
2300 | 2300 | |
|
2301 | 2301 | edits = opts[b'edits'] |
|
2302 | 2302 | maxhunklines = opts[b'max_hunk_lines'] |
|
2303 | 2303 | |
|
2304 | 2304 | maxb1 = 100000 |
|
2305 | 2305 | random.seed(0) |
|
2306 | 2306 | randint = random.randint |
|
2307 | 2307 | currentlines = 0 |
|
2308 | 2308 | arglist = [] |
|
2309 | 2309 | for rev in _xrange(edits): |
|
2310 | 2310 | a1 = randint(0, currentlines) |
|
2311 | 2311 | a2 = randint(a1, min(currentlines, a1 + maxhunklines)) |
|
2312 | 2312 | b1 = randint(0, maxb1) |
|
2313 | 2313 | b2 = randint(b1, b1 + maxhunklines) |
|
2314 | 2314 | currentlines += (b2 - b1) - (a2 - a1) |
|
2315 | 2315 | arglist.append((rev, a1, a2, b1, b2)) |
|
2316 | 2316 | |
|
2317 | 2317 | def d(): |
|
2318 | 2318 | ll = linelog.linelog() |
|
2319 | 2319 | for args in arglist: |
|
2320 | 2320 | ll.replacelines(*args) |
|
2321 | 2321 | |
|
2322 | 2322 | timer, fm = gettimer(ui, opts) |
|
2323 | 2323 | timer(d) |
|
2324 | 2324 | fm.end() |
|
2325 | 2325 | |
|
2326 | 2326 | |
|
2327 | 2327 | @command(b'perf::revrange|perfrevrange', formatteropts) |
|
2328 | 2328 | def perfrevrange(ui, repo, *specs, **opts): |
|
2329 | 2329 | opts = _byteskwargs(opts) |
|
2330 | 2330 | timer, fm = gettimer(ui, opts) |
|
2331 | 2331 | revrange = scmutil.revrange |
|
2332 | 2332 | timer(lambda: len(revrange(repo, specs))) |
|
2333 | 2333 | fm.end() |
|
2334 | 2334 | |
|
2335 | 2335 | |
|
2336 | 2336 | @command(b'perf::nodelookup|perfnodelookup', formatteropts) |
|
2337 | 2337 | def perfnodelookup(ui, repo, rev, **opts): |
|
2338 | 2338 | opts = _byteskwargs(opts) |
|
2339 | 2339 | timer, fm = gettimer(ui, opts) |
|
2340 | 2340 | import mercurial.revlog |
|
2341 | 2341 | |
|
2342 | 2342 | mercurial.revlog._prereadsize = 2 ** 24 # disable lazy parser in old hg |
|
2343 | 2343 | n = scmutil.revsingle(repo, rev).node() |
|
2344 | 2344 | |
|
2345 | 2345 | try: |
|
2346 | 2346 | cl = revlog(getsvfs(repo), radix=b"00changelog") |
|
2347 | 2347 | except TypeError: |
|
2348 | 2348 | cl = revlog(getsvfs(repo), indexfile=b"00changelog.i") |
|
2349 | 2349 | |
|
2350 | 2350 | def d(): |
|
2351 | 2351 | cl.rev(n) |
|
2352 | 2352 | clearcaches(cl) |
|
2353 | 2353 | |
|
2354 | 2354 | timer(d) |
|
2355 | 2355 | fm.end() |
|
2356 | 2356 | |
|
2357 | 2357 | |
|
2358 | 2358 | @command( |
|
2359 | 2359 | b'perf::log|perflog', |
|
2360 | 2360 | [(b'', b'rename', False, b'ask log to follow renames')] + formatteropts, |
|
2361 | 2361 | ) |
|
2362 | 2362 | def perflog(ui, repo, rev=None, **opts): |
|
2363 | 2363 | opts = _byteskwargs(opts) |
|
2364 | 2364 | if rev is None: |
|
2365 | 2365 | rev = [] |
|
2366 | 2366 | timer, fm = gettimer(ui, opts) |
|
2367 | 2367 | ui.pushbuffer() |
|
2368 | 2368 | timer( |
|
2369 | 2369 | lambda: commands.log( |
|
2370 | 2370 | ui, repo, rev=rev, date=b'', user=b'', copies=opts.get(b'rename') |
|
2371 | 2371 | ) |
|
2372 | 2372 | ) |
|
2373 | 2373 | ui.popbuffer() |
|
2374 | 2374 | fm.end() |
|
2375 | 2375 | |
|
2376 | 2376 | |
|
2377 | 2377 | @command(b'perf::moonwalk|perfmoonwalk', formatteropts) |
|
2378 | 2378 | def perfmoonwalk(ui, repo, **opts): |
|
2379 | 2379 | """benchmark walking the changelog backwards |
|
2380 | 2380 | |
|
2381 | 2381 | This also loads the changelog data for each revision in the changelog. |
|
2382 | 2382 | """ |
|
2383 | 2383 | opts = _byteskwargs(opts) |
|
2384 | 2384 | timer, fm = gettimer(ui, opts) |
|
2385 | 2385 | |
|
2386 | 2386 | def moonwalk(): |
|
2387 | 2387 | for i in repo.changelog.revs(start=(len(repo) - 1), stop=-1): |
|
2388 | 2388 | ctx = repo[i] |
|
2389 | 2389 | ctx.branch() # read changelog data (in addition to the index) |
|
2390 | 2390 | |
|
2391 | 2391 | timer(moonwalk) |
|
2392 | 2392 | fm.end() |
|
2393 | 2393 | |
|
2394 | 2394 | |
|
2395 | 2395 | @command( |
|
2396 | 2396 | b'perf::templating|perftemplating', |
|
2397 | 2397 | [ |
|
2398 | 2398 | (b'r', b'rev', [], b'revisions to run the template on'), |
|
2399 | 2399 | ] |
|
2400 | 2400 | + formatteropts, |
|
2401 | 2401 | ) |
|
2402 | 2402 | def perftemplating(ui, repo, testedtemplate=None, **opts): |
|
2403 | 2403 | """test the rendering time of a given template""" |
|
2404 | 2404 | if makelogtemplater is None: |
|
2405 | 2405 | raise error.Abort( |
|
2406 | 2406 | b"perftemplating not available with this Mercurial", |
|
2407 | 2407 | hint=b"use 4.3 or later", |
|
2408 | 2408 | ) |
|
2409 | 2409 | |
|
2410 | 2410 | opts = _byteskwargs(opts) |
|
2411 | 2411 | |
|
2412 | 2412 | nullui = ui.copy() |
|
2413 | 2413 | nullui.fout = open(os.devnull, 'wb') |
|
2414 | 2414 | nullui.disablepager() |
|
2415 | 2415 | revs = opts.get(b'rev') |
|
2416 | 2416 | if not revs: |
|
2417 | 2417 | revs = [b'all()'] |
|
2418 | 2418 | revs = list(scmutil.revrange(repo, revs)) |
|
2419 | 2419 | |
|
2420 | 2420 | defaulttemplate = ( |
|
2421 | 2421 | b'{date|shortdate} [{rev}:{node|short}]' |
|
2422 | 2422 | b' {author|person}: {desc|firstline}\n' |
|
2423 | 2423 | ) |
|
2424 | 2424 | if testedtemplate is None: |
|
2425 | 2425 | testedtemplate = defaulttemplate |
|
2426 | 2426 | displayer = makelogtemplater(nullui, repo, testedtemplate) |
|
2427 | 2427 | |
|
2428 | 2428 | def format(): |
|
2429 | 2429 | for r in revs: |
|
2430 | 2430 | ctx = repo[r] |
|
2431 | 2431 | displayer.show(ctx) |
|
2432 | 2432 | displayer.flush(ctx) |
|
2433 | 2433 | |
|
2434 | 2434 | timer, fm = gettimer(ui, opts) |
|
2435 | 2435 | timer(format) |
|
2436 | 2436 | fm.end() |
|
2437 | 2437 | |
|
2438 | 2438 | |
|
2439 | 2439 | def _displaystats(ui, opts, entries, data): |
|
2440 | 2440 | # use a second formatter because the data are quite different, not sure |
|
2441 | 2441 | # how it flies with the templater. |
|
2442 | 2442 | fm = ui.formatter(b'perf-stats', opts) |
|
2443 | 2443 | for key, title in entries: |
|
2444 | 2444 | values = data[key] |
|
2445 | 2445 | nbvalues = len(data) |
|
2446 | 2446 | values.sort() |
|
2447 | 2447 | stats = { |
|
2448 | 2448 | 'key': key, |
|
2449 | 2449 | 'title': title, |
|
2450 | 2450 | 'nbitems': len(values), |
|
2451 | 2451 | 'min': values[0][0], |
|
2452 | 2452 | '10%': values[(nbvalues * 10) // 100][0], |
|
2453 | 2453 | '25%': values[(nbvalues * 25) // 100][0], |
|
2454 | 2454 | '50%': values[(nbvalues * 50) // 100][0], |
|
2455 | 2455 | '75%': values[(nbvalues * 75) // 100][0], |
|
2456 | 2456 | '80%': values[(nbvalues * 80) // 100][0], |
|
2457 | 2457 | '85%': values[(nbvalues * 85) // 100][0], |
|
2458 | 2458 | '90%': values[(nbvalues * 90) // 100][0], |
|
2459 | 2459 | '95%': values[(nbvalues * 95) // 100][0], |
|
2460 | 2460 | '99%': values[(nbvalues * 99) // 100][0], |
|
2461 | 2461 | 'max': values[-1][0], |
|
2462 | 2462 | } |
|
2463 | 2463 | fm.startitem() |
|
2464 | 2464 | fm.data(**stats) |
|
2465 | 2465 | # make node pretty for the human output |
|
2466 | 2466 | fm.plain('### %s (%d items)\n' % (title, len(values))) |
|
2467 | 2467 | lines = [ |
|
2468 | 2468 | 'min', |
|
2469 | 2469 | '10%', |
|
2470 | 2470 | '25%', |
|
2471 | 2471 | '50%', |
|
2472 | 2472 | '75%', |
|
2473 | 2473 | '80%', |
|
2474 | 2474 | '85%', |
|
2475 | 2475 | '90%', |
|
2476 | 2476 | '95%', |
|
2477 | 2477 | '99%', |
|
2478 | 2478 | 'max', |
|
2479 | 2479 | ] |
|
2480 | 2480 | for l in lines: |
|
2481 | 2481 | fm.plain('%s: %s\n' % (l, stats[l])) |
|
2482 | 2482 | fm.end() |
|
2483 | 2483 | |
|
2484 | 2484 | |
|
2485 | 2485 | @command( |
|
2486 | 2486 | b'perf::helper-mergecopies|perfhelper-mergecopies', |
|
2487 | 2487 | formatteropts |
|
2488 | 2488 | + [ |
|
2489 | 2489 | (b'r', b'revs', [], b'restrict search to these revisions'), |
|
2490 | 2490 | (b'', b'timing', False, b'provides extra data (costly)'), |
|
2491 | 2491 | (b'', b'stats', False, b'provides statistic about the measured data'), |
|
2492 | 2492 | ], |
|
2493 | 2493 | ) |
|
2494 | 2494 | def perfhelpermergecopies(ui, repo, revs=[], **opts): |
|
2495 | 2495 | """find statistics about potential parameters for `perfmergecopies` |
|
2496 | 2496 | |
|
2497 | 2497 | This command find (base, p1, p2) triplet relevant for copytracing |
|
2498 | 2498 | benchmarking in the context of a merge. It reports values for some of the |
|
2499 | 2499 | parameters that impact merge copy tracing time during merge. |
|
2500 | 2500 | |
|
2501 | 2501 | If `--timing` is set, rename detection is run and the associated timing |
|
2502 | 2502 | will be reported. The extra details come at the cost of slower command |
|
2503 | 2503 | execution. |
|
2504 | 2504 | |
|
2505 | 2505 | Since rename detection is only run once, other factors might easily |
|
2506 | 2506 | affect the precision of the timing. However it should give a good |
|
2507 | 2507 | approximation of which revision triplets are very costly. |
|
2508 | 2508 | """ |
|
2509 | 2509 | opts = _byteskwargs(opts) |
|
2510 | 2510 | fm = ui.formatter(b'perf', opts) |
|
2511 | 2511 | dotiming = opts[b'timing'] |
|
2512 | 2512 | dostats = opts[b'stats'] |
|
2513 | 2513 | |
|
2514 | 2514 | output_template = [ |
|
2515 | 2515 | ("base", "%(base)12s"), |
|
2516 | 2516 | ("p1", "%(p1.node)12s"), |
|
2517 | 2517 | ("p2", "%(p2.node)12s"), |
|
2518 | 2518 | ("p1.nb-revs", "%(p1.nbrevs)12d"), |
|
2519 | 2519 | ("p1.nb-files", "%(p1.nbmissingfiles)12d"), |
|
2520 | 2520 | ("p1.renames", "%(p1.renamedfiles)12d"), |
|
2521 | 2521 | ("p1.time", "%(p1.time)12.3f"), |
|
2522 | 2522 | ("p2.nb-revs", "%(p2.nbrevs)12d"), |
|
2523 | 2523 | ("p2.nb-files", "%(p2.nbmissingfiles)12d"), |
|
2524 | 2524 | ("p2.renames", "%(p2.renamedfiles)12d"), |
|
2525 | 2525 | ("p2.time", "%(p2.time)12.3f"), |
|
2526 | 2526 | ("renames", "%(nbrenamedfiles)12d"), |
|
2527 | 2527 | ("total.time", "%(time)12.3f"), |
|
2528 | 2528 | ] |
|
2529 | 2529 | if not dotiming: |
|
2530 | 2530 | output_template = [ |
|
2531 | 2531 | i |
|
2532 | 2532 | for i in output_template |
|
2533 | 2533 | if not ('time' in i[0] or 'renames' in i[0]) |
|
2534 | 2534 | ] |
|
2535 | 2535 | header_names = [h for (h, v) in output_template] |
|
2536 | 2536 | output = ' '.join([v for (h, v) in output_template]) + '\n' |
|
2537 | 2537 | header = ' '.join(['%12s'] * len(header_names)) + '\n' |
|
2538 | 2538 | fm.plain(header % tuple(header_names)) |
|
2539 | 2539 | |
|
2540 | 2540 | if not revs: |
|
2541 | 2541 | revs = ['all()'] |
|
2542 | 2542 | revs = scmutil.revrange(repo, revs) |
|
2543 | 2543 | |
|
2544 | 2544 | if dostats: |
|
2545 | 2545 | alldata = { |
|
2546 | 2546 | 'nbrevs': [], |
|
2547 | 2547 | 'nbmissingfiles': [], |
|
2548 | 2548 | } |
|
2549 | 2549 | if dotiming: |
|
2550 | 2550 | alldata['parentnbrenames'] = [] |
|
2551 | 2551 | alldata['totalnbrenames'] = [] |
|
2552 | 2552 | alldata['parenttime'] = [] |
|
2553 | 2553 | alldata['totaltime'] = [] |
|
2554 | 2554 | |
|
2555 | 2555 | roi = repo.revs('merge() and %ld', revs) |
|
2556 | 2556 | for r in roi: |
|
2557 | 2557 | ctx = repo[r] |
|
2558 | 2558 | p1 = ctx.p1() |
|
2559 | 2559 | p2 = ctx.p2() |
|
2560 | 2560 | bases = repo.changelog._commonancestorsheads(p1.rev(), p2.rev()) |
|
2561 | 2561 | for b in bases: |
|
2562 | 2562 | b = repo[b] |
|
2563 | 2563 | p1missing = copies._computeforwardmissing(b, p1) |
|
2564 | 2564 | p2missing = copies._computeforwardmissing(b, p2) |
|
2565 | 2565 | data = { |
|
2566 | 2566 | b'base': b.hex(), |
|
2567 | 2567 | b'p1.node': p1.hex(), |
|
2568 | 2568 | b'p1.nbrevs': len(repo.revs('only(%d, %d)', p1.rev(), b.rev())), |
|
2569 | 2569 | b'p1.nbmissingfiles': len(p1missing), |
|
2570 | 2570 | b'p2.node': p2.hex(), |
|
2571 | 2571 | b'p2.nbrevs': len(repo.revs('only(%d, %d)', p2.rev(), b.rev())), |
|
2572 | 2572 | b'p2.nbmissingfiles': len(p2missing), |
|
2573 | 2573 | } |
|
2574 | 2574 | if dostats: |
|
2575 | 2575 | if p1missing: |
|
2576 | 2576 | alldata['nbrevs'].append( |
|
2577 | 2577 | (data['p1.nbrevs'], b.hex(), p1.hex()) |
|
2578 | 2578 | ) |
|
2579 | 2579 | alldata['nbmissingfiles'].append( |
|
2580 | 2580 | (data['p1.nbmissingfiles'], b.hex(), p1.hex()) |
|
2581 | 2581 | ) |
|
2582 | 2582 | if p2missing: |
|
2583 | 2583 | alldata['nbrevs'].append( |
|
2584 | 2584 | (data['p2.nbrevs'], b.hex(), p2.hex()) |
|
2585 | 2585 | ) |
|
2586 | 2586 | alldata['nbmissingfiles'].append( |
|
2587 | 2587 | (data['p2.nbmissingfiles'], b.hex(), p2.hex()) |
|
2588 | 2588 | ) |
|
2589 | 2589 | if dotiming: |
|
2590 | 2590 | begin = util.timer() |
|
2591 | 2591 | mergedata = copies.mergecopies(repo, p1, p2, b) |
|
2592 | 2592 | end = util.timer() |
|
2593 | 2593 | # not very stable timing since we did only one run |
|
2594 | 2594 | data['time'] = end - begin |
|
2595 | 2595 | # mergedata contains five dicts: "copy", "movewithdir", |
|
2596 | 2596 | # "diverge", "renamedelete" and "dirmove". |
|
2597 | 2597 | # The first 4 are about renamed file so lets count that. |
|
2598 | 2598 | renames = len(mergedata[0]) |
|
2599 | 2599 | renames += len(mergedata[1]) |
|
2600 | 2600 | renames += len(mergedata[2]) |
|
2601 | 2601 | renames += len(mergedata[3]) |
|
2602 | 2602 | data['nbrenamedfiles'] = renames |
|
2603 | 2603 | begin = util.timer() |
|
2604 | 2604 | p1renames = copies.pathcopies(b, p1) |
|
2605 | 2605 | end = util.timer() |
|
2606 | 2606 | data['p1.time'] = end - begin |
|
2607 | 2607 | begin = util.timer() |
|
2608 | 2608 | p2renames = copies.pathcopies(b, p2) |
|
2609 | 2609 | end = util.timer() |
|
2610 | 2610 | data['p2.time'] = end - begin |
|
2611 | 2611 | data['p1.renamedfiles'] = len(p1renames) |
|
2612 | 2612 | data['p2.renamedfiles'] = len(p2renames) |
|
2613 | 2613 | |
|
2614 | 2614 | if dostats: |
|
2615 | 2615 | if p1missing: |
|
2616 | 2616 | alldata['parentnbrenames'].append( |
|
2617 | 2617 | (data['p1.renamedfiles'], b.hex(), p1.hex()) |
|
2618 | 2618 | ) |
|
2619 | 2619 | alldata['parenttime'].append( |
|
2620 | 2620 | (data['p1.time'], b.hex(), p1.hex()) |
|
2621 | 2621 | ) |
|
2622 | 2622 | if p2missing: |
|
2623 | 2623 | alldata['parentnbrenames'].append( |
|
2624 | 2624 | (data['p2.renamedfiles'], b.hex(), p2.hex()) |
|
2625 | 2625 | ) |
|
2626 | 2626 | alldata['parenttime'].append( |
|
2627 | 2627 | (data['p2.time'], b.hex(), p2.hex()) |
|
2628 | 2628 | ) |
|
2629 | 2629 | if p1missing or p2missing: |
|
2630 | 2630 | alldata['totalnbrenames'].append( |
|
2631 | 2631 | ( |
|
2632 | 2632 | data['nbrenamedfiles'], |
|
2633 | 2633 | b.hex(), |
|
2634 | 2634 | p1.hex(), |
|
2635 | 2635 | p2.hex(), |
|
2636 | 2636 | ) |
|
2637 | 2637 | ) |
|
2638 | 2638 | alldata['totaltime'].append( |
|
2639 | 2639 | (data['time'], b.hex(), p1.hex(), p2.hex()) |
|
2640 | 2640 | ) |
|
2641 | 2641 | fm.startitem() |
|
2642 | 2642 | fm.data(**data) |
|
2643 | 2643 | # make node pretty for the human output |
|
2644 | 2644 | out = data.copy() |
|
2645 | 2645 | out['base'] = fm.hexfunc(b.node()) |
|
2646 | 2646 | out['p1.node'] = fm.hexfunc(p1.node()) |
|
2647 | 2647 | out['p2.node'] = fm.hexfunc(p2.node()) |
|
2648 | 2648 | fm.plain(output % out) |
|
2649 | 2649 | |
|
2650 | 2650 | fm.end() |
|
2651 | 2651 | if dostats: |
|
2652 | 2652 | # use a second formatter because the data are quite different, not sure |
|
2653 | 2653 | # how it flies with the templater. |
|
2654 | 2654 | entries = [ |
|
2655 | 2655 | ('nbrevs', 'number of revision covered'), |
|
2656 | 2656 | ('nbmissingfiles', 'number of missing files at head'), |
|
2657 | 2657 | ] |
|
2658 | 2658 | if dotiming: |
|
2659 | 2659 | entries.append( |
|
2660 | 2660 | ('parentnbrenames', 'rename from one parent to base') |
|
2661 | 2661 | ) |
|
2662 | 2662 | entries.append(('totalnbrenames', 'total number of renames')) |
|
2663 | 2663 | entries.append(('parenttime', 'time for one parent')) |
|
2664 | 2664 | entries.append(('totaltime', 'time for both parents')) |
|
2665 | 2665 | _displaystats(ui, opts, entries, alldata) |
|
2666 | 2666 | |
|
2667 | 2667 | |
|
2668 | 2668 | @command( |
|
2669 | 2669 | b'perf::helper-pathcopies|perfhelper-pathcopies', |
|
2670 | 2670 | formatteropts |
|
2671 | 2671 | + [ |
|
2672 | 2672 | (b'r', b'revs', [], b'restrict search to these revisions'), |
|
2673 | 2673 | (b'', b'timing', False, b'provides extra data (costly)'), |
|
2674 | 2674 | (b'', b'stats', False, b'provides statistic about the measured data'), |
|
2675 | 2675 | ], |
|
2676 | 2676 | ) |
|
2677 | 2677 | def perfhelperpathcopies(ui, repo, revs=[], **opts): |
|
2678 | 2678 | """find statistic about potential parameters for the `perftracecopies` |
|
2679 | 2679 | |
|
2680 | 2680 | This command find source-destination pair relevant for copytracing testing. |
|
2681 | 2681 | It report value for some of the parameters that impact copy tracing time. |
|
2682 | 2682 | |
|
2683 | 2683 | If `--timing` is set, rename detection is run and the associated timing |
|
2684 | 2684 | will be reported. The extra details comes at the cost of a slower command |
|
2685 | 2685 | execution. |
|
2686 | 2686 | |
|
2687 | 2687 | Since the rename detection is only run once, other factors might easily |
|
2688 | 2688 | affect the precision of the timing. However it should give a good |
|
2689 | 2689 | approximation of which revision pairs are very costly. |
|
2690 | 2690 | """ |
|
2691 | 2691 | opts = _byteskwargs(opts) |
|
2692 | 2692 | fm = ui.formatter(b'perf', opts) |
|
2693 | 2693 | dotiming = opts[b'timing'] |
|
2694 | 2694 | dostats = opts[b'stats'] |
|
2695 | 2695 | |
|
2696 | 2696 | if dotiming: |
|
2697 | 2697 | header = '%12s %12s %12s %12s %12s %12s\n' |
|
2698 | 2698 | output = ( |
|
2699 | 2699 | "%(source)12s %(destination)12s " |
|
2700 | 2700 | "%(nbrevs)12d %(nbmissingfiles)12d " |
|
2701 | 2701 | "%(nbrenamedfiles)12d %(time)18.5f\n" |
|
2702 | 2702 | ) |
|
2703 | 2703 | header_names = ( |
|
2704 | 2704 | "source", |
|
2705 | 2705 | "destination", |
|
2706 | 2706 | "nb-revs", |
|
2707 | 2707 | "nb-files", |
|
2708 | 2708 | "nb-renames", |
|
2709 | 2709 | "time", |
|
2710 | 2710 | ) |
|
2711 | 2711 | fm.plain(header % header_names) |
|
2712 | 2712 | else: |
|
2713 | 2713 | header = '%12s %12s %12s %12s\n' |
|
2714 | 2714 | output = ( |
|
2715 | 2715 | "%(source)12s %(destination)12s " |
|
2716 | 2716 | "%(nbrevs)12d %(nbmissingfiles)12d\n" |
|
2717 | 2717 | ) |
|
2718 | 2718 | fm.plain(header % ("source", "destination", "nb-revs", "nb-files")) |
|
2719 | 2719 | |
|
2720 | 2720 | if not revs: |
|
2721 | 2721 | revs = ['all()'] |
|
2722 | 2722 | revs = scmutil.revrange(repo, revs) |
|
2723 | 2723 | |
|
2724 | 2724 | if dostats: |
|
2725 | 2725 | alldata = { |
|
2726 | 2726 | 'nbrevs': [], |
|
2727 | 2727 | 'nbmissingfiles': [], |
|
2728 | 2728 | } |
|
2729 | 2729 | if dotiming: |
|
2730 | 2730 | alldata['nbrenames'] = [] |
|
2731 | 2731 | alldata['time'] = [] |
|
2732 | 2732 | |
|
2733 | 2733 | roi = repo.revs('merge() and %ld', revs) |
|
2734 | 2734 | for r in roi: |
|
2735 | 2735 | ctx = repo[r] |
|
2736 | 2736 | p1 = ctx.p1().rev() |
|
2737 | 2737 | p2 = ctx.p2().rev() |
|
2738 | 2738 | bases = repo.changelog._commonancestorsheads(p1, p2) |
|
2739 | 2739 | for p in (p1, p2): |
|
2740 | 2740 | for b in bases: |
|
2741 | 2741 | base = repo[b] |
|
2742 | 2742 | parent = repo[p] |
|
2743 | 2743 | missing = copies._computeforwardmissing(base, parent) |
|
2744 | 2744 | if not missing: |
|
2745 | 2745 | continue |
|
2746 | 2746 | data = { |
|
2747 | 2747 | b'source': base.hex(), |
|
2748 | 2748 | b'destination': parent.hex(), |
|
2749 | 2749 | b'nbrevs': len(repo.revs('only(%d, %d)', p, b)), |
|
2750 | 2750 | b'nbmissingfiles': len(missing), |
|
2751 | 2751 | } |
|
2752 | 2752 | if dostats: |
|
2753 | 2753 | alldata['nbrevs'].append( |
|
2754 | 2754 | ( |
|
2755 | 2755 | data['nbrevs'], |
|
2756 | 2756 | base.hex(), |
|
2757 | 2757 | parent.hex(), |
|
2758 | 2758 | ) |
|
2759 | 2759 | ) |
|
2760 | 2760 | alldata['nbmissingfiles'].append( |
|
2761 | 2761 | ( |
|
2762 | 2762 | data['nbmissingfiles'], |
|
2763 | 2763 | base.hex(), |
|
2764 | 2764 | parent.hex(), |
|
2765 | 2765 | ) |
|
2766 | 2766 | ) |
|
2767 | 2767 | if dotiming: |
|
2768 | 2768 | begin = util.timer() |
|
2769 | 2769 | renames = copies.pathcopies(base, parent) |
|
2770 | 2770 | end = util.timer() |
|
2771 | 2771 | # not very stable timing since we did only one run |
|
2772 | 2772 | data['time'] = end - begin |
|
2773 | 2773 | data['nbrenamedfiles'] = len(renames) |
|
2774 | 2774 | if dostats: |
|
2775 | 2775 | alldata['time'].append( |
|
2776 | 2776 | ( |
|
2777 | 2777 | data['time'], |
|
2778 | 2778 | base.hex(), |
|
2779 | 2779 | parent.hex(), |
|
2780 | 2780 | ) |
|
2781 | 2781 | ) |
|
2782 | 2782 | alldata['nbrenames'].append( |
|
2783 | 2783 | ( |
|
2784 | 2784 | data['nbrenamedfiles'], |
|
2785 | 2785 | base.hex(), |
|
2786 | 2786 | parent.hex(), |
|
2787 | 2787 | ) |
|
2788 | 2788 | ) |
|
2789 | 2789 | fm.startitem() |
|
2790 | 2790 | fm.data(**data) |
|
2791 | 2791 | out = data.copy() |
|
2792 | 2792 | out['source'] = fm.hexfunc(base.node()) |
|
2793 | 2793 | out['destination'] = fm.hexfunc(parent.node()) |
|
2794 | 2794 | fm.plain(output % out) |
|
2795 | 2795 | |
|
2796 | 2796 | fm.end() |
|
2797 | 2797 | if dostats: |
|
2798 | 2798 | entries = [ |
|
2799 | 2799 | ('nbrevs', 'number of revision covered'), |
|
2800 | 2800 | ('nbmissingfiles', 'number of missing files at head'), |
|
2801 | 2801 | ] |
|
2802 | 2802 | if dotiming: |
|
2803 | 2803 | entries.append(('nbrenames', 'renamed files')) |
|
2804 | 2804 | entries.append(('time', 'time')) |
|
2805 | 2805 | _displaystats(ui, opts, entries, alldata) |
|
2806 | 2806 | |
|
2807 | 2807 | |
|
2808 | 2808 | @command(b'perf::cca|perfcca', formatteropts) |
|
2809 | 2809 | def perfcca(ui, repo, **opts): |
|
2810 | 2810 | opts = _byteskwargs(opts) |
|
2811 | 2811 | timer, fm = gettimer(ui, opts) |
|
2812 | 2812 | timer(lambda: scmutil.casecollisionauditor(ui, False, repo.dirstate)) |
|
2813 | 2813 | fm.end() |
|
2814 | 2814 | |
|
2815 | 2815 | |
|
2816 | 2816 | @command(b'perf::fncacheload|perffncacheload', formatteropts) |
|
2817 | 2817 | def perffncacheload(ui, repo, **opts): |
|
2818 | 2818 | opts = _byteskwargs(opts) |
|
2819 | 2819 | timer, fm = gettimer(ui, opts) |
|
2820 | 2820 | s = repo.store |
|
2821 | 2821 | |
|
2822 | 2822 | def d(): |
|
2823 | 2823 | s.fncache._load() |
|
2824 | 2824 | |
|
2825 | 2825 | timer(d) |
|
2826 | 2826 | fm.end() |
|
2827 | 2827 | |
|
2828 | 2828 | |
|
2829 | 2829 | @command(b'perf::fncachewrite|perffncachewrite', formatteropts) |
|
2830 | 2830 | def perffncachewrite(ui, repo, **opts): |
|
2831 | 2831 | opts = _byteskwargs(opts) |
|
2832 | 2832 | timer, fm = gettimer(ui, opts) |
|
2833 | 2833 | s = repo.store |
|
2834 | 2834 | lock = repo.lock() |
|
2835 | 2835 | s.fncache._load() |
|
2836 | 2836 | tr = repo.transaction(b'perffncachewrite') |
|
2837 | 2837 | tr.addbackup(b'fncache') |
|
2838 | 2838 | |
|
2839 | 2839 | def d(): |
|
2840 | 2840 | s.fncache._dirty = True |
|
2841 | 2841 | s.fncache.write(tr) |
|
2842 | 2842 | |
|
2843 | 2843 | timer(d) |
|
2844 | 2844 | tr.close() |
|
2845 | 2845 | lock.release() |
|
2846 | 2846 | fm.end() |
|
2847 | 2847 | |
|
2848 | 2848 | |
|
2849 | 2849 | @command(b'perf::fncacheencode|perffncacheencode', formatteropts) |
|
2850 | 2850 | def perffncacheencode(ui, repo, **opts): |
|
2851 | 2851 | opts = _byteskwargs(opts) |
|
2852 | 2852 | timer, fm = gettimer(ui, opts) |
|
2853 | 2853 | s = repo.store |
|
2854 | 2854 | s.fncache._load() |
|
2855 | 2855 | |
|
2856 | 2856 | def d(): |
|
2857 | 2857 | for p in s.fncache.entries: |
|
2858 | 2858 | s.encode(p) |
|
2859 | 2859 | |
|
2860 | 2860 | timer(d) |
|
2861 | 2861 | fm.end() |
|
2862 | 2862 | |
|
2863 | 2863 | |
|
2864 | 2864 | def _bdiffworker(q, blocks, xdiff, ready, done): |
|
2865 | 2865 | while not done.is_set(): |
|
2866 | 2866 | pair = q.get() |
|
2867 | 2867 | while pair is not None: |
|
2868 | 2868 | if xdiff: |
|
2869 | 2869 | mdiff.bdiff.xdiffblocks(*pair) |
|
2870 | 2870 | elif blocks: |
|
2871 | 2871 | mdiff.bdiff.blocks(*pair) |
|
2872 | 2872 | else: |
|
2873 | 2873 | mdiff.textdiff(*pair) |
|
2874 | 2874 | q.task_done() |
|
2875 | 2875 | pair = q.get() |
|
2876 | 2876 | q.task_done() # for the None one |
|
2877 | 2877 | with ready: |
|
2878 | 2878 | ready.wait() |
|
2879 | 2879 | |
|
2880 | 2880 | |
|
2881 | 2881 | def _manifestrevision(repo, mnode): |
|
2882 | 2882 | ml = repo.manifestlog |
|
2883 | 2883 | |
|
2884 | 2884 | if util.safehasattr(ml, b'getstorage'): |
|
2885 | 2885 | store = ml.getstorage(b'') |
|
2886 | 2886 | else: |
|
2887 | 2887 | store = ml._revlog |
|
2888 | 2888 | |
|
2889 | 2889 | return store.revision(mnode) |
|
2890 | 2890 | |
|
2891 | 2891 | |
|
2892 | 2892 | @command( |
|
2893 | 2893 | b'perf::bdiff|perfbdiff', |
|
2894 | 2894 | revlogopts |
|
2895 | 2895 | + formatteropts |
|
2896 | 2896 | + [ |
|
2897 | 2897 | ( |
|
2898 | 2898 | b'', |
|
2899 | 2899 | b'count', |
|
2900 | 2900 | 1, |
|
2901 | 2901 | b'number of revisions to test (when using --startrev)', |
|
2902 | 2902 | ), |
|
2903 | 2903 | (b'', b'alldata', False, b'test bdiffs for all associated revisions'), |
|
2904 | 2904 | (b'', b'threads', 0, b'number of thread to use (disable with 0)'), |
|
2905 | 2905 | (b'', b'blocks', False, b'test computing diffs into blocks'), |
|
2906 | 2906 | (b'', b'xdiff', False, b'use xdiff algorithm'), |
|
2907 | 2907 | ], |
|
2908 | 2908 | b'-c|-m|FILE REV', |
|
2909 | 2909 | ) |
|
2910 | 2910 | def perfbdiff(ui, repo, file_, rev=None, count=None, threads=0, **opts): |
|
2911 | 2911 | """benchmark a bdiff between revisions |
|
2912 | 2912 | |
|
2913 | 2913 | By default, benchmark a bdiff between its delta parent and itself. |
|
2914 | 2914 | |
|
2915 | 2915 | With ``--count``, benchmark bdiffs between delta parents and self for N |
|
2916 | 2916 | revisions starting at the specified revision. |
|
2917 | 2917 | |
|
2918 | 2918 | With ``--alldata``, assume the requested revision is a changeset and |
|
2919 | 2919 | measure bdiffs for all changes related to that changeset (manifest |
|
2920 | 2920 | and filelogs). |
|
2921 | 2921 | """ |
|
2922 | 2922 | opts = _byteskwargs(opts) |
|
2923 | 2923 | |
|
2924 | 2924 | if opts[b'xdiff'] and not opts[b'blocks']: |
|
2925 | 2925 | raise error.CommandError(b'perfbdiff', b'--xdiff requires --blocks') |
|
2926 | 2926 | |
|
2927 | 2927 | if opts[b'alldata']: |
|
2928 | 2928 | opts[b'changelog'] = True |
|
2929 | 2929 | |
|
2930 | 2930 | if opts.get(b'changelog') or opts.get(b'manifest'): |
|
2931 | 2931 | file_, rev = None, file_ |
|
2932 | 2932 | elif rev is None: |
|
2933 | 2933 | raise error.CommandError(b'perfbdiff', b'invalid arguments') |
|
2934 | 2934 | |
|
2935 | 2935 | blocks = opts[b'blocks'] |
|
2936 | 2936 | xdiff = opts[b'xdiff'] |
|
2937 | 2937 | textpairs = [] |
|
2938 | 2938 | |
|
2939 | 2939 | r = cmdutil.openrevlog(repo, b'perfbdiff', file_, opts) |
|
2940 | 2940 | |
|
2941 | 2941 | startrev = r.rev(r.lookup(rev)) |
|
2942 | 2942 | for rev in range(startrev, min(startrev + count, len(r) - 1)): |
|
2943 | 2943 | if opts[b'alldata']: |
|
2944 | 2944 | # Load revisions associated with changeset. |
|
2945 | 2945 | ctx = repo[rev] |
|
2946 | 2946 | mtext = _manifestrevision(repo, ctx.manifestnode()) |
|
2947 | 2947 | for pctx in ctx.parents(): |
|
2948 | 2948 | pman = _manifestrevision(repo, pctx.manifestnode()) |
|
2949 | 2949 | textpairs.append((pman, mtext)) |
|
2950 | 2950 | |
|
2951 | 2951 | # Load filelog revisions by iterating manifest delta. |
|
2952 | 2952 | man = ctx.manifest() |
|
2953 | 2953 | pman = ctx.p1().manifest() |
|
2954 | 2954 | for filename, change in pman.diff(man).items(): |
|
2955 | 2955 | fctx = repo.file(filename) |
|
2956 | 2956 | f1 = fctx.revision(change[0][0] or -1) |
|
2957 | 2957 | f2 = fctx.revision(change[1][0] or -1) |
|
2958 | 2958 | textpairs.append((f1, f2)) |
|
2959 | 2959 | else: |
|
2960 | 2960 | dp = r.deltaparent(rev) |
|
2961 | 2961 | textpairs.append((r.revision(dp), r.revision(rev))) |
|
2962 | 2962 | |
|
2963 | 2963 | withthreads = threads > 0 |
|
2964 | 2964 | if not withthreads: |
|
2965 | 2965 | |
|
2966 | 2966 | def d(): |
|
2967 | 2967 | for pair in textpairs: |
|
2968 | 2968 | if xdiff: |
|
2969 | 2969 | mdiff.bdiff.xdiffblocks(*pair) |
|
2970 | 2970 | elif blocks: |
|
2971 | 2971 | mdiff.bdiff.blocks(*pair) |
|
2972 | 2972 | else: |
|
2973 | 2973 | mdiff.textdiff(*pair) |
|
2974 | 2974 | |
|
2975 | 2975 | else: |
|
2976 | 2976 | q = queue() |
|
2977 | 2977 | for i in _xrange(threads): |
|
2978 | 2978 | q.put(None) |
|
2979 | 2979 | ready = threading.Condition() |
|
2980 | 2980 | done = threading.Event() |
|
2981 | 2981 | for i in _xrange(threads): |
|
2982 | 2982 | threading.Thread( |
|
2983 | 2983 | target=_bdiffworker, args=(q, blocks, xdiff, ready, done) |
|
2984 | 2984 | ).start() |
|
2985 | 2985 | q.join() |
|
2986 | 2986 | |
|
2987 | 2987 | def d(): |
|
2988 | 2988 | for pair in textpairs: |
|
2989 | 2989 | q.put(pair) |
|
2990 | 2990 | for i in _xrange(threads): |
|
2991 | 2991 | q.put(None) |
|
2992 | 2992 | with ready: |
|
2993 | 2993 | ready.notify_all() |
|
2994 | 2994 | q.join() |
|
2995 | 2995 | |
|
2996 | 2996 | timer, fm = gettimer(ui, opts) |
|
2997 | 2997 | timer(d) |
|
2998 | 2998 | fm.end() |
|
2999 | 2999 | |
|
3000 | 3000 | if withthreads: |
|
3001 | 3001 | done.set() |
|
3002 | 3002 | for i in _xrange(threads): |
|
3003 | 3003 | q.put(None) |
|
3004 | 3004 | with ready: |
|
3005 | 3005 | ready.notify_all() |
|
3006 | 3006 | |
|
3007 | 3007 | |
|
3008 | 3008 | @command( |
|
3009 | 3009 | b'perf::unbundle', |
|
3010 | 3010 | formatteropts, |
|
3011 | 3011 | b'BUNDLE_FILE', |
|
3012 | 3012 | ) |
|
3013 | 3013 | def perf_unbundle(ui, repo, fname, **opts): |
|
3014 | 3014 | """benchmark application of a bundle in a repository. |
|
3015 | 3015 | |
|
3016 | 3016 | This does not include the final transaction processing""" |
|
3017 | 3017 | |
|
3018 | 3018 | from mercurial import exchange |
|
3019 | 3019 | from mercurial import bundle2 |
|
3020 | 3020 | from mercurial import transaction |
|
3021 | 3021 | |
|
3022 | 3022 | opts = _byteskwargs(opts) |
|
3023 | 3023 | |
|
3024 | 3024 | ### some compatibility hotfix |
|
3025 | 3025 | # |
|
3026 | 3026 | # the data attribute is dropped in 63edc384d3b7 a changeset introducing a |
|
3027 | 3027 | # critical regression that break transaction rollback for files that are |
|
3028 | 3028 | # de-inlined. |
|
3029 | 3029 | method = transaction.transaction._addentry |
|
3030 | 3030 | pre_63edc384d3b7 = "data" in getargspec(method).args |
|
3031 | 3031 | # the `detailed_exit_code` attribute is introduced in 33c0c25d0b0f |
|
3032 | 3032 | # a changeset that is a close descendant of 18415fc918a1, the changeset |
|
3033 | 3033 | # that conclude the fix run for the bug introduced in 63edc384d3b7. |
|
3034 | 3034 | args = getargspec(error.Abort.__init__).args |
|
3035 | 3035 | post_18415fc918a1 = "detailed_exit_code" in args |
|
3036 | 3036 | |
|
3037 | 3037 | old_max_inline = None |
|
3038 | 3038 | try: |
|
3039 | 3039 | if not (pre_63edc384d3b7 or post_18415fc918a1): |
|
3040 | 3040 | # disable inlining |
|
3041 | 3041 | old_max_inline = mercurial.revlog._maxinline |
|
3042 | 3042 | # large enough to never happen |
|
3043 | 3043 | mercurial.revlog._maxinline = 2 ** 50 |
|
3044 | 3044 | |
|
3045 | 3045 | with repo.lock(): |
|
3046 | 3046 | bundle = [None, None] |
|
3047 | 3047 | orig_quiet = repo.ui.quiet |
|
3048 | 3048 | try: |
|
3049 | 3049 | repo.ui.quiet = True |
|
3050 | 3050 | with open(fname, mode="rb") as f: |
|
3051 | 3051 | |
|
3052 | 3052 | def noop_report(*args, **kwargs): |
|
3053 | 3053 | pass |
|
3054 | 3054 | |
|
3055 | 3055 | def setup(): |
|
3056 | 3056 | gen, tr = bundle |
|
3057 | 3057 | if tr is not None: |
|
3058 | 3058 | tr.abort() |
|
3059 | 3059 | bundle[:] = [None, None] |
|
3060 | 3060 | f.seek(0) |
|
3061 | 3061 | bundle[0] = exchange.readbundle(ui, f, fname) |
|
3062 | 3062 | bundle[1] = repo.transaction(b'perf::unbundle') |
|
3063 | 3063 | # silence the transaction |
|
3064 | 3064 | bundle[1]._report = noop_report |
|
3065 | 3065 | |
|
3066 | 3066 | def apply(): |
|
3067 | 3067 | gen, tr = bundle |
|
3068 | 3068 | bundle2.applybundle( |
|
3069 | 3069 | repo, |
|
3070 | 3070 | gen, |
|
3071 | 3071 | tr, |
|
3072 | 3072 | source=b'perf::unbundle', |
|
3073 | 3073 | url=fname, |
|
3074 | 3074 | ) |
|
3075 | 3075 | |
|
3076 | 3076 | timer, fm = gettimer(ui, opts) |
|
3077 | 3077 | timer(apply, setup=setup) |
|
3078 | 3078 | fm.end() |
|
3079 | 3079 | finally: |
|
3080 | 3080 | repo.ui.quiet == orig_quiet |
|
3081 | 3081 | gen, tr = bundle |
|
3082 | 3082 | if tr is not None: |
|
3083 | 3083 | tr.abort() |
|
3084 | 3084 | finally: |
|
3085 | 3085 | if old_max_inline is not None: |
|
3086 | 3086 | mercurial.revlog._maxinline = old_max_inline |
|
3087 | 3087 | |
|
3088 | 3088 | |
|
3089 | 3089 | @command( |
|
3090 | 3090 | b'perf::unidiff|perfunidiff', |
|
3091 | 3091 | revlogopts |
|
3092 | 3092 | + formatteropts |
|
3093 | 3093 | + [ |
|
3094 | 3094 | ( |
|
3095 | 3095 | b'', |
|
3096 | 3096 | b'count', |
|
3097 | 3097 | 1, |
|
3098 | 3098 | b'number of revisions to test (when using --startrev)', |
|
3099 | 3099 | ), |
|
3100 | 3100 | (b'', b'alldata', False, b'test unidiffs for all associated revisions'), |
|
3101 | 3101 | ], |
|
3102 | 3102 | b'-c|-m|FILE REV', |
|
3103 | 3103 | ) |
|
3104 | 3104 | def perfunidiff(ui, repo, file_, rev=None, count=None, **opts): |
|
3105 | 3105 | """benchmark a unified diff between revisions |
|
3106 | 3106 | |
|
3107 | 3107 | This doesn't include any copy tracing - it's just a unified diff |
|
3108 | 3108 | of the texts. |
|
3109 | 3109 | |
|
3110 | 3110 | By default, benchmark a diff between its delta parent and itself. |
|
3111 | 3111 | |
|
3112 | 3112 | With ``--count``, benchmark diffs between delta parents and self for N |
|
3113 | 3113 | revisions starting at the specified revision. |
|
3114 | 3114 | |
|
3115 | 3115 | With ``--alldata``, assume the requested revision is a changeset and |
|
3116 | 3116 | measure diffs for all changes related to that changeset (manifest |
|
3117 | 3117 | and filelogs). |
|
3118 | 3118 | """ |
|
3119 | 3119 | opts = _byteskwargs(opts) |
|
3120 | 3120 | if opts[b'alldata']: |
|
3121 | 3121 | opts[b'changelog'] = True |
|
3122 | 3122 | |
|
3123 | 3123 | if opts.get(b'changelog') or opts.get(b'manifest'): |
|
3124 | 3124 | file_, rev = None, file_ |
|
3125 | 3125 | elif rev is None: |
|
3126 | 3126 | raise error.CommandError(b'perfunidiff', b'invalid arguments') |
|
3127 | 3127 | |
|
3128 | 3128 | textpairs = [] |
|
3129 | 3129 | |
|
3130 | 3130 | r = cmdutil.openrevlog(repo, b'perfunidiff', file_, opts) |
|
3131 | 3131 | |
|
3132 | 3132 | startrev = r.rev(r.lookup(rev)) |
|
3133 | 3133 | for rev in range(startrev, min(startrev + count, len(r) - 1)): |
|
3134 | 3134 | if opts[b'alldata']: |
|
3135 | 3135 | # Load revisions associated with changeset. |
|
3136 | 3136 | ctx = repo[rev] |
|
3137 | 3137 | mtext = _manifestrevision(repo, ctx.manifestnode()) |
|
3138 | 3138 | for pctx in ctx.parents(): |
|
3139 | 3139 | pman = _manifestrevision(repo, pctx.manifestnode()) |
|
3140 | 3140 | textpairs.append((pman, mtext)) |
|
3141 | 3141 | |
|
3142 | 3142 | # Load filelog revisions by iterating manifest delta. |
|
3143 | 3143 | man = ctx.manifest() |
|
3144 | 3144 | pman = ctx.p1().manifest() |
|
3145 | 3145 | for filename, change in pman.diff(man).items(): |
|
3146 | 3146 | fctx = repo.file(filename) |
|
3147 | 3147 | f1 = fctx.revision(change[0][0] or -1) |
|
3148 | 3148 | f2 = fctx.revision(change[1][0] or -1) |
|
3149 | 3149 | textpairs.append((f1, f2)) |
|
3150 | 3150 | else: |
|
3151 | 3151 | dp = r.deltaparent(rev) |
|
3152 | 3152 | textpairs.append((r.revision(dp), r.revision(rev))) |
|
3153 | 3153 | |
|
3154 | 3154 | def d(): |
|
3155 | 3155 | for left, right in textpairs: |
|
3156 | 3156 | # The date strings don't matter, so we pass empty strings. |
|
3157 | 3157 | headerlines, hunks = mdiff.unidiff( |
|
3158 | 3158 | left, b'', right, b'', b'left', b'right', binary=False |
|
3159 | 3159 | ) |
|
3160 | 3160 | # consume iterators in roughly the way patch.py does |
|
3161 | 3161 | b'\n'.join(headerlines) |
|
3162 | 3162 | b''.join(sum((list(hlines) for hrange, hlines in hunks), [])) |
|
3163 | 3163 | |
|
3164 | 3164 | timer, fm = gettimer(ui, opts) |
|
3165 | 3165 | timer(d) |
|
3166 | 3166 | fm.end() |
|
3167 | 3167 | |
|
3168 | 3168 | |
|
3169 | 3169 | @command(b'perf::diffwd|perfdiffwd', formatteropts) |
|
3170 | 3170 | def perfdiffwd(ui, repo, **opts): |
|
3171 | 3171 | """Profile diff of working directory changes""" |
|
3172 | 3172 | opts = _byteskwargs(opts) |
|
3173 | 3173 | timer, fm = gettimer(ui, opts) |
|
3174 | 3174 | options = { |
|
3175 | 3175 | 'w': 'ignore_all_space', |
|
3176 | 3176 | 'b': 'ignore_space_change', |
|
3177 | 3177 | 'B': 'ignore_blank_lines', |
|
3178 | 3178 | } |
|
3179 | 3179 | |
|
3180 | 3180 | for diffopt in ('', 'w', 'b', 'B', 'wB'): |
|
3181 | 3181 | opts = {options[c]: b'1' for c in diffopt} |
|
3182 | 3182 | |
|
3183 | 3183 | def d(): |
|
3184 | 3184 | ui.pushbuffer() |
|
3185 | 3185 | commands.diff(ui, repo, **opts) |
|
3186 | 3186 | ui.popbuffer() |
|
3187 | 3187 | |
|
3188 | 3188 | diffopt = diffopt.encode('ascii') |
|
3189 | 3189 | title = b'diffopts: %s' % (diffopt and (b'-' + diffopt) or b'none') |
|
3190 | 3190 | timer(d, title=title) |
|
3191 | 3191 | fm.end() |
|
3192 | 3192 | |
|
3193 | 3193 | |
|
3194 | 3194 | @command( |
|
3195 | 3195 | b'perf::revlogindex|perfrevlogindex', |
|
3196 | 3196 | revlogopts + formatteropts, |
|
3197 | 3197 | b'-c|-m|FILE', |
|
3198 | 3198 | ) |
|
3199 | 3199 | def perfrevlogindex(ui, repo, file_=None, **opts): |
|
3200 | 3200 | """Benchmark operations against a revlog index. |
|
3201 | 3201 | |
|
3202 | 3202 | This tests constructing a revlog instance, reading index data, |
|
3203 | 3203 | parsing index data, and performing various operations related to |
|
3204 | 3204 | index data. |
|
3205 | 3205 | """ |
|
3206 | 3206 | |
|
3207 | 3207 | opts = _byteskwargs(opts) |
|
3208 | 3208 | |
|
3209 | 3209 | rl = cmdutil.openrevlog(repo, b'perfrevlogindex', file_, opts) |
|
3210 | 3210 | |
|
3211 | 3211 | opener = getattr(rl, 'opener') # trick linter |
|
3212 | 3212 | # compat with hg <= 5.8 |
|
3213 | 3213 | radix = getattr(rl, 'radix', None) |
|
3214 | 3214 | indexfile = getattr(rl, '_indexfile', None) |
|
3215 | 3215 | if indexfile is None: |
|
3216 | 3216 | # compatibility with <= hg-5.8 |
|
3217 | 3217 | indexfile = getattr(rl, 'indexfile') |
|
3218 | 3218 | data = opener.read(indexfile) |
|
3219 | 3219 | |
|
3220 | 3220 | header = struct.unpack(b'>I', data[0:4])[0] |
|
3221 | 3221 | version = header & 0xFFFF |
|
3222 | 3222 | if version == 1: |
|
3223 | 3223 | inline = header & (1 << 16) |
|
3224 | 3224 | else: |
|
3225 | 3225 | raise error.Abort(b'unsupported revlog version: %d' % version) |
|
3226 | 3226 | |
|
3227 | 3227 | parse_index_v1 = getattr(mercurial.revlog, 'parse_index_v1', None) |
|
3228 | 3228 | if parse_index_v1 is None: |
|
3229 | 3229 | parse_index_v1 = mercurial.revlog.revlogio().parseindex |
|
3230 | 3230 | |
|
3231 | 3231 | rllen = len(rl) |
|
3232 | 3232 | |
|
3233 | 3233 | node0 = rl.node(0) |
|
3234 | 3234 | node25 = rl.node(rllen // 4) |
|
3235 | 3235 | node50 = rl.node(rllen // 2) |
|
3236 | 3236 | node75 = rl.node(rllen // 4 * 3) |
|
3237 | 3237 | node100 = rl.node(rllen - 1) |
|
3238 | 3238 | |
|
3239 | 3239 | allrevs = range(rllen) |
|
3240 | 3240 | allrevsrev = list(reversed(allrevs)) |
|
3241 | 3241 | allnodes = [rl.node(rev) for rev in range(rllen)] |
|
3242 | 3242 | allnodesrev = list(reversed(allnodes)) |
|
3243 | 3243 | |
|
3244 | 3244 | def constructor(): |
|
3245 | 3245 | if radix is not None: |
|
3246 | 3246 | revlog(opener, radix=radix) |
|
3247 | 3247 | else: |
|
3248 | 3248 | # hg <= 5.8 |
|
3249 | 3249 | revlog(opener, indexfile=indexfile) |
|
3250 | 3250 | |
|
3251 | 3251 | def read(): |
|
3252 | 3252 | with opener(indexfile) as fh: |
|
3253 | 3253 | fh.read() |
|
3254 | 3254 | |
|
3255 | 3255 | def parseindex(): |
|
3256 | 3256 | parse_index_v1(data, inline) |
|
3257 | 3257 | |
|
3258 | 3258 | def getentry(revornode): |
|
3259 | 3259 | index = parse_index_v1(data, inline)[0] |
|
3260 | 3260 | index[revornode] |
|
3261 | 3261 | |
|
3262 | 3262 | def getentries(revs, count=1): |
|
3263 | 3263 | index = parse_index_v1(data, inline)[0] |
|
3264 | 3264 | |
|
3265 | 3265 | for i in range(count): |
|
3266 | 3266 | for rev in revs: |
|
3267 | 3267 | index[rev] |
|
3268 | 3268 | |
|
3269 | 3269 | def resolvenode(node): |
|
3270 | 3270 | index = parse_index_v1(data, inline)[0] |
|
3271 | 3271 | rev = getattr(index, 'rev', None) |
|
3272 | 3272 | if rev is None: |
|
3273 | 3273 | nodemap = getattr(parse_index_v1(data, inline)[0], 'nodemap', None) |
|
3274 | 3274 | # This only works for the C code. |
|
3275 | 3275 | if nodemap is None: |
|
3276 | 3276 | return |
|
3277 | 3277 | rev = nodemap.__getitem__ |
|
3278 | 3278 | |
|
3279 | 3279 | try: |
|
3280 | 3280 | rev(node) |
|
3281 | 3281 | except error.RevlogError: |
|
3282 | 3282 | pass |
|
3283 | 3283 | |
|
3284 | 3284 | def resolvenodes(nodes, count=1): |
|
3285 | 3285 | index = parse_index_v1(data, inline)[0] |
|
3286 | 3286 | rev = getattr(index, 'rev', None) |
|
3287 | 3287 | if rev is None: |
|
3288 | 3288 | nodemap = getattr(parse_index_v1(data, inline)[0], 'nodemap', None) |
|
3289 | 3289 | # This only works for the C code. |
|
3290 | 3290 | if nodemap is None: |
|
3291 | 3291 | return |
|
3292 | 3292 | rev = nodemap.__getitem__ |
|
3293 | 3293 | |
|
3294 | 3294 | for i in range(count): |
|
3295 | 3295 | for node in nodes: |
|
3296 | 3296 | try: |
|
3297 | 3297 | rev(node) |
|
3298 | 3298 | except error.RevlogError: |
|
3299 | 3299 | pass |
|
3300 | 3300 | |
|
3301 | 3301 | benches = [ |
|
3302 | 3302 | (constructor, b'revlog constructor'), |
|
3303 | 3303 | (read, b'read'), |
|
3304 | 3304 | (parseindex, b'create index object'), |
|
3305 | 3305 | (lambda: getentry(0), b'retrieve index entry for rev 0'), |
|
3306 | 3306 | (lambda: resolvenode(b'a' * 20), b'look up missing node'), |
|
3307 | 3307 | (lambda: resolvenode(node0), b'look up node at rev 0'), |
|
3308 | 3308 | (lambda: resolvenode(node25), b'look up node at 1/4 len'), |
|
3309 | 3309 | (lambda: resolvenode(node50), b'look up node at 1/2 len'), |
|
3310 | 3310 | (lambda: resolvenode(node75), b'look up node at 3/4 len'), |
|
3311 | 3311 | (lambda: resolvenode(node100), b'look up node at tip'), |
|
3312 | 3312 | # 2x variation is to measure caching impact. |
|
3313 | 3313 | (lambda: resolvenodes(allnodes), b'look up all nodes (forward)'), |
|
3314 | 3314 | (lambda: resolvenodes(allnodes, 2), b'look up all nodes 2x (forward)'), |
|
3315 | 3315 | (lambda: resolvenodes(allnodesrev), b'look up all nodes (reverse)'), |
|
3316 | 3316 | ( |
|
3317 | 3317 | lambda: resolvenodes(allnodesrev, 2), |
|
3318 | 3318 | b'look up all nodes 2x (reverse)', |
|
3319 | 3319 | ), |
|
3320 | 3320 | (lambda: getentries(allrevs), b'retrieve all index entries (forward)'), |
|
3321 | 3321 | ( |
|
3322 | 3322 | lambda: getentries(allrevs, 2), |
|
3323 | 3323 | b'retrieve all index entries 2x (forward)', |
|
3324 | 3324 | ), |
|
3325 | 3325 | ( |
|
3326 | 3326 | lambda: getentries(allrevsrev), |
|
3327 | 3327 | b'retrieve all index entries (reverse)', |
|
3328 | 3328 | ), |
|
3329 | 3329 | ( |
|
3330 | 3330 | lambda: getentries(allrevsrev, 2), |
|
3331 | 3331 | b'retrieve all index entries 2x (reverse)', |
|
3332 | 3332 | ), |
|
3333 | 3333 | ] |
|
3334 | 3334 | |
|
3335 | 3335 | for fn, title in benches: |
|
3336 | 3336 | timer, fm = gettimer(ui, opts) |
|
3337 | 3337 | timer(fn, title=title) |
|
3338 | 3338 | fm.end() |
|
3339 | 3339 | |
|
3340 | 3340 | |
|
3341 | 3341 | @command( |
|
3342 | 3342 | b'perf::revlogrevisions|perfrevlogrevisions', |
|
3343 | 3343 | revlogopts |
|
3344 | 3344 | + formatteropts |
|
3345 | 3345 | + [ |
|
3346 | 3346 | (b'd', b'dist', 100, b'distance between the revisions'), |
|
3347 | 3347 | (b's', b'startrev', 0, b'revision to start reading at'), |
|
3348 | 3348 | (b'', b'reverse', False, b'read in reverse'), |
|
3349 | 3349 | ], |
|
3350 | 3350 | b'-c|-m|FILE', |
|
3351 | 3351 | ) |
|
3352 | 3352 | def perfrevlogrevisions( |
|
3353 | 3353 | ui, repo, file_=None, startrev=0, reverse=False, **opts |
|
3354 | 3354 | ): |
|
3355 | 3355 | """Benchmark reading a series of revisions from a revlog. |
|
3356 | 3356 | |
|
3357 | 3357 | By default, we read every ``-d/--dist`` revision from 0 to tip of |
|
3358 | 3358 | the specified revlog. |
|
3359 | 3359 | |
|
3360 | 3360 | The start revision can be defined via ``-s/--startrev``. |
|
3361 | 3361 | """ |
|
3362 | 3362 | opts = _byteskwargs(opts) |
|
3363 | 3363 | |
|
3364 | 3364 | rl = cmdutil.openrevlog(repo, b'perfrevlogrevisions', file_, opts) |
|
3365 | 3365 | rllen = getlen(ui)(rl) |
|
3366 | 3366 | |
|
3367 | 3367 | if startrev < 0: |
|
3368 | 3368 | startrev = rllen + startrev |
|
3369 | 3369 | |
|
3370 | 3370 | def d(): |
|
3371 | 3371 | rl.clearcaches() |
|
3372 | 3372 | |
|
3373 | 3373 | beginrev = startrev |
|
3374 | 3374 | endrev = rllen |
|
3375 | 3375 | dist = opts[b'dist'] |
|
3376 | 3376 | |
|
3377 | 3377 | if reverse: |
|
3378 | 3378 | beginrev, endrev = endrev - 1, beginrev - 1 |
|
3379 | 3379 | dist = -1 * dist |
|
3380 | 3380 | |
|
3381 | 3381 | for x in _xrange(beginrev, endrev, dist): |
|
3382 | 3382 | # Old revisions don't support passing int. |
|
3383 | 3383 | n = rl.node(x) |
|
3384 | 3384 | rl.revision(n) |
|
3385 | 3385 | |
|
3386 | 3386 | timer, fm = gettimer(ui, opts) |
|
3387 | 3387 | timer(d) |
|
3388 | 3388 | fm.end() |
|
3389 | 3389 | |
|
3390 | 3390 | |
|
3391 | 3391 | @command( |
|
3392 | 3392 | b'perf::revlogwrite|perfrevlogwrite', |
|
3393 | 3393 | revlogopts |
|
3394 | 3394 | + formatteropts |
|
3395 | 3395 | + [ |
|
3396 | 3396 | (b's', b'startrev', 1000, b'revision to start writing at'), |
|
3397 | 3397 | (b'', b'stoprev', -1, b'last revision to write'), |
|
3398 | 3398 | (b'', b'count', 3, b'number of passes to perform'), |
|
3399 | 3399 | (b'', b'details', False, b'print timing for every revisions tested'), |
|
3400 | 3400 | (b'', b'source', b'full', b'the kind of data feed in the revlog'), |
|
3401 | 3401 | (b'', b'lazydeltabase', True, b'try the provided delta first'), |
|
3402 | 3402 | (b'', b'clear-caches', True, b'clear revlog cache between calls'), |
|
3403 | 3403 | ], |
|
3404 | 3404 | b'-c|-m|FILE', |
|
3405 | 3405 | ) |
|
3406 | 3406 | def perfrevlogwrite(ui, repo, file_=None, startrev=1000, stoprev=-1, **opts): |
|
3407 | 3407 | """Benchmark writing a series of revisions to a revlog. |
|
3408 | 3408 | |
|
3409 | 3409 | Possible source values are: |
|
3410 | 3410 | * `full`: add from a full text (default). |
|
3411 | 3411 | * `parent-1`: add from a delta to the first parent |
|
3412 | 3412 | * `parent-2`: add from a delta to the second parent if it exists |
|
3413 | 3413 | (use a delta from the first parent otherwise) |
|
3414 | 3414 | * `parent-smallest`: add from the smallest delta (either p1 or p2) |
|
3415 | 3415 | * `storage`: add from the existing precomputed deltas |
|
3416 | 3416 | |
|
3417 | 3417 | Note: This performance command measures performance in a custom way. As a |
|
3418 | 3418 | result some of the global configuration of the 'perf' command does not |
|
3419 | 3419 | apply to it: |
|
3420 | 3420 | |
|
3421 | 3421 | * ``pre-run``: disabled |
|
3422 | 3422 | |
|
3423 | 3423 | * ``profile-benchmark``: disabled |
|
3424 | 3424 | |
|
3425 | 3425 | * ``run-limits``: disabled use --count instead |
|
3426 | 3426 | """ |
|
3427 | 3427 | opts = _byteskwargs(opts) |
|
3428 | 3428 | |
|
3429 | 3429 | rl = cmdutil.openrevlog(repo, b'perfrevlogwrite', file_, opts) |
|
3430 | 3430 | rllen = getlen(ui)(rl) |
|
3431 | 3431 | if startrev < 0: |
|
3432 | 3432 | startrev = rllen + startrev |
|
3433 | 3433 | if stoprev < 0: |
|
3434 | 3434 | stoprev = rllen + stoprev |
|
3435 | 3435 | |
|
3436 | 3436 | lazydeltabase = opts['lazydeltabase'] |
|
3437 | 3437 | source = opts['source'] |
|
3438 | 3438 | clearcaches = opts['clear_caches'] |
|
3439 | 3439 | validsource = ( |
|
3440 | 3440 | b'full', |
|
3441 | 3441 | b'parent-1', |
|
3442 | 3442 | b'parent-2', |
|
3443 | 3443 | b'parent-smallest', |
|
3444 | 3444 | b'storage', |
|
3445 | 3445 | ) |
|
3446 | 3446 | if source not in validsource: |
|
3447 | 3447 | raise error.Abort('invalid source type: %s' % source) |
|
3448 | 3448 | |
|
3449 | 3449 | ### actually gather results |
|
3450 | 3450 | count = opts['count'] |
|
3451 | 3451 | if count <= 0: |
|
3452 | 3452 | raise error.Abort('invalide run count: %d' % count) |
|
3453 | 3453 | allresults = [] |
|
3454 | 3454 | for c in range(count): |
|
3455 | 3455 | timing = _timeonewrite( |
|
3456 | 3456 | ui, |
|
3457 | 3457 | rl, |
|
3458 | 3458 | source, |
|
3459 | 3459 | startrev, |
|
3460 | 3460 | stoprev, |
|
3461 | 3461 | c + 1, |
|
3462 | 3462 | lazydeltabase=lazydeltabase, |
|
3463 | 3463 | clearcaches=clearcaches, |
|
3464 | 3464 | ) |
|
3465 | 3465 | allresults.append(timing) |
|
3466 | 3466 | |
|
3467 | 3467 | ### consolidate the results in a single list |
|
3468 | 3468 | results = [] |
|
3469 | 3469 | for idx, (rev, t) in enumerate(allresults[0]): |
|
3470 | 3470 | ts = [t] |
|
3471 | 3471 | for other in allresults[1:]: |
|
3472 | 3472 | orev, ot = other[idx] |
|
3473 | 3473 | assert orev == rev |
|
3474 | 3474 | ts.append(ot) |
|
3475 | 3475 | results.append((rev, ts)) |
|
3476 | 3476 | resultcount = len(results) |
|
3477 | 3477 | |
|
3478 | 3478 | ### Compute and display relevant statistics |
|
3479 | 3479 | |
|
3480 | 3480 | # get a formatter |
|
3481 | 3481 | fm = ui.formatter(b'perf', opts) |
|
3482 | 3482 | displayall = ui.configbool(b"perf", b"all-timing", True) |
|
3483 | 3483 | |
|
3484 | 3484 | # print individual details if requested |
|
3485 | 3485 | if opts['details']: |
|
3486 | 3486 | for idx, item in enumerate(results, 1): |
|
3487 | 3487 | rev, data = item |
|
3488 | 3488 | title = 'revisions #%d of %d, rev %d' % (idx, resultcount, rev) |
|
3489 | 3489 | formatone(fm, data, title=title, displayall=displayall) |
|
3490 | 3490 | |
|
3491 | 3491 | # sorts results by median time |
|
3492 | 3492 | results.sort(key=lambda x: sorted(x[1])[len(x[1]) // 2]) |
|
3493 | 3493 | # list of (name, index) to display) |
|
3494 | 3494 | relevants = [ |
|
3495 | 3495 | ("min", 0), |
|
3496 | 3496 | ("10%", resultcount * 10 // 100), |
|
3497 | 3497 | ("25%", resultcount * 25 // 100), |
|
3498 | 3498 | ("50%", resultcount * 70 // 100), |
|
3499 | 3499 | ("75%", resultcount * 75 // 100), |
|
3500 | 3500 | ("90%", resultcount * 90 // 100), |
|
3501 | 3501 | ("95%", resultcount * 95 // 100), |
|
3502 | 3502 | ("99%", resultcount * 99 // 100), |
|
3503 | 3503 | ("99.9%", resultcount * 999 // 1000), |
|
3504 | 3504 | ("99.99%", resultcount * 9999 // 10000), |
|
3505 | 3505 | ("99.999%", resultcount * 99999 // 100000), |
|
3506 | 3506 | ("max", -1), |
|
3507 | 3507 | ] |
|
3508 | 3508 | if not ui.quiet: |
|
3509 | 3509 | for name, idx in relevants: |
|
3510 | 3510 | data = results[idx] |
|
3511 | 3511 | title = '%s of %d, rev %d' % (name, resultcount, data[0]) |
|
3512 | 3512 | formatone(fm, data[1], title=title, displayall=displayall) |
|
3513 | 3513 | |
|
3514 | 3514 | # XXX summing that many float will not be very precise, we ignore this fact |
|
3515 | 3515 | # for now |
|
3516 | 3516 | totaltime = [] |
|
3517 | 3517 | for item in allresults: |
|
3518 | 3518 | totaltime.append( |
|
3519 | 3519 | ( |
|
3520 | 3520 | sum(x[1][0] for x in item), |
|
3521 | 3521 | sum(x[1][1] for x in item), |
|
3522 | 3522 | sum(x[1][2] for x in item), |
|
3523 | 3523 | ) |
|
3524 | 3524 | ) |
|
3525 | 3525 | formatone( |
|
3526 | 3526 | fm, |
|
3527 | 3527 | totaltime, |
|
3528 | 3528 | title="total time (%d revs)" % resultcount, |
|
3529 | 3529 | displayall=displayall, |
|
3530 | 3530 | ) |
|
3531 | 3531 | fm.end() |
|
3532 | 3532 | |
|
3533 | 3533 | |
|
3534 | 3534 | class _faketr: |
|
3535 | 3535 | def add(s, x, y, z=None): |
|
3536 | 3536 | return None |
|
3537 | 3537 | |
|
3538 | 3538 | |
|
3539 | 3539 | def _timeonewrite( |
|
3540 | 3540 | ui, |
|
3541 | 3541 | orig, |
|
3542 | 3542 | source, |
|
3543 | 3543 | startrev, |
|
3544 | 3544 | stoprev, |
|
3545 | 3545 | runidx=None, |
|
3546 | 3546 | lazydeltabase=True, |
|
3547 | 3547 | clearcaches=True, |
|
3548 | 3548 | ): |
|
3549 | 3549 | timings = [] |
|
3550 | 3550 | tr = _faketr() |
|
3551 | 3551 | with _temprevlog(ui, orig, startrev) as dest: |
|
3552 | 3552 | if hasattr(dest, "delta_config"): |
|
3553 | 3553 | dest.delta_config.lazy_delta_base = lazydeltabase |
|
3554 | 3554 | else: |
|
3555 | 3555 | dest._lazydeltabase = lazydeltabase |
|
3556 | 3556 | revs = list(orig.revs(startrev, stoprev)) |
|
3557 | 3557 | total = len(revs) |
|
3558 | 3558 | topic = 'adding' |
|
3559 | 3559 | if runidx is not None: |
|
3560 | 3560 | topic += ' (run #%d)' % runidx |
|
3561 | 3561 | # Support both old and new progress API |
|
3562 | 3562 | if util.safehasattr(ui, 'makeprogress'): |
|
3563 | 3563 | progress = ui.makeprogress(topic, unit='revs', total=total) |
|
3564 | 3564 | |
|
3565 | 3565 | def updateprogress(pos): |
|
3566 | 3566 | progress.update(pos) |
|
3567 | 3567 | |
|
3568 | 3568 | def completeprogress(): |
|
3569 | 3569 | progress.complete() |
|
3570 | 3570 | |
|
3571 | 3571 | else: |
|
3572 | 3572 | |
|
3573 | 3573 | def updateprogress(pos): |
|
3574 | 3574 | ui.progress(topic, pos, unit='revs', total=total) |
|
3575 | 3575 | |
|
3576 | 3576 | def completeprogress(): |
|
3577 | 3577 | ui.progress(topic, None, unit='revs', total=total) |
|
3578 | 3578 | |
|
3579 | 3579 | for idx, rev in enumerate(revs): |
|
3580 | 3580 | updateprogress(idx) |
|
3581 | 3581 | addargs, addkwargs = _getrevisionseed(orig, rev, tr, source) |
|
3582 | 3582 | if clearcaches: |
|
3583 | 3583 | dest.index.clearcaches() |
|
3584 | 3584 | dest.clearcaches() |
|
3585 | 3585 | with timeone() as r: |
|
3586 | 3586 | dest.addrawrevision(*addargs, **addkwargs) |
|
3587 | 3587 | timings.append((rev, r[0])) |
|
3588 | 3588 | updateprogress(total) |
|
3589 | 3589 | completeprogress() |
|
3590 | 3590 | return timings |
|
3591 | 3591 | |
|
3592 | 3592 | |
|
3593 | 3593 | def _getrevisionseed(orig, rev, tr, source): |
|
3594 | 3594 | from mercurial.node import nullid |
|
3595 | 3595 | |
|
3596 | 3596 | linkrev = orig.linkrev(rev) |
|
3597 | 3597 | node = orig.node(rev) |
|
3598 | 3598 | p1, p2 = orig.parents(node) |
|
3599 | 3599 | flags = orig.flags(rev) |
|
3600 | 3600 | cachedelta = None |
|
3601 | 3601 | text = None |
|
3602 | 3602 | |
|
3603 | 3603 | if source == b'full': |
|
3604 | 3604 | text = orig.revision(rev) |
|
3605 | 3605 | elif source == b'parent-1': |
|
3606 | 3606 | baserev = orig.rev(p1) |
|
3607 | 3607 | cachedelta = (baserev, orig.revdiff(p1, rev)) |
|
3608 | 3608 | elif source == b'parent-2': |
|
3609 | 3609 | parent = p2 |
|
3610 | 3610 | if p2 == nullid: |
|
3611 | 3611 | parent = p1 |
|
3612 | 3612 | baserev = orig.rev(parent) |
|
3613 | 3613 | cachedelta = (baserev, orig.revdiff(parent, rev)) |
|
3614 | 3614 | elif source == b'parent-smallest': |
|
3615 | 3615 | p1diff = orig.revdiff(p1, rev) |
|
3616 | 3616 | parent = p1 |
|
3617 | 3617 | diff = p1diff |
|
3618 | 3618 | if p2 != nullid: |
|
3619 | 3619 | p2diff = orig.revdiff(p2, rev) |
|
3620 | 3620 | if len(p1diff) > len(p2diff): |
|
3621 | 3621 | parent = p2 |
|
3622 | 3622 | diff = p2diff |
|
3623 | 3623 | baserev = orig.rev(parent) |
|
3624 | 3624 | cachedelta = (baserev, diff) |
|
3625 | 3625 | elif source == b'storage': |
|
3626 | 3626 | baserev = orig.deltaparent(rev) |
|
3627 | 3627 | cachedelta = (baserev, orig.revdiff(orig.node(baserev), rev)) |
|
3628 | 3628 | |
|
3629 | 3629 | return ( |
|
3630 | 3630 | (text, tr, linkrev, p1, p2), |
|
3631 | 3631 | {'node': node, 'flags': flags, 'cachedelta': cachedelta}, |
|
3632 | 3632 | ) |
|
3633 | 3633 | |
|
3634 | 3634 | |
|
3635 | 3635 | @contextlib.contextmanager |
|
3636 | 3636 | def _temprevlog(ui, orig, truncaterev): |
|
3637 | 3637 | from mercurial import vfs as vfsmod |
|
3638 | 3638 | |
|
3639 | 3639 | if orig._inline: |
|
3640 | 3640 | raise error.Abort('not supporting inline revlog (yet)') |
|
3641 | 3641 | revlogkwargs = {} |
|
3642 | 3642 | k = 'upperboundcomp' |
|
3643 | 3643 | if util.safehasattr(orig, k): |
|
3644 | 3644 | revlogkwargs[k] = getattr(orig, k) |
|
3645 | 3645 | |
|
3646 | 3646 | indexfile = getattr(orig, '_indexfile', None) |
|
3647 | 3647 | if indexfile is None: |
|
3648 | 3648 | # compatibility with <= hg-5.8 |
|
3649 | 3649 | indexfile = getattr(orig, 'indexfile') |
|
3650 | 3650 | origindexpath = orig.opener.join(indexfile) |
|
3651 | 3651 | |
|
3652 | 3652 | datafile = getattr(orig, '_datafile', getattr(orig, 'datafile')) |
|
3653 | 3653 | origdatapath = orig.opener.join(datafile) |
|
3654 | 3654 | radix = b'revlog' |
|
3655 | 3655 | indexname = b'revlog.i' |
|
3656 | 3656 | dataname = b'revlog.d' |
|
3657 | 3657 | |
|
3658 | 3658 | tmpdir = tempfile.mkdtemp(prefix='tmp-hgperf-') |
|
3659 | 3659 | try: |
|
3660 | 3660 | # copy the data file in a temporary directory |
|
3661 | 3661 | ui.debug('copying data in %s\n' % tmpdir) |
|
3662 | 3662 | destindexpath = os.path.join(tmpdir, 'revlog.i') |
|
3663 | 3663 | destdatapath = os.path.join(tmpdir, 'revlog.d') |
|
3664 | 3664 | shutil.copyfile(origindexpath, destindexpath) |
|
3665 | 3665 | shutil.copyfile(origdatapath, destdatapath) |
|
3666 | 3666 | |
|
3667 | 3667 | # remove the data we want to add again |
|
3668 | 3668 | ui.debug('truncating data to be rewritten\n') |
|
3669 | 3669 | with open(destindexpath, 'ab') as index: |
|
3670 | 3670 | index.seek(0) |
|
3671 | 3671 | index.truncate(truncaterev * orig._io.size) |
|
3672 | 3672 | with open(destdatapath, 'ab') as data: |
|
3673 | 3673 | data.seek(0) |
|
3674 | 3674 | data.truncate(orig.start(truncaterev)) |
|
3675 | 3675 | |
|
3676 | 3676 | # instantiate a new revlog from the temporary copy |
|
3677 | 3677 | ui.debug('truncating adding to be rewritten\n') |
|
3678 | 3678 | vfs = vfsmod.vfs(tmpdir) |
|
3679 | 3679 | vfs.options = getattr(orig.opener, 'options', None) |
|
3680 | 3680 | |
|
3681 | 3681 | try: |
|
3682 | 3682 | dest = revlog(vfs, radix=radix, **revlogkwargs) |
|
3683 | 3683 | except TypeError: |
|
3684 | 3684 | dest = revlog( |
|
3685 | 3685 | vfs, indexfile=indexname, datafile=dataname, **revlogkwargs |
|
3686 | 3686 | ) |
|
3687 | 3687 | if dest._inline: |
|
3688 | 3688 | raise error.Abort('not supporting inline revlog (yet)') |
|
3689 | 3689 | # make sure internals are initialized |
|
3690 | 3690 | dest.revision(len(dest) - 1) |
|
3691 | 3691 | yield dest |
|
3692 | 3692 | del dest, vfs |
|
3693 | 3693 | finally: |
|
3694 | 3694 | shutil.rmtree(tmpdir, True) |
|
3695 | 3695 | |
|
3696 | 3696 | |
|
3697 | 3697 | @command( |
|
3698 | 3698 | b'perf::revlogchunks|perfrevlogchunks', |
|
3699 | 3699 | revlogopts |
|
3700 | 3700 | + formatteropts |
|
3701 | 3701 | + [ |
|
3702 | 3702 | (b'e', b'engines', b'', b'compression engines to use'), |
|
3703 | 3703 | (b's', b'startrev', 0, b'revision to start at'), |
|
3704 | 3704 | ], |
|
3705 | 3705 | b'-c|-m|FILE', |
|
3706 | 3706 | ) |
|
3707 | 3707 | def perfrevlogchunks(ui, repo, file_=None, engines=None, startrev=0, **opts): |
|
3708 | 3708 | """Benchmark operations on revlog chunks. |
|
3709 | 3709 | |
|
3710 | 3710 | Logically, each revlog is a collection of fulltext revisions. However, |
|
3711 | 3711 | stored within each revlog are "chunks" of possibly compressed data. This |
|
3712 | 3712 | data needs to be read and decompressed or compressed and written. |
|
3713 | 3713 | |
|
3714 | 3714 | This command measures the time it takes to read+decompress and recompress |
|
3715 | 3715 | chunks in a revlog. It effectively isolates I/O and compression performance. |
|
3716 | 3716 | For measurements of higher-level operations like resolving revisions, |
|
3717 | 3717 | see ``perfrevlogrevisions`` and ``perfrevlogrevision``. |
|
3718 | 3718 | """ |
|
3719 | 3719 | opts = _byteskwargs(opts) |
|
3720 | 3720 | |
|
3721 | 3721 | rl = cmdutil.openrevlog(repo, b'perfrevlogchunks', file_, opts) |
|
3722 | 3722 | |
|
3723 | 3723 | # - _chunkraw was renamed to _getsegmentforrevs |
|
3724 | 3724 | # - _getsegmentforrevs was moved on the inner object |
|
3725 | 3725 | try: |
|
3726 | 3726 | segmentforrevs = rl._inner.get_segment_for_revs |
|
3727 | 3727 | except AttributeError: |
|
3728 | 3728 | try: |
|
3729 | 3729 | segmentforrevs = rl._getsegmentforrevs |
|
3730 | 3730 | except AttributeError: |
|
3731 | 3731 | segmentforrevs = rl._chunkraw |
|
3732 | 3732 | |
|
3733 | 3733 | # Verify engines argument. |
|
3734 | 3734 | if engines: |
|
3735 | 3735 | engines = {e.strip() for e in engines.split(b',')} |
|
3736 | 3736 | for engine in engines: |
|
3737 | 3737 | try: |
|
3738 | 3738 | util.compressionengines[engine] |
|
3739 | 3739 | except KeyError: |
|
3740 | 3740 | raise error.Abort(b'unknown compression engine: %s' % engine) |
|
3741 | 3741 | else: |
|
3742 | 3742 | engines = [] |
|
3743 | 3743 | for e in util.compengines: |
|
3744 | 3744 | engine = util.compengines[e] |
|
3745 | 3745 | try: |
|
3746 | 3746 | if engine.available(): |
|
3747 | 3747 | engine.revlogcompressor().compress(b'dummy') |
|
3748 | 3748 | engines.append(e) |
|
3749 | 3749 | except NotImplementedError: |
|
3750 | 3750 | pass |
|
3751 | 3751 | |
|
3752 | 3752 | revs = list(rl.revs(startrev, len(rl) - 1)) |
|
3753 | 3753 | |
|
3754 | 3754 | @contextlib.contextmanager |
|
3755 | 3755 | def reading(rl): |
|
3756 | 3756 | if getattr(rl, 'reading', None) is not None: |
|
3757 | 3757 | with rl.reading(): |
|
3758 | 3758 | yield None |
|
3759 | 3759 | elif rl._inline: |
|
3760 | 3760 | indexfile = getattr(rl, '_indexfile', None) |
|
3761 | 3761 | if indexfile is None: |
|
3762 | 3762 | # compatibility with <= hg-5.8 |
|
3763 | 3763 | indexfile = getattr(rl, 'indexfile') |
|
3764 | 3764 | yield getsvfs(repo)(indexfile) |
|
3765 | 3765 | else: |
|
3766 | 3766 | datafile = getattr(rl, 'datafile', getattr(rl, 'datafile')) |
|
3767 | 3767 | yield getsvfs(repo)(datafile) |
|
3768 | 3768 | |
|
3769 | 3769 | if getattr(rl, 'reading', None) is not None: |
|
3770 | 3770 | |
|
3771 | 3771 | @contextlib.contextmanager |
|
3772 | 3772 | def lazy_reading(rl): |
|
3773 | 3773 | with rl.reading(): |
|
3774 | 3774 | yield |
|
3775 | 3775 | |
|
3776 | 3776 | else: |
|
3777 | 3777 | |
|
3778 | 3778 | @contextlib.contextmanager |
|
3779 | 3779 | def lazy_reading(rl): |
|
3780 | 3780 | yield |
|
3781 | 3781 | |
|
3782 | 3782 | def doread(): |
|
3783 | 3783 | rl.clearcaches() |
|
3784 | 3784 | for rev in revs: |
|
3785 | 3785 | with lazy_reading(rl): |
|
3786 | 3786 | segmentforrevs(rev, rev) |
|
3787 | 3787 | |
|
3788 | 3788 | def doreadcachedfh(): |
|
3789 | 3789 | rl.clearcaches() |
|
3790 | 3790 | with reading(rl) as fh: |
|
3791 | 3791 | if fh is not None: |
|
3792 | 3792 | for rev in revs: |
|
3793 | 3793 | segmentforrevs(rev, rev, df=fh) |
|
3794 | 3794 | else: |
|
3795 | 3795 | for rev in revs: |
|
3796 | 3796 | segmentforrevs(rev, rev) |
|
3797 | 3797 | |
|
3798 | 3798 | def doreadbatch(): |
|
3799 | 3799 | rl.clearcaches() |
|
3800 | 3800 | with lazy_reading(rl): |
|
3801 | 3801 | segmentforrevs(revs[0], revs[-1]) |
|
3802 | 3802 | |
|
3803 | 3803 | def doreadbatchcachedfh(): |
|
3804 | 3804 | rl.clearcaches() |
|
3805 | 3805 | with reading(rl) as fh: |
|
3806 | 3806 | if fh is not None: |
|
3807 | 3807 | segmentforrevs(revs[0], revs[-1], df=fh) |
|
3808 | 3808 | else: |
|
3809 | 3809 | segmentforrevs(revs[0], revs[-1]) |
|
3810 | 3810 | |
|
3811 | 3811 | def dochunk(): |
|
3812 | 3812 | rl.clearcaches() |
|
3813 | # chunk used to be available directly on the revlog | |
|
3814 | _chunk = getattr(rl, '_inner', rl)._chunk | |
|
3813 | 3815 | with reading(rl) as fh: |
|
3814 | 3816 | if fh is not None: |
|
3815 | 3817 | for rev in revs: |
|
3816 |
|
|
|
3818 | _chunk(rev, df=fh) | |
|
3817 | 3819 | else: |
|
3818 | 3820 | for rev in revs: |
|
3819 |
|
|
|
3821 | _chunk(rev) | |
|
3820 | 3822 | |
|
3821 | 3823 | chunks = [None] |
|
3822 | 3824 | |
|
3823 | 3825 | def dochunkbatch(): |
|
3824 | 3826 | rl.clearcaches() |
|
3825 | 3827 | with reading(rl) as fh: |
|
3826 | 3828 | if fh is not None: |
|
3827 | 3829 | # Save chunks as a side-effect. |
|
3828 | 3830 | chunks[0] = rl._chunks(revs, df=fh) |
|
3829 | 3831 | else: |
|
3830 | 3832 | # Save chunks as a side-effect. |
|
3831 | 3833 | chunks[0] = rl._chunks(revs) |
|
3832 | 3834 | |
|
3833 | 3835 | def docompress(compressor): |
|
3834 | 3836 | rl.clearcaches() |
|
3835 | 3837 | |
|
3836 | 3838 | compressor_holder = getattr(rl, '_inner', rl) |
|
3837 | 3839 | |
|
3838 | 3840 | try: |
|
3839 | 3841 | # Swap in the requested compression engine. |
|
3840 | 3842 | oldcompressor = compressor_holder._compressor |
|
3841 | 3843 | compressor_holder._compressor = compressor |
|
3842 | 3844 | for chunk in chunks[0]: |
|
3843 | 3845 | rl.compress(chunk) |
|
3844 | 3846 | finally: |
|
3845 | 3847 | compressor_holder._compressor = oldcompressor |
|
3846 | 3848 | |
|
3847 | 3849 | benches = [ |
|
3848 | 3850 | (lambda: doread(), b'read'), |
|
3849 | 3851 | (lambda: doreadcachedfh(), b'read w/ reused fd'), |
|
3850 | 3852 | (lambda: doreadbatch(), b'read batch'), |
|
3851 | 3853 | (lambda: doreadbatchcachedfh(), b'read batch w/ reused fd'), |
|
3852 | 3854 | (lambda: dochunk(), b'chunk'), |
|
3853 | 3855 | (lambda: dochunkbatch(), b'chunk batch'), |
|
3854 | 3856 | ] |
|
3855 | 3857 | |
|
3856 | 3858 | for engine in sorted(engines): |
|
3857 | 3859 | compressor = util.compengines[engine].revlogcompressor() |
|
3858 | 3860 | benches.append( |
|
3859 | 3861 | ( |
|
3860 | 3862 | functools.partial(docompress, compressor), |
|
3861 | 3863 | b'compress w/ %s' % engine, |
|
3862 | 3864 | ) |
|
3863 | 3865 | ) |
|
3864 | 3866 | |
|
3865 | 3867 | for fn, title in benches: |
|
3866 | 3868 | timer, fm = gettimer(ui, opts) |
|
3867 | 3869 | timer(fn, title=title) |
|
3868 | 3870 | fm.end() |
|
3869 | 3871 | |
|
3870 | 3872 | |
|
3871 | 3873 | @command( |
|
3872 | 3874 | b'perf::revlogrevision|perfrevlogrevision', |
|
3873 | 3875 | revlogopts |
|
3874 | 3876 | + formatteropts |
|
3875 | 3877 | + [(b'', b'cache', False, b'use caches instead of clearing')], |
|
3876 | 3878 | b'-c|-m|FILE REV', |
|
3877 | 3879 | ) |
|
3878 | 3880 | def perfrevlogrevision(ui, repo, file_, rev=None, cache=None, **opts): |
|
3879 | 3881 | """Benchmark obtaining a revlog revision. |
|
3880 | 3882 | |
|
3881 | 3883 | Obtaining a revlog revision consists of roughly the following steps: |
|
3882 | 3884 | |
|
3883 | 3885 | 1. Compute the delta chain |
|
3884 | 3886 | 2. Slice the delta chain if applicable |
|
3885 | 3887 | 3. Obtain the raw chunks for that delta chain |
|
3886 | 3888 | 4. Decompress each raw chunk |
|
3887 | 3889 | 5. Apply binary patches to obtain fulltext |
|
3888 | 3890 | 6. Verify hash of fulltext |
|
3889 | 3891 | |
|
3890 | 3892 | This command measures the time spent in each of these phases. |
|
3891 | 3893 | """ |
|
3892 | 3894 | opts = _byteskwargs(opts) |
|
3893 | 3895 | |
|
3894 | 3896 | if opts.get(b'changelog') or opts.get(b'manifest'): |
|
3895 | 3897 | file_, rev = None, file_ |
|
3896 | 3898 | elif rev is None: |
|
3897 | 3899 | raise error.CommandError(b'perfrevlogrevision', b'invalid arguments') |
|
3898 | 3900 | |
|
3899 | 3901 | r = cmdutil.openrevlog(repo, b'perfrevlogrevision', file_, opts) |
|
3900 | 3902 | |
|
3901 | 3903 | # _chunkraw was renamed to _getsegmentforrevs. |
|
3902 | 3904 | try: |
|
3903 | 3905 | segmentforrevs = r._inner.get_segment_for_revs |
|
3904 | 3906 | except AttributeError: |
|
3905 | 3907 | try: |
|
3906 | 3908 | segmentforrevs = r._getsegmentforrevs |
|
3907 | 3909 | except AttributeError: |
|
3908 | 3910 | segmentforrevs = r._chunkraw |
|
3909 | 3911 | |
|
3910 | 3912 | node = r.lookup(rev) |
|
3911 | 3913 | rev = r.rev(node) |
|
3912 | 3914 | |
|
3913 | 3915 | if getattr(r, 'reading', None) is not None: |
|
3914 | 3916 | |
|
3915 | 3917 | @contextlib.contextmanager |
|
3916 | 3918 | def lazy_reading(r): |
|
3917 | 3919 | with r.reading(): |
|
3918 | 3920 | yield |
|
3919 | 3921 | |
|
3920 | 3922 | else: |
|
3921 | 3923 | |
|
3922 | 3924 | @contextlib.contextmanager |
|
3923 | 3925 | def lazy_reading(r): |
|
3924 | 3926 | yield |
|
3925 | 3927 | |
|
3926 | 3928 | def getrawchunks(data, chain): |
|
3927 | 3929 | start = r.start |
|
3928 | 3930 | length = r.length |
|
3929 | 3931 | inline = r._inline |
|
3930 | 3932 | try: |
|
3931 | 3933 | iosize = r.index.entry_size |
|
3932 | 3934 | except AttributeError: |
|
3933 | 3935 | iosize = r._io.size |
|
3934 | 3936 | buffer = util.buffer |
|
3935 | 3937 | |
|
3936 | 3938 | chunks = [] |
|
3937 | 3939 | ladd = chunks.append |
|
3938 | 3940 | for idx, item in enumerate(chain): |
|
3939 | 3941 | offset = start(item[0]) |
|
3940 | 3942 | bits = data[idx] |
|
3941 | 3943 | for rev in item: |
|
3942 | 3944 | chunkstart = start(rev) |
|
3943 | 3945 | if inline: |
|
3944 | 3946 | chunkstart += (rev + 1) * iosize |
|
3945 | 3947 | chunklength = length(rev) |
|
3946 | 3948 | ladd(buffer(bits, chunkstart - offset, chunklength)) |
|
3947 | 3949 | |
|
3948 | 3950 | return chunks |
|
3949 | 3951 | |
|
3950 | 3952 | def dodeltachain(rev): |
|
3951 | 3953 | if not cache: |
|
3952 | 3954 | r.clearcaches() |
|
3953 | 3955 | r._deltachain(rev) |
|
3954 | 3956 | |
|
3955 | 3957 | def doread(chain): |
|
3956 | 3958 | if not cache: |
|
3957 | 3959 | r.clearcaches() |
|
3958 | 3960 | for item in slicedchain: |
|
3959 | 3961 | with lazy_reading(r): |
|
3960 | 3962 | segmentforrevs(item[0], item[-1]) |
|
3961 | 3963 | |
|
3962 | 3964 | def doslice(r, chain, size): |
|
3963 | 3965 | for s in slicechunk(r, chain, targetsize=size): |
|
3964 | 3966 | pass |
|
3965 | 3967 | |
|
3966 | 3968 | def dorawchunks(data, chain): |
|
3967 | 3969 | if not cache: |
|
3968 | 3970 | r.clearcaches() |
|
3969 | 3971 | getrawchunks(data, chain) |
|
3970 | 3972 | |
|
3971 | 3973 | def dodecompress(chunks): |
|
3972 | 3974 | decomp = r.decompress |
|
3973 | 3975 | for chunk in chunks: |
|
3974 | 3976 | decomp(chunk) |
|
3975 | 3977 | |
|
3976 | 3978 | def dopatch(text, bins): |
|
3977 | 3979 | if not cache: |
|
3978 | 3980 | r.clearcaches() |
|
3979 | 3981 | mdiff.patches(text, bins) |
|
3980 | 3982 | |
|
3981 | 3983 | def dohash(text): |
|
3982 | 3984 | if not cache: |
|
3983 | 3985 | r.clearcaches() |
|
3984 | 3986 | r.checkhash(text, node, rev=rev) |
|
3985 | 3987 | |
|
3986 | 3988 | def dorevision(): |
|
3987 | 3989 | if not cache: |
|
3988 | 3990 | r.clearcaches() |
|
3989 | 3991 | r.revision(node) |
|
3990 | 3992 | |
|
3991 | 3993 | try: |
|
3992 | 3994 | from mercurial.revlogutils.deltas import slicechunk |
|
3993 | 3995 | except ImportError: |
|
3994 | 3996 | slicechunk = getattr(revlog, '_slicechunk', None) |
|
3995 | 3997 | |
|
3996 | 3998 | size = r.length(rev) |
|
3997 | 3999 | chain = r._deltachain(rev)[0] |
|
3998 | 4000 | |
|
3999 | 4001 | with_sparse_read = False |
|
4000 | 4002 | if hasattr(r, 'data_config'): |
|
4001 | 4003 | with_sparse_read = r.data_config.with_sparse_read |
|
4002 | 4004 | elif hasattr(r, '_withsparseread'): |
|
4003 | 4005 | with_sparse_read = r._withsparseread |
|
4004 | 4006 | if with_sparse_read: |
|
4005 | 4007 | slicedchain = (chain,) |
|
4006 | 4008 | else: |
|
4007 | 4009 | slicedchain = tuple(slicechunk(r, chain, targetsize=size)) |
|
4008 | 4010 | data = [segmentforrevs(seg[0], seg[-1])[1] for seg in slicedchain] |
|
4009 | 4011 | rawchunks = getrawchunks(data, slicedchain) |
|
4010 | 4012 | bins = r._chunks(chain) |
|
4011 | 4013 | text = bytes(bins[0]) |
|
4012 | 4014 | bins = bins[1:] |
|
4013 | 4015 | text = mdiff.patches(text, bins) |
|
4014 | 4016 | |
|
4015 | 4017 | benches = [ |
|
4016 | 4018 | (lambda: dorevision(), b'full'), |
|
4017 | 4019 | (lambda: dodeltachain(rev), b'deltachain'), |
|
4018 | 4020 | (lambda: doread(chain), b'read'), |
|
4019 | 4021 | ] |
|
4020 | 4022 | |
|
4021 | 4023 | if with_sparse_read: |
|
4022 | 4024 | slicing = (lambda: doslice(r, chain, size), b'slice-sparse-chain') |
|
4023 | 4025 | benches.append(slicing) |
|
4024 | 4026 | |
|
4025 | 4027 | benches.extend( |
|
4026 | 4028 | [ |
|
4027 | 4029 | (lambda: dorawchunks(data, slicedchain), b'rawchunks'), |
|
4028 | 4030 | (lambda: dodecompress(rawchunks), b'decompress'), |
|
4029 | 4031 | (lambda: dopatch(text, bins), b'patch'), |
|
4030 | 4032 | (lambda: dohash(text), b'hash'), |
|
4031 | 4033 | ] |
|
4032 | 4034 | ) |
|
4033 | 4035 | |
|
4034 | 4036 | timer, fm = gettimer(ui, opts) |
|
4035 | 4037 | for fn, title in benches: |
|
4036 | 4038 | timer(fn, title=title) |
|
4037 | 4039 | fm.end() |
|
4038 | 4040 | |
|
4039 | 4041 | |
|
4040 | 4042 | @command( |
|
4041 | 4043 | b'perf::revset|perfrevset', |
|
4042 | 4044 | [ |
|
4043 | 4045 | (b'C', b'clear', False, b'clear volatile cache between each call.'), |
|
4044 | 4046 | (b'', b'contexts', False, b'obtain changectx for each revision'), |
|
4045 | 4047 | ] |
|
4046 | 4048 | + formatteropts, |
|
4047 | 4049 | b"REVSET", |
|
4048 | 4050 | ) |
|
4049 | 4051 | def perfrevset(ui, repo, expr, clear=False, contexts=False, **opts): |
|
4050 | 4052 | """benchmark the execution time of a revset |
|
4051 | 4053 | |
|
4052 | 4054 | Use the --clean option if need to evaluate the impact of build volatile |
|
4053 | 4055 | revisions set cache on the revset execution. Volatile cache hold filtered |
|
4054 | 4056 | and obsolete related cache.""" |
|
4055 | 4057 | opts = _byteskwargs(opts) |
|
4056 | 4058 | |
|
4057 | 4059 | timer, fm = gettimer(ui, opts) |
|
4058 | 4060 | |
|
4059 | 4061 | def d(): |
|
4060 | 4062 | if clear: |
|
4061 | 4063 | repo.invalidatevolatilesets() |
|
4062 | 4064 | if contexts: |
|
4063 | 4065 | for ctx in repo.set(expr): |
|
4064 | 4066 | pass |
|
4065 | 4067 | else: |
|
4066 | 4068 | for r in repo.revs(expr): |
|
4067 | 4069 | pass |
|
4068 | 4070 | |
|
4069 | 4071 | timer(d) |
|
4070 | 4072 | fm.end() |
|
4071 | 4073 | |
|
4072 | 4074 | |
|
4073 | 4075 | @command( |
|
4074 | 4076 | b'perf::volatilesets|perfvolatilesets', |
|
4075 | 4077 | [ |
|
4076 | 4078 | (b'', b'clear-obsstore', False, b'drop obsstore between each call.'), |
|
4077 | 4079 | ] |
|
4078 | 4080 | + formatteropts, |
|
4079 | 4081 | ) |
|
4080 | 4082 | def perfvolatilesets(ui, repo, *names, **opts): |
|
4081 | 4083 | """benchmark the computation of various volatile set |
|
4082 | 4084 | |
|
4083 | 4085 | Volatile set computes element related to filtering and obsolescence.""" |
|
4084 | 4086 | opts = _byteskwargs(opts) |
|
4085 | 4087 | timer, fm = gettimer(ui, opts) |
|
4086 | 4088 | repo = repo.unfiltered() |
|
4087 | 4089 | |
|
4088 | 4090 | def getobs(name): |
|
4089 | 4091 | def d(): |
|
4090 | 4092 | repo.invalidatevolatilesets() |
|
4091 | 4093 | if opts[b'clear_obsstore']: |
|
4092 | 4094 | clearfilecache(repo, b'obsstore') |
|
4093 | 4095 | obsolete.getrevs(repo, name) |
|
4094 | 4096 | |
|
4095 | 4097 | return d |
|
4096 | 4098 | |
|
4097 | 4099 | allobs = sorted(obsolete.cachefuncs) |
|
4098 | 4100 | if names: |
|
4099 | 4101 | allobs = [n for n in allobs if n in names] |
|
4100 | 4102 | |
|
4101 | 4103 | for name in allobs: |
|
4102 | 4104 | timer(getobs(name), title=name) |
|
4103 | 4105 | |
|
4104 | 4106 | def getfiltered(name): |
|
4105 | 4107 | def d(): |
|
4106 | 4108 | repo.invalidatevolatilesets() |
|
4107 | 4109 | if opts[b'clear_obsstore']: |
|
4108 | 4110 | clearfilecache(repo, b'obsstore') |
|
4109 | 4111 | repoview.filterrevs(repo, name) |
|
4110 | 4112 | |
|
4111 | 4113 | return d |
|
4112 | 4114 | |
|
4113 | 4115 | allfilter = sorted(repoview.filtertable) |
|
4114 | 4116 | if names: |
|
4115 | 4117 | allfilter = [n for n in allfilter if n in names] |
|
4116 | 4118 | |
|
4117 | 4119 | for name in allfilter: |
|
4118 | 4120 | timer(getfiltered(name), title=name) |
|
4119 | 4121 | fm.end() |
|
4120 | 4122 | |
|
4121 | 4123 | |
|
4122 | 4124 | @command( |
|
4123 | 4125 | b'perf::branchmap|perfbranchmap', |
|
4124 | 4126 | [ |
|
4125 | 4127 | (b'f', b'full', False, b'Includes build time of subset'), |
|
4126 | 4128 | ( |
|
4127 | 4129 | b'', |
|
4128 | 4130 | b'clear-revbranch', |
|
4129 | 4131 | False, |
|
4130 | 4132 | b'purge the revbranch cache between computation', |
|
4131 | 4133 | ), |
|
4132 | 4134 | ] |
|
4133 | 4135 | + formatteropts, |
|
4134 | 4136 | ) |
|
4135 | 4137 | def perfbranchmap(ui, repo, *filternames, **opts): |
|
4136 | 4138 | """benchmark the update of a branchmap |
|
4137 | 4139 | |
|
4138 | 4140 | This benchmarks the full repo.branchmap() call with read and write disabled |
|
4139 | 4141 | """ |
|
4140 | 4142 | opts = _byteskwargs(opts) |
|
4141 | 4143 | full = opts.get(b"full", False) |
|
4142 | 4144 | clear_revbranch = opts.get(b"clear_revbranch", False) |
|
4143 | 4145 | timer, fm = gettimer(ui, opts) |
|
4144 | 4146 | |
|
4145 | 4147 | def getbranchmap(filtername): |
|
4146 | 4148 | """generate a benchmark function for the filtername""" |
|
4147 | 4149 | if filtername is None: |
|
4148 | 4150 | view = repo |
|
4149 | 4151 | else: |
|
4150 | 4152 | view = repo.filtered(filtername) |
|
4151 | 4153 | if util.safehasattr(view._branchcaches, '_per_filter'): |
|
4152 | 4154 | filtered = view._branchcaches._per_filter |
|
4153 | 4155 | else: |
|
4154 | 4156 | # older versions |
|
4155 | 4157 | filtered = view._branchcaches |
|
4156 | 4158 | |
|
4157 | 4159 | def d(): |
|
4158 | 4160 | if clear_revbranch: |
|
4159 | 4161 | repo.revbranchcache()._clear() |
|
4160 | 4162 | if full: |
|
4161 | 4163 | view._branchcaches.clear() |
|
4162 | 4164 | else: |
|
4163 | 4165 | filtered.pop(filtername, None) |
|
4164 | 4166 | view.branchmap() |
|
4165 | 4167 | |
|
4166 | 4168 | return d |
|
4167 | 4169 | |
|
4168 | 4170 | # add filter in smaller subset to bigger subset |
|
4169 | 4171 | possiblefilters = set(repoview.filtertable) |
|
4170 | 4172 | if filternames: |
|
4171 | 4173 | possiblefilters &= set(filternames) |
|
4172 | 4174 | subsettable = getbranchmapsubsettable() |
|
4173 | 4175 | allfilters = [] |
|
4174 | 4176 | while possiblefilters: |
|
4175 | 4177 | for name in possiblefilters: |
|
4176 | 4178 | subset = subsettable.get(name) |
|
4177 | 4179 | if subset not in possiblefilters: |
|
4178 | 4180 | break |
|
4179 | 4181 | else: |
|
4180 | 4182 | assert False, b'subset cycle %s!' % possiblefilters |
|
4181 | 4183 | allfilters.append(name) |
|
4182 | 4184 | possiblefilters.remove(name) |
|
4183 | 4185 | |
|
4184 | 4186 | # warm the cache |
|
4185 | 4187 | if not full: |
|
4186 | 4188 | for name in allfilters: |
|
4187 | 4189 | repo.filtered(name).branchmap() |
|
4188 | 4190 | if not filternames or b'unfiltered' in filternames: |
|
4189 | 4191 | # add unfiltered |
|
4190 | 4192 | allfilters.append(None) |
|
4191 | 4193 | |
|
4192 | 4194 | if util.safehasattr(branchmap.branchcache, 'fromfile'): |
|
4193 | 4195 | branchcacheread = safeattrsetter(branchmap.branchcache, b'fromfile') |
|
4194 | 4196 | branchcacheread.set(classmethod(lambda *args: None)) |
|
4195 | 4197 | else: |
|
4196 | 4198 | # older versions |
|
4197 | 4199 | branchcacheread = safeattrsetter(branchmap, b'read') |
|
4198 | 4200 | branchcacheread.set(lambda *args: None) |
|
4199 | 4201 | branchcachewrite = safeattrsetter(branchmap.branchcache, b'write') |
|
4200 | 4202 | branchcachewrite.set(lambda *args: None) |
|
4201 | 4203 | try: |
|
4202 | 4204 | for name in allfilters: |
|
4203 | 4205 | printname = name |
|
4204 | 4206 | if name is None: |
|
4205 | 4207 | printname = b'unfiltered' |
|
4206 | 4208 | timer(getbranchmap(name), title=printname) |
|
4207 | 4209 | finally: |
|
4208 | 4210 | branchcacheread.restore() |
|
4209 | 4211 | branchcachewrite.restore() |
|
4210 | 4212 | fm.end() |
|
4211 | 4213 | |
|
4212 | 4214 | |
|
4213 | 4215 | @command( |
|
4214 | 4216 | b'perf::branchmapupdate|perfbranchmapupdate', |
|
4215 | 4217 | [ |
|
4216 | 4218 | (b'', b'base', [], b'subset of revision to start from'), |
|
4217 | 4219 | (b'', b'target', [], b'subset of revision to end with'), |
|
4218 | 4220 | (b'', b'clear-caches', False, b'clear cache between each runs'), |
|
4219 | 4221 | ] |
|
4220 | 4222 | + formatteropts, |
|
4221 | 4223 | ) |
|
4222 | 4224 | def perfbranchmapupdate(ui, repo, base=(), target=(), **opts): |
|
4223 | 4225 | """benchmark branchmap update from for <base> revs to <target> revs |
|
4224 | 4226 | |
|
4225 | 4227 | If `--clear-caches` is passed, the following items will be reset before |
|
4226 | 4228 | each update: |
|
4227 | 4229 | * the changelog instance and associated indexes |
|
4228 | 4230 | * the rev-branch-cache instance |
|
4229 | 4231 | |
|
4230 | 4232 | Examples: |
|
4231 | 4233 | |
|
4232 | 4234 | # update for the one last revision |
|
4233 | 4235 | $ hg perfbranchmapupdate --base 'not tip' --target 'tip' |
|
4234 | 4236 | |
|
4235 | 4237 | $ update for change coming with a new branch |
|
4236 | 4238 | $ hg perfbranchmapupdate --base 'stable' --target 'default' |
|
4237 | 4239 | """ |
|
4238 | 4240 | from mercurial import branchmap |
|
4239 | 4241 | from mercurial import repoview |
|
4240 | 4242 | |
|
4241 | 4243 | opts = _byteskwargs(opts) |
|
4242 | 4244 | timer, fm = gettimer(ui, opts) |
|
4243 | 4245 | clearcaches = opts[b'clear_caches'] |
|
4244 | 4246 | unfi = repo.unfiltered() |
|
4245 | 4247 | x = [None] # used to pass data between closure |
|
4246 | 4248 | |
|
4247 | 4249 | # we use a `list` here to avoid possible side effect from smartset |
|
4248 | 4250 | baserevs = list(scmutil.revrange(repo, base)) |
|
4249 | 4251 | targetrevs = list(scmutil.revrange(repo, target)) |
|
4250 | 4252 | if not baserevs: |
|
4251 | 4253 | raise error.Abort(b'no revisions selected for --base') |
|
4252 | 4254 | if not targetrevs: |
|
4253 | 4255 | raise error.Abort(b'no revisions selected for --target') |
|
4254 | 4256 | |
|
4255 | 4257 | # make sure the target branchmap also contains the one in the base |
|
4256 | 4258 | targetrevs = list(set(baserevs) | set(targetrevs)) |
|
4257 | 4259 | targetrevs.sort() |
|
4258 | 4260 | |
|
4259 | 4261 | cl = repo.changelog |
|
4260 | 4262 | allbaserevs = list(cl.ancestors(baserevs, inclusive=True)) |
|
4261 | 4263 | allbaserevs.sort() |
|
4262 | 4264 | alltargetrevs = frozenset(cl.ancestors(targetrevs, inclusive=True)) |
|
4263 | 4265 | |
|
4264 | 4266 | newrevs = list(alltargetrevs.difference(allbaserevs)) |
|
4265 | 4267 | newrevs.sort() |
|
4266 | 4268 | |
|
4267 | 4269 | allrevs = frozenset(unfi.changelog.revs()) |
|
4268 | 4270 | basefilterrevs = frozenset(allrevs.difference(allbaserevs)) |
|
4269 | 4271 | targetfilterrevs = frozenset(allrevs.difference(alltargetrevs)) |
|
4270 | 4272 | |
|
4271 | 4273 | def basefilter(repo, visibilityexceptions=None): |
|
4272 | 4274 | return basefilterrevs |
|
4273 | 4275 | |
|
4274 | 4276 | def targetfilter(repo, visibilityexceptions=None): |
|
4275 | 4277 | return targetfilterrevs |
|
4276 | 4278 | |
|
4277 | 4279 | msg = b'benchmark of branchmap with %d revisions with %d new ones\n' |
|
4278 | 4280 | ui.status(msg % (len(allbaserevs), len(newrevs))) |
|
4279 | 4281 | if targetfilterrevs: |
|
4280 | 4282 | msg = b'(%d revisions still filtered)\n' |
|
4281 | 4283 | ui.status(msg % len(targetfilterrevs)) |
|
4282 | 4284 | |
|
4283 | 4285 | try: |
|
4284 | 4286 | repoview.filtertable[b'__perf_branchmap_update_base'] = basefilter |
|
4285 | 4287 | repoview.filtertable[b'__perf_branchmap_update_target'] = targetfilter |
|
4286 | 4288 | |
|
4287 | 4289 | baserepo = repo.filtered(b'__perf_branchmap_update_base') |
|
4288 | 4290 | targetrepo = repo.filtered(b'__perf_branchmap_update_target') |
|
4289 | 4291 | |
|
4290 | 4292 | # try to find an existing branchmap to reuse |
|
4291 | 4293 | subsettable = getbranchmapsubsettable() |
|
4292 | 4294 | candidatefilter = subsettable.get(None) |
|
4293 | 4295 | while candidatefilter is not None: |
|
4294 | 4296 | candidatebm = repo.filtered(candidatefilter).branchmap() |
|
4295 | 4297 | if candidatebm.validfor(baserepo): |
|
4296 | 4298 | filtered = repoview.filterrevs(repo, candidatefilter) |
|
4297 | 4299 | missing = [r for r in allbaserevs if r in filtered] |
|
4298 | 4300 | base = candidatebm.copy() |
|
4299 | 4301 | base.update(baserepo, missing) |
|
4300 | 4302 | break |
|
4301 | 4303 | candidatefilter = subsettable.get(candidatefilter) |
|
4302 | 4304 | else: |
|
4303 | 4305 | # no suitable subset where found |
|
4304 | 4306 | base = branchmap.branchcache() |
|
4305 | 4307 | base.update(baserepo, allbaserevs) |
|
4306 | 4308 | |
|
4307 | 4309 | def setup(): |
|
4308 | 4310 | x[0] = base.copy() |
|
4309 | 4311 | if clearcaches: |
|
4310 | 4312 | unfi._revbranchcache = None |
|
4311 | 4313 | clearchangelog(repo) |
|
4312 | 4314 | |
|
4313 | 4315 | def bench(): |
|
4314 | 4316 | x[0].update(targetrepo, newrevs) |
|
4315 | 4317 | |
|
4316 | 4318 | timer(bench, setup=setup) |
|
4317 | 4319 | fm.end() |
|
4318 | 4320 | finally: |
|
4319 | 4321 | repoview.filtertable.pop(b'__perf_branchmap_update_base', None) |
|
4320 | 4322 | repoview.filtertable.pop(b'__perf_branchmap_update_target', None) |
|
4321 | 4323 | |
|
4322 | 4324 | |
|
4323 | 4325 | @command( |
|
4324 | 4326 | b'perf::branchmapload|perfbranchmapload', |
|
4325 | 4327 | [ |
|
4326 | 4328 | (b'f', b'filter', b'', b'Specify repoview filter'), |
|
4327 | 4329 | (b'', b'list', False, b'List brachmap filter caches'), |
|
4328 | 4330 | (b'', b'clear-revlogs', False, b'refresh changelog and manifest'), |
|
4329 | 4331 | ] |
|
4330 | 4332 | + formatteropts, |
|
4331 | 4333 | ) |
|
4332 | 4334 | def perfbranchmapload(ui, repo, filter=b'', list=False, **opts): |
|
4333 | 4335 | """benchmark reading the branchmap""" |
|
4334 | 4336 | opts = _byteskwargs(opts) |
|
4335 | 4337 | clearrevlogs = opts[b'clear_revlogs'] |
|
4336 | 4338 | |
|
4337 | 4339 | if list: |
|
4338 | 4340 | for name, kind, st in repo.cachevfs.readdir(stat=True): |
|
4339 | 4341 | if name.startswith(b'branch2'): |
|
4340 | 4342 | filtername = name.partition(b'-')[2] or b'unfiltered' |
|
4341 | 4343 | ui.status( |
|
4342 | 4344 | b'%s - %s\n' % (filtername, util.bytecount(st.st_size)) |
|
4343 | 4345 | ) |
|
4344 | 4346 | return |
|
4345 | 4347 | if not filter: |
|
4346 | 4348 | filter = None |
|
4347 | 4349 | subsettable = getbranchmapsubsettable() |
|
4348 | 4350 | if filter is None: |
|
4349 | 4351 | repo = repo.unfiltered() |
|
4350 | 4352 | else: |
|
4351 | 4353 | repo = repoview.repoview(repo, filter) |
|
4352 | 4354 | |
|
4353 | 4355 | repo.branchmap() # make sure we have a relevant, up to date branchmap |
|
4354 | 4356 | |
|
4355 | 4357 | try: |
|
4356 | 4358 | fromfile = branchmap.branchcache.fromfile |
|
4357 | 4359 | except AttributeError: |
|
4358 | 4360 | # older versions |
|
4359 | 4361 | fromfile = branchmap.read |
|
4360 | 4362 | |
|
4361 | 4363 | currentfilter = filter |
|
4362 | 4364 | # try once without timer, the filter may not be cached |
|
4363 | 4365 | while fromfile(repo) is None: |
|
4364 | 4366 | currentfilter = subsettable.get(currentfilter) |
|
4365 | 4367 | if currentfilter is None: |
|
4366 | 4368 | raise error.Abort( |
|
4367 | 4369 | b'No branchmap cached for %s repo' % (filter or b'unfiltered') |
|
4368 | 4370 | ) |
|
4369 | 4371 | repo = repo.filtered(currentfilter) |
|
4370 | 4372 | timer, fm = gettimer(ui, opts) |
|
4371 | 4373 | |
|
4372 | 4374 | def setup(): |
|
4373 | 4375 | if clearrevlogs: |
|
4374 | 4376 | clearchangelog(repo) |
|
4375 | 4377 | |
|
4376 | 4378 | def bench(): |
|
4377 | 4379 | fromfile(repo) |
|
4378 | 4380 | |
|
4379 | 4381 | timer(bench, setup=setup) |
|
4380 | 4382 | fm.end() |
|
4381 | 4383 | |
|
4382 | 4384 | |
|
4383 | 4385 | @command(b'perf::loadmarkers|perfloadmarkers') |
|
4384 | 4386 | def perfloadmarkers(ui, repo): |
|
4385 | 4387 | """benchmark the time to parse the on-disk markers for a repo |
|
4386 | 4388 | |
|
4387 | 4389 | Result is the number of markers in the repo.""" |
|
4388 | 4390 | timer, fm = gettimer(ui) |
|
4389 | 4391 | svfs = getsvfs(repo) |
|
4390 | 4392 | timer(lambda: len(obsolete.obsstore(repo, svfs))) |
|
4391 | 4393 | fm.end() |
|
4392 | 4394 | |
|
4393 | 4395 | |
|
4394 | 4396 | @command( |
|
4395 | 4397 | b'perf::lrucachedict|perflrucachedict', |
|
4396 | 4398 | formatteropts |
|
4397 | 4399 | + [ |
|
4398 | 4400 | (b'', b'costlimit', 0, b'maximum total cost of items in cache'), |
|
4399 | 4401 | (b'', b'mincost', 0, b'smallest cost of items in cache'), |
|
4400 | 4402 | (b'', b'maxcost', 100, b'maximum cost of items in cache'), |
|
4401 | 4403 | (b'', b'size', 4, b'size of cache'), |
|
4402 | 4404 | (b'', b'gets', 10000, b'number of key lookups'), |
|
4403 | 4405 | (b'', b'sets', 10000, b'number of key sets'), |
|
4404 | 4406 | (b'', b'mixed', 10000, b'number of mixed mode operations'), |
|
4405 | 4407 | ( |
|
4406 | 4408 | b'', |
|
4407 | 4409 | b'mixedgetfreq', |
|
4408 | 4410 | 50, |
|
4409 | 4411 | b'frequency of get vs set ops in mixed mode', |
|
4410 | 4412 | ), |
|
4411 | 4413 | ], |
|
4412 | 4414 | norepo=True, |
|
4413 | 4415 | ) |
|
4414 | 4416 | def perflrucache( |
|
4415 | 4417 | ui, |
|
4416 | 4418 | mincost=0, |
|
4417 | 4419 | maxcost=100, |
|
4418 | 4420 | costlimit=0, |
|
4419 | 4421 | size=4, |
|
4420 | 4422 | gets=10000, |
|
4421 | 4423 | sets=10000, |
|
4422 | 4424 | mixed=10000, |
|
4423 | 4425 | mixedgetfreq=50, |
|
4424 | 4426 | **opts |
|
4425 | 4427 | ): |
|
4426 | 4428 | opts = _byteskwargs(opts) |
|
4427 | 4429 | |
|
4428 | 4430 | def doinit(): |
|
4429 | 4431 | for i in _xrange(10000): |
|
4430 | 4432 | util.lrucachedict(size) |
|
4431 | 4433 | |
|
4432 | 4434 | costrange = list(range(mincost, maxcost + 1)) |
|
4433 | 4435 | |
|
4434 | 4436 | values = [] |
|
4435 | 4437 | for i in _xrange(size): |
|
4436 | 4438 | values.append(random.randint(0, _maxint)) |
|
4437 | 4439 | |
|
4438 | 4440 | # Get mode fills the cache and tests raw lookup performance with no |
|
4439 | 4441 | # eviction. |
|
4440 | 4442 | getseq = [] |
|
4441 | 4443 | for i in _xrange(gets): |
|
4442 | 4444 | getseq.append(random.choice(values)) |
|
4443 | 4445 | |
|
4444 | 4446 | def dogets(): |
|
4445 | 4447 | d = util.lrucachedict(size) |
|
4446 | 4448 | for v in values: |
|
4447 | 4449 | d[v] = v |
|
4448 | 4450 | for key in getseq: |
|
4449 | 4451 | value = d[key] |
|
4450 | 4452 | value # silence pyflakes warning |
|
4451 | 4453 | |
|
4452 | 4454 | def dogetscost(): |
|
4453 | 4455 | d = util.lrucachedict(size, maxcost=costlimit) |
|
4454 | 4456 | for i, v in enumerate(values): |
|
4455 | 4457 | d.insert(v, v, cost=costs[i]) |
|
4456 | 4458 | for key in getseq: |
|
4457 | 4459 | try: |
|
4458 | 4460 | value = d[key] |
|
4459 | 4461 | value # silence pyflakes warning |
|
4460 | 4462 | except KeyError: |
|
4461 | 4463 | pass |
|
4462 | 4464 | |
|
4463 | 4465 | # Set mode tests insertion speed with cache eviction. |
|
4464 | 4466 | setseq = [] |
|
4465 | 4467 | costs = [] |
|
4466 | 4468 | for i in _xrange(sets): |
|
4467 | 4469 | setseq.append(random.randint(0, _maxint)) |
|
4468 | 4470 | costs.append(random.choice(costrange)) |
|
4469 | 4471 | |
|
4470 | 4472 | def doinserts(): |
|
4471 | 4473 | d = util.lrucachedict(size) |
|
4472 | 4474 | for v in setseq: |
|
4473 | 4475 | d.insert(v, v) |
|
4474 | 4476 | |
|
4475 | 4477 | def doinsertscost(): |
|
4476 | 4478 | d = util.lrucachedict(size, maxcost=costlimit) |
|
4477 | 4479 | for i, v in enumerate(setseq): |
|
4478 | 4480 | d.insert(v, v, cost=costs[i]) |
|
4479 | 4481 | |
|
4480 | 4482 | def dosets(): |
|
4481 | 4483 | d = util.lrucachedict(size) |
|
4482 | 4484 | for v in setseq: |
|
4483 | 4485 | d[v] = v |
|
4484 | 4486 | |
|
4485 | 4487 | # Mixed mode randomly performs gets and sets with eviction. |
|
4486 | 4488 | mixedops = [] |
|
4487 | 4489 | for i in _xrange(mixed): |
|
4488 | 4490 | r = random.randint(0, 100) |
|
4489 | 4491 | if r < mixedgetfreq: |
|
4490 | 4492 | op = 0 |
|
4491 | 4493 | else: |
|
4492 | 4494 | op = 1 |
|
4493 | 4495 | |
|
4494 | 4496 | mixedops.append( |
|
4495 | 4497 | (op, random.randint(0, size * 2), random.choice(costrange)) |
|
4496 | 4498 | ) |
|
4497 | 4499 | |
|
4498 | 4500 | def domixed(): |
|
4499 | 4501 | d = util.lrucachedict(size) |
|
4500 | 4502 | |
|
4501 | 4503 | for op, v, cost in mixedops: |
|
4502 | 4504 | if op == 0: |
|
4503 | 4505 | try: |
|
4504 | 4506 | d[v] |
|
4505 | 4507 | except KeyError: |
|
4506 | 4508 | pass |
|
4507 | 4509 | else: |
|
4508 | 4510 | d[v] = v |
|
4509 | 4511 | |
|
4510 | 4512 | def domixedcost(): |
|
4511 | 4513 | d = util.lrucachedict(size, maxcost=costlimit) |
|
4512 | 4514 | |
|
4513 | 4515 | for op, v, cost in mixedops: |
|
4514 | 4516 | if op == 0: |
|
4515 | 4517 | try: |
|
4516 | 4518 | d[v] |
|
4517 | 4519 | except KeyError: |
|
4518 | 4520 | pass |
|
4519 | 4521 | else: |
|
4520 | 4522 | d.insert(v, v, cost=cost) |
|
4521 | 4523 | |
|
4522 | 4524 | benches = [ |
|
4523 | 4525 | (doinit, b'init'), |
|
4524 | 4526 | ] |
|
4525 | 4527 | |
|
4526 | 4528 | if costlimit: |
|
4527 | 4529 | benches.extend( |
|
4528 | 4530 | [ |
|
4529 | 4531 | (dogetscost, b'gets w/ cost limit'), |
|
4530 | 4532 | (doinsertscost, b'inserts w/ cost limit'), |
|
4531 | 4533 | (domixedcost, b'mixed w/ cost limit'), |
|
4532 | 4534 | ] |
|
4533 | 4535 | ) |
|
4534 | 4536 | else: |
|
4535 | 4537 | benches.extend( |
|
4536 | 4538 | [ |
|
4537 | 4539 | (dogets, b'gets'), |
|
4538 | 4540 | (doinserts, b'inserts'), |
|
4539 | 4541 | (dosets, b'sets'), |
|
4540 | 4542 | (domixed, b'mixed'), |
|
4541 | 4543 | ] |
|
4542 | 4544 | ) |
|
4543 | 4545 | |
|
4544 | 4546 | for fn, title in benches: |
|
4545 | 4547 | timer, fm = gettimer(ui, opts) |
|
4546 | 4548 | timer(fn, title=title) |
|
4547 | 4549 | fm.end() |
|
4548 | 4550 | |
|
4549 | 4551 | |
|
4550 | 4552 | @command( |
|
4551 | 4553 | b'perf::write|perfwrite', |
|
4552 | 4554 | formatteropts |
|
4553 | 4555 | + [ |
|
4554 | 4556 | (b'', b'write-method', b'write', b'ui write method'), |
|
4555 | 4557 | (b'', b'nlines', 100, b'number of lines'), |
|
4556 | 4558 | (b'', b'nitems', 100, b'number of items (per line)'), |
|
4557 | 4559 | (b'', b'item', b'x', b'item that is written'), |
|
4558 | 4560 | (b'', b'batch-line', None, b'pass whole line to write method at once'), |
|
4559 | 4561 | (b'', b'flush-line', None, b'flush after each line'), |
|
4560 | 4562 | ], |
|
4561 | 4563 | ) |
|
4562 | 4564 | def perfwrite(ui, repo, **opts): |
|
4563 | 4565 | """microbenchmark ui.write (and others)""" |
|
4564 | 4566 | opts = _byteskwargs(opts) |
|
4565 | 4567 | |
|
4566 | 4568 | write = getattr(ui, _sysstr(opts[b'write_method'])) |
|
4567 | 4569 | nlines = int(opts[b'nlines']) |
|
4568 | 4570 | nitems = int(opts[b'nitems']) |
|
4569 | 4571 | item = opts[b'item'] |
|
4570 | 4572 | batch_line = opts.get(b'batch_line') |
|
4571 | 4573 | flush_line = opts.get(b'flush_line') |
|
4572 | 4574 | |
|
4573 | 4575 | if batch_line: |
|
4574 | 4576 | line = item * nitems + b'\n' |
|
4575 | 4577 | |
|
4576 | 4578 | def benchmark(): |
|
4577 | 4579 | for i in pycompat.xrange(nlines): |
|
4578 | 4580 | if batch_line: |
|
4579 | 4581 | write(line) |
|
4580 | 4582 | else: |
|
4581 | 4583 | for i in pycompat.xrange(nitems): |
|
4582 | 4584 | write(item) |
|
4583 | 4585 | write(b'\n') |
|
4584 | 4586 | if flush_line: |
|
4585 | 4587 | ui.flush() |
|
4586 | 4588 | ui.flush() |
|
4587 | 4589 | |
|
4588 | 4590 | timer, fm = gettimer(ui, opts) |
|
4589 | 4591 | timer(benchmark) |
|
4590 | 4592 | fm.end() |
|
4591 | 4593 | |
|
4592 | 4594 | |
|
4593 | 4595 | def uisetup(ui): |
|
4594 | 4596 | if util.safehasattr(cmdutil, b'openrevlog') and not util.safehasattr( |
|
4595 | 4597 | commands, b'debugrevlogopts' |
|
4596 | 4598 | ): |
|
4597 | 4599 | # for "historical portability": |
|
4598 | 4600 | # In this case, Mercurial should be 1.9 (or a79fea6b3e77) - |
|
4599 | 4601 | # 3.7 (or 5606f7d0d063). Therefore, '--dir' option for |
|
4600 | 4602 | # openrevlog() should cause failure, because it has been |
|
4601 | 4603 | # available since 3.5 (or 49c583ca48c4). |
|
4602 | 4604 | def openrevlog(orig, repo, cmd, file_, opts): |
|
4603 | 4605 | if opts.get(b'dir') and not util.safehasattr(repo, b'dirlog'): |
|
4604 | 4606 | raise error.Abort( |
|
4605 | 4607 | b"This version doesn't support --dir option", |
|
4606 | 4608 | hint=b"use 3.5 or later", |
|
4607 | 4609 | ) |
|
4608 | 4610 | return orig(repo, cmd, file_, opts) |
|
4609 | 4611 | |
|
4610 | 4612 | name = _sysstr(b'openrevlog') |
|
4611 | 4613 | extensions.wrapfunction(cmdutil, name, openrevlog) |
|
4612 | 4614 | |
|
4613 | 4615 | |
|
4614 | 4616 | @command( |
|
4615 | 4617 | b'perf::progress|perfprogress', |
|
4616 | 4618 | formatteropts |
|
4617 | 4619 | + [ |
|
4618 | 4620 | (b'', b'topic', b'topic', b'topic for progress messages'), |
|
4619 | 4621 | (b'c', b'total', 1000000, b'total value we are progressing to'), |
|
4620 | 4622 | ], |
|
4621 | 4623 | norepo=True, |
|
4622 | 4624 | ) |
|
4623 | 4625 | def perfprogress(ui, topic=None, total=None, **opts): |
|
4624 | 4626 | """printing of progress bars""" |
|
4625 | 4627 | opts = _byteskwargs(opts) |
|
4626 | 4628 | |
|
4627 | 4629 | timer, fm = gettimer(ui, opts) |
|
4628 | 4630 | |
|
4629 | 4631 | def doprogress(): |
|
4630 | 4632 | with ui.makeprogress(topic, total=total) as progress: |
|
4631 | 4633 | for i in _xrange(total): |
|
4632 | 4634 | progress.increment() |
|
4633 | 4635 | |
|
4634 | 4636 | timer(doprogress) |
|
4635 | 4637 | fm.end() |
@@ -1,3948 +1,3948 b'' | |||
|
1 | 1 | # revlog.py - storage back-end for mercurial |
|
2 | 2 | # coding: utf8 |
|
3 | 3 | # |
|
4 | 4 | # Copyright 2005-2007 Olivia Mackall <olivia@selenic.com> |
|
5 | 5 | # |
|
6 | 6 | # This software may be used and distributed according to the terms of the |
|
7 | 7 | # GNU General Public License version 2 or any later version. |
|
8 | 8 | |
|
9 | 9 | """Storage back-end for Mercurial. |
|
10 | 10 | |
|
11 | 11 | This provides efficient delta storage with O(1) retrieve and append |
|
12 | 12 | and O(changes) merge between branches. |
|
13 | 13 | """ |
|
14 | 14 | |
|
15 | 15 | |
|
16 | 16 | import binascii |
|
17 | 17 | import collections |
|
18 | 18 | import contextlib |
|
19 | 19 | import io |
|
20 | 20 | import os |
|
21 | 21 | import struct |
|
22 | 22 | import weakref |
|
23 | 23 | import zlib |
|
24 | 24 | |
|
25 | 25 | # import stuff from node for others to import from revlog |
|
26 | 26 | from .node import ( |
|
27 | 27 | bin, |
|
28 | 28 | hex, |
|
29 | 29 | nullrev, |
|
30 | 30 | sha1nodeconstants, |
|
31 | 31 | short, |
|
32 | 32 | wdirrev, |
|
33 | 33 | ) |
|
34 | 34 | from .i18n import _ |
|
35 | 35 | from .revlogutils.constants import ( |
|
36 | 36 | ALL_KINDS, |
|
37 | 37 | CHANGELOGV2, |
|
38 | 38 | COMP_MODE_DEFAULT, |
|
39 | 39 | COMP_MODE_INLINE, |
|
40 | 40 | COMP_MODE_PLAIN, |
|
41 | 41 | DELTA_BASE_REUSE_NO, |
|
42 | 42 | DELTA_BASE_REUSE_TRY, |
|
43 | 43 | ENTRY_RANK, |
|
44 | 44 | FEATURES_BY_VERSION, |
|
45 | 45 | FLAG_GENERALDELTA, |
|
46 | 46 | FLAG_INLINE_DATA, |
|
47 | 47 | INDEX_HEADER, |
|
48 | 48 | KIND_CHANGELOG, |
|
49 | 49 | KIND_FILELOG, |
|
50 | 50 | RANK_UNKNOWN, |
|
51 | 51 | REVLOGV0, |
|
52 | 52 | REVLOGV1, |
|
53 | 53 | REVLOGV1_FLAGS, |
|
54 | 54 | REVLOGV2, |
|
55 | 55 | REVLOGV2_FLAGS, |
|
56 | 56 | REVLOG_DEFAULT_FLAGS, |
|
57 | 57 | REVLOG_DEFAULT_FORMAT, |
|
58 | 58 | REVLOG_DEFAULT_VERSION, |
|
59 | 59 | SUPPORTED_FLAGS, |
|
60 | 60 | ) |
|
61 | 61 | from .revlogutils.flagutil import ( |
|
62 | 62 | REVIDX_DEFAULT_FLAGS, |
|
63 | 63 | REVIDX_ELLIPSIS, |
|
64 | 64 | REVIDX_EXTSTORED, |
|
65 | 65 | REVIDX_FLAGS_ORDER, |
|
66 | 66 | REVIDX_HASCOPIESINFO, |
|
67 | 67 | REVIDX_ISCENSORED, |
|
68 | 68 | REVIDX_RAWTEXT_CHANGING_FLAGS, |
|
69 | 69 | ) |
|
70 | 70 | from .thirdparty import attr |
|
71 | 71 | from . import ( |
|
72 | 72 | ancestor, |
|
73 | 73 | dagop, |
|
74 | 74 | error, |
|
75 | 75 | mdiff, |
|
76 | 76 | policy, |
|
77 | 77 | pycompat, |
|
78 | 78 | revlogutils, |
|
79 | 79 | templatefilters, |
|
80 | 80 | util, |
|
81 | 81 | ) |
|
82 | 82 | from .interfaces import ( |
|
83 | 83 | repository, |
|
84 | 84 | util as interfaceutil, |
|
85 | 85 | ) |
|
86 | 86 | from .revlogutils import ( |
|
87 | 87 | deltas as deltautil, |
|
88 | 88 | docket as docketutil, |
|
89 | 89 | flagutil, |
|
90 | 90 | nodemap as nodemaputil, |
|
91 | 91 | randomaccessfile, |
|
92 | 92 | revlogv0, |
|
93 | 93 | rewrite, |
|
94 | 94 | sidedata as sidedatautil, |
|
95 | 95 | ) |
|
96 | 96 | from .utils import ( |
|
97 | 97 | storageutil, |
|
98 | 98 | stringutil, |
|
99 | 99 | ) |
|
100 | 100 | |
|
101 | 101 | # blanked usage of all the name to prevent pyflakes constraints |
|
102 | 102 | # We need these name available in the module for extensions. |
|
103 | 103 | |
|
104 | 104 | REVLOGV0 |
|
105 | 105 | REVLOGV1 |
|
106 | 106 | REVLOGV2 |
|
107 | 107 | CHANGELOGV2 |
|
108 | 108 | FLAG_INLINE_DATA |
|
109 | 109 | FLAG_GENERALDELTA |
|
110 | 110 | REVLOG_DEFAULT_FLAGS |
|
111 | 111 | REVLOG_DEFAULT_FORMAT |
|
112 | 112 | REVLOG_DEFAULT_VERSION |
|
113 | 113 | REVLOGV1_FLAGS |
|
114 | 114 | REVLOGV2_FLAGS |
|
115 | 115 | REVIDX_ISCENSORED |
|
116 | 116 | REVIDX_ELLIPSIS |
|
117 | 117 | REVIDX_HASCOPIESINFO |
|
118 | 118 | REVIDX_EXTSTORED |
|
119 | 119 | REVIDX_DEFAULT_FLAGS |
|
120 | 120 | REVIDX_FLAGS_ORDER |
|
121 | 121 | REVIDX_RAWTEXT_CHANGING_FLAGS |
|
122 | 122 | |
|
123 | 123 | parsers = policy.importmod('parsers') |
|
124 | 124 | rustancestor = policy.importrust('ancestor') |
|
125 | 125 | rustdagop = policy.importrust('dagop') |
|
126 | 126 | rustrevlog = policy.importrust('revlog') |
|
127 | 127 | |
|
128 | 128 | # Aliased for performance. |
|
129 | 129 | _zlibdecompress = zlib.decompress |
|
130 | 130 | |
|
131 | 131 | # max size of inline data embedded into a revlog |
|
132 | 132 | _maxinline = 131072 |
|
133 | 133 | |
|
134 | 134 | # Flag processors for REVIDX_ELLIPSIS. |
|
135 | 135 | def ellipsisreadprocessor(rl, text): |
|
136 | 136 | return text, False |
|
137 | 137 | |
|
138 | 138 | |
|
139 | 139 | def ellipsiswriteprocessor(rl, text): |
|
140 | 140 | return text, False |
|
141 | 141 | |
|
142 | 142 | |
|
143 | 143 | def ellipsisrawprocessor(rl, text): |
|
144 | 144 | return False |
|
145 | 145 | |
|
146 | 146 | |
|
147 | 147 | ellipsisprocessor = ( |
|
148 | 148 | ellipsisreadprocessor, |
|
149 | 149 | ellipsiswriteprocessor, |
|
150 | 150 | ellipsisrawprocessor, |
|
151 | 151 | ) |
|
152 | 152 | |
|
153 | 153 | |
|
154 | 154 | def _verify_revision(rl, skipflags, state, node): |
|
155 | 155 | """Verify the integrity of the given revlog ``node`` while providing a hook |
|
156 | 156 | point for extensions to influence the operation.""" |
|
157 | 157 | if skipflags: |
|
158 | 158 | state[b'skipread'].add(node) |
|
159 | 159 | else: |
|
160 | 160 | # Side-effect: read content and verify hash. |
|
161 | 161 | rl.revision(node) |
|
162 | 162 | |
|
163 | 163 | |
|
164 | 164 | # True if a fast implementation for persistent-nodemap is available |
|
165 | 165 | # |
|
166 | 166 | # We also consider we have a "fast" implementation in "pure" python because |
|
167 | 167 | # people using pure don't really have performance consideration (and a |
|
168 | 168 | # wheelbarrow of other slowness source) |
|
169 | 169 | HAS_FAST_PERSISTENT_NODEMAP = rustrevlog is not None or hasattr( |
|
170 | 170 | parsers, 'BaseIndexObject' |
|
171 | 171 | ) |
|
172 | 172 | |
|
173 | 173 | |
|
174 | 174 | @interfaceutil.implementer(repository.irevisiondelta) |
|
175 | 175 | @attr.s(slots=True) |
|
176 | 176 | class revlogrevisiondelta: |
|
177 | 177 | node = attr.ib() |
|
178 | 178 | p1node = attr.ib() |
|
179 | 179 | p2node = attr.ib() |
|
180 | 180 | basenode = attr.ib() |
|
181 | 181 | flags = attr.ib() |
|
182 | 182 | baserevisionsize = attr.ib() |
|
183 | 183 | revision = attr.ib() |
|
184 | 184 | delta = attr.ib() |
|
185 | 185 | sidedata = attr.ib() |
|
186 | 186 | protocol_flags = attr.ib() |
|
187 | 187 | linknode = attr.ib(default=None) |
|
188 | 188 | |
|
189 | 189 | |
|
190 | 190 | @interfaceutil.implementer(repository.iverifyproblem) |
|
191 | 191 | @attr.s(frozen=True) |
|
192 | 192 | class revlogproblem: |
|
193 | 193 | warning = attr.ib(default=None) |
|
194 | 194 | error = attr.ib(default=None) |
|
195 | 195 | node = attr.ib(default=None) |
|
196 | 196 | |
|
197 | 197 | |
|
198 | 198 | def parse_index_v1(data, inline): |
|
199 | 199 | # call the C implementation to parse the index data |
|
200 | 200 | index, cache = parsers.parse_index2(data, inline) |
|
201 | 201 | return index, cache |
|
202 | 202 | |
|
203 | 203 | |
|
204 | 204 | def parse_index_v2(data, inline): |
|
205 | 205 | # call the C implementation to parse the index data |
|
206 | 206 | index, cache = parsers.parse_index2(data, inline, format=REVLOGV2) |
|
207 | 207 | return index, cache |
|
208 | 208 | |
|
209 | 209 | |
|
210 | 210 | def parse_index_cl_v2(data, inline): |
|
211 | 211 | # call the C implementation to parse the index data |
|
212 | 212 | index, cache = parsers.parse_index2(data, inline, format=CHANGELOGV2) |
|
213 | 213 | return index, cache |
|
214 | 214 | |
|
215 | 215 | |
|
216 | 216 | if hasattr(parsers, 'parse_index_devel_nodemap'): |
|
217 | 217 | |
|
218 | 218 | def parse_index_v1_nodemap(data, inline): |
|
219 | 219 | index, cache = parsers.parse_index_devel_nodemap(data, inline) |
|
220 | 220 | return index, cache |
|
221 | 221 | |
|
222 | 222 | |
|
223 | 223 | else: |
|
224 | 224 | parse_index_v1_nodemap = None |
|
225 | 225 | |
|
226 | 226 | |
|
227 | 227 | def parse_index_v1_mixed(data, inline): |
|
228 | 228 | index, cache = parse_index_v1(data, inline) |
|
229 | 229 | return rustrevlog.MixedIndex(index), cache |
|
230 | 230 | |
|
231 | 231 | |
|
232 | 232 | # corresponds to uncompressed length of indexformatng (2 gigs, 4-byte |
|
233 | 233 | # signed integer) |
|
234 | 234 | _maxentrysize = 0x7FFFFFFF |
|
235 | 235 | |
|
236 | 236 | FILE_TOO_SHORT_MSG = _( |
|
237 | 237 | b'cannot read from revlog %s;' |
|
238 | 238 | b' expected %d bytes from offset %d, data size is %d' |
|
239 | 239 | ) |
|
240 | 240 | |
|
241 | 241 | hexdigits = b'0123456789abcdefABCDEF' |
|
242 | 242 | |
|
243 | 243 | |
|
244 | 244 | class _Config: |
|
245 | 245 | def copy(self): |
|
246 | 246 | return self.__class__(**self.__dict__) |
|
247 | 247 | |
|
248 | 248 | |
|
249 | 249 | @attr.s() |
|
250 | 250 | class FeatureConfig(_Config): |
|
251 | 251 | """Hold configuration values about the available revlog features""" |
|
252 | 252 | |
|
253 | 253 | # the default compression engine |
|
254 | 254 | compression_engine = attr.ib(default=b'zlib') |
|
255 | 255 | # compression engines options |
|
256 | 256 | compression_engine_options = attr.ib(default=attr.Factory(dict)) |
|
257 | 257 | |
|
258 | 258 | # can we use censor on this revlog |
|
259 | 259 | censorable = attr.ib(default=False) |
|
260 | 260 | # does this revlog use the "side data" feature |
|
261 | 261 | has_side_data = attr.ib(default=False) |
|
262 | 262 | # might remove rank configuration once the computation has no impact |
|
263 | 263 | compute_rank = attr.ib(default=False) |
|
264 | 264 | # parent order is supposed to be semantically irrelevant, so we |
|
265 | 265 | # normally resort parents to ensure that the first parent is non-null, |
|
266 | 266 | # if there is a non-null parent at all. |
|
267 | 267 | # filelog abuses the parent order as flag to mark some instances of |
|
268 | 268 | # meta-encoded files, so allow it to disable this behavior. |
|
269 | 269 | canonical_parent_order = attr.ib(default=False) |
|
270 | 270 | # can ellipsis commit be used |
|
271 | 271 | enable_ellipsis = attr.ib(default=False) |
|
272 | 272 | |
|
273 | 273 | def copy(self): |
|
274 | 274 | new = super().copy() |
|
275 | 275 | new.compression_engine_options = self.compression_engine_options.copy() |
|
276 | 276 | return new |
|
277 | 277 | |
|
278 | 278 | |
|
279 | 279 | @attr.s() |
|
280 | 280 | class DataConfig(_Config): |
|
281 | 281 | """Hold configuration value about how the revlog data are read""" |
|
282 | 282 | |
|
283 | 283 | # should we try to open the "pending" version of the revlog |
|
284 | 284 | try_pending = attr.ib(default=False) |
|
285 | 285 | # should we try to open the "splitted" version of the revlog |
|
286 | 286 | try_split = attr.ib(default=False) |
|
287 | 287 | # When True, indexfile should be opened with checkambig=True at writing, |
|
288 | 288 | # to avoid file stat ambiguity. |
|
289 | 289 | check_ambig = attr.ib(default=False) |
|
290 | 290 | |
|
291 | 291 | # If true, use mmap instead of reading to deal with large index |
|
292 | 292 | mmap_large_index = attr.ib(default=False) |
|
293 | 293 | # how much data is large |
|
294 | 294 | mmap_index_threshold = attr.ib(default=None) |
|
295 | 295 | # How much data to read and cache into the raw revlog data cache. |
|
296 | 296 | chunk_cache_size = attr.ib(default=65536) |
|
297 | 297 | |
|
298 | 298 | # Allow sparse reading of the revlog data |
|
299 | 299 | with_sparse_read = attr.ib(default=False) |
|
300 | 300 | # minimal density of a sparse read chunk |
|
301 | 301 | sr_density_threshold = attr.ib(default=0.50) |
|
302 | 302 | # minimal size of data we skip when performing sparse read |
|
303 | 303 | sr_min_gap_size = attr.ib(default=262144) |
|
304 | 304 | |
|
305 | 305 | # are delta encoded against arbitrary bases. |
|
306 | 306 | generaldelta = attr.ib(default=False) |
|
307 | 307 | |
|
308 | 308 | |
|
309 | 309 | @attr.s() |
|
310 | 310 | class DeltaConfig(_Config): |
|
311 | 311 | """Hold configuration value about how new delta are computed |
|
312 | 312 | |
|
313 | 313 | Some attributes are duplicated from DataConfig to help havign each object |
|
314 | 314 | self contained. |
|
315 | 315 | """ |
|
316 | 316 | |
|
317 | 317 | # can delta be encoded against arbitrary bases. |
|
318 | 318 | general_delta = attr.ib(default=False) |
|
319 | 319 | # Allow sparse writing of the revlog data |
|
320 | 320 | sparse_revlog = attr.ib(default=False) |
|
321 | 321 | # maximum length of a delta chain |
|
322 | 322 | max_chain_len = attr.ib(default=None) |
|
323 | 323 | # Maximum distance between delta chain base start and end |
|
324 | 324 | max_deltachain_span = attr.ib(default=-1) |
|
325 | 325 | # If `upper_bound_comp` is not None, this is the expected maximal gain from |
|
326 | 326 | # compression for the data content. |
|
327 | 327 | upper_bound_comp = attr.ib(default=None) |
|
328 | 328 | # Should we try a delta against both parent |
|
329 | 329 | delta_both_parents = attr.ib(default=True) |
|
330 | 330 | # Test delta base candidate group by chunk of this maximal size. |
|
331 | 331 | candidate_group_chunk_size = attr.ib(default=0) |
|
332 | 332 | # Should we display debug information about delta computation |
|
333 | 333 | debug_delta = attr.ib(default=False) |
|
334 | 334 | # trust incoming delta by default |
|
335 | 335 | lazy_delta = attr.ib(default=True) |
|
336 | 336 | # trust the base of incoming delta by default |
|
337 | 337 | lazy_delta_base = attr.ib(default=False) |
|
338 | 338 | |
|
339 | 339 | |
|
340 | 340 | class _InnerRevlog: |
|
341 | 341 | """An inner layer of the revlog object |
|
342 | 342 | |
|
343 | 343 | That layer exist to be able to delegate some operation to Rust, its |
|
344 | 344 | boundaries are arbitrary and based on what we can delegate to Rust. |
|
345 | 345 | """ |
|
346 | 346 | |
|
347 | 347 | def __init__( |
|
348 | 348 | self, |
|
349 | 349 | opener, |
|
350 | 350 | index, |
|
351 | 351 | index_file, |
|
352 | 352 | data_file, |
|
353 | 353 | sidedata_file, |
|
354 | 354 | inline, |
|
355 | 355 | data_config, |
|
356 | 356 | feature_config, |
|
357 | 357 | chunk_cache, |
|
358 | 358 | default_compression_header, |
|
359 | 359 | ): |
|
360 | 360 | self.opener = opener |
|
361 | 361 | self.index = index |
|
362 | 362 | |
|
363 | 363 | self.__index_file = index_file |
|
364 | 364 | self.data_file = data_file |
|
365 | 365 | self.sidedata_file = sidedata_file |
|
366 | 366 | self.inline = inline |
|
367 | 367 | self.data_config = data_config |
|
368 | 368 | self.feature_config = feature_config |
|
369 | 369 | |
|
370 | 370 | self._default_compression_header = default_compression_header |
|
371 | 371 | |
|
372 | 372 | # index |
|
373 | 373 | |
|
374 | 374 | # 3-tuple of file handles being used for active writing. |
|
375 | 375 | self._writinghandles = None |
|
376 | 376 | |
|
377 | 377 | self._segmentfile = randomaccessfile.randomaccessfile( |
|
378 | 378 | self.opener, |
|
379 | 379 | (self.index_file if self.inline else self.data_file), |
|
380 | 380 | self.data_config.chunk_cache_size, |
|
381 | 381 | chunk_cache, |
|
382 | 382 | ) |
|
383 | 383 | self._segmentfile_sidedata = randomaccessfile.randomaccessfile( |
|
384 | 384 | self.opener, |
|
385 | 385 | self.sidedata_file, |
|
386 | 386 | self.data_config.chunk_cache_size, |
|
387 | 387 | ) |
|
388 | 388 | |
|
389 | 389 | # revlog header -> revlog compressor |
|
390 | 390 | self._decompressors = {} |
|
391 | 391 | |
|
392 | 392 | @property |
|
393 | 393 | def index_file(self): |
|
394 | 394 | return self.__index_file |
|
395 | 395 | |
|
396 | 396 | @index_file.setter |
|
397 | 397 | def index_file(self, new_index_file): |
|
398 | 398 | self.__index_file = new_index_file |
|
399 | 399 | if self.inline: |
|
400 | 400 | self._segmentfile.filename = new_index_file |
|
401 | 401 | |
|
402 | 402 | # Derived from index values. |
|
403 | 403 | |
|
404 | 404 | def start(self, rev): |
|
405 | 405 | """the offset of the data chunk for this revision""" |
|
406 | 406 | return int(self.index[rev][0] >> 16) |
|
407 | 407 | |
|
408 | 408 | def length(self, rev): |
|
409 | 409 | """the length of the data chunk for this revision""" |
|
410 | 410 | return self.index[rev][1] |
|
411 | 411 | |
|
412 | 412 | def end(self, rev): |
|
413 | 413 | """the end of the data chunk for this revision""" |
|
414 | 414 | return self.start(rev) + self.length(rev) |
|
415 | 415 | |
|
416 | 416 | @util.propertycache |
|
417 | 417 | def _compressor(self): |
|
418 | 418 | engine = util.compengines[self.feature_config.compression_engine] |
|
419 | 419 | return engine.revlogcompressor( |
|
420 | 420 | self.feature_config.compression_engine_options |
|
421 | 421 | ) |
|
422 | 422 | |
|
423 | 423 | @util.propertycache |
|
424 | 424 | def _decompressor(self): |
|
425 | 425 | """the default decompressor""" |
|
426 | 426 | if self._default_compression_header is None: |
|
427 | 427 | return None |
|
428 | 428 | t = self._default_compression_header |
|
429 | 429 | c = self._get_decompressor(t) |
|
430 | 430 | return c.decompress |
|
431 | 431 | |
|
432 | 432 | def _get_decompressor(self, t): |
|
433 | 433 | try: |
|
434 | 434 | compressor = self._decompressors[t] |
|
435 | 435 | except KeyError: |
|
436 | 436 | try: |
|
437 | 437 | engine = util.compengines.forrevlogheader(t) |
|
438 | 438 | compressor = engine.revlogcompressor( |
|
439 | 439 | self.feature_config.compression_engine_options |
|
440 | 440 | ) |
|
441 | 441 | self._decompressors[t] = compressor |
|
442 | 442 | except KeyError: |
|
443 | 443 | raise error.RevlogError( |
|
444 | 444 | _(b'unknown compression type %s') % binascii.hexlify(t) |
|
445 | 445 | ) |
|
446 | 446 | return compressor |
|
447 | 447 | |
|
448 | 448 | def compress(self, data): |
|
449 | 449 | """Generate a possibly-compressed representation of data.""" |
|
450 | 450 | if not data: |
|
451 | 451 | return b'', data |
|
452 | 452 | |
|
453 | 453 | compressed = self._compressor.compress(data) |
|
454 | 454 | |
|
455 | 455 | if compressed: |
|
456 | 456 | # The revlog compressor added the header in the returned data. |
|
457 | 457 | return b'', compressed |
|
458 | 458 | |
|
459 | 459 | if data[0:1] == b'\0': |
|
460 | 460 | return b'', data |
|
461 | 461 | return b'u', data |
|
462 | 462 | |
|
463 | 463 | def decompress(self, data): |
|
464 | 464 | """Decompress a revlog chunk. |
|
465 | 465 | |
|
466 | 466 | The chunk is expected to begin with a header identifying the |
|
467 | 467 | format type so it can be routed to an appropriate decompressor. |
|
468 | 468 | """ |
|
469 | 469 | if not data: |
|
470 | 470 | return data |
|
471 | 471 | |
|
472 | 472 | # Revlogs are read much more frequently than they are written and many |
|
473 | 473 | # chunks only take microseconds to decompress, so performance is |
|
474 | 474 | # important here. |
|
475 | 475 | # |
|
476 | 476 | # We can make a few assumptions about revlogs: |
|
477 | 477 | # |
|
478 | 478 | # 1) the majority of chunks will be compressed (as opposed to inline |
|
479 | 479 | # raw data). |
|
480 | 480 | # 2) decompressing *any* data will likely by at least 10x slower than |
|
481 | 481 | # returning raw inline data. |
|
482 | 482 | # 3) we want to prioritize common and officially supported compression |
|
483 | 483 | # engines |
|
484 | 484 | # |
|
485 | 485 | # It follows that we want to optimize for "decompress compressed data |
|
486 | 486 | # when encoded with common and officially supported compression engines" |
|
487 | 487 | # case over "raw data" and "data encoded by less common or non-official |
|
488 | 488 | # compression engines." That is why we have the inline lookup first |
|
489 | 489 | # followed by the compengines lookup. |
|
490 | 490 | # |
|
491 | 491 | # According to `hg perfrevlogchunks`, this is ~0.5% faster for zlib |
|
492 | 492 | # compressed chunks. And this matters for changelog and manifest reads. |
|
493 | 493 | t = data[0:1] |
|
494 | 494 | |
|
495 | 495 | if t == b'x': |
|
496 | 496 | try: |
|
497 | 497 | return _zlibdecompress(data) |
|
498 | 498 | except zlib.error as e: |
|
499 | 499 | raise error.RevlogError( |
|
500 | 500 | _(b'revlog decompress error: %s') |
|
501 | 501 | % stringutil.forcebytestr(e) |
|
502 | 502 | ) |
|
503 | 503 | # '\0' is more common than 'u' so it goes first. |
|
504 | 504 | elif t == b'\0': |
|
505 | 505 | return data |
|
506 | 506 | elif t == b'u': |
|
507 | 507 | return util.buffer(data, 1) |
|
508 | 508 | |
|
509 | 509 | compressor = self._get_decompressor(t) |
|
510 | 510 | |
|
511 | 511 | return compressor.decompress(data) |
|
512 | 512 | |
|
513 | 513 | @contextlib.contextmanager |
|
514 | 514 | def reading(self): |
|
515 | 515 | """Context manager that keeps data and sidedata files open for reading""" |
|
516 | 516 | if len(self.index) == 0: |
|
517 | 517 | yield # nothing to be read |
|
518 | 518 | else: |
|
519 | 519 | with self._segmentfile.reading(): |
|
520 | 520 | with self._segmentfile_sidedata.reading(): |
|
521 | 521 | yield |
|
522 | 522 | |
|
523 | 523 | @property |
|
524 | 524 | def is_writing(self): |
|
525 | 525 | """True is a writing context is open""" |
|
526 | 526 | return self._writinghandles is not None |
|
527 | 527 | |
|
528 | 528 | @contextlib.contextmanager |
|
529 | 529 | def writing(self, transaction, data_end=None, sidedata_end=None): |
|
530 | 530 | """Open the revlog files for writing |
|
531 | 531 | |
|
532 | 532 | Add content to a revlog should be done within such context. |
|
533 | 533 | """ |
|
534 | 534 | if self.is_writing: |
|
535 | 535 | yield |
|
536 | 536 | else: |
|
537 | 537 | ifh = dfh = sdfh = None |
|
538 | 538 | try: |
|
539 | 539 | r = len(self.index) |
|
540 | 540 | # opening the data file. |
|
541 | 541 | dsize = 0 |
|
542 | 542 | if r: |
|
543 | 543 | dsize = self.end(r - 1) |
|
544 | 544 | dfh = None |
|
545 | 545 | if not self.inline: |
|
546 | 546 | try: |
|
547 | 547 | dfh = self.opener(self.data_file, mode=b"r+") |
|
548 | 548 | if data_end is None: |
|
549 | 549 | dfh.seek(0, os.SEEK_END) |
|
550 | 550 | else: |
|
551 | 551 | dfh.seek(data_end, os.SEEK_SET) |
|
552 | 552 | except FileNotFoundError: |
|
553 | 553 | dfh = self.opener(self.data_file, mode=b"w+") |
|
554 | 554 | transaction.add(self.data_file, dsize) |
|
555 | 555 | if self.sidedata_file is not None: |
|
556 | 556 | assert sidedata_end is not None |
|
557 | 557 | # revlog-v2 does not inline, help Pytype |
|
558 | 558 | assert dfh is not None |
|
559 | 559 | try: |
|
560 | 560 | sdfh = self.opener(self.sidedata_file, mode=b"r+") |
|
561 | 561 | dfh.seek(sidedata_end, os.SEEK_SET) |
|
562 | 562 | except FileNotFoundError: |
|
563 | 563 | sdfh = self.opener(self.sidedata_file, mode=b"w+") |
|
564 | 564 | transaction.add(self.sidedata_file, sidedata_end) |
|
565 | 565 | |
|
566 | 566 | # opening the index file. |
|
567 | 567 | isize = r * self.index.entry_size |
|
568 | 568 | ifh = self.__index_write_fp() |
|
569 | 569 | if self.inline: |
|
570 | 570 | transaction.add(self.index_file, dsize + isize) |
|
571 | 571 | else: |
|
572 | 572 | transaction.add(self.index_file, isize) |
|
573 | 573 | # exposing all file handle for writing. |
|
574 | 574 | self._writinghandles = (ifh, dfh, sdfh) |
|
575 | 575 | self._segmentfile.writing_handle = ifh if self.inline else dfh |
|
576 | 576 | self._segmentfile_sidedata.writing_handle = sdfh |
|
577 | 577 | yield |
|
578 | 578 | finally: |
|
579 | 579 | self._writinghandles = None |
|
580 | 580 | self._segmentfile.writing_handle = None |
|
581 | 581 | self._segmentfile_sidedata.writing_handle = None |
|
582 | 582 | if dfh is not None: |
|
583 | 583 | dfh.close() |
|
584 | 584 | if sdfh is not None: |
|
585 | 585 | sdfh.close() |
|
586 | 586 | # closing the index file last to avoid exposing referent to |
|
587 | 587 | # potential unflushed data content. |
|
588 | 588 | if ifh is not None: |
|
589 | 589 | ifh.close() |
|
590 | 590 | |
|
591 | 591 | def __index_write_fp(self, index_end=None): |
|
592 | 592 | """internal method to open the index file for writing |
|
593 | 593 | |
|
594 | 594 | You should not use this directly and use `_writing` instead |
|
595 | 595 | """ |
|
596 | 596 | try: |
|
597 | 597 | f = self.opener( |
|
598 | 598 | self.index_file, |
|
599 | 599 | mode=b"r+", |
|
600 | 600 | checkambig=self.data_config.check_ambig, |
|
601 | 601 | ) |
|
602 | 602 | if index_end is None: |
|
603 | 603 | f.seek(0, os.SEEK_END) |
|
604 | 604 | else: |
|
605 | 605 | f.seek(index_end, os.SEEK_SET) |
|
606 | 606 | return f |
|
607 | 607 | except FileNotFoundError: |
|
608 | 608 | return self.opener( |
|
609 | 609 | self.index_file, |
|
610 | 610 | mode=b"w+", |
|
611 | 611 | checkambig=self.data_config.check_ambig, |
|
612 | 612 | ) |
|
613 | 613 | |
|
614 | 614 | def __index_new_fp(self): |
|
615 | 615 | """internal method to create a new index file for writing |
|
616 | 616 | |
|
617 | 617 | You should not use this unless you are upgrading from inline revlog |
|
618 | 618 | """ |
|
619 | 619 | return self.opener( |
|
620 | 620 | self.index_file, |
|
621 | 621 | mode=b"w", |
|
622 | 622 | checkambig=self.data_config.check_ambig, |
|
623 | 623 | atomictemp=True, |
|
624 | 624 | ) |
|
625 | 625 | |
|
626 | 626 | def split_inline(self, tr, header, new_index_file_path=None): |
|
627 | 627 | """split the data of an inline revlog into an index and a data file""" |
|
628 | 628 | existing_handles = False |
|
629 | 629 | if self._writinghandles is not None: |
|
630 | 630 | existing_handles = True |
|
631 | 631 | fp = self._writinghandles[0] |
|
632 | 632 | fp.flush() |
|
633 | 633 | fp.close() |
|
634 | 634 | # We can't use the cached file handle after close(). So prevent |
|
635 | 635 | # its usage. |
|
636 | 636 | self._writinghandles = None |
|
637 | 637 | self._segmentfile.writing_handle = None |
|
638 | 638 | # No need to deal with sidedata writing handle as it is only |
|
639 | 639 | # relevant with revlog-v2 which is never inline, not reaching |
|
640 | 640 | # this code |
|
641 | 641 | |
|
642 | 642 | new_dfh = self.opener(self.data_file, mode=b"w+") |
|
643 | 643 | new_dfh.truncate(0) # drop any potentially existing data |
|
644 | 644 | try: |
|
645 | 645 | with self.reading(): |
|
646 | 646 | for r in range(len(self.index)): |
|
647 | 647 | new_dfh.write(self.get_segment_for_revs(r, r)[1]) |
|
648 | 648 | new_dfh.flush() |
|
649 | 649 | |
|
650 | 650 | if new_index_file_path is not None: |
|
651 | 651 | self.index_file = new_index_file_path |
|
652 | 652 | with self.__index_new_fp() as fp: |
|
653 | 653 | self.inline = False |
|
654 | 654 | for i in range(len(self.index)): |
|
655 | 655 | e = self.index.entry_binary(i) |
|
656 | 656 | if i == 0: |
|
657 | 657 | packed_header = self.index.pack_header(header) |
|
658 | 658 | e = packed_header + e |
|
659 | 659 | fp.write(e) |
|
660 | 660 | |
|
661 | 661 | # If we don't use side-write, the temp file replace the real |
|
662 | 662 | # index when we exit the context manager |
|
663 | 663 | |
|
664 | 664 | self._segmentfile = randomaccessfile.randomaccessfile( |
|
665 | 665 | self.opener, |
|
666 | 666 | self.data_file, |
|
667 | 667 | self.data_config.chunk_cache_size, |
|
668 | 668 | ) |
|
669 | 669 | |
|
670 | 670 | if existing_handles: |
|
671 | 671 | # switched from inline to conventional reopen the index |
|
672 | 672 | ifh = self.__index_write_fp() |
|
673 | 673 | self._writinghandles = (ifh, new_dfh, None) |
|
674 | 674 | self._segmentfile.writing_handle = new_dfh |
|
675 | 675 | new_dfh = None |
|
676 | 676 | # No need to deal with sidedata writing handle as it is only |
|
677 | 677 | # relevant with revlog-v2 which is never inline, not reaching |
|
678 | 678 | # this code |
|
679 | 679 | finally: |
|
680 | 680 | if new_dfh is not None: |
|
681 | 681 | new_dfh.close() |
|
682 | 682 | return self.index_file |
|
683 | 683 | |
|
684 | 684 | def get_segment_for_revs(self, startrev, endrev): |
|
685 | 685 | """Obtain a segment of raw data corresponding to a range of revisions. |
|
686 | 686 | |
|
687 | 687 | Accepts the start and end revisions and an optional already-open |
|
688 | 688 | file handle to be used for reading. If the file handle is read, its |
|
689 | 689 | seek position will not be preserved. |
|
690 | 690 | |
|
691 | 691 | Requests for data may be satisfied by a cache. |
|
692 | 692 | |
|
693 | 693 | Returns a 2-tuple of (offset, data) for the requested range of |
|
694 | 694 | revisions. Offset is the integer offset from the beginning of the |
|
695 | 695 | revlog and data is a str or buffer of the raw byte data. |
|
696 | 696 | |
|
697 | 697 | Callers will need to call ``self.start(rev)`` and ``self.length(rev)`` |
|
698 | 698 | to determine where each revision's data begins and ends. |
|
699 | 699 | |
|
700 | 700 | API: we should consider making this a private part of the InnerRevlog |
|
701 | 701 | at some point. |
|
702 | 702 | """ |
|
703 | 703 | # Inlined self.start(startrev) & self.end(endrev) for perf reasons |
|
704 | 704 | # (functions are expensive). |
|
705 | 705 | index = self.index |
|
706 | 706 | istart = index[startrev] |
|
707 | 707 | start = int(istart[0] >> 16) |
|
708 | 708 | if startrev == endrev: |
|
709 | 709 | end = start + istart[1] |
|
710 | 710 | else: |
|
711 | 711 | iend = index[endrev] |
|
712 | 712 | end = int(iend[0] >> 16) + iend[1] |
|
713 | 713 | |
|
714 | 714 | if self.inline: |
|
715 | 715 | start += (startrev + 1) * self.index.entry_size |
|
716 | 716 | end += (endrev + 1) * self.index.entry_size |
|
717 | 717 | length = end - start |
|
718 | 718 | |
|
719 | 719 | return start, self._segmentfile.read_chunk(start, length) |
|
720 | 720 | |
|
721 | def _chunk(self, rev): | |
|
722 | """Obtain a single decompressed chunk for a revision. | |
|
723 | ||
|
724 | Accepts an integer revision and an optional already-open file handle | |
|
725 | to be used for reading. If used, the seek position of the file will not | |
|
726 | be preserved. | |
|
727 | ||
|
728 | Returns a str holding uncompressed data for the requested revision. | |
|
729 | """ | |
|
730 | compression_mode = self.index[rev][10] | |
|
731 | data = self.get_segment_for_revs(rev, rev)[1] | |
|
732 | if compression_mode == COMP_MODE_PLAIN: | |
|
733 | return data | |
|
734 | elif compression_mode == COMP_MODE_DEFAULT: | |
|
735 | return self._decompressor(data) | |
|
736 | elif compression_mode == COMP_MODE_INLINE: | |
|
737 | return self.decompress(data) | |
|
738 | else: | |
|
739 | msg = b'unknown compression mode %d' | |
|
740 | msg %= compression_mode | |
|
741 | raise error.RevlogError(msg) | |
|
742 | ||
|
721 | 743 | |
|
722 | 744 | class revlog: |
|
723 | 745 | """ |
|
724 | 746 | the underlying revision storage object |
|
725 | 747 | |
|
726 | 748 | A revlog consists of two parts, an index and the revision data. |
|
727 | 749 | |
|
728 | 750 | The index is a file with a fixed record size containing |
|
729 | 751 | information on each revision, including its nodeid (hash), the |
|
730 | 752 | nodeids of its parents, the position and offset of its data within |
|
731 | 753 | the data file, and the revision it's based on. Finally, each entry |
|
732 | 754 | contains a linkrev entry that can serve as a pointer to external |
|
733 | 755 | data. |
|
734 | 756 | |
|
735 | 757 | The revision data itself is a linear collection of data chunks. |
|
736 | 758 | Each chunk represents a revision and is usually represented as a |
|
737 | 759 | delta against the previous chunk. To bound lookup time, runs of |
|
738 | 760 | deltas are limited to about 2 times the length of the original |
|
739 | 761 | version data. This makes retrieval of a version proportional to |
|
740 | 762 | its size, or O(1) relative to the number of revisions. |
|
741 | 763 | |
|
742 | 764 | Both pieces of the revlog are written to in an append-only |
|
743 | 765 | fashion, which means we never need to rewrite a file to insert or |
|
744 | 766 | remove data, and can use some simple techniques to avoid the need |
|
745 | 767 | for locking while reading. |
|
746 | 768 | |
|
747 | 769 | If checkambig, indexfile is opened with checkambig=True at |
|
748 | 770 | writing, to avoid file stat ambiguity. |
|
749 | 771 | |
|
750 | 772 | If mmaplargeindex is True, and an mmapindexthreshold is set, the |
|
751 | 773 | index will be mmapped rather than read if it is larger than the |
|
752 | 774 | configured threshold. |
|
753 | 775 | |
|
754 | 776 | If censorable is True, the revlog can have censored revisions. |
|
755 | 777 | |
|
756 | 778 | If `upperboundcomp` is not None, this is the expected maximal gain from |
|
757 | 779 | compression for the data content. |
|
758 | 780 | |
|
759 | 781 | `concurrencychecker` is an optional function that receives 3 arguments: a |
|
760 | 782 | file handle, a filename, and an expected position. It should check whether |
|
761 | 783 | the current position in the file handle is valid, and log/warn/fail (by |
|
762 | 784 | raising). |
|
763 | 785 | |
|
764 | 786 | See mercurial/revlogutils/contants.py for details about the content of an |
|
765 | 787 | index entry. |
|
766 | 788 | """ |
|
767 | 789 | |
|
768 | 790 | _flagserrorclass = error.RevlogError |
|
769 | 791 | |
|
770 | 792 | @staticmethod |
|
771 | 793 | def is_inline_index(header_bytes): |
|
772 | 794 | """Determine if a revlog is inline from the initial bytes of the index""" |
|
773 | 795 | header = INDEX_HEADER.unpack(header_bytes)[0] |
|
774 | 796 | |
|
775 | 797 | _format_flags = header & ~0xFFFF |
|
776 | 798 | _format_version = header & 0xFFFF |
|
777 | 799 | |
|
778 | 800 | features = FEATURES_BY_VERSION[_format_version] |
|
779 | 801 | return features[b'inline'](_format_flags) |
|
780 | 802 | |
|
781 | 803 | def __init__( |
|
782 | 804 | self, |
|
783 | 805 | opener, |
|
784 | 806 | target, |
|
785 | 807 | radix, |
|
786 | 808 | postfix=None, # only exist for `tmpcensored` now |
|
787 | 809 | checkambig=False, |
|
788 | 810 | mmaplargeindex=False, |
|
789 | 811 | censorable=False, |
|
790 | 812 | upperboundcomp=None, |
|
791 | 813 | persistentnodemap=False, |
|
792 | 814 | concurrencychecker=None, |
|
793 | 815 | trypending=False, |
|
794 | 816 | try_split=False, |
|
795 | 817 | canonical_parent_order=True, |
|
796 | 818 | ): |
|
797 | 819 | """ |
|
798 | 820 | create a revlog object |
|
799 | 821 | |
|
800 | 822 | opener is a function that abstracts the file opening operation |
|
801 | 823 | and can be used to implement COW semantics or the like. |
|
802 | 824 | |
|
803 | 825 | `target`: a (KIND, ID) tuple that identify the content stored in |
|
804 | 826 | this revlog. It help the rest of the code to understand what the revlog |
|
805 | 827 | is about without having to resort to heuristic and index filename |
|
806 | 828 | analysis. Note: that this must be reliably be set by normal code, but |
|
807 | 829 | that test, debug, or performance measurement code might not set this to |
|
808 | 830 | accurate value. |
|
809 | 831 | """ |
|
810 | 832 | |
|
811 | 833 | self.radix = radix |
|
812 | 834 | |
|
813 | 835 | self._docket_file = None |
|
814 | 836 | self._indexfile = None |
|
815 | 837 | self._datafile = None |
|
816 | 838 | self._sidedatafile = None |
|
817 | 839 | self._nodemap_file = None |
|
818 | 840 | self.postfix = postfix |
|
819 | 841 | self._trypending = trypending |
|
820 | 842 | self._try_split = try_split |
|
821 | 843 | self.opener = opener |
|
822 | 844 | if persistentnodemap: |
|
823 | 845 | self._nodemap_file = nodemaputil.get_nodemap_file(self) |
|
824 | 846 | |
|
825 | 847 | assert target[0] in ALL_KINDS |
|
826 | 848 | assert len(target) == 2 |
|
827 | 849 | self.target = target |
|
828 | 850 | if b'feature-config' in self.opener.options: |
|
829 | 851 | self.feature_config = self.opener.options[b'feature-config'].copy() |
|
830 | 852 | else: |
|
831 | 853 | self.feature_config = FeatureConfig() |
|
832 | 854 | self.feature_config.censorable = censorable |
|
833 | 855 | self.feature_config.canonical_parent_order = canonical_parent_order |
|
834 | 856 | if b'data-config' in self.opener.options: |
|
835 | 857 | self.data_config = self.opener.options[b'data-config'].copy() |
|
836 | 858 | else: |
|
837 | 859 | self.data_config = DataConfig() |
|
838 | 860 | self.data_config.check_ambig = checkambig |
|
839 | 861 | self.data_config.mmap_large_index = mmaplargeindex |
|
840 | 862 | if b'delta-config' in self.opener.options: |
|
841 | 863 | self.delta_config = self.opener.options[b'delta-config'].copy() |
|
842 | 864 | else: |
|
843 | 865 | self.delta_config = DeltaConfig() |
|
844 | 866 | self.delta_config.upper_bound_comp = upperboundcomp |
|
845 | 867 | |
|
846 | 868 | # 3-tuple of (node, rev, text) for a raw revision. |
|
847 | 869 | self._revisioncache = None |
|
848 | 870 | # Maps rev to chain base rev. |
|
849 | 871 | self._chainbasecache = util.lrucachedict(100) |
|
850 | 872 | |
|
851 | 873 | self.index = None |
|
852 | 874 | self._docket = None |
|
853 | 875 | self._nodemap_docket = None |
|
854 | 876 | # Mapping of partial identifiers to full nodes. |
|
855 | 877 | self._pcache = {} |
|
856 | 878 | |
|
857 | 879 | # other optionnals features |
|
858 | 880 | |
|
859 | 881 | # Make copy of flag processors so each revlog instance can support |
|
860 | 882 | # custom flags. |
|
861 | 883 | self._flagprocessors = dict(flagutil.flagprocessors) |
|
862 | 884 | # prevent nesting of addgroup |
|
863 | 885 | self._adding_group = None |
|
864 | 886 | |
|
865 | 887 | chunk_cache = self._loadindex() |
|
866 | 888 | self._load_inner(chunk_cache) |
|
867 | 889 | |
|
868 | 890 | self._concurrencychecker = concurrencychecker |
|
869 | 891 | |
|
870 | 892 | @property |
|
871 | 893 | def _generaldelta(self): |
|
872 | 894 | """temporary compatibility proxy""" |
|
873 | 895 | util.nouideprecwarn( |
|
874 | 896 | b"use revlog.delta_config.general_delta", b"6.6", stacklevel=2 |
|
875 | 897 | ) |
|
876 | 898 | return self.delta_config.general_delta |
|
877 | 899 | |
|
878 | 900 | @property |
|
879 | 901 | def _checkambig(self): |
|
880 | 902 | """temporary compatibility proxy""" |
|
881 | 903 | util.nouideprecwarn( |
|
882 | 904 | b"use revlog.data_config.checkambig", b"6.6", stacklevel=2 |
|
883 | 905 | ) |
|
884 | 906 | return self.data_config.check_ambig |
|
885 | 907 | |
|
886 | 908 | @property |
|
887 | 909 | def _mmaplargeindex(self): |
|
888 | 910 | """temporary compatibility proxy""" |
|
889 | 911 | util.nouideprecwarn( |
|
890 | 912 | b"use revlog.data_config.mmap_large_index", b"6.6", stacklevel=2 |
|
891 | 913 | ) |
|
892 | 914 | return self.data_config.mmap_large_index |
|
893 | 915 | |
|
894 | 916 | @property |
|
895 | 917 | def _censorable(self): |
|
896 | 918 | """temporary compatibility proxy""" |
|
897 | 919 | util.nouideprecwarn( |
|
898 | 920 | b"use revlog.feature_config.censorable", b"6.6", stacklevel=2 |
|
899 | 921 | ) |
|
900 | 922 | return self.feature_config.censorable |
|
901 | 923 | |
|
902 | 924 | @property |
|
903 | 925 | def _chunkcachesize(self): |
|
904 | 926 | """temporary compatibility proxy""" |
|
905 | 927 | util.nouideprecwarn( |
|
906 | 928 | b"use revlog.data_config.chunk_cache_size", b"6.6", stacklevel=2 |
|
907 | 929 | ) |
|
908 | 930 | return self.data_config.chunk_cache_size |
|
909 | 931 | |
|
910 | 932 | @property |
|
911 | 933 | def _maxchainlen(self): |
|
912 | 934 | """temporary compatibility proxy""" |
|
913 | 935 | util.nouideprecwarn( |
|
914 | 936 | b"use revlog.delta_config.max_chain_len", b"6.6", stacklevel=2 |
|
915 | 937 | ) |
|
916 | 938 | return self.delta_config.max_chain_len |
|
917 | 939 | |
|
918 | 940 | @property |
|
919 | 941 | def _deltabothparents(self): |
|
920 | 942 | """temporary compatibility proxy""" |
|
921 | 943 | util.nouideprecwarn( |
|
922 | 944 | b"use revlog.delta_config.delta_both_parents", b"6.6", stacklevel=2 |
|
923 | 945 | ) |
|
924 | 946 | return self.delta_config.delta_both_parents |
|
925 | 947 | |
|
926 | 948 | @property |
|
927 | 949 | def _candidate_group_chunk_size(self): |
|
928 | 950 | """temporary compatibility proxy""" |
|
929 | 951 | util.nouideprecwarn( |
|
930 | 952 | b"use revlog.delta_config.candidate_group_chunk_size", |
|
931 | 953 | b"6.6", |
|
932 | 954 | stacklevel=2, |
|
933 | 955 | ) |
|
934 | 956 | return self.delta_config.candidate_group_chunk_size |
|
935 | 957 | |
|
936 | 958 | @property |
|
937 | 959 | def _debug_delta(self): |
|
938 | 960 | """temporary compatibility proxy""" |
|
939 | 961 | util.nouideprecwarn( |
|
940 | 962 | b"use revlog.delta_config.debug_delta", b"6.6", stacklevel=2 |
|
941 | 963 | ) |
|
942 | 964 | return self.delta_config.debug_delta |
|
943 | 965 | |
|
944 | 966 | @property |
|
945 | 967 | def _compengine(self): |
|
946 | 968 | """temporary compatibility proxy""" |
|
947 | 969 | util.nouideprecwarn( |
|
948 | 970 | b"use revlog.feature_config.compression_engine", |
|
949 | 971 | b"6.6", |
|
950 | 972 | stacklevel=2, |
|
951 | 973 | ) |
|
952 | 974 | return self.feature_config.compression_engine |
|
953 | 975 | |
|
954 | 976 | @property |
|
955 | 977 | def upperboundcomp(self): |
|
956 | 978 | """temporary compatibility proxy""" |
|
957 | 979 | util.nouideprecwarn( |
|
958 | 980 | b"use revlog.delta_config.upper_bound_comp", |
|
959 | 981 | b"6.6", |
|
960 | 982 | stacklevel=2, |
|
961 | 983 | ) |
|
962 | 984 | return self.delta_config.upper_bound_comp |
|
963 | 985 | |
|
964 | 986 | @property |
|
965 | 987 | def _compengineopts(self): |
|
966 | 988 | """temporary compatibility proxy""" |
|
967 | 989 | util.nouideprecwarn( |
|
968 | 990 | b"use revlog.feature_config.compression_engine_options", |
|
969 | 991 | b"6.6", |
|
970 | 992 | stacklevel=2, |
|
971 | 993 | ) |
|
972 | 994 | return self.feature_config.compression_engine_options |
|
973 | 995 | |
|
974 | 996 | @property |
|
975 | 997 | def _maxdeltachainspan(self): |
|
976 | 998 | """temporary compatibility proxy""" |
|
977 | 999 | util.nouideprecwarn( |
|
978 | 1000 | b"use revlog.delta_config.max_deltachain_span", b"6.6", stacklevel=2 |
|
979 | 1001 | ) |
|
980 | 1002 | return self.delta_config.max_deltachain_span |
|
981 | 1003 | |
|
982 | 1004 | @property |
|
983 | 1005 | def _withsparseread(self): |
|
984 | 1006 | """temporary compatibility proxy""" |
|
985 | 1007 | util.nouideprecwarn( |
|
986 | 1008 | b"use revlog.data_config.with_sparse_read", b"6.6", stacklevel=2 |
|
987 | 1009 | ) |
|
988 | 1010 | return self.data_config.with_sparse_read |
|
989 | 1011 | |
|
990 | 1012 | @property |
|
991 | 1013 | def _sparserevlog(self): |
|
992 | 1014 | """temporary compatibility proxy""" |
|
993 | 1015 | util.nouideprecwarn( |
|
994 | 1016 | b"use revlog.delta_config.sparse_revlog", b"6.6", stacklevel=2 |
|
995 | 1017 | ) |
|
996 | 1018 | return self.delta_config.sparse_revlog |
|
997 | 1019 | |
|
998 | 1020 | @property |
|
999 | 1021 | def hassidedata(self): |
|
1000 | 1022 | """temporary compatibility proxy""" |
|
1001 | 1023 | util.nouideprecwarn( |
|
1002 | 1024 | b"use revlog.feature_config.has_side_data", b"6.6", stacklevel=2 |
|
1003 | 1025 | ) |
|
1004 | 1026 | return self.feature_config.has_side_data |
|
1005 | 1027 | |
|
1006 | 1028 | @property |
|
1007 | 1029 | def _srdensitythreshold(self): |
|
1008 | 1030 | """temporary compatibility proxy""" |
|
1009 | 1031 | util.nouideprecwarn( |
|
1010 | 1032 | b"use revlog.data_config.sr_density_threshold", |
|
1011 | 1033 | b"6.6", |
|
1012 | 1034 | stacklevel=2, |
|
1013 | 1035 | ) |
|
1014 | 1036 | return self.data_config.sr_density_threshold |
|
1015 | 1037 | |
|
1016 | 1038 | @property |
|
1017 | 1039 | def _srmingapsize(self): |
|
1018 | 1040 | """temporary compatibility proxy""" |
|
1019 | 1041 | util.nouideprecwarn( |
|
1020 | 1042 | b"use revlog.data_config.sr_min_gap_size", b"6.6", stacklevel=2 |
|
1021 | 1043 | ) |
|
1022 | 1044 | return self.data_config.sr_min_gap_size |
|
1023 | 1045 | |
|
1024 | 1046 | @property |
|
1025 | 1047 | def _compute_rank(self): |
|
1026 | 1048 | """temporary compatibility proxy""" |
|
1027 | 1049 | util.nouideprecwarn( |
|
1028 | 1050 | b"use revlog.feature_config.compute_rank", b"6.6", stacklevel=2 |
|
1029 | 1051 | ) |
|
1030 | 1052 | return self.feature_config.compute_rank |
|
1031 | 1053 | |
|
1032 | 1054 | @property |
|
1033 | 1055 | def canonical_parent_order(self): |
|
1034 | 1056 | """temporary compatibility proxy""" |
|
1035 | 1057 | util.nouideprecwarn( |
|
1036 | 1058 | b"use revlog.feature_config.canonical_parent_order", |
|
1037 | 1059 | b"6.6", |
|
1038 | 1060 | stacklevel=2, |
|
1039 | 1061 | ) |
|
1040 | 1062 | return self.feature_config.canonical_parent_order |
|
1041 | 1063 | |
|
1042 | 1064 | @property |
|
1043 | 1065 | def _lazydelta(self): |
|
1044 | 1066 | """temporary compatibility proxy""" |
|
1045 | 1067 | util.nouideprecwarn( |
|
1046 | 1068 | b"use revlog.delta_config.lazy_delta", b"6.6", stacklevel=2 |
|
1047 | 1069 | ) |
|
1048 | 1070 | return self.delta_config.lazy_delta |
|
1049 | 1071 | |
|
1050 | 1072 | @property |
|
1051 | 1073 | def _lazydeltabase(self): |
|
1052 | 1074 | """temporary compatibility proxy""" |
|
1053 | 1075 | util.nouideprecwarn( |
|
1054 | 1076 | b"use revlog.delta_config.lazy_delta_base", b"6.6", stacklevel=2 |
|
1055 | 1077 | ) |
|
1056 | 1078 | return self.delta_config.lazy_delta_base |
|
1057 | 1079 | |
|
1058 | 1080 | def _init_opts(self): |
|
1059 | 1081 | """process options (from above/config) to setup associated default revlog mode |
|
1060 | 1082 | |
|
1061 | 1083 | These values might be affected when actually reading on disk information. |
|
1062 | 1084 | |
|
1063 | 1085 | The relevant values are returned for use in _loadindex(). |
|
1064 | 1086 | |
|
1065 | 1087 | * newversionflags: |
|
1066 | 1088 | version header to use if we need to create a new revlog |
|
1067 | 1089 | |
|
1068 | 1090 | * mmapindexthreshold: |
|
1069 | 1091 | minimal index size for start to use mmap |
|
1070 | 1092 | |
|
1071 | 1093 | * force_nodemap: |
|
1072 | 1094 | force the usage of a "development" version of the nodemap code |
|
1073 | 1095 | """ |
|
1074 | 1096 | opts = self.opener.options |
|
1075 | 1097 | |
|
1076 | 1098 | if b'changelogv2' in opts and self.revlog_kind == KIND_CHANGELOG: |
|
1077 | 1099 | new_header = CHANGELOGV2 |
|
1078 | 1100 | compute_rank = opts.get(b'changelogv2.compute-rank', True) |
|
1079 | 1101 | self.feature_config.compute_rank = compute_rank |
|
1080 | 1102 | elif b'revlogv2' in opts: |
|
1081 | 1103 | new_header = REVLOGV2 |
|
1082 | 1104 | elif b'revlogv1' in opts: |
|
1083 | 1105 | new_header = REVLOGV1 | FLAG_INLINE_DATA |
|
1084 | 1106 | if b'generaldelta' in opts: |
|
1085 | 1107 | new_header |= FLAG_GENERALDELTA |
|
1086 | 1108 | elif b'revlogv0' in self.opener.options: |
|
1087 | 1109 | new_header = REVLOGV0 |
|
1088 | 1110 | else: |
|
1089 | 1111 | new_header = REVLOG_DEFAULT_VERSION |
|
1090 | 1112 | |
|
1091 | 1113 | mmapindexthreshold = None |
|
1092 | 1114 | if self.data_config.mmap_large_index: |
|
1093 | 1115 | mmapindexthreshold = self.data_config.mmap_index_threshold |
|
1094 | 1116 | if self.feature_config.enable_ellipsis: |
|
1095 | 1117 | self._flagprocessors[REVIDX_ELLIPSIS] = ellipsisprocessor |
|
1096 | 1118 | |
|
1097 | 1119 | # revlog v0 doesn't have flag processors |
|
1098 | 1120 | for flag, processor in opts.get(b'flagprocessors', {}).items(): |
|
1099 | 1121 | flagutil.insertflagprocessor(flag, processor, self._flagprocessors) |
|
1100 | 1122 | |
|
1101 | 1123 | chunk_cache_size = self.data_config.chunk_cache_size |
|
1102 | 1124 | if chunk_cache_size <= 0: |
|
1103 | 1125 | raise error.RevlogError( |
|
1104 | 1126 | _(b'revlog chunk cache size %r is not greater than 0') |
|
1105 | 1127 | % chunk_cache_size |
|
1106 | 1128 | ) |
|
1107 | 1129 | elif chunk_cache_size & (chunk_cache_size - 1): |
|
1108 | 1130 | raise error.RevlogError( |
|
1109 | 1131 | _(b'revlog chunk cache size %r is not a power of 2') |
|
1110 | 1132 | % chunk_cache_size |
|
1111 | 1133 | ) |
|
1112 | 1134 | force_nodemap = opts.get(b'devel-force-nodemap', False) |
|
1113 | 1135 | return new_header, mmapindexthreshold, force_nodemap |
|
1114 | 1136 | |
|
1115 | 1137 | def _get_data(self, filepath, mmap_threshold, size=None): |
|
1116 | 1138 | """return a file content with or without mmap |
|
1117 | 1139 | |
|
1118 | 1140 | If the file is missing return the empty string""" |
|
1119 | 1141 | try: |
|
1120 | 1142 | with self.opener(filepath) as fp: |
|
1121 | 1143 | if mmap_threshold is not None: |
|
1122 | 1144 | file_size = self.opener.fstat(fp).st_size |
|
1123 | 1145 | if file_size >= mmap_threshold: |
|
1124 | 1146 | if size is not None: |
|
1125 | 1147 | # avoid potentiel mmap crash |
|
1126 | 1148 | size = min(file_size, size) |
|
1127 | 1149 | # TODO: should .close() to release resources without |
|
1128 | 1150 | # relying on Python GC |
|
1129 | 1151 | if size is None: |
|
1130 | 1152 | return util.buffer(util.mmapread(fp)) |
|
1131 | 1153 | else: |
|
1132 | 1154 | return util.buffer(util.mmapread(fp, size)) |
|
1133 | 1155 | if size is None: |
|
1134 | 1156 | return fp.read() |
|
1135 | 1157 | else: |
|
1136 | 1158 | return fp.read(size) |
|
1137 | 1159 | except FileNotFoundError: |
|
1138 | 1160 | return b'' |
|
1139 | 1161 | |
|
1140 | 1162 | def get_streams(self, max_linkrev, force_inline=False): |
|
1141 | 1163 | """return a list of streams that represent this revlog |
|
1142 | 1164 | |
|
1143 | 1165 | This is used by stream-clone to do bytes to bytes copies of a repository. |
|
1144 | 1166 | |
|
1145 | 1167 | This streams data for all revisions that refer to a changelog revision up |
|
1146 | 1168 | to `max_linkrev`. |
|
1147 | 1169 | |
|
1148 | 1170 | If `force_inline` is set, it enforces that the stream will represent an inline revlog. |
|
1149 | 1171 | |
|
1150 | 1172 | It returns is a list of three-tuple: |
|
1151 | 1173 | |
|
1152 | 1174 | [ |
|
1153 | 1175 | (filename, bytes_stream, stream_size), |
|
1154 | 1176 | β¦ |
|
1155 | 1177 | ] |
|
1156 | 1178 | """ |
|
1157 | 1179 | n = len(self) |
|
1158 | 1180 | index = self.index |
|
1159 | 1181 | while n > 0: |
|
1160 | 1182 | linkrev = index[n - 1][4] |
|
1161 | 1183 | if linkrev < max_linkrev: |
|
1162 | 1184 | break |
|
1163 | 1185 | # note: this loop will rarely go through multiple iterations, since |
|
1164 | 1186 | # it only traverses commits created during the current streaming |
|
1165 | 1187 | # pull operation. |
|
1166 | 1188 | # |
|
1167 | 1189 | # If this become a problem, using a binary search should cap the |
|
1168 | 1190 | # runtime of this. |
|
1169 | 1191 | n = n - 1 |
|
1170 | 1192 | if n == 0: |
|
1171 | 1193 | # no data to send |
|
1172 | 1194 | return [] |
|
1173 | 1195 | index_size = n * index.entry_size |
|
1174 | 1196 | data_size = self.end(n - 1) |
|
1175 | 1197 | |
|
1176 | 1198 | # XXX we might have been split (or stripped) since the object |
|
1177 | 1199 | # initialization, We need to close this race too, but having a way to |
|
1178 | 1200 | # pre-open the file we feed to the revlog and never closing them before |
|
1179 | 1201 | # we are done streaming. |
|
1180 | 1202 | |
|
1181 | 1203 | if self._inline: |
|
1182 | 1204 | |
|
1183 | 1205 | def get_stream(): |
|
1184 | 1206 | with self.opener(self._indexfile, mode=b"r") as fp: |
|
1185 | 1207 | yield None |
|
1186 | 1208 | size = index_size + data_size |
|
1187 | 1209 | if size <= 65536: |
|
1188 | 1210 | yield fp.read(size) |
|
1189 | 1211 | else: |
|
1190 | 1212 | yield from util.filechunkiter(fp, limit=size) |
|
1191 | 1213 | |
|
1192 | 1214 | inline_stream = get_stream() |
|
1193 | 1215 | next(inline_stream) |
|
1194 | 1216 | return [ |
|
1195 | 1217 | (self._indexfile, inline_stream, index_size + data_size), |
|
1196 | 1218 | ] |
|
1197 | 1219 | elif force_inline: |
|
1198 | 1220 | |
|
1199 | 1221 | def get_stream(): |
|
1200 | 1222 | with self.reading(): |
|
1201 | 1223 | yield None |
|
1202 | 1224 | |
|
1203 | 1225 | for rev in range(n): |
|
1204 | 1226 | idx = self.index.entry_binary(rev) |
|
1205 | 1227 | if rev == 0 and self._docket is None: |
|
1206 | 1228 | # re-inject the inline flag |
|
1207 | 1229 | header = self._format_flags |
|
1208 | 1230 | header |= self._format_version |
|
1209 | 1231 | header |= FLAG_INLINE_DATA |
|
1210 | 1232 | header = self.index.pack_header(header) |
|
1211 | 1233 | idx = header + idx |
|
1212 | 1234 | yield idx |
|
1213 | 1235 | yield self._inner.get_segment_for_revs(rev, rev)[1] |
|
1214 | 1236 | |
|
1215 | 1237 | inline_stream = get_stream() |
|
1216 | 1238 | next(inline_stream) |
|
1217 | 1239 | return [ |
|
1218 | 1240 | (self._indexfile, inline_stream, index_size + data_size), |
|
1219 | 1241 | ] |
|
1220 | 1242 | else: |
|
1221 | 1243 | |
|
1222 | 1244 | def get_index_stream(): |
|
1223 | 1245 | with self.opener(self._indexfile, mode=b"r") as fp: |
|
1224 | 1246 | yield None |
|
1225 | 1247 | if index_size <= 65536: |
|
1226 | 1248 | yield fp.read(index_size) |
|
1227 | 1249 | else: |
|
1228 | 1250 | yield from util.filechunkiter(fp, limit=index_size) |
|
1229 | 1251 | |
|
1230 | 1252 | def get_data_stream(): |
|
1231 | 1253 | with self._datafp() as fp: |
|
1232 | 1254 | yield None |
|
1233 | 1255 | if data_size <= 65536: |
|
1234 | 1256 | yield fp.read(data_size) |
|
1235 | 1257 | else: |
|
1236 | 1258 | yield from util.filechunkiter(fp, limit=data_size) |
|
1237 | 1259 | |
|
1238 | 1260 | index_stream = get_index_stream() |
|
1239 | 1261 | next(index_stream) |
|
1240 | 1262 | data_stream = get_data_stream() |
|
1241 | 1263 | next(data_stream) |
|
1242 | 1264 | return [ |
|
1243 | 1265 | (self._datafile, data_stream, data_size), |
|
1244 | 1266 | (self._indexfile, index_stream, index_size), |
|
1245 | 1267 | ] |
|
1246 | 1268 | |
|
1247 | 1269 | def _loadindex(self, docket=None): |
|
1248 | 1270 | |
|
1249 | 1271 | new_header, mmapindexthreshold, force_nodemap = self._init_opts() |
|
1250 | 1272 | |
|
1251 | 1273 | if self.postfix is not None: |
|
1252 | 1274 | entry_point = b'%s.i.%s' % (self.radix, self.postfix) |
|
1253 | 1275 | elif self._trypending and self.opener.exists(b'%s.i.a' % self.radix): |
|
1254 | 1276 | entry_point = b'%s.i.a' % self.radix |
|
1255 | 1277 | elif self._try_split and self.opener.exists(self._split_index_file): |
|
1256 | 1278 | entry_point = self._split_index_file |
|
1257 | 1279 | else: |
|
1258 | 1280 | entry_point = b'%s.i' % self.radix |
|
1259 | 1281 | |
|
1260 | 1282 | if docket is not None: |
|
1261 | 1283 | self._docket = docket |
|
1262 | 1284 | self._docket_file = entry_point |
|
1263 | 1285 | else: |
|
1264 | 1286 | self._initempty = True |
|
1265 | 1287 | entry_data = self._get_data(entry_point, mmapindexthreshold) |
|
1266 | 1288 | if len(entry_data) > 0: |
|
1267 | 1289 | header = INDEX_HEADER.unpack(entry_data[:4])[0] |
|
1268 | 1290 | self._initempty = False |
|
1269 | 1291 | else: |
|
1270 | 1292 | header = new_header |
|
1271 | 1293 | |
|
1272 | 1294 | self._format_flags = header & ~0xFFFF |
|
1273 | 1295 | self._format_version = header & 0xFFFF |
|
1274 | 1296 | |
|
1275 | 1297 | supported_flags = SUPPORTED_FLAGS.get(self._format_version) |
|
1276 | 1298 | if supported_flags is None: |
|
1277 | 1299 | msg = _(b'unknown version (%d) in revlog %s') |
|
1278 | 1300 | msg %= (self._format_version, self.display_id) |
|
1279 | 1301 | raise error.RevlogError(msg) |
|
1280 | 1302 | elif self._format_flags & ~supported_flags: |
|
1281 | 1303 | msg = _(b'unknown flags (%#04x) in version %d revlog %s') |
|
1282 | 1304 | display_flag = self._format_flags >> 16 |
|
1283 | 1305 | msg %= (display_flag, self._format_version, self.display_id) |
|
1284 | 1306 | raise error.RevlogError(msg) |
|
1285 | 1307 | |
|
1286 | 1308 | features = FEATURES_BY_VERSION[self._format_version] |
|
1287 | 1309 | self._inline = features[b'inline'](self._format_flags) |
|
1288 | 1310 | self.delta_config.general_delta = features[b'generaldelta']( |
|
1289 | 1311 | self._format_flags |
|
1290 | 1312 | ) |
|
1291 | 1313 | self.feature_config.has_side_data = features[b'sidedata'] |
|
1292 | 1314 | |
|
1293 | 1315 | if not features[b'docket']: |
|
1294 | 1316 | self._indexfile = entry_point |
|
1295 | 1317 | index_data = entry_data |
|
1296 | 1318 | else: |
|
1297 | 1319 | self._docket_file = entry_point |
|
1298 | 1320 | if self._initempty: |
|
1299 | 1321 | self._docket = docketutil.default_docket(self, header) |
|
1300 | 1322 | else: |
|
1301 | 1323 | self._docket = docketutil.parse_docket( |
|
1302 | 1324 | self, entry_data, use_pending=self._trypending |
|
1303 | 1325 | ) |
|
1304 | 1326 | |
|
1305 | 1327 | if self._docket is not None: |
|
1306 | 1328 | self._indexfile = self._docket.index_filepath() |
|
1307 | 1329 | index_data = b'' |
|
1308 | 1330 | index_size = self._docket.index_end |
|
1309 | 1331 | if index_size > 0: |
|
1310 | 1332 | index_data = self._get_data( |
|
1311 | 1333 | self._indexfile, mmapindexthreshold, size=index_size |
|
1312 | 1334 | ) |
|
1313 | 1335 | if len(index_data) < index_size: |
|
1314 | 1336 | msg = _(b'too few index data for %s: got %d, expected %d') |
|
1315 | 1337 | msg %= (self.display_id, len(index_data), index_size) |
|
1316 | 1338 | raise error.RevlogError(msg) |
|
1317 | 1339 | |
|
1318 | 1340 | self._inline = False |
|
1319 | 1341 | # generaldelta implied by version 2 revlogs. |
|
1320 | 1342 | self.delta_config.general_delta = True |
|
1321 | 1343 | # the logic for persistent nodemap will be dealt with within the |
|
1322 | 1344 | # main docket, so disable it for now. |
|
1323 | 1345 | self._nodemap_file = None |
|
1324 | 1346 | |
|
1325 | 1347 | if self._docket is not None: |
|
1326 | 1348 | self._datafile = self._docket.data_filepath() |
|
1327 | 1349 | self._sidedatafile = self._docket.sidedata_filepath() |
|
1328 | 1350 | elif self.postfix is None: |
|
1329 | 1351 | self._datafile = b'%s.d' % self.radix |
|
1330 | 1352 | else: |
|
1331 | 1353 | self._datafile = b'%s.d.%s' % (self.radix, self.postfix) |
|
1332 | 1354 | |
|
1333 | 1355 | self.nodeconstants = sha1nodeconstants |
|
1334 | 1356 | self.nullid = self.nodeconstants.nullid |
|
1335 | 1357 | |
|
1336 | 1358 | # sparse-revlog can't be on without general-delta (issue6056) |
|
1337 | 1359 | if not self.delta_config.general_delta: |
|
1338 | 1360 | self.delta_config.sparse_revlog = False |
|
1339 | 1361 | |
|
1340 | 1362 | self._storedeltachains = True |
|
1341 | 1363 | |
|
1342 | 1364 | devel_nodemap = ( |
|
1343 | 1365 | self._nodemap_file |
|
1344 | 1366 | and force_nodemap |
|
1345 | 1367 | and parse_index_v1_nodemap is not None |
|
1346 | 1368 | ) |
|
1347 | 1369 | |
|
1348 | 1370 | use_rust_index = False |
|
1349 | 1371 | if rustrevlog is not None: |
|
1350 | 1372 | if self._nodemap_file is not None: |
|
1351 | 1373 | use_rust_index = True |
|
1352 | 1374 | else: |
|
1353 | 1375 | use_rust_index = self.opener.options.get(b'rust.index') |
|
1354 | 1376 | |
|
1355 | 1377 | self._parse_index = parse_index_v1 |
|
1356 | 1378 | if self._format_version == REVLOGV0: |
|
1357 | 1379 | self._parse_index = revlogv0.parse_index_v0 |
|
1358 | 1380 | elif self._format_version == REVLOGV2: |
|
1359 | 1381 | self._parse_index = parse_index_v2 |
|
1360 | 1382 | elif self._format_version == CHANGELOGV2: |
|
1361 | 1383 | self._parse_index = parse_index_cl_v2 |
|
1362 | 1384 | elif devel_nodemap: |
|
1363 | 1385 | self._parse_index = parse_index_v1_nodemap |
|
1364 | 1386 | elif use_rust_index: |
|
1365 | 1387 | self._parse_index = parse_index_v1_mixed |
|
1366 | 1388 | try: |
|
1367 | 1389 | d = self._parse_index(index_data, self._inline) |
|
1368 | 1390 | index, chunkcache = d |
|
1369 | 1391 | use_nodemap = ( |
|
1370 | 1392 | not self._inline |
|
1371 | 1393 | and self._nodemap_file is not None |
|
1372 | 1394 | and hasattr(index, 'update_nodemap_data') |
|
1373 | 1395 | ) |
|
1374 | 1396 | if use_nodemap: |
|
1375 | 1397 | nodemap_data = nodemaputil.persisted_data(self) |
|
1376 | 1398 | if nodemap_data is not None: |
|
1377 | 1399 | docket = nodemap_data[0] |
|
1378 | 1400 | if ( |
|
1379 | 1401 | len(d[0]) > docket.tip_rev |
|
1380 | 1402 | and d[0][docket.tip_rev][7] == docket.tip_node |
|
1381 | 1403 | ): |
|
1382 | 1404 | # no changelog tampering |
|
1383 | 1405 | self._nodemap_docket = docket |
|
1384 | 1406 | index.update_nodemap_data(*nodemap_data) |
|
1385 | 1407 | except (ValueError, IndexError): |
|
1386 | 1408 | raise error.RevlogError( |
|
1387 | 1409 | _(b"index %s is corrupted") % self.display_id |
|
1388 | 1410 | ) |
|
1389 | 1411 | self.index = index |
|
1390 | 1412 | # revnum -> (chain-length, sum-delta-length) |
|
1391 | 1413 | self._chaininfocache = util.lrucachedict(500) |
|
1392 | 1414 | |
|
1393 | 1415 | return chunkcache |
|
1394 | 1416 | |
|
1395 | 1417 | def _load_inner(self, chunk_cache): |
|
1396 | 1418 | if self._docket is None: |
|
1397 | 1419 | default_compression_header = None |
|
1398 | 1420 | else: |
|
1399 | 1421 | default_compression_header = self._docket.default_compression_header |
|
1400 | 1422 | |
|
1401 | 1423 | self._inner = _InnerRevlog( |
|
1402 | 1424 | opener=self.opener, |
|
1403 | 1425 | index=self.index, |
|
1404 | 1426 | index_file=self._indexfile, |
|
1405 | 1427 | data_file=self._datafile, |
|
1406 | 1428 | sidedata_file=self._sidedatafile, |
|
1407 | 1429 | inline=self._inline, |
|
1408 | 1430 | data_config=self.data_config, |
|
1409 | 1431 | feature_config=self.feature_config, |
|
1410 | 1432 | chunk_cache=chunk_cache, |
|
1411 | 1433 | default_compression_header=default_compression_header, |
|
1412 | 1434 | ) |
|
1413 | 1435 | |
|
1414 | 1436 | def get_revlog(self): |
|
1415 | 1437 | """simple function to mirror API of other not-really-revlog API""" |
|
1416 | 1438 | return self |
|
1417 | 1439 | |
|
1418 | 1440 | @util.propertycache |
|
1419 | 1441 | def revlog_kind(self): |
|
1420 | 1442 | return self.target[0] |
|
1421 | 1443 | |
|
1422 | 1444 | @util.propertycache |
|
1423 | 1445 | def display_id(self): |
|
1424 | 1446 | """The public facing "ID" of the revlog that we use in message""" |
|
1425 | 1447 | if self.revlog_kind == KIND_FILELOG: |
|
1426 | 1448 | # Reference the file without the "data/" prefix, so it is familiar |
|
1427 | 1449 | # to the user. |
|
1428 | 1450 | return self.target[1] |
|
1429 | 1451 | else: |
|
1430 | 1452 | return self.radix |
|
1431 | 1453 | |
|
1432 | 1454 | def _datafp(self, mode=b'r'): |
|
1433 | 1455 | """file object for the revlog's data file""" |
|
1434 | 1456 | return self.opener(self._datafile, mode=mode) |
|
1435 | 1457 | |
|
1436 | 1458 | def tiprev(self): |
|
1437 | 1459 | return len(self.index) - 1 |
|
1438 | 1460 | |
|
1439 | 1461 | def tip(self): |
|
1440 | 1462 | return self.node(self.tiprev()) |
|
1441 | 1463 | |
|
1442 | 1464 | def __contains__(self, rev): |
|
1443 | 1465 | return 0 <= rev < len(self) |
|
1444 | 1466 | |
|
1445 | 1467 | def __len__(self): |
|
1446 | 1468 | return len(self.index) |
|
1447 | 1469 | |
|
1448 | 1470 | def __iter__(self): |
|
1449 | 1471 | return iter(range(len(self))) |
|
1450 | 1472 | |
|
1451 | 1473 | def revs(self, start=0, stop=None): |
|
1452 | 1474 | """iterate over all rev in this revlog (from start to stop)""" |
|
1453 | 1475 | return storageutil.iterrevs(len(self), start=start, stop=stop) |
|
1454 | 1476 | |
|
1455 | 1477 | def hasnode(self, node): |
|
1456 | 1478 | try: |
|
1457 | 1479 | self.rev(node) |
|
1458 | 1480 | return True |
|
1459 | 1481 | except KeyError: |
|
1460 | 1482 | return False |
|
1461 | 1483 | |
|
1462 | 1484 | def _candelta(self, baserev, rev): |
|
1463 | 1485 | """whether two revisions (baserev, rev) can be delta-ed or not""" |
|
1464 | 1486 | # Disable delta if either rev requires a content-changing flag |
|
1465 | 1487 | # processor (ex. LFS). This is because such flag processor can alter |
|
1466 | 1488 | # the rawtext content that the delta will be based on, and two clients |
|
1467 | 1489 | # could have a same revlog node with different flags (i.e. different |
|
1468 | 1490 | # rawtext contents) and the delta could be incompatible. |
|
1469 | 1491 | if (self.flags(baserev) & REVIDX_RAWTEXT_CHANGING_FLAGS) or ( |
|
1470 | 1492 | self.flags(rev) & REVIDX_RAWTEXT_CHANGING_FLAGS |
|
1471 | 1493 | ): |
|
1472 | 1494 | return False |
|
1473 | 1495 | return True |
|
1474 | 1496 | |
|
1475 | 1497 | def update_caches(self, transaction): |
|
1476 | 1498 | """update on disk cache |
|
1477 | 1499 | |
|
1478 | 1500 | If a transaction is passed, the update may be delayed to transaction |
|
1479 | 1501 | commit.""" |
|
1480 | 1502 | if self._nodemap_file is not None: |
|
1481 | 1503 | if transaction is None: |
|
1482 | 1504 | nodemaputil.update_persistent_nodemap(self) |
|
1483 | 1505 | else: |
|
1484 | 1506 | nodemaputil.setup_persistent_nodemap(transaction, self) |
|
1485 | 1507 | |
|
1486 | 1508 | def clearcaches(self): |
|
1487 | 1509 | """Clear in-memory caches""" |
|
1488 | 1510 | self._revisioncache = None |
|
1489 | 1511 | self._chainbasecache.clear() |
|
1490 | 1512 | self._inner._segmentfile.clear_cache() |
|
1491 | 1513 | self._inner._segmentfile_sidedata.clear_cache() |
|
1492 | 1514 | self._pcache = {} |
|
1493 | 1515 | self._nodemap_docket = None |
|
1494 | 1516 | self.index.clearcaches() |
|
1495 | 1517 | # The python code is the one responsible for validating the docket, we |
|
1496 | 1518 | # end up having to refresh it here. |
|
1497 | 1519 | use_nodemap = ( |
|
1498 | 1520 | not self._inline |
|
1499 | 1521 | and self._nodemap_file is not None |
|
1500 | 1522 | and hasattr(self.index, 'update_nodemap_data') |
|
1501 | 1523 | ) |
|
1502 | 1524 | if use_nodemap: |
|
1503 | 1525 | nodemap_data = nodemaputil.persisted_data(self) |
|
1504 | 1526 | if nodemap_data is not None: |
|
1505 | 1527 | self._nodemap_docket = nodemap_data[0] |
|
1506 | 1528 | self.index.update_nodemap_data(*nodemap_data) |
|
1507 | 1529 | |
|
1508 | 1530 | def rev(self, node): |
|
1509 | 1531 | """return the revision number associated with a <nodeid>""" |
|
1510 | 1532 | try: |
|
1511 | 1533 | return self.index.rev(node) |
|
1512 | 1534 | except TypeError: |
|
1513 | 1535 | raise |
|
1514 | 1536 | except error.RevlogError: |
|
1515 | 1537 | # parsers.c radix tree lookup failed |
|
1516 | 1538 | if ( |
|
1517 | 1539 | node == self.nodeconstants.wdirid |
|
1518 | 1540 | or node in self.nodeconstants.wdirfilenodeids |
|
1519 | 1541 | ): |
|
1520 | 1542 | raise error.WdirUnsupported |
|
1521 | 1543 | raise error.LookupError(node, self.display_id, _(b'no node')) |
|
1522 | 1544 | |
|
1523 | 1545 | # Accessors for index entries. |
|
1524 | 1546 | |
|
1525 | 1547 | # First tuple entry is 8 bytes. First 6 bytes are offset. Last 2 bytes |
|
1526 | 1548 | # are flags. |
|
1527 | 1549 | def start(self, rev): |
|
1528 | 1550 | return int(self.index[rev][0] >> 16) |
|
1529 | 1551 | |
|
1530 | 1552 | def sidedata_cut_off(self, rev): |
|
1531 | 1553 | sd_cut_off = self.index[rev][8] |
|
1532 | 1554 | if sd_cut_off != 0: |
|
1533 | 1555 | return sd_cut_off |
|
1534 | 1556 | # This is some annoying dance, because entries without sidedata |
|
1535 | 1557 | # currently use 0 as their ofsset. (instead of previous-offset + |
|
1536 | 1558 | # previous-size) |
|
1537 | 1559 | # |
|
1538 | 1560 | # We should reconsider this sidedata β 0 sidata_offset policy. |
|
1539 | 1561 | # In the meantime, we need this. |
|
1540 | 1562 | while 0 <= rev: |
|
1541 | 1563 | e = self.index[rev] |
|
1542 | 1564 | if e[9] != 0: |
|
1543 | 1565 | return e[8] + e[9] |
|
1544 | 1566 | rev -= 1 |
|
1545 | 1567 | return 0 |
|
1546 | 1568 | |
|
1547 | 1569 | def flags(self, rev): |
|
1548 | 1570 | return self.index[rev][0] & 0xFFFF |
|
1549 | 1571 | |
|
1550 | 1572 | def length(self, rev): |
|
1551 | 1573 | return self.index[rev][1] |
|
1552 | 1574 | |
|
1553 | 1575 | def sidedata_length(self, rev): |
|
1554 | 1576 | if not self.feature_config.has_side_data: |
|
1555 | 1577 | return 0 |
|
1556 | 1578 | return self.index[rev][9] |
|
1557 | 1579 | |
|
1558 | 1580 | def rawsize(self, rev): |
|
1559 | 1581 | """return the length of the uncompressed text for a given revision""" |
|
1560 | 1582 | l = self.index[rev][2] |
|
1561 | 1583 | if l >= 0: |
|
1562 | 1584 | return l |
|
1563 | 1585 | |
|
1564 | 1586 | t = self.rawdata(rev) |
|
1565 | 1587 | return len(t) |
|
1566 | 1588 | |
|
1567 | 1589 | def size(self, rev): |
|
1568 | 1590 | """length of non-raw text (processed by a "read" flag processor)""" |
|
1569 | 1591 | # fast path: if no "read" flag processor could change the content, |
|
1570 | 1592 | # size is rawsize. note: ELLIPSIS is known to not change the content. |
|
1571 | 1593 | flags = self.flags(rev) |
|
1572 | 1594 | if flags & (flagutil.REVIDX_KNOWN_FLAGS ^ REVIDX_ELLIPSIS) == 0: |
|
1573 | 1595 | return self.rawsize(rev) |
|
1574 | 1596 | |
|
1575 | 1597 | return len(self.revision(rev)) |
|
1576 | 1598 | |
|
1577 | 1599 | def fast_rank(self, rev): |
|
1578 | 1600 | """Return the rank of a revision if already known, or None otherwise. |
|
1579 | 1601 | |
|
1580 | 1602 | The rank of a revision is the size of the sub-graph it defines as a |
|
1581 | 1603 | head. Equivalently, the rank of a revision `r` is the size of the set |
|
1582 | 1604 | `ancestors(r)`, `r` included. |
|
1583 | 1605 | |
|
1584 | 1606 | This method returns the rank retrieved from the revlog in constant |
|
1585 | 1607 | time. It makes no attempt at computing unknown values for versions of |
|
1586 | 1608 | the revlog which do not persist the rank. |
|
1587 | 1609 | """ |
|
1588 | 1610 | rank = self.index[rev][ENTRY_RANK] |
|
1589 | 1611 | if self._format_version != CHANGELOGV2 or rank == RANK_UNKNOWN: |
|
1590 | 1612 | return None |
|
1591 | 1613 | if rev == nullrev: |
|
1592 | 1614 | return 0 # convention |
|
1593 | 1615 | return rank |
|
1594 | 1616 | |
|
1595 | 1617 | def chainbase(self, rev): |
|
1596 | 1618 | base = self._chainbasecache.get(rev) |
|
1597 | 1619 | if base is not None: |
|
1598 | 1620 | return base |
|
1599 | 1621 | |
|
1600 | 1622 | index = self.index |
|
1601 | 1623 | iterrev = rev |
|
1602 | 1624 | base = index[iterrev][3] |
|
1603 | 1625 | while base != iterrev: |
|
1604 | 1626 | iterrev = base |
|
1605 | 1627 | base = index[iterrev][3] |
|
1606 | 1628 | |
|
1607 | 1629 | self._chainbasecache[rev] = base |
|
1608 | 1630 | return base |
|
1609 | 1631 | |
|
1610 | 1632 | def linkrev(self, rev): |
|
1611 | 1633 | return self.index[rev][4] |
|
1612 | 1634 | |
|
1613 | 1635 | def parentrevs(self, rev): |
|
1614 | 1636 | try: |
|
1615 | 1637 | entry = self.index[rev] |
|
1616 | 1638 | except IndexError: |
|
1617 | 1639 | if rev == wdirrev: |
|
1618 | 1640 | raise error.WdirUnsupported |
|
1619 | 1641 | raise |
|
1620 | 1642 | |
|
1621 | 1643 | if self.feature_config.canonical_parent_order and entry[5] == nullrev: |
|
1622 | 1644 | return entry[6], entry[5] |
|
1623 | 1645 | else: |
|
1624 | 1646 | return entry[5], entry[6] |
|
1625 | 1647 | |
|
1626 | 1648 | # fast parentrevs(rev) where rev isn't filtered |
|
1627 | 1649 | _uncheckedparentrevs = parentrevs |
|
1628 | 1650 | |
|
1629 | 1651 | def node(self, rev): |
|
1630 | 1652 | try: |
|
1631 | 1653 | return self.index[rev][7] |
|
1632 | 1654 | except IndexError: |
|
1633 | 1655 | if rev == wdirrev: |
|
1634 | 1656 | raise error.WdirUnsupported |
|
1635 | 1657 | raise |
|
1636 | 1658 | |
|
1637 | 1659 | # Derived from index values. |
|
1638 | 1660 | |
|
1639 | 1661 | def end(self, rev): |
|
1640 | 1662 | return self.start(rev) + self.length(rev) |
|
1641 | 1663 | |
|
1642 | 1664 | def parents(self, node): |
|
1643 | 1665 | i = self.index |
|
1644 | 1666 | d = i[self.rev(node)] |
|
1645 | 1667 | # inline node() to avoid function call overhead |
|
1646 | 1668 | if self.feature_config.canonical_parent_order and d[5] == self.nullid: |
|
1647 | 1669 | return i[d[6]][7], i[d[5]][7] |
|
1648 | 1670 | else: |
|
1649 | 1671 | return i[d[5]][7], i[d[6]][7] |
|
1650 | 1672 | |
|
1651 | 1673 | def chainlen(self, rev): |
|
1652 | 1674 | return self._chaininfo(rev)[0] |
|
1653 | 1675 | |
|
1654 | 1676 | def _chaininfo(self, rev): |
|
1655 | 1677 | chaininfocache = self._chaininfocache |
|
1656 | 1678 | if rev in chaininfocache: |
|
1657 | 1679 | return chaininfocache[rev] |
|
1658 | 1680 | index = self.index |
|
1659 | 1681 | generaldelta = self.delta_config.general_delta |
|
1660 | 1682 | iterrev = rev |
|
1661 | 1683 | e = index[iterrev] |
|
1662 | 1684 | clen = 0 |
|
1663 | 1685 | compresseddeltalen = 0 |
|
1664 | 1686 | while iterrev != e[3]: |
|
1665 | 1687 | clen += 1 |
|
1666 | 1688 | compresseddeltalen += e[1] |
|
1667 | 1689 | if generaldelta: |
|
1668 | 1690 | iterrev = e[3] |
|
1669 | 1691 | else: |
|
1670 | 1692 | iterrev -= 1 |
|
1671 | 1693 | if iterrev in chaininfocache: |
|
1672 | 1694 | t = chaininfocache[iterrev] |
|
1673 | 1695 | clen += t[0] |
|
1674 | 1696 | compresseddeltalen += t[1] |
|
1675 | 1697 | break |
|
1676 | 1698 | e = index[iterrev] |
|
1677 | 1699 | else: |
|
1678 | 1700 | # Add text length of base since decompressing that also takes |
|
1679 | 1701 | # work. For cache hits the length is already included. |
|
1680 | 1702 | compresseddeltalen += e[1] |
|
1681 | 1703 | r = (clen, compresseddeltalen) |
|
1682 | 1704 | chaininfocache[rev] = r |
|
1683 | 1705 | return r |
|
1684 | 1706 | |
|
1685 | 1707 | def _deltachain(self, rev, stoprev=None): |
|
1686 | 1708 | """Obtain the delta chain for a revision. |
|
1687 | 1709 | |
|
1688 | 1710 | ``stoprev`` specifies a revision to stop at. If not specified, we |
|
1689 | 1711 | stop at the base of the chain. |
|
1690 | 1712 | |
|
1691 | 1713 | Returns a 2-tuple of (chain, stopped) where ``chain`` is a list of |
|
1692 | 1714 | revs in ascending order and ``stopped`` is a bool indicating whether |
|
1693 | 1715 | ``stoprev`` was hit. |
|
1694 | 1716 | """ |
|
1695 | 1717 | generaldelta = self.delta_config.general_delta |
|
1696 | 1718 | # Try C implementation. |
|
1697 | 1719 | try: |
|
1698 | 1720 | return self.index.deltachain(rev, stoprev, generaldelta) |
|
1699 | 1721 | except AttributeError: |
|
1700 | 1722 | pass |
|
1701 | 1723 | |
|
1702 | 1724 | chain = [] |
|
1703 | 1725 | |
|
1704 | 1726 | # Alias to prevent attribute lookup in tight loop. |
|
1705 | 1727 | index = self.index |
|
1706 | 1728 | |
|
1707 | 1729 | iterrev = rev |
|
1708 | 1730 | e = index[iterrev] |
|
1709 | 1731 | while iterrev != e[3] and iterrev != stoprev: |
|
1710 | 1732 | chain.append(iterrev) |
|
1711 | 1733 | if generaldelta: |
|
1712 | 1734 | iterrev = e[3] |
|
1713 | 1735 | else: |
|
1714 | 1736 | iterrev -= 1 |
|
1715 | 1737 | e = index[iterrev] |
|
1716 | 1738 | |
|
1717 | 1739 | if iterrev == stoprev: |
|
1718 | 1740 | stopped = True |
|
1719 | 1741 | else: |
|
1720 | 1742 | chain.append(iterrev) |
|
1721 | 1743 | stopped = False |
|
1722 | 1744 | |
|
1723 | 1745 | chain.reverse() |
|
1724 | 1746 | return chain, stopped |
|
1725 | 1747 | |
|
1726 | 1748 | def ancestors(self, revs, stoprev=0, inclusive=False): |
|
1727 | 1749 | """Generate the ancestors of 'revs' in reverse revision order. |
|
1728 | 1750 | Does not generate revs lower than stoprev. |
|
1729 | 1751 | |
|
1730 | 1752 | See the documentation for ancestor.lazyancestors for more details.""" |
|
1731 | 1753 | |
|
1732 | 1754 | # first, make sure start revisions aren't filtered |
|
1733 | 1755 | revs = list(revs) |
|
1734 | 1756 | checkrev = self.node |
|
1735 | 1757 | for r in revs: |
|
1736 | 1758 | checkrev(r) |
|
1737 | 1759 | # and we're sure ancestors aren't filtered as well |
|
1738 | 1760 | |
|
1739 | 1761 | if rustancestor is not None and self.index.rust_ext_compat: |
|
1740 | 1762 | lazyancestors = rustancestor.LazyAncestors |
|
1741 | 1763 | arg = self.index |
|
1742 | 1764 | else: |
|
1743 | 1765 | lazyancestors = ancestor.lazyancestors |
|
1744 | 1766 | arg = self._uncheckedparentrevs |
|
1745 | 1767 | return lazyancestors(arg, revs, stoprev=stoprev, inclusive=inclusive) |
|
1746 | 1768 | |
|
1747 | 1769 | def descendants(self, revs): |
|
1748 | 1770 | return dagop.descendantrevs(revs, self.revs, self.parentrevs) |
|
1749 | 1771 | |
|
1750 | 1772 | def findcommonmissing(self, common=None, heads=None): |
|
1751 | 1773 | """Return a tuple of the ancestors of common and the ancestors of heads |
|
1752 | 1774 | that are not ancestors of common. In revset terminology, we return the |
|
1753 | 1775 | tuple: |
|
1754 | 1776 | |
|
1755 | 1777 | ::common, (::heads) - (::common) |
|
1756 | 1778 | |
|
1757 | 1779 | The list is sorted by revision number, meaning it is |
|
1758 | 1780 | topologically sorted. |
|
1759 | 1781 | |
|
1760 | 1782 | 'heads' and 'common' are both lists of node IDs. If heads is |
|
1761 | 1783 | not supplied, uses all of the revlog's heads. If common is not |
|
1762 | 1784 | supplied, uses nullid.""" |
|
1763 | 1785 | if common is None: |
|
1764 | 1786 | common = [self.nullid] |
|
1765 | 1787 | if heads is None: |
|
1766 | 1788 | heads = self.heads() |
|
1767 | 1789 | |
|
1768 | 1790 | common = [self.rev(n) for n in common] |
|
1769 | 1791 | heads = [self.rev(n) for n in heads] |
|
1770 | 1792 | |
|
1771 | 1793 | # we want the ancestors, but inclusive |
|
1772 | 1794 | class lazyset: |
|
1773 | 1795 | def __init__(self, lazyvalues): |
|
1774 | 1796 | self.addedvalues = set() |
|
1775 | 1797 | self.lazyvalues = lazyvalues |
|
1776 | 1798 | |
|
1777 | 1799 | def __contains__(self, value): |
|
1778 | 1800 | return value in self.addedvalues or value in self.lazyvalues |
|
1779 | 1801 | |
|
1780 | 1802 | def __iter__(self): |
|
1781 | 1803 | added = self.addedvalues |
|
1782 | 1804 | for r in added: |
|
1783 | 1805 | yield r |
|
1784 | 1806 | for r in self.lazyvalues: |
|
1785 | 1807 | if not r in added: |
|
1786 | 1808 | yield r |
|
1787 | 1809 | |
|
1788 | 1810 | def add(self, value): |
|
1789 | 1811 | self.addedvalues.add(value) |
|
1790 | 1812 | |
|
1791 | 1813 | def update(self, values): |
|
1792 | 1814 | self.addedvalues.update(values) |
|
1793 | 1815 | |
|
1794 | 1816 | has = lazyset(self.ancestors(common)) |
|
1795 | 1817 | has.add(nullrev) |
|
1796 | 1818 | has.update(common) |
|
1797 | 1819 | |
|
1798 | 1820 | # take all ancestors from heads that aren't in has |
|
1799 | 1821 | missing = set() |
|
1800 | 1822 | visit = collections.deque(r for r in heads if r not in has) |
|
1801 | 1823 | while visit: |
|
1802 | 1824 | r = visit.popleft() |
|
1803 | 1825 | if r in missing: |
|
1804 | 1826 | continue |
|
1805 | 1827 | else: |
|
1806 | 1828 | missing.add(r) |
|
1807 | 1829 | for p in self.parentrevs(r): |
|
1808 | 1830 | if p not in has: |
|
1809 | 1831 | visit.append(p) |
|
1810 | 1832 | missing = list(missing) |
|
1811 | 1833 | missing.sort() |
|
1812 | 1834 | return has, [self.node(miss) for miss in missing] |
|
1813 | 1835 | |
|
1814 | 1836 | def incrementalmissingrevs(self, common=None): |
|
1815 | 1837 | """Return an object that can be used to incrementally compute the |
|
1816 | 1838 | revision numbers of the ancestors of arbitrary sets that are not |
|
1817 | 1839 | ancestors of common. This is an ancestor.incrementalmissingancestors |
|
1818 | 1840 | object. |
|
1819 | 1841 | |
|
1820 | 1842 | 'common' is a list of revision numbers. If common is not supplied, uses |
|
1821 | 1843 | nullrev. |
|
1822 | 1844 | """ |
|
1823 | 1845 | if common is None: |
|
1824 | 1846 | common = [nullrev] |
|
1825 | 1847 | |
|
1826 | 1848 | if rustancestor is not None and self.index.rust_ext_compat: |
|
1827 | 1849 | return rustancestor.MissingAncestors(self.index, common) |
|
1828 | 1850 | return ancestor.incrementalmissingancestors(self.parentrevs, common) |
|
1829 | 1851 | |
|
1830 | 1852 | def findmissingrevs(self, common=None, heads=None): |
|
1831 | 1853 | """Return the revision numbers of the ancestors of heads that |
|
1832 | 1854 | are not ancestors of common. |
|
1833 | 1855 | |
|
1834 | 1856 | More specifically, return a list of revision numbers corresponding to |
|
1835 | 1857 | nodes N such that every N satisfies the following constraints: |
|
1836 | 1858 | |
|
1837 | 1859 | 1. N is an ancestor of some node in 'heads' |
|
1838 | 1860 | 2. N is not an ancestor of any node in 'common' |
|
1839 | 1861 | |
|
1840 | 1862 | The list is sorted by revision number, meaning it is |
|
1841 | 1863 | topologically sorted. |
|
1842 | 1864 | |
|
1843 | 1865 | 'heads' and 'common' are both lists of revision numbers. If heads is |
|
1844 | 1866 | not supplied, uses all of the revlog's heads. If common is not |
|
1845 | 1867 | supplied, uses nullid.""" |
|
1846 | 1868 | if common is None: |
|
1847 | 1869 | common = [nullrev] |
|
1848 | 1870 | if heads is None: |
|
1849 | 1871 | heads = self.headrevs() |
|
1850 | 1872 | |
|
1851 | 1873 | inc = self.incrementalmissingrevs(common=common) |
|
1852 | 1874 | return inc.missingancestors(heads) |
|
1853 | 1875 | |
|
1854 | 1876 | def findmissing(self, common=None, heads=None): |
|
1855 | 1877 | """Return the ancestors of heads that are not ancestors of common. |
|
1856 | 1878 | |
|
1857 | 1879 | More specifically, return a list of nodes N such that every N |
|
1858 | 1880 | satisfies the following constraints: |
|
1859 | 1881 | |
|
1860 | 1882 | 1. N is an ancestor of some node in 'heads' |
|
1861 | 1883 | 2. N is not an ancestor of any node in 'common' |
|
1862 | 1884 | |
|
1863 | 1885 | The list is sorted by revision number, meaning it is |
|
1864 | 1886 | topologically sorted. |
|
1865 | 1887 | |
|
1866 | 1888 | 'heads' and 'common' are both lists of node IDs. If heads is |
|
1867 | 1889 | not supplied, uses all of the revlog's heads. If common is not |
|
1868 | 1890 | supplied, uses nullid.""" |
|
1869 | 1891 | if common is None: |
|
1870 | 1892 | common = [self.nullid] |
|
1871 | 1893 | if heads is None: |
|
1872 | 1894 | heads = self.heads() |
|
1873 | 1895 | |
|
1874 | 1896 | common = [self.rev(n) for n in common] |
|
1875 | 1897 | heads = [self.rev(n) for n in heads] |
|
1876 | 1898 | |
|
1877 | 1899 | inc = self.incrementalmissingrevs(common=common) |
|
1878 | 1900 | return [self.node(r) for r in inc.missingancestors(heads)] |
|
1879 | 1901 | |
|
1880 | 1902 | def nodesbetween(self, roots=None, heads=None): |
|
1881 | 1903 | """Return a topological path from 'roots' to 'heads'. |
|
1882 | 1904 | |
|
1883 | 1905 | Return a tuple (nodes, outroots, outheads) where 'nodes' is a |
|
1884 | 1906 | topologically sorted list of all nodes N that satisfy both of |
|
1885 | 1907 | these constraints: |
|
1886 | 1908 | |
|
1887 | 1909 | 1. N is a descendant of some node in 'roots' |
|
1888 | 1910 | 2. N is an ancestor of some node in 'heads' |
|
1889 | 1911 | |
|
1890 | 1912 | Every node is considered to be both a descendant and an ancestor |
|
1891 | 1913 | of itself, so every reachable node in 'roots' and 'heads' will be |
|
1892 | 1914 | included in 'nodes'. |
|
1893 | 1915 | |
|
1894 | 1916 | 'outroots' is the list of reachable nodes in 'roots', i.e., the |
|
1895 | 1917 | subset of 'roots' that is returned in 'nodes'. Likewise, |
|
1896 | 1918 | 'outheads' is the subset of 'heads' that is also in 'nodes'. |
|
1897 | 1919 | |
|
1898 | 1920 | 'roots' and 'heads' are both lists of node IDs. If 'roots' is |
|
1899 | 1921 | unspecified, uses nullid as the only root. If 'heads' is |
|
1900 | 1922 | unspecified, uses list of all of the revlog's heads.""" |
|
1901 | 1923 | nonodes = ([], [], []) |
|
1902 | 1924 | if roots is not None: |
|
1903 | 1925 | roots = list(roots) |
|
1904 | 1926 | if not roots: |
|
1905 | 1927 | return nonodes |
|
1906 | 1928 | lowestrev = min([self.rev(n) for n in roots]) |
|
1907 | 1929 | else: |
|
1908 | 1930 | roots = [self.nullid] # Everybody's a descendant of nullid |
|
1909 | 1931 | lowestrev = nullrev |
|
1910 | 1932 | if (lowestrev == nullrev) and (heads is None): |
|
1911 | 1933 | # We want _all_ the nodes! |
|
1912 | 1934 | return ( |
|
1913 | 1935 | [self.node(r) for r in self], |
|
1914 | 1936 | [self.nullid], |
|
1915 | 1937 | list(self.heads()), |
|
1916 | 1938 | ) |
|
1917 | 1939 | if heads is None: |
|
1918 | 1940 | # All nodes are ancestors, so the latest ancestor is the last |
|
1919 | 1941 | # node. |
|
1920 | 1942 | highestrev = len(self) - 1 |
|
1921 | 1943 | # Set ancestors to None to signal that every node is an ancestor. |
|
1922 | 1944 | ancestors = None |
|
1923 | 1945 | # Set heads to an empty dictionary for later discovery of heads |
|
1924 | 1946 | heads = {} |
|
1925 | 1947 | else: |
|
1926 | 1948 | heads = list(heads) |
|
1927 | 1949 | if not heads: |
|
1928 | 1950 | return nonodes |
|
1929 | 1951 | ancestors = set() |
|
1930 | 1952 | # Turn heads into a dictionary so we can remove 'fake' heads. |
|
1931 | 1953 | # Also, later we will be using it to filter out the heads we can't |
|
1932 | 1954 | # find from roots. |
|
1933 | 1955 | heads = dict.fromkeys(heads, False) |
|
1934 | 1956 | # Start at the top and keep marking parents until we're done. |
|
1935 | 1957 | nodestotag = set(heads) |
|
1936 | 1958 | # Remember where the top was so we can use it as a limit later. |
|
1937 | 1959 | highestrev = max([self.rev(n) for n in nodestotag]) |
|
1938 | 1960 | while nodestotag: |
|
1939 | 1961 | # grab a node to tag |
|
1940 | 1962 | n = nodestotag.pop() |
|
1941 | 1963 | # Never tag nullid |
|
1942 | 1964 | if n == self.nullid: |
|
1943 | 1965 | continue |
|
1944 | 1966 | # A node's revision number represents its place in a |
|
1945 | 1967 | # topologically sorted list of nodes. |
|
1946 | 1968 | r = self.rev(n) |
|
1947 | 1969 | if r >= lowestrev: |
|
1948 | 1970 | if n not in ancestors: |
|
1949 | 1971 | # If we are possibly a descendant of one of the roots |
|
1950 | 1972 | # and we haven't already been marked as an ancestor |
|
1951 | 1973 | ancestors.add(n) # Mark as ancestor |
|
1952 | 1974 | # Add non-nullid parents to list of nodes to tag. |
|
1953 | 1975 | nodestotag.update( |
|
1954 | 1976 | [p for p in self.parents(n) if p != self.nullid] |
|
1955 | 1977 | ) |
|
1956 | 1978 | elif n in heads: # We've seen it before, is it a fake head? |
|
1957 | 1979 | # So it is, real heads should not be the ancestors of |
|
1958 | 1980 | # any other heads. |
|
1959 | 1981 | heads.pop(n) |
|
1960 | 1982 | if not ancestors: |
|
1961 | 1983 | return nonodes |
|
1962 | 1984 | # Now that we have our set of ancestors, we want to remove any |
|
1963 | 1985 | # roots that are not ancestors. |
|
1964 | 1986 | |
|
1965 | 1987 | # If one of the roots was nullid, everything is included anyway. |
|
1966 | 1988 | if lowestrev > nullrev: |
|
1967 | 1989 | # But, since we weren't, let's recompute the lowest rev to not |
|
1968 | 1990 | # include roots that aren't ancestors. |
|
1969 | 1991 | |
|
1970 | 1992 | # Filter out roots that aren't ancestors of heads |
|
1971 | 1993 | roots = [root for root in roots if root in ancestors] |
|
1972 | 1994 | # Recompute the lowest revision |
|
1973 | 1995 | if roots: |
|
1974 | 1996 | lowestrev = min([self.rev(root) for root in roots]) |
|
1975 | 1997 | else: |
|
1976 | 1998 | # No more roots? Return empty list |
|
1977 | 1999 | return nonodes |
|
1978 | 2000 | else: |
|
1979 | 2001 | # We are descending from nullid, and don't need to care about |
|
1980 | 2002 | # any other roots. |
|
1981 | 2003 | lowestrev = nullrev |
|
1982 | 2004 | roots = [self.nullid] |
|
1983 | 2005 | # Transform our roots list into a set. |
|
1984 | 2006 | descendants = set(roots) |
|
1985 | 2007 | # Also, keep the original roots so we can filter out roots that aren't |
|
1986 | 2008 | # 'real' roots (i.e. are descended from other roots). |
|
1987 | 2009 | roots = descendants.copy() |
|
1988 | 2010 | # Our topologically sorted list of output nodes. |
|
1989 | 2011 | orderedout = [] |
|
1990 | 2012 | # Don't start at nullid since we don't want nullid in our output list, |
|
1991 | 2013 | # and if nullid shows up in descendants, empty parents will look like |
|
1992 | 2014 | # they're descendants. |
|
1993 | 2015 | for r in self.revs(start=max(lowestrev, 0), stop=highestrev + 1): |
|
1994 | 2016 | n = self.node(r) |
|
1995 | 2017 | isdescendant = False |
|
1996 | 2018 | if lowestrev == nullrev: # Everybody is a descendant of nullid |
|
1997 | 2019 | isdescendant = True |
|
1998 | 2020 | elif n in descendants: |
|
1999 | 2021 | # n is already a descendant |
|
2000 | 2022 | isdescendant = True |
|
2001 | 2023 | # This check only needs to be done here because all the roots |
|
2002 | 2024 | # will start being marked is descendants before the loop. |
|
2003 | 2025 | if n in roots: |
|
2004 | 2026 | # If n was a root, check if it's a 'real' root. |
|
2005 | 2027 | p = tuple(self.parents(n)) |
|
2006 | 2028 | # If any of its parents are descendants, it's not a root. |
|
2007 | 2029 | if (p[0] in descendants) or (p[1] in descendants): |
|
2008 | 2030 | roots.remove(n) |
|
2009 | 2031 | else: |
|
2010 | 2032 | p = tuple(self.parents(n)) |
|
2011 | 2033 | # A node is a descendant if either of its parents are |
|
2012 | 2034 | # descendants. (We seeded the dependents list with the roots |
|
2013 | 2035 | # up there, remember?) |
|
2014 | 2036 | if (p[0] in descendants) or (p[1] in descendants): |
|
2015 | 2037 | descendants.add(n) |
|
2016 | 2038 | isdescendant = True |
|
2017 | 2039 | if isdescendant and ((ancestors is None) or (n in ancestors)): |
|
2018 | 2040 | # Only include nodes that are both descendants and ancestors. |
|
2019 | 2041 | orderedout.append(n) |
|
2020 | 2042 | if (ancestors is not None) and (n in heads): |
|
2021 | 2043 | # We're trying to figure out which heads are reachable |
|
2022 | 2044 | # from roots. |
|
2023 | 2045 | # Mark this head as having been reached |
|
2024 | 2046 | heads[n] = True |
|
2025 | 2047 | elif ancestors is None: |
|
2026 | 2048 | # Otherwise, we're trying to discover the heads. |
|
2027 | 2049 | # Assume this is a head because if it isn't, the next step |
|
2028 | 2050 | # will eventually remove it. |
|
2029 | 2051 | heads[n] = True |
|
2030 | 2052 | # But, obviously its parents aren't. |
|
2031 | 2053 | for p in self.parents(n): |
|
2032 | 2054 | heads.pop(p, None) |
|
2033 | 2055 | heads = [head for head, flag in heads.items() if flag] |
|
2034 | 2056 | roots = list(roots) |
|
2035 | 2057 | assert orderedout |
|
2036 | 2058 | assert roots |
|
2037 | 2059 | assert heads |
|
2038 | 2060 | return (orderedout, roots, heads) |
|
2039 | 2061 | |
|
2040 | 2062 | def headrevs(self, revs=None): |
|
2041 | 2063 | if revs is None: |
|
2042 | 2064 | try: |
|
2043 | 2065 | return self.index.headrevs() |
|
2044 | 2066 | except AttributeError: |
|
2045 | 2067 | return self._headrevs() |
|
2046 | 2068 | if rustdagop is not None and self.index.rust_ext_compat: |
|
2047 | 2069 | return rustdagop.headrevs(self.index, revs) |
|
2048 | 2070 | return dagop.headrevs(revs, self._uncheckedparentrevs) |
|
2049 | 2071 | |
|
2050 | 2072 | def computephases(self, roots): |
|
2051 | 2073 | return self.index.computephasesmapsets(roots) |
|
2052 | 2074 | |
|
2053 | 2075 | def _headrevs(self): |
|
2054 | 2076 | count = len(self) |
|
2055 | 2077 | if not count: |
|
2056 | 2078 | return [nullrev] |
|
2057 | 2079 | # we won't iter over filtered rev so nobody is a head at start |
|
2058 | 2080 | ishead = [0] * (count + 1) |
|
2059 | 2081 | index = self.index |
|
2060 | 2082 | for r in self: |
|
2061 | 2083 | ishead[r] = 1 # I may be an head |
|
2062 | 2084 | e = index[r] |
|
2063 | 2085 | ishead[e[5]] = ishead[e[6]] = 0 # my parent are not |
|
2064 | 2086 | return [r for r, val in enumerate(ishead) if val] |
|
2065 | 2087 | |
|
2066 | 2088 | def heads(self, start=None, stop=None): |
|
2067 | 2089 | """return the list of all nodes that have no children |
|
2068 | 2090 | |
|
2069 | 2091 | if start is specified, only heads that are descendants of |
|
2070 | 2092 | start will be returned |
|
2071 | 2093 | if stop is specified, it will consider all the revs from stop |
|
2072 | 2094 | as if they had no children |
|
2073 | 2095 | """ |
|
2074 | 2096 | if start is None and stop is None: |
|
2075 | 2097 | if not len(self): |
|
2076 | 2098 | return [self.nullid] |
|
2077 | 2099 | return [self.node(r) for r in self.headrevs()] |
|
2078 | 2100 | |
|
2079 | 2101 | if start is None: |
|
2080 | 2102 | start = nullrev |
|
2081 | 2103 | else: |
|
2082 | 2104 | start = self.rev(start) |
|
2083 | 2105 | |
|
2084 | 2106 | stoprevs = {self.rev(n) for n in stop or []} |
|
2085 | 2107 | |
|
2086 | 2108 | revs = dagop.headrevssubset( |
|
2087 | 2109 | self.revs, self.parentrevs, startrev=start, stoprevs=stoprevs |
|
2088 | 2110 | ) |
|
2089 | 2111 | |
|
2090 | 2112 | return [self.node(rev) for rev in revs] |
|
2091 | 2113 | |
|
2092 | 2114 | def children(self, node): |
|
2093 | 2115 | """find the children of a given node""" |
|
2094 | 2116 | c = [] |
|
2095 | 2117 | p = self.rev(node) |
|
2096 | 2118 | for r in self.revs(start=p + 1): |
|
2097 | 2119 | prevs = [pr for pr in self.parentrevs(r) if pr != nullrev] |
|
2098 | 2120 | if prevs: |
|
2099 | 2121 | for pr in prevs: |
|
2100 | 2122 | if pr == p: |
|
2101 | 2123 | c.append(self.node(r)) |
|
2102 | 2124 | elif p == nullrev: |
|
2103 | 2125 | c.append(self.node(r)) |
|
2104 | 2126 | return c |
|
2105 | 2127 | |
|
2106 | 2128 | def commonancestorsheads(self, a, b): |
|
2107 | 2129 | """calculate all the heads of the common ancestors of nodes a and b""" |
|
2108 | 2130 | a, b = self.rev(a), self.rev(b) |
|
2109 | 2131 | ancs = self._commonancestorsheads(a, b) |
|
2110 | 2132 | return pycompat.maplist(self.node, ancs) |
|
2111 | 2133 | |
|
2112 | 2134 | def _commonancestorsheads(self, *revs): |
|
2113 | 2135 | """calculate all the heads of the common ancestors of revs""" |
|
2114 | 2136 | try: |
|
2115 | 2137 | ancs = self.index.commonancestorsheads(*revs) |
|
2116 | 2138 | except (AttributeError, OverflowError): # C implementation failed |
|
2117 | 2139 | ancs = ancestor.commonancestorsheads(self.parentrevs, *revs) |
|
2118 | 2140 | return ancs |
|
2119 | 2141 | |
|
2120 | 2142 | def isancestor(self, a, b): |
|
2121 | 2143 | """return True if node a is an ancestor of node b |
|
2122 | 2144 | |
|
2123 | 2145 | A revision is considered an ancestor of itself.""" |
|
2124 | 2146 | a, b = self.rev(a), self.rev(b) |
|
2125 | 2147 | return self.isancestorrev(a, b) |
|
2126 | 2148 | |
|
2127 | 2149 | def isancestorrev(self, a, b): |
|
2128 | 2150 | """return True if revision a is an ancestor of revision b |
|
2129 | 2151 | |
|
2130 | 2152 | A revision is considered an ancestor of itself. |
|
2131 | 2153 | |
|
2132 | 2154 | The implementation of this is trivial but the use of |
|
2133 | 2155 | reachableroots is not.""" |
|
2134 | 2156 | if a == nullrev: |
|
2135 | 2157 | return True |
|
2136 | 2158 | elif a == b: |
|
2137 | 2159 | return True |
|
2138 | 2160 | elif a > b: |
|
2139 | 2161 | return False |
|
2140 | 2162 | return bool(self.reachableroots(a, [b], [a], includepath=False)) |
|
2141 | 2163 | |
|
2142 | 2164 | def reachableroots(self, minroot, heads, roots, includepath=False): |
|
2143 | 2165 | """return (heads(::(<roots> and <roots>::<heads>))) |
|
2144 | 2166 | |
|
2145 | 2167 | If includepath is True, return (<roots>::<heads>).""" |
|
2146 | 2168 | try: |
|
2147 | 2169 | return self.index.reachableroots2( |
|
2148 | 2170 | minroot, heads, roots, includepath |
|
2149 | 2171 | ) |
|
2150 | 2172 | except AttributeError: |
|
2151 | 2173 | return dagop._reachablerootspure( |
|
2152 | 2174 | self.parentrevs, minroot, roots, heads, includepath |
|
2153 | 2175 | ) |
|
2154 | 2176 | |
|
2155 | 2177 | def ancestor(self, a, b): |
|
2156 | 2178 | """calculate the "best" common ancestor of nodes a and b""" |
|
2157 | 2179 | |
|
2158 | 2180 | a, b = self.rev(a), self.rev(b) |
|
2159 | 2181 | try: |
|
2160 | 2182 | ancs = self.index.ancestors(a, b) |
|
2161 | 2183 | except (AttributeError, OverflowError): |
|
2162 | 2184 | ancs = ancestor.ancestors(self.parentrevs, a, b) |
|
2163 | 2185 | if ancs: |
|
2164 | 2186 | # choose a consistent winner when there's a tie |
|
2165 | 2187 | return min(map(self.node, ancs)) |
|
2166 | 2188 | return self.nullid |
|
2167 | 2189 | |
|
2168 | 2190 | def _match(self, id): |
|
2169 | 2191 | if isinstance(id, int): |
|
2170 | 2192 | # rev |
|
2171 | 2193 | return self.node(id) |
|
2172 | 2194 | if len(id) == self.nodeconstants.nodelen: |
|
2173 | 2195 | # possibly a binary node |
|
2174 | 2196 | # odds of a binary node being all hex in ASCII are 1 in 10**25 |
|
2175 | 2197 | try: |
|
2176 | 2198 | node = id |
|
2177 | 2199 | self.rev(node) # quick search the index |
|
2178 | 2200 | return node |
|
2179 | 2201 | except error.LookupError: |
|
2180 | 2202 | pass # may be partial hex id |
|
2181 | 2203 | try: |
|
2182 | 2204 | # str(rev) |
|
2183 | 2205 | rev = int(id) |
|
2184 | 2206 | if b"%d" % rev != id: |
|
2185 | 2207 | raise ValueError |
|
2186 | 2208 | if rev < 0: |
|
2187 | 2209 | rev = len(self) + rev |
|
2188 | 2210 | if rev < 0 or rev >= len(self): |
|
2189 | 2211 | raise ValueError |
|
2190 | 2212 | return self.node(rev) |
|
2191 | 2213 | except (ValueError, OverflowError): |
|
2192 | 2214 | pass |
|
2193 | 2215 | if len(id) == 2 * self.nodeconstants.nodelen: |
|
2194 | 2216 | try: |
|
2195 | 2217 | # a full hex nodeid? |
|
2196 | 2218 | node = bin(id) |
|
2197 | 2219 | self.rev(node) |
|
2198 | 2220 | return node |
|
2199 | 2221 | except (binascii.Error, error.LookupError): |
|
2200 | 2222 | pass |
|
2201 | 2223 | |
|
2202 | 2224 | def _partialmatch(self, id): |
|
2203 | 2225 | # we don't care wdirfilenodeids as they should be always full hash |
|
2204 | 2226 | maybewdir = self.nodeconstants.wdirhex.startswith(id) |
|
2205 | 2227 | ambiguous = False |
|
2206 | 2228 | try: |
|
2207 | 2229 | partial = self.index.partialmatch(id) |
|
2208 | 2230 | if partial and self.hasnode(partial): |
|
2209 | 2231 | if maybewdir: |
|
2210 | 2232 | # single 'ff...' match in radix tree, ambiguous with wdir |
|
2211 | 2233 | ambiguous = True |
|
2212 | 2234 | else: |
|
2213 | 2235 | return partial |
|
2214 | 2236 | elif maybewdir: |
|
2215 | 2237 | # no 'ff...' match in radix tree, wdir identified |
|
2216 | 2238 | raise error.WdirUnsupported |
|
2217 | 2239 | else: |
|
2218 | 2240 | return None |
|
2219 | 2241 | except error.RevlogError: |
|
2220 | 2242 | # parsers.c radix tree lookup gave multiple matches |
|
2221 | 2243 | # fast path: for unfiltered changelog, radix tree is accurate |
|
2222 | 2244 | if not getattr(self, 'filteredrevs', None): |
|
2223 | 2245 | ambiguous = True |
|
2224 | 2246 | # fall through to slow path that filters hidden revisions |
|
2225 | 2247 | except (AttributeError, ValueError): |
|
2226 | 2248 | # we are pure python, or key is not hex |
|
2227 | 2249 | pass |
|
2228 | 2250 | if ambiguous: |
|
2229 | 2251 | raise error.AmbiguousPrefixLookupError( |
|
2230 | 2252 | id, self.display_id, _(b'ambiguous identifier') |
|
2231 | 2253 | ) |
|
2232 | 2254 | |
|
2233 | 2255 | if id in self._pcache: |
|
2234 | 2256 | return self._pcache[id] |
|
2235 | 2257 | |
|
2236 | 2258 | if len(id) <= 40: |
|
2237 | 2259 | # hex(node)[:...] |
|
2238 | 2260 | l = len(id) // 2 * 2 # grab an even number of digits |
|
2239 | 2261 | try: |
|
2240 | 2262 | # we're dropping the last digit, so let's check that it's hex, |
|
2241 | 2263 | # to avoid the expensive computation below if it's not |
|
2242 | 2264 | if len(id) % 2 > 0: |
|
2243 | 2265 | if not (id[-1] in hexdigits): |
|
2244 | 2266 | return None |
|
2245 | 2267 | prefix = bin(id[:l]) |
|
2246 | 2268 | except binascii.Error: |
|
2247 | 2269 | pass |
|
2248 | 2270 | else: |
|
2249 | 2271 | nl = [e[7] for e in self.index if e[7].startswith(prefix)] |
|
2250 | 2272 | nl = [ |
|
2251 | 2273 | n for n in nl if hex(n).startswith(id) and self.hasnode(n) |
|
2252 | 2274 | ] |
|
2253 | 2275 | if self.nodeconstants.nullhex.startswith(id): |
|
2254 | 2276 | nl.append(self.nullid) |
|
2255 | 2277 | if len(nl) > 0: |
|
2256 | 2278 | if len(nl) == 1 and not maybewdir: |
|
2257 | 2279 | self._pcache[id] = nl[0] |
|
2258 | 2280 | return nl[0] |
|
2259 | 2281 | raise error.AmbiguousPrefixLookupError( |
|
2260 | 2282 | id, self.display_id, _(b'ambiguous identifier') |
|
2261 | 2283 | ) |
|
2262 | 2284 | if maybewdir: |
|
2263 | 2285 | raise error.WdirUnsupported |
|
2264 | 2286 | return None |
|
2265 | 2287 | |
|
2266 | 2288 | def lookup(self, id): |
|
2267 | 2289 | """locate a node based on: |
|
2268 | 2290 | - revision number or str(revision number) |
|
2269 | 2291 | - nodeid or subset of hex nodeid |
|
2270 | 2292 | """ |
|
2271 | 2293 | n = self._match(id) |
|
2272 | 2294 | if n is not None: |
|
2273 | 2295 | return n |
|
2274 | 2296 | n = self._partialmatch(id) |
|
2275 | 2297 | if n: |
|
2276 | 2298 | return n |
|
2277 | 2299 | |
|
2278 | 2300 | raise error.LookupError(id, self.display_id, _(b'no match found')) |
|
2279 | 2301 | |
|
2280 | 2302 | def shortest(self, node, minlength=1): |
|
2281 | 2303 | """Find the shortest unambiguous prefix that matches node.""" |
|
2282 | 2304 | |
|
2283 | 2305 | def isvalid(prefix): |
|
2284 | 2306 | try: |
|
2285 | 2307 | matchednode = self._partialmatch(prefix) |
|
2286 | 2308 | except error.AmbiguousPrefixLookupError: |
|
2287 | 2309 | return False |
|
2288 | 2310 | except error.WdirUnsupported: |
|
2289 | 2311 | # single 'ff...' match |
|
2290 | 2312 | return True |
|
2291 | 2313 | if matchednode is None: |
|
2292 | 2314 | raise error.LookupError(node, self.display_id, _(b'no node')) |
|
2293 | 2315 | return True |
|
2294 | 2316 | |
|
2295 | 2317 | def maybewdir(prefix): |
|
2296 | 2318 | return all(c == b'f' for c in pycompat.iterbytestr(prefix)) |
|
2297 | 2319 | |
|
2298 | 2320 | hexnode = hex(node) |
|
2299 | 2321 | |
|
2300 | 2322 | def disambiguate(hexnode, minlength): |
|
2301 | 2323 | """Disambiguate against wdirid.""" |
|
2302 | 2324 | for length in range(minlength, len(hexnode) + 1): |
|
2303 | 2325 | prefix = hexnode[:length] |
|
2304 | 2326 | if not maybewdir(prefix): |
|
2305 | 2327 | return prefix |
|
2306 | 2328 | |
|
2307 | 2329 | if not getattr(self, 'filteredrevs', None): |
|
2308 | 2330 | try: |
|
2309 | 2331 | length = max(self.index.shortest(node), minlength) |
|
2310 | 2332 | return disambiguate(hexnode, length) |
|
2311 | 2333 | except error.RevlogError: |
|
2312 | 2334 | if node != self.nodeconstants.wdirid: |
|
2313 | 2335 | raise error.LookupError( |
|
2314 | 2336 | node, self.display_id, _(b'no node') |
|
2315 | 2337 | ) |
|
2316 | 2338 | except AttributeError: |
|
2317 | 2339 | # Fall through to pure code |
|
2318 | 2340 | pass |
|
2319 | 2341 | |
|
2320 | 2342 | if node == self.nodeconstants.wdirid: |
|
2321 | 2343 | for length in range(minlength, len(hexnode) + 1): |
|
2322 | 2344 | prefix = hexnode[:length] |
|
2323 | 2345 | if isvalid(prefix): |
|
2324 | 2346 | return prefix |
|
2325 | 2347 | |
|
2326 | 2348 | for length in range(minlength, len(hexnode) + 1): |
|
2327 | 2349 | prefix = hexnode[:length] |
|
2328 | 2350 | if isvalid(prefix): |
|
2329 | 2351 | return disambiguate(hexnode, length) |
|
2330 | 2352 | |
|
2331 | 2353 | def cmp(self, node, text): |
|
2332 | 2354 | """compare text with a given file revision |
|
2333 | 2355 | |
|
2334 | 2356 | returns True if text is different than what is stored. |
|
2335 | 2357 | """ |
|
2336 | 2358 | p1, p2 = self.parents(node) |
|
2337 | 2359 | return storageutil.hashrevisionsha1(text, p1, p2) != node |
|
2338 | 2360 | |
|
2339 | def _chunk(self, rev): | |
|
2340 | """Obtain a single decompressed chunk for a revision. | |
|
2341 | ||
|
2342 | Accepts an integer revision and an optional already-open file handle | |
|
2343 | to be used for reading. If used, the seek position of the file will not | |
|
2344 | be preserved. | |
|
2345 | ||
|
2346 | Returns a str holding uncompressed data for the requested revision. | |
|
2347 | """ | |
|
2348 | compression_mode = self.index[rev][10] | |
|
2349 | data = self._inner.get_segment_for_revs(rev, rev)[1] | |
|
2350 | if compression_mode == COMP_MODE_PLAIN: | |
|
2351 | return data | |
|
2352 | elif compression_mode == COMP_MODE_DEFAULT: | |
|
2353 | return self._inner._decompressor(data) | |
|
2354 | elif compression_mode == COMP_MODE_INLINE: | |
|
2355 | return self._inner.decompress(data) | |
|
2356 | else: | |
|
2357 | msg = b'unknown compression mode %d' | |
|
2358 | msg %= compression_mode | |
|
2359 | raise error.RevlogError(msg) | |
|
2360 | ||
|
2361 | 2361 | def _chunks(self, revs, targetsize=None): |
|
2362 | 2362 | """Obtain decompressed chunks for the specified revisions. |
|
2363 | 2363 | |
|
2364 | 2364 | Accepts an iterable of numeric revisions that are assumed to be in |
|
2365 | 2365 | ascending order. Also accepts an optional already-open file handle |
|
2366 | 2366 | to be used for reading. If used, the seek position of the file will |
|
2367 | 2367 | not be preserved. |
|
2368 | 2368 | |
|
2369 | 2369 | This function is similar to calling ``self._chunk()`` multiple times, |
|
2370 | 2370 | but is faster. |
|
2371 | 2371 | |
|
2372 | 2372 | Returns a list with decompressed data for each requested revision. |
|
2373 | 2373 | """ |
|
2374 | 2374 | if not revs: |
|
2375 | 2375 | return [] |
|
2376 | 2376 | start = self.start |
|
2377 | 2377 | length = self.length |
|
2378 | 2378 | inline = self._inline |
|
2379 | 2379 | iosize = self.index.entry_size |
|
2380 | 2380 | buffer = util.buffer |
|
2381 | 2381 | |
|
2382 | 2382 | l = [] |
|
2383 | 2383 | ladd = l.append |
|
2384 | 2384 | |
|
2385 | 2385 | if not self.data_config.with_sparse_read: |
|
2386 | 2386 | slicedchunks = (revs,) |
|
2387 | 2387 | else: |
|
2388 | 2388 | slicedchunks = deltautil.slicechunk( |
|
2389 | 2389 | self, revs, targetsize=targetsize |
|
2390 | 2390 | ) |
|
2391 | 2391 | |
|
2392 | 2392 | for revschunk in slicedchunks: |
|
2393 | 2393 | firstrev = revschunk[0] |
|
2394 | 2394 | # Skip trailing revisions with empty diff |
|
2395 | 2395 | for lastrev in revschunk[::-1]: |
|
2396 | 2396 | if length(lastrev) != 0: |
|
2397 | 2397 | break |
|
2398 | 2398 | |
|
2399 | 2399 | try: |
|
2400 | 2400 | offset, data = self._inner.get_segment_for_revs( |
|
2401 | 2401 | firstrev, |
|
2402 | 2402 | lastrev, |
|
2403 | 2403 | ) |
|
2404 | 2404 | except OverflowError: |
|
2405 | 2405 | # issue4215 - we can't cache a run of chunks greater than |
|
2406 | 2406 | # 2G on Windows |
|
2407 | return [self._chunk(rev) for rev in revschunk] | |
|
2407 | return [self._inner._chunk(rev) for rev in revschunk] | |
|
2408 | 2408 | |
|
2409 | 2409 | decomp = self._inner.decompress |
|
2410 | 2410 | # self._decompressor might be None, but will not be used in that case |
|
2411 | 2411 | def_decomp = self._inner._decompressor |
|
2412 | 2412 | for rev in revschunk: |
|
2413 | 2413 | chunkstart = start(rev) |
|
2414 | 2414 | if inline: |
|
2415 | 2415 | chunkstart += (rev + 1) * iosize |
|
2416 | 2416 | chunklength = length(rev) |
|
2417 | 2417 | comp_mode = self.index[rev][10] |
|
2418 | 2418 | c = buffer(data, chunkstart - offset, chunklength) |
|
2419 | 2419 | if comp_mode == COMP_MODE_PLAIN: |
|
2420 | 2420 | ladd(c) |
|
2421 | 2421 | elif comp_mode == COMP_MODE_INLINE: |
|
2422 | 2422 | ladd(decomp(c)) |
|
2423 | 2423 | elif comp_mode == COMP_MODE_DEFAULT: |
|
2424 | 2424 | ladd(def_decomp(c)) |
|
2425 | 2425 | else: |
|
2426 | 2426 | msg = b'unknown compression mode %d' |
|
2427 | 2427 | msg %= comp_mode |
|
2428 | 2428 | raise error.RevlogError(msg) |
|
2429 | 2429 | |
|
2430 | 2430 | return l |
|
2431 | 2431 | |
|
2432 | 2432 | def deltaparent(self, rev): |
|
2433 | 2433 | """return deltaparent of the given revision""" |
|
2434 | 2434 | base = self.index[rev][3] |
|
2435 | 2435 | if base == rev: |
|
2436 | 2436 | return nullrev |
|
2437 | 2437 | elif self.delta_config.general_delta: |
|
2438 | 2438 | return base |
|
2439 | 2439 | else: |
|
2440 | 2440 | return rev - 1 |
|
2441 | 2441 | |
|
2442 | 2442 | def issnapshot(self, rev): |
|
2443 | 2443 | """tells whether rev is a snapshot""" |
|
2444 | 2444 | if not self.delta_config.sparse_revlog: |
|
2445 | 2445 | return self.deltaparent(rev) == nullrev |
|
2446 | 2446 | elif hasattr(self.index, 'issnapshot'): |
|
2447 | 2447 | # directly assign the method to cache the testing and access |
|
2448 | 2448 | self.issnapshot = self.index.issnapshot |
|
2449 | 2449 | return self.issnapshot(rev) |
|
2450 | 2450 | if rev == nullrev: |
|
2451 | 2451 | return True |
|
2452 | 2452 | entry = self.index[rev] |
|
2453 | 2453 | base = entry[3] |
|
2454 | 2454 | if base == rev: |
|
2455 | 2455 | return True |
|
2456 | 2456 | if base == nullrev: |
|
2457 | 2457 | return True |
|
2458 | 2458 | p1 = entry[5] |
|
2459 | 2459 | while self.length(p1) == 0: |
|
2460 | 2460 | b = self.deltaparent(p1) |
|
2461 | 2461 | if b == p1: |
|
2462 | 2462 | break |
|
2463 | 2463 | p1 = b |
|
2464 | 2464 | p2 = entry[6] |
|
2465 | 2465 | while self.length(p2) == 0: |
|
2466 | 2466 | b = self.deltaparent(p2) |
|
2467 | 2467 | if b == p2: |
|
2468 | 2468 | break |
|
2469 | 2469 | p2 = b |
|
2470 | 2470 | if base == p1 or base == p2: |
|
2471 | 2471 | return False |
|
2472 | 2472 | return self.issnapshot(base) |
|
2473 | 2473 | |
|
2474 | 2474 | def snapshotdepth(self, rev): |
|
2475 | 2475 | """number of snapshot in the chain before this one""" |
|
2476 | 2476 | if not self.issnapshot(rev): |
|
2477 | 2477 | raise error.ProgrammingError(b'revision %d not a snapshot') |
|
2478 | 2478 | return len(self._deltachain(rev)[0]) - 1 |
|
2479 | 2479 | |
|
2480 | 2480 | def revdiff(self, rev1, rev2): |
|
2481 | 2481 | """return or calculate a delta between two revisions |
|
2482 | 2482 | |
|
2483 | 2483 | The delta calculated is in binary form and is intended to be written to |
|
2484 | 2484 | revlog data directly. So this function needs raw revision data. |
|
2485 | 2485 | """ |
|
2486 | 2486 | if rev1 != nullrev and self.deltaparent(rev2) == rev1: |
|
2487 | return bytes(self._chunk(rev2)) | |
|
2487 | return bytes(self._inner._chunk(rev2)) | |
|
2488 | 2488 | |
|
2489 | 2489 | return mdiff.textdiff(self.rawdata(rev1), self.rawdata(rev2)) |
|
2490 | 2490 | |
|
2491 | 2491 | def revision(self, nodeorrev): |
|
2492 | 2492 | """return an uncompressed revision of a given node or revision |
|
2493 | 2493 | number. |
|
2494 | 2494 | """ |
|
2495 | 2495 | return self._revisiondata(nodeorrev) |
|
2496 | 2496 | |
|
2497 | 2497 | def sidedata(self, nodeorrev): |
|
2498 | 2498 | """a map of extra data related to the changeset but not part of the hash |
|
2499 | 2499 | |
|
2500 | 2500 | This function currently return a dictionary. However, more advanced |
|
2501 | 2501 | mapping object will likely be used in the future for a more |
|
2502 | 2502 | efficient/lazy code. |
|
2503 | 2503 | """ |
|
2504 | 2504 | # deal with <nodeorrev> argument type |
|
2505 | 2505 | if isinstance(nodeorrev, int): |
|
2506 | 2506 | rev = nodeorrev |
|
2507 | 2507 | else: |
|
2508 | 2508 | rev = self.rev(nodeorrev) |
|
2509 | 2509 | return self._sidedata(rev) |
|
2510 | 2510 | |
|
2511 | 2511 | def _revisiondata(self, nodeorrev, raw=False): |
|
2512 | 2512 | # deal with <nodeorrev> argument type |
|
2513 | 2513 | if isinstance(nodeorrev, int): |
|
2514 | 2514 | rev = nodeorrev |
|
2515 | 2515 | node = self.node(rev) |
|
2516 | 2516 | else: |
|
2517 | 2517 | node = nodeorrev |
|
2518 | 2518 | rev = None |
|
2519 | 2519 | |
|
2520 | 2520 | # fast path the special `nullid` rev |
|
2521 | 2521 | if node == self.nullid: |
|
2522 | 2522 | return b"" |
|
2523 | 2523 | |
|
2524 | 2524 | # ``rawtext`` is the text as stored inside the revlog. Might be the |
|
2525 | 2525 | # revision or might need to be processed to retrieve the revision. |
|
2526 | 2526 | rev, rawtext, validated = self._rawtext(node, rev) |
|
2527 | 2527 | |
|
2528 | 2528 | if raw and validated: |
|
2529 | 2529 | # if we don't want to process the raw text and that raw |
|
2530 | 2530 | # text is cached, we can exit early. |
|
2531 | 2531 | return rawtext |
|
2532 | 2532 | if rev is None: |
|
2533 | 2533 | rev = self.rev(node) |
|
2534 | 2534 | # the revlog's flag for this revision |
|
2535 | 2535 | # (usually alter its state or content) |
|
2536 | 2536 | flags = self.flags(rev) |
|
2537 | 2537 | |
|
2538 | 2538 | if validated and flags == REVIDX_DEFAULT_FLAGS: |
|
2539 | 2539 | # no extra flags set, no flag processor runs, text = rawtext |
|
2540 | 2540 | return rawtext |
|
2541 | 2541 | |
|
2542 | 2542 | if raw: |
|
2543 | 2543 | validatehash = flagutil.processflagsraw(self, rawtext, flags) |
|
2544 | 2544 | text = rawtext |
|
2545 | 2545 | else: |
|
2546 | 2546 | r = flagutil.processflagsread(self, rawtext, flags) |
|
2547 | 2547 | text, validatehash = r |
|
2548 | 2548 | if validatehash: |
|
2549 | 2549 | self.checkhash(text, node, rev=rev) |
|
2550 | 2550 | if not validated: |
|
2551 | 2551 | self._revisioncache = (node, rev, rawtext) |
|
2552 | 2552 | |
|
2553 | 2553 | return text |
|
2554 | 2554 | |
|
2555 | 2555 | def _rawtext(self, node, rev): |
|
2556 | 2556 | """return the possibly unvalidated rawtext for a revision |
|
2557 | 2557 | |
|
2558 | 2558 | returns (rev, rawtext, validated) |
|
2559 | 2559 | """ |
|
2560 | 2560 | |
|
2561 | 2561 | # revision in the cache (could be useful to apply delta) |
|
2562 | 2562 | cachedrev = None |
|
2563 | 2563 | # An intermediate text to apply deltas to |
|
2564 | 2564 | basetext = None |
|
2565 | 2565 | |
|
2566 | 2566 | # Check if we have the entry in cache |
|
2567 | 2567 | # The cache entry looks like (node, rev, rawtext) |
|
2568 | 2568 | if self._revisioncache: |
|
2569 | 2569 | if self._revisioncache[0] == node: |
|
2570 | 2570 | return (rev, self._revisioncache[2], True) |
|
2571 | 2571 | cachedrev = self._revisioncache[1] |
|
2572 | 2572 | |
|
2573 | 2573 | if rev is None: |
|
2574 | 2574 | rev = self.rev(node) |
|
2575 | 2575 | |
|
2576 | 2576 | chain, stopped = self._deltachain(rev, stoprev=cachedrev) |
|
2577 | 2577 | if stopped: |
|
2578 | 2578 | basetext = self._revisioncache[2] |
|
2579 | 2579 | |
|
2580 | 2580 | # drop cache to save memory, the caller is expected to |
|
2581 | 2581 | # update self._revisioncache after validating the text |
|
2582 | 2582 | self._revisioncache = None |
|
2583 | 2583 | |
|
2584 | 2584 | targetsize = None |
|
2585 | 2585 | rawsize = self.index[rev][2] |
|
2586 | 2586 | if 0 <= rawsize: |
|
2587 | 2587 | targetsize = 4 * rawsize |
|
2588 | 2588 | |
|
2589 | 2589 | bins = self._chunks(chain, targetsize=targetsize) |
|
2590 | 2590 | if basetext is None: |
|
2591 | 2591 | basetext = bytes(bins[0]) |
|
2592 | 2592 | bins = bins[1:] |
|
2593 | 2593 | |
|
2594 | 2594 | rawtext = mdiff.patches(basetext, bins) |
|
2595 | 2595 | del basetext # let us have a chance to free memory early |
|
2596 | 2596 | return (rev, rawtext, False) |
|
2597 | 2597 | |
|
2598 | 2598 | def _sidedata(self, rev): |
|
2599 | 2599 | """Return the sidedata for a given revision number.""" |
|
2600 | 2600 | index_entry = self.index[rev] |
|
2601 | 2601 | sidedata_offset = index_entry[8] |
|
2602 | 2602 | sidedata_size = index_entry[9] |
|
2603 | 2603 | |
|
2604 | 2604 | if self._inline: |
|
2605 | 2605 | sidedata_offset += self.index.entry_size * (1 + rev) |
|
2606 | 2606 | if sidedata_size == 0: |
|
2607 | 2607 | return {} |
|
2608 | 2608 | |
|
2609 | 2609 | if self._docket.sidedata_end < sidedata_offset + sidedata_size: |
|
2610 | 2610 | filename = self._sidedatafile |
|
2611 | 2611 | end = self._docket.sidedata_end |
|
2612 | 2612 | offset = sidedata_offset |
|
2613 | 2613 | length = sidedata_size |
|
2614 | 2614 | m = FILE_TOO_SHORT_MSG % (filename, length, offset, end) |
|
2615 | 2615 | raise error.RevlogError(m) |
|
2616 | 2616 | |
|
2617 | 2617 | comp_segment = self._inner._segmentfile_sidedata.read_chunk( |
|
2618 | 2618 | sidedata_offset, sidedata_size |
|
2619 | 2619 | ) |
|
2620 | 2620 | |
|
2621 | 2621 | comp = self.index[rev][11] |
|
2622 | 2622 | if comp == COMP_MODE_PLAIN: |
|
2623 | 2623 | segment = comp_segment |
|
2624 | 2624 | elif comp == COMP_MODE_DEFAULT: |
|
2625 | 2625 | segment = self._inner._decompressor(comp_segment) |
|
2626 | 2626 | elif comp == COMP_MODE_INLINE: |
|
2627 | 2627 | segment = self._inner.decompress(comp_segment) |
|
2628 | 2628 | else: |
|
2629 | 2629 | msg = b'unknown compression mode %d' |
|
2630 | 2630 | msg %= comp |
|
2631 | 2631 | raise error.RevlogError(msg) |
|
2632 | 2632 | |
|
2633 | 2633 | sidedata = sidedatautil.deserialize_sidedata(segment) |
|
2634 | 2634 | return sidedata |
|
2635 | 2635 | |
|
2636 | 2636 | def rawdata(self, nodeorrev): |
|
2637 | 2637 | """return an uncompressed raw data of a given node or revision number.""" |
|
2638 | 2638 | return self._revisiondata(nodeorrev, raw=True) |
|
2639 | 2639 | |
|
2640 | 2640 | def hash(self, text, p1, p2): |
|
2641 | 2641 | """Compute a node hash. |
|
2642 | 2642 | |
|
2643 | 2643 | Available as a function so that subclasses can replace the hash |
|
2644 | 2644 | as needed. |
|
2645 | 2645 | """ |
|
2646 | 2646 | return storageutil.hashrevisionsha1(text, p1, p2) |
|
2647 | 2647 | |
|
2648 | 2648 | def checkhash(self, text, node, p1=None, p2=None, rev=None): |
|
2649 | 2649 | """Check node hash integrity. |
|
2650 | 2650 | |
|
2651 | 2651 | Available as a function so that subclasses can extend hash mismatch |
|
2652 | 2652 | behaviors as needed. |
|
2653 | 2653 | """ |
|
2654 | 2654 | try: |
|
2655 | 2655 | if p1 is None and p2 is None: |
|
2656 | 2656 | p1, p2 = self.parents(node) |
|
2657 | 2657 | if node != self.hash(text, p1, p2): |
|
2658 | 2658 | # Clear the revision cache on hash failure. The revision cache |
|
2659 | 2659 | # only stores the raw revision and clearing the cache does have |
|
2660 | 2660 | # the side-effect that we won't have a cache hit when the raw |
|
2661 | 2661 | # revision data is accessed. But this case should be rare and |
|
2662 | 2662 | # it is extra work to teach the cache about the hash |
|
2663 | 2663 | # verification state. |
|
2664 | 2664 | if self._revisioncache and self._revisioncache[0] == node: |
|
2665 | 2665 | self._revisioncache = None |
|
2666 | 2666 | |
|
2667 | 2667 | revornode = rev |
|
2668 | 2668 | if revornode is None: |
|
2669 | 2669 | revornode = templatefilters.short(hex(node)) |
|
2670 | 2670 | raise error.RevlogError( |
|
2671 | 2671 | _(b"integrity check failed on %s:%s") |
|
2672 | 2672 | % (self.display_id, pycompat.bytestr(revornode)) |
|
2673 | 2673 | ) |
|
2674 | 2674 | except error.RevlogError: |
|
2675 | 2675 | if self.feature_config.censorable and storageutil.iscensoredtext( |
|
2676 | 2676 | text |
|
2677 | 2677 | ): |
|
2678 | 2678 | raise error.CensoredNodeError(self.display_id, node, text) |
|
2679 | 2679 | raise |
|
2680 | 2680 | |
|
2681 | 2681 | @property |
|
2682 | 2682 | def _split_index_file(self): |
|
2683 | 2683 | """the path where to expect the index of an ongoing splitting operation |
|
2684 | 2684 | |
|
2685 | 2685 | The file will only exist if a splitting operation is in progress, but |
|
2686 | 2686 | it is always expected at the same location.""" |
|
2687 | 2687 | parts = self.radix.split(b'/') |
|
2688 | 2688 | if len(parts) > 1: |
|
2689 | 2689 | # adds a '-s' prefix to the ``data/` or `meta/` base |
|
2690 | 2690 | head = parts[0] + b'-s' |
|
2691 | 2691 | mids = parts[1:-1] |
|
2692 | 2692 | tail = parts[-1] + b'.i' |
|
2693 | 2693 | pieces = [head] + mids + [tail] |
|
2694 | 2694 | return b'/'.join(pieces) |
|
2695 | 2695 | else: |
|
2696 | 2696 | # the revlog is stored at the root of the store (changelog or |
|
2697 | 2697 | # manifest), no risk of collision. |
|
2698 | 2698 | return self.radix + b'.i.s' |
|
2699 | 2699 | |
|
2700 | 2700 | def _enforceinlinesize(self, tr, side_write=True): |
|
2701 | 2701 | """Check if the revlog is too big for inline and convert if so. |
|
2702 | 2702 | |
|
2703 | 2703 | This should be called after revisions are added to the revlog. If the |
|
2704 | 2704 | revlog has grown too large to be an inline revlog, it will convert it |
|
2705 | 2705 | to use multiple index and data files. |
|
2706 | 2706 | """ |
|
2707 | 2707 | tiprev = len(self) - 1 |
|
2708 | 2708 | total_size = self.start(tiprev) + self.length(tiprev) |
|
2709 | 2709 | if not self._inline or total_size < _maxinline: |
|
2710 | 2710 | return |
|
2711 | 2711 | |
|
2712 | 2712 | if self._docket is not None: |
|
2713 | 2713 | msg = b"inline revlog should not have a docket" |
|
2714 | 2714 | raise error.ProgrammingError(msg) |
|
2715 | 2715 | |
|
2716 | 2716 | troffset = tr.findoffset(self._indexfile) |
|
2717 | 2717 | if troffset is None: |
|
2718 | 2718 | raise error.RevlogError( |
|
2719 | 2719 | _(b"%s not found in the transaction") % self._indexfile |
|
2720 | 2720 | ) |
|
2721 | 2721 | if troffset: |
|
2722 | 2722 | tr.addbackup(self._indexfile, for_offset=True) |
|
2723 | 2723 | tr.add(self._datafile, 0) |
|
2724 | 2724 | |
|
2725 | 2725 | new_index_file_path = None |
|
2726 | 2726 | if side_write: |
|
2727 | 2727 | old_index_file_path = self._indexfile |
|
2728 | 2728 | new_index_file_path = self._split_index_file |
|
2729 | 2729 | opener = self.opener |
|
2730 | 2730 | weak_self = weakref.ref(self) |
|
2731 | 2731 | |
|
2732 | 2732 | # the "split" index replace the real index when the transaction is |
|
2733 | 2733 | # finalized |
|
2734 | 2734 | def finalize_callback(tr): |
|
2735 | 2735 | opener.rename( |
|
2736 | 2736 | new_index_file_path, |
|
2737 | 2737 | old_index_file_path, |
|
2738 | 2738 | checkambig=True, |
|
2739 | 2739 | ) |
|
2740 | 2740 | maybe_self = weak_self() |
|
2741 | 2741 | if maybe_self is not None: |
|
2742 | 2742 | maybe_self._indexfile = old_index_file_path |
|
2743 | 2743 | maybe_self._inner.index_file = maybe_self._indexfile |
|
2744 | 2744 | |
|
2745 | 2745 | def abort_callback(tr): |
|
2746 | 2746 | maybe_self = weak_self() |
|
2747 | 2747 | if maybe_self is not None: |
|
2748 | 2748 | maybe_self._indexfile = old_index_file_path |
|
2749 | 2749 | maybe_self._inner.inline = True |
|
2750 | 2750 | maybe_self._inner.index_file = old_index_file_path |
|
2751 | 2751 | |
|
2752 | 2752 | tr.registertmp(new_index_file_path) |
|
2753 | 2753 | if self.target[1] is not None: |
|
2754 | 2754 | callback_id = b'000-revlog-split-%d-%s' % self.target |
|
2755 | 2755 | else: |
|
2756 | 2756 | callback_id = b'000-revlog-split-%d' % self.target[0] |
|
2757 | 2757 | tr.addfinalize(callback_id, finalize_callback) |
|
2758 | 2758 | tr.addabort(callback_id, abort_callback) |
|
2759 | 2759 | |
|
2760 | 2760 | self._format_flags &= ~FLAG_INLINE_DATA |
|
2761 | 2761 | self._inner.split_inline( |
|
2762 | 2762 | tr, |
|
2763 | 2763 | self._format_flags | self._format_version, |
|
2764 | 2764 | new_index_file_path=new_index_file_path, |
|
2765 | 2765 | ) |
|
2766 | 2766 | |
|
2767 | 2767 | self._inline = False |
|
2768 | 2768 | if new_index_file_path is not None: |
|
2769 | 2769 | self._indexfile = new_index_file_path |
|
2770 | 2770 | |
|
2771 | 2771 | nodemaputil.setup_persistent_nodemap(tr, self) |
|
2772 | 2772 | |
|
2773 | 2773 | def _nodeduplicatecallback(self, transaction, node): |
|
2774 | 2774 | """called when trying to add a node already stored.""" |
|
2775 | 2775 | |
|
2776 | 2776 | @contextlib.contextmanager |
|
2777 | 2777 | def reading(self): |
|
2778 | 2778 | with self._inner.reading(): |
|
2779 | 2779 | yield |
|
2780 | 2780 | |
|
2781 | 2781 | @contextlib.contextmanager |
|
2782 | 2782 | def _writing(self, transaction): |
|
2783 | 2783 | if self._trypending: |
|
2784 | 2784 | msg = b'try to write in a `trypending` revlog: %s' |
|
2785 | 2785 | msg %= self.display_id |
|
2786 | 2786 | raise error.ProgrammingError(msg) |
|
2787 | 2787 | if self._inner.is_writing: |
|
2788 | 2788 | yield |
|
2789 | 2789 | else: |
|
2790 | 2790 | data_end = None |
|
2791 | 2791 | sidedata_end = None |
|
2792 | 2792 | if self._docket is not None: |
|
2793 | 2793 | data_end = self._docket.data_end |
|
2794 | 2794 | sidedata_end = self._docket.sidedata_end |
|
2795 | 2795 | with self._inner.writing( |
|
2796 | 2796 | transaction, |
|
2797 | 2797 | data_end=data_end, |
|
2798 | 2798 | sidedata_end=sidedata_end, |
|
2799 | 2799 | ): |
|
2800 | 2800 | yield |
|
2801 | 2801 | if self._docket is not None: |
|
2802 | 2802 | self._write_docket(transaction) |
|
2803 | 2803 | |
|
2804 | 2804 | def _write_docket(self, transaction): |
|
2805 | 2805 | """write the current docket on disk |
|
2806 | 2806 | |
|
2807 | 2807 | Exist as a method to help changelog to implement transaction logic |
|
2808 | 2808 | |
|
2809 | 2809 | We could also imagine using the same transaction logic for all revlog |
|
2810 | 2810 | since docket are cheap.""" |
|
2811 | 2811 | self._docket.write(transaction) |
|
2812 | 2812 | |
|
2813 | 2813 | def addrevision( |
|
2814 | 2814 | self, |
|
2815 | 2815 | text, |
|
2816 | 2816 | transaction, |
|
2817 | 2817 | link, |
|
2818 | 2818 | p1, |
|
2819 | 2819 | p2, |
|
2820 | 2820 | cachedelta=None, |
|
2821 | 2821 | node=None, |
|
2822 | 2822 | flags=REVIDX_DEFAULT_FLAGS, |
|
2823 | 2823 | deltacomputer=None, |
|
2824 | 2824 | sidedata=None, |
|
2825 | 2825 | ): |
|
2826 | 2826 | """add a revision to the log |
|
2827 | 2827 | |
|
2828 | 2828 | text - the revision data to add |
|
2829 | 2829 | transaction - the transaction object used for rollback |
|
2830 | 2830 | link - the linkrev data to add |
|
2831 | 2831 | p1, p2 - the parent nodeids of the revision |
|
2832 | 2832 | cachedelta - an optional precomputed delta |
|
2833 | 2833 | node - nodeid of revision; typically node is not specified, and it is |
|
2834 | 2834 | computed by default as hash(text, p1, p2), however subclasses might |
|
2835 | 2835 | use different hashing method (and override checkhash() in such case) |
|
2836 | 2836 | flags - the known flags to set on the revision |
|
2837 | 2837 | deltacomputer - an optional deltacomputer instance shared between |
|
2838 | 2838 | multiple calls |
|
2839 | 2839 | """ |
|
2840 | 2840 | if link == nullrev: |
|
2841 | 2841 | raise error.RevlogError( |
|
2842 | 2842 | _(b"attempted to add linkrev -1 to %s") % self.display_id |
|
2843 | 2843 | ) |
|
2844 | 2844 | |
|
2845 | 2845 | if sidedata is None: |
|
2846 | 2846 | sidedata = {} |
|
2847 | 2847 | elif sidedata and not self.feature_config.has_side_data: |
|
2848 | 2848 | raise error.ProgrammingError( |
|
2849 | 2849 | _(b"trying to add sidedata to a revlog who don't support them") |
|
2850 | 2850 | ) |
|
2851 | 2851 | |
|
2852 | 2852 | if flags: |
|
2853 | 2853 | node = node or self.hash(text, p1, p2) |
|
2854 | 2854 | |
|
2855 | 2855 | rawtext, validatehash = flagutil.processflagswrite(self, text, flags) |
|
2856 | 2856 | |
|
2857 | 2857 | # If the flag processor modifies the revision data, ignore any provided |
|
2858 | 2858 | # cachedelta. |
|
2859 | 2859 | if rawtext != text: |
|
2860 | 2860 | cachedelta = None |
|
2861 | 2861 | |
|
2862 | 2862 | if len(rawtext) > _maxentrysize: |
|
2863 | 2863 | raise error.RevlogError( |
|
2864 | 2864 | _( |
|
2865 | 2865 | b"%s: size of %d bytes exceeds maximum revlog storage of 2GiB" |
|
2866 | 2866 | ) |
|
2867 | 2867 | % (self.display_id, len(rawtext)) |
|
2868 | 2868 | ) |
|
2869 | 2869 | |
|
2870 | 2870 | node = node or self.hash(rawtext, p1, p2) |
|
2871 | 2871 | rev = self.index.get_rev(node) |
|
2872 | 2872 | if rev is not None: |
|
2873 | 2873 | return rev |
|
2874 | 2874 | |
|
2875 | 2875 | if validatehash: |
|
2876 | 2876 | self.checkhash(rawtext, node, p1=p1, p2=p2) |
|
2877 | 2877 | |
|
2878 | 2878 | return self.addrawrevision( |
|
2879 | 2879 | rawtext, |
|
2880 | 2880 | transaction, |
|
2881 | 2881 | link, |
|
2882 | 2882 | p1, |
|
2883 | 2883 | p2, |
|
2884 | 2884 | node, |
|
2885 | 2885 | flags, |
|
2886 | 2886 | cachedelta=cachedelta, |
|
2887 | 2887 | deltacomputer=deltacomputer, |
|
2888 | 2888 | sidedata=sidedata, |
|
2889 | 2889 | ) |
|
2890 | 2890 | |
|
2891 | 2891 | def addrawrevision( |
|
2892 | 2892 | self, |
|
2893 | 2893 | rawtext, |
|
2894 | 2894 | transaction, |
|
2895 | 2895 | link, |
|
2896 | 2896 | p1, |
|
2897 | 2897 | p2, |
|
2898 | 2898 | node, |
|
2899 | 2899 | flags, |
|
2900 | 2900 | cachedelta=None, |
|
2901 | 2901 | deltacomputer=None, |
|
2902 | 2902 | sidedata=None, |
|
2903 | 2903 | ): |
|
2904 | 2904 | """add a raw revision with known flags, node and parents |
|
2905 | 2905 | useful when reusing a revision not stored in this revlog (ex: received |
|
2906 | 2906 | over wire, or read from an external bundle). |
|
2907 | 2907 | """ |
|
2908 | 2908 | with self._writing(transaction): |
|
2909 | 2909 | return self._addrevision( |
|
2910 | 2910 | node, |
|
2911 | 2911 | rawtext, |
|
2912 | 2912 | transaction, |
|
2913 | 2913 | link, |
|
2914 | 2914 | p1, |
|
2915 | 2915 | p2, |
|
2916 | 2916 | flags, |
|
2917 | 2917 | cachedelta, |
|
2918 | 2918 | deltacomputer=deltacomputer, |
|
2919 | 2919 | sidedata=sidedata, |
|
2920 | 2920 | ) |
|
2921 | 2921 | |
|
2922 | 2922 | def compress(self, data): |
|
2923 | 2923 | return self._inner.compress(data) |
|
2924 | 2924 | |
|
2925 | 2925 | def decompress(self, data): |
|
2926 | 2926 | return self._inner.decompress(data) |
|
2927 | 2927 | |
|
2928 | 2928 | def _addrevision( |
|
2929 | 2929 | self, |
|
2930 | 2930 | node, |
|
2931 | 2931 | rawtext, |
|
2932 | 2932 | transaction, |
|
2933 | 2933 | link, |
|
2934 | 2934 | p1, |
|
2935 | 2935 | p2, |
|
2936 | 2936 | flags, |
|
2937 | 2937 | cachedelta, |
|
2938 | 2938 | alwayscache=False, |
|
2939 | 2939 | deltacomputer=None, |
|
2940 | 2940 | sidedata=None, |
|
2941 | 2941 | ): |
|
2942 | 2942 | """internal function to add revisions to the log |
|
2943 | 2943 | |
|
2944 | 2944 | see addrevision for argument descriptions. |
|
2945 | 2945 | |
|
2946 | 2946 | note: "addrevision" takes non-raw text, "_addrevision" takes raw text. |
|
2947 | 2947 | |
|
2948 | 2948 | if "deltacomputer" is not provided or None, a defaultdeltacomputer will |
|
2949 | 2949 | be used. |
|
2950 | 2950 | |
|
2951 | 2951 | invariants: |
|
2952 | 2952 | - rawtext is optional (can be None); if not set, cachedelta must be set. |
|
2953 | 2953 | if both are set, they must correspond to each other. |
|
2954 | 2954 | """ |
|
2955 | 2955 | if node == self.nullid: |
|
2956 | 2956 | raise error.RevlogError( |
|
2957 | 2957 | _(b"%s: attempt to add null revision") % self.display_id |
|
2958 | 2958 | ) |
|
2959 | 2959 | if ( |
|
2960 | 2960 | node == self.nodeconstants.wdirid |
|
2961 | 2961 | or node in self.nodeconstants.wdirfilenodeids |
|
2962 | 2962 | ): |
|
2963 | 2963 | raise error.RevlogError( |
|
2964 | 2964 | _(b"%s: attempt to add wdir revision") % self.display_id |
|
2965 | 2965 | ) |
|
2966 | 2966 | if self._inner._writinghandles is None: |
|
2967 | 2967 | msg = b'adding revision outside `revlog._writing` context' |
|
2968 | 2968 | raise error.ProgrammingError(msg) |
|
2969 | 2969 | |
|
2970 | 2970 | btext = [rawtext] |
|
2971 | 2971 | |
|
2972 | 2972 | curr = len(self) |
|
2973 | 2973 | prev = curr - 1 |
|
2974 | 2974 | |
|
2975 | 2975 | offset = self._get_data_offset(prev) |
|
2976 | 2976 | |
|
2977 | 2977 | if self._concurrencychecker: |
|
2978 | 2978 | ifh, dfh, sdfh = self._inner._writinghandles |
|
2979 | 2979 | # XXX no checking for the sidedata file |
|
2980 | 2980 | if self._inline: |
|
2981 | 2981 | # offset is "as if" it were in the .d file, so we need to add on |
|
2982 | 2982 | # the size of the entry metadata. |
|
2983 | 2983 | self._concurrencychecker( |
|
2984 | 2984 | ifh, self._indexfile, offset + curr * self.index.entry_size |
|
2985 | 2985 | ) |
|
2986 | 2986 | else: |
|
2987 | 2987 | # Entries in the .i are a consistent size. |
|
2988 | 2988 | self._concurrencychecker( |
|
2989 | 2989 | ifh, self._indexfile, curr * self.index.entry_size |
|
2990 | 2990 | ) |
|
2991 | 2991 | self._concurrencychecker(dfh, self._datafile, offset) |
|
2992 | 2992 | |
|
2993 | 2993 | p1r, p2r = self.rev(p1), self.rev(p2) |
|
2994 | 2994 | |
|
2995 | 2995 | # full versions are inserted when the needed deltas |
|
2996 | 2996 | # become comparable to the uncompressed text |
|
2997 | 2997 | if rawtext is None: |
|
2998 | 2998 | # need rawtext size, before changed by flag processors, which is |
|
2999 | 2999 | # the non-raw size. use revlog explicitly to avoid filelog's extra |
|
3000 | 3000 | # logic that might remove metadata size. |
|
3001 | 3001 | textlen = mdiff.patchedsize( |
|
3002 | 3002 | revlog.size(self, cachedelta[0]), cachedelta[1] |
|
3003 | 3003 | ) |
|
3004 | 3004 | else: |
|
3005 | 3005 | textlen = len(rawtext) |
|
3006 | 3006 | |
|
3007 | 3007 | if deltacomputer is None: |
|
3008 | 3008 | write_debug = None |
|
3009 | 3009 | if self.delta_config.debug_delta: |
|
3010 | 3010 | write_debug = transaction._report |
|
3011 | 3011 | deltacomputer = deltautil.deltacomputer( |
|
3012 | 3012 | self, write_debug=write_debug |
|
3013 | 3013 | ) |
|
3014 | 3014 | |
|
3015 | 3015 | if cachedelta is not None and len(cachedelta) == 2: |
|
3016 | 3016 | # If the cached delta has no information about how it should be |
|
3017 | 3017 | # reused, add the default reuse instruction according to the |
|
3018 | 3018 | # revlog's configuration. |
|
3019 | 3019 | if ( |
|
3020 | 3020 | self.delta_config.general_delta |
|
3021 | 3021 | and self.delta_config.lazy_delta_base |
|
3022 | 3022 | ): |
|
3023 | 3023 | delta_base_reuse = DELTA_BASE_REUSE_TRY |
|
3024 | 3024 | else: |
|
3025 | 3025 | delta_base_reuse = DELTA_BASE_REUSE_NO |
|
3026 | 3026 | cachedelta = (cachedelta[0], cachedelta[1], delta_base_reuse) |
|
3027 | 3027 | |
|
3028 | 3028 | revinfo = revlogutils.revisioninfo( |
|
3029 | 3029 | node, |
|
3030 | 3030 | p1, |
|
3031 | 3031 | p2, |
|
3032 | 3032 | btext, |
|
3033 | 3033 | textlen, |
|
3034 | 3034 | cachedelta, |
|
3035 | 3035 | flags, |
|
3036 | 3036 | ) |
|
3037 | 3037 | |
|
3038 | 3038 | deltainfo = deltacomputer.finddeltainfo(revinfo) |
|
3039 | 3039 | |
|
3040 | 3040 | compression_mode = COMP_MODE_INLINE |
|
3041 | 3041 | if self._docket is not None: |
|
3042 | 3042 | default_comp = self._docket.default_compression_header |
|
3043 | 3043 | r = deltautil.delta_compression(default_comp, deltainfo) |
|
3044 | 3044 | compression_mode, deltainfo = r |
|
3045 | 3045 | |
|
3046 | 3046 | sidedata_compression_mode = COMP_MODE_INLINE |
|
3047 | 3047 | if sidedata and self.feature_config.has_side_data: |
|
3048 | 3048 | sidedata_compression_mode = COMP_MODE_PLAIN |
|
3049 | 3049 | serialized_sidedata = sidedatautil.serialize_sidedata(sidedata) |
|
3050 | 3050 | sidedata_offset = self._docket.sidedata_end |
|
3051 | 3051 | h, comp_sidedata = self._inner.compress(serialized_sidedata) |
|
3052 | 3052 | if ( |
|
3053 | 3053 | h != b'u' |
|
3054 | 3054 | and comp_sidedata[0:1] != b'\0' |
|
3055 | 3055 | and len(comp_sidedata) < len(serialized_sidedata) |
|
3056 | 3056 | ): |
|
3057 | 3057 | assert not h |
|
3058 | 3058 | if ( |
|
3059 | 3059 | comp_sidedata[0:1] |
|
3060 | 3060 | == self._docket.default_compression_header |
|
3061 | 3061 | ): |
|
3062 | 3062 | sidedata_compression_mode = COMP_MODE_DEFAULT |
|
3063 | 3063 | serialized_sidedata = comp_sidedata |
|
3064 | 3064 | else: |
|
3065 | 3065 | sidedata_compression_mode = COMP_MODE_INLINE |
|
3066 | 3066 | serialized_sidedata = comp_sidedata |
|
3067 | 3067 | else: |
|
3068 | 3068 | serialized_sidedata = b"" |
|
3069 | 3069 | # Don't store the offset if the sidedata is empty, that way |
|
3070 | 3070 | # we can easily detect empty sidedata and they will be no different |
|
3071 | 3071 | # than ones we manually add. |
|
3072 | 3072 | sidedata_offset = 0 |
|
3073 | 3073 | |
|
3074 | 3074 | rank = RANK_UNKNOWN |
|
3075 | 3075 | if self.feature_config.compute_rank: |
|
3076 | 3076 | if (p1r, p2r) == (nullrev, nullrev): |
|
3077 | 3077 | rank = 1 |
|
3078 | 3078 | elif p1r != nullrev and p2r == nullrev: |
|
3079 | 3079 | rank = 1 + self.fast_rank(p1r) |
|
3080 | 3080 | elif p1r == nullrev and p2r != nullrev: |
|
3081 | 3081 | rank = 1 + self.fast_rank(p2r) |
|
3082 | 3082 | else: # merge node |
|
3083 | 3083 | if rustdagop is not None and self.index.rust_ext_compat: |
|
3084 | 3084 | rank = rustdagop.rank(self.index, p1r, p2r) |
|
3085 | 3085 | else: |
|
3086 | 3086 | pmin, pmax = sorted((p1r, p2r)) |
|
3087 | 3087 | rank = 1 + self.fast_rank(pmax) |
|
3088 | 3088 | rank += sum(1 for _ in self.findmissingrevs([pmax], [pmin])) |
|
3089 | 3089 | |
|
3090 | 3090 | e = revlogutils.entry( |
|
3091 | 3091 | flags=flags, |
|
3092 | 3092 | data_offset=offset, |
|
3093 | 3093 | data_compressed_length=deltainfo.deltalen, |
|
3094 | 3094 | data_uncompressed_length=textlen, |
|
3095 | 3095 | data_compression_mode=compression_mode, |
|
3096 | 3096 | data_delta_base=deltainfo.base, |
|
3097 | 3097 | link_rev=link, |
|
3098 | 3098 | parent_rev_1=p1r, |
|
3099 | 3099 | parent_rev_2=p2r, |
|
3100 | 3100 | node_id=node, |
|
3101 | 3101 | sidedata_offset=sidedata_offset, |
|
3102 | 3102 | sidedata_compressed_length=len(serialized_sidedata), |
|
3103 | 3103 | sidedata_compression_mode=sidedata_compression_mode, |
|
3104 | 3104 | rank=rank, |
|
3105 | 3105 | ) |
|
3106 | 3106 | |
|
3107 | 3107 | self.index.append(e) |
|
3108 | 3108 | entry = self.index.entry_binary(curr) |
|
3109 | 3109 | if curr == 0 and self._docket is None: |
|
3110 | 3110 | header = self._format_flags | self._format_version |
|
3111 | 3111 | header = self.index.pack_header(header) |
|
3112 | 3112 | entry = header + entry |
|
3113 | 3113 | self._writeentry( |
|
3114 | 3114 | transaction, |
|
3115 | 3115 | entry, |
|
3116 | 3116 | deltainfo.data, |
|
3117 | 3117 | link, |
|
3118 | 3118 | offset, |
|
3119 | 3119 | serialized_sidedata, |
|
3120 | 3120 | sidedata_offset, |
|
3121 | 3121 | ) |
|
3122 | 3122 | |
|
3123 | 3123 | rawtext = btext[0] |
|
3124 | 3124 | |
|
3125 | 3125 | if alwayscache and rawtext is None: |
|
3126 | 3126 | rawtext = deltacomputer.buildtext(revinfo) |
|
3127 | 3127 | |
|
3128 | 3128 | if type(rawtext) == bytes: # only accept immutable objects |
|
3129 | 3129 | self._revisioncache = (node, curr, rawtext) |
|
3130 | 3130 | self._chainbasecache[curr] = deltainfo.chainbase |
|
3131 | 3131 | return curr |
|
3132 | 3132 | |
|
3133 | 3133 | def _get_data_offset(self, prev): |
|
3134 | 3134 | """Returns the current offset in the (in-transaction) data file. |
|
3135 | 3135 | Versions < 2 of the revlog can get this 0(1), revlog v2 needs a docket |
|
3136 | 3136 | file to store that information: since sidedata can be rewritten to the |
|
3137 | 3137 | end of the data file within a transaction, you can have cases where, for |
|
3138 | 3138 | example, rev `n` does not have sidedata while rev `n - 1` does, leading |
|
3139 | 3139 | to `n - 1`'s sidedata being written after `n`'s data. |
|
3140 | 3140 | |
|
3141 | 3141 | TODO cache this in a docket file before getting out of experimental.""" |
|
3142 | 3142 | if self._docket is None: |
|
3143 | 3143 | return self.end(prev) |
|
3144 | 3144 | else: |
|
3145 | 3145 | return self._docket.data_end |
|
3146 | 3146 | |
|
3147 | 3147 | def _writeentry( |
|
3148 | 3148 | self, transaction, entry, data, link, offset, sidedata, sidedata_offset |
|
3149 | 3149 | ): |
|
3150 | 3150 | # Files opened in a+ mode have inconsistent behavior on various |
|
3151 | 3151 | # platforms. Windows requires that a file positioning call be made |
|
3152 | 3152 | # when the file handle transitions between reads and writes. See |
|
3153 | 3153 | # 3686fa2b8eee and the mixedfilemodewrapper in windows.py. On other |
|
3154 | 3154 | # platforms, Python or the platform itself can be buggy. Some versions |
|
3155 | 3155 | # of Solaris have been observed to not append at the end of the file |
|
3156 | 3156 | # if the file was seeked to before the end. See issue4943 for more. |
|
3157 | 3157 | # |
|
3158 | 3158 | # We work around this issue by inserting a seek() before writing. |
|
3159 | 3159 | # Note: This is likely not necessary on Python 3. However, because |
|
3160 | 3160 | # the file handle is reused for reads and may be seeked there, we need |
|
3161 | 3161 | # to be careful before changing this. |
|
3162 | 3162 | if self._inner._writinghandles is None: |
|
3163 | 3163 | msg = b'adding revision outside `revlog._writing` context' |
|
3164 | 3164 | raise error.ProgrammingError(msg) |
|
3165 | 3165 | ifh, dfh, sdfh = self._inner._writinghandles |
|
3166 | 3166 | if self._docket is None: |
|
3167 | 3167 | ifh.seek(0, os.SEEK_END) |
|
3168 | 3168 | else: |
|
3169 | 3169 | ifh.seek(self._docket.index_end, os.SEEK_SET) |
|
3170 | 3170 | if dfh: |
|
3171 | 3171 | if self._docket is None: |
|
3172 | 3172 | dfh.seek(0, os.SEEK_END) |
|
3173 | 3173 | else: |
|
3174 | 3174 | dfh.seek(self._docket.data_end, os.SEEK_SET) |
|
3175 | 3175 | if sdfh: |
|
3176 | 3176 | sdfh.seek(self._docket.sidedata_end, os.SEEK_SET) |
|
3177 | 3177 | |
|
3178 | 3178 | curr = len(self) - 1 |
|
3179 | 3179 | if not self._inline: |
|
3180 | 3180 | transaction.add(self._datafile, offset) |
|
3181 | 3181 | if self._sidedatafile: |
|
3182 | 3182 | transaction.add(self._sidedatafile, sidedata_offset) |
|
3183 | 3183 | transaction.add(self._indexfile, curr * len(entry)) |
|
3184 | 3184 | if data[0]: |
|
3185 | 3185 | dfh.write(data[0]) |
|
3186 | 3186 | dfh.write(data[1]) |
|
3187 | 3187 | if sidedata: |
|
3188 | 3188 | sdfh.write(sidedata) |
|
3189 | 3189 | ifh.write(entry) |
|
3190 | 3190 | else: |
|
3191 | 3191 | offset += curr * self.index.entry_size |
|
3192 | 3192 | transaction.add(self._indexfile, offset) |
|
3193 | 3193 | ifh.write(entry) |
|
3194 | 3194 | ifh.write(data[0]) |
|
3195 | 3195 | ifh.write(data[1]) |
|
3196 | 3196 | assert not sidedata |
|
3197 | 3197 | self._enforceinlinesize(transaction) |
|
3198 | 3198 | if self._docket is not None: |
|
3199 | 3199 | # revlog-v2 always has 3 writing handles, help Pytype |
|
3200 | 3200 | wh1 = self._inner._writinghandles[0] |
|
3201 | 3201 | wh2 = self._inner._writinghandles[1] |
|
3202 | 3202 | wh3 = self._inner._writinghandles[2] |
|
3203 | 3203 | assert wh1 is not None |
|
3204 | 3204 | assert wh2 is not None |
|
3205 | 3205 | assert wh3 is not None |
|
3206 | 3206 | self._docket.index_end = wh1.tell() |
|
3207 | 3207 | self._docket.data_end = wh2.tell() |
|
3208 | 3208 | self._docket.sidedata_end = wh3.tell() |
|
3209 | 3209 | |
|
3210 | 3210 | nodemaputil.setup_persistent_nodemap(transaction, self) |
|
3211 | 3211 | |
|
3212 | 3212 | def addgroup( |
|
3213 | 3213 | self, |
|
3214 | 3214 | deltas, |
|
3215 | 3215 | linkmapper, |
|
3216 | 3216 | transaction, |
|
3217 | 3217 | alwayscache=False, |
|
3218 | 3218 | addrevisioncb=None, |
|
3219 | 3219 | duplicaterevisioncb=None, |
|
3220 | 3220 | debug_info=None, |
|
3221 | 3221 | delta_base_reuse_policy=None, |
|
3222 | 3222 | ): |
|
3223 | 3223 | """ |
|
3224 | 3224 | add a delta group |
|
3225 | 3225 | |
|
3226 | 3226 | given a set of deltas, add them to the revision log. the |
|
3227 | 3227 | first delta is against its parent, which should be in our |
|
3228 | 3228 | log, the rest are against the previous delta. |
|
3229 | 3229 | |
|
3230 | 3230 | If ``addrevisioncb`` is defined, it will be called with arguments of |
|
3231 | 3231 | this revlog and the node that was added. |
|
3232 | 3232 | """ |
|
3233 | 3233 | |
|
3234 | 3234 | if self._adding_group: |
|
3235 | 3235 | raise error.ProgrammingError(b'cannot nest addgroup() calls') |
|
3236 | 3236 | |
|
3237 | 3237 | # read the default delta-base reuse policy from revlog config if the |
|
3238 | 3238 | # group did not specify one. |
|
3239 | 3239 | if delta_base_reuse_policy is None: |
|
3240 | 3240 | if ( |
|
3241 | 3241 | self.delta_config.general_delta |
|
3242 | 3242 | and self.delta_config.lazy_delta_base |
|
3243 | 3243 | ): |
|
3244 | 3244 | delta_base_reuse_policy = DELTA_BASE_REUSE_TRY |
|
3245 | 3245 | else: |
|
3246 | 3246 | delta_base_reuse_policy = DELTA_BASE_REUSE_NO |
|
3247 | 3247 | |
|
3248 | 3248 | self._adding_group = True |
|
3249 | 3249 | empty = True |
|
3250 | 3250 | try: |
|
3251 | 3251 | with self._writing(transaction): |
|
3252 | 3252 | write_debug = None |
|
3253 | 3253 | if self.delta_config.debug_delta: |
|
3254 | 3254 | write_debug = transaction._report |
|
3255 | 3255 | deltacomputer = deltautil.deltacomputer( |
|
3256 | 3256 | self, |
|
3257 | 3257 | write_debug=write_debug, |
|
3258 | 3258 | debug_info=debug_info, |
|
3259 | 3259 | ) |
|
3260 | 3260 | # loop through our set of deltas |
|
3261 | 3261 | for data in deltas: |
|
3262 | 3262 | ( |
|
3263 | 3263 | node, |
|
3264 | 3264 | p1, |
|
3265 | 3265 | p2, |
|
3266 | 3266 | linknode, |
|
3267 | 3267 | deltabase, |
|
3268 | 3268 | delta, |
|
3269 | 3269 | flags, |
|
3270 | 3270 | sidedata, |
|
3271 | 3271 | ) = data |
|
3272 | 3272 | link = linkmapper(linknode) |
|
3273 | 3273 | flags = flags or REVIDX_DEFAULT_FLAGS |
|
3274 | 3274 | |
|
3275 | 3275 | rev = self.index.get_rev(node) |
|
3276 | 3276 | if rev is not None: |
|
3277 | 3277 | # this can happen if two branches make the same change |
|
3278 | 3278 | self._nodeduplicatecallback(transaction, rev) |
|
3279 | 3279 | if duplicaterevisioncb: |
|
3280 | 3280 | duplicaterevisioncb(self, rev) |
|
3281 | 3281 | empty = False |
|
3282 | 3282 | continue |
|
3283 | 3283 | |
|
3284 | 3284 | for p in (p1, p2): |
|
3285 | 3285 | if not self.index.has_node(p): |
|
3286 | 3286 | raise error.LookupError( |
|
3287 | 3287 | p, self.radix, _(b'unknown parent') |
|
3288 | 3288 | ) |
|
3289 | 3289 | |
|
3290 | 3290 | if not self.index.has_node(deltabase): |
|
3291 | 3291 | raise error.LookupError( |
|
3292 | 3292 | deltabase, self.display_id, _(b'unknown delta base') |
|
3293 | 3293 | ) |
|
3294 | 3294 | |
|
3295 | 3295 | baserev = self.rev(deltabase) |
|
3296 | 3296 | |
|
3297 | 3297 | if baserev != nullrev and self.iscensored(baserev): |
|
3298 | 3298 | # if base is censored, delta must be full replacement in a |
|
3299 | 3299 | # single patch operation |
|
3300 | 3300 | hlen = struct.calcsize(b">lll") |
|
3301 | 3301 | oldlen = self.rawsize(baserev) |
|
3302 | 3302 | newlen = len(delta) - hlen |
|
3303 | 3303 | if delta[:hlen] != mdiff.replacediffheader( |
|
3304 | 3304 | oldlen, newlen |
|
3305 | 3305 | ): |
|
3306 | 3306 | raise error.CensoredBaseError( |
|
3307 | 3307 | self.display_id, self.node(baserev) |
|
3308 | 3308 | ) |
|
3309 | 3309 | |
|
3310 | 3310 | if not flags and self._peek_iscensored(baserev, delta): |
|
3311 | 3311 | flags |= REVIDX_ISCENSORED |
|
3312 | 3312 | |
|
3313 | 3313 | # We assume consumers of addrevisioncb will want to retrieve |
|
3314 | 3314 | # the added revision, which will require a call to |
|
3315 | 3315 | # revision(). revision() will fast path if there is a cache |
|
3316 | 3316 | # hit. So, we tell _addrevision() to always cache in this case. |
|
3317 | 3317 | # We're only using addgroup() in the context of changegroup |
|
3318 | 3318 | # generation so the revision data can always be handled as raw |
|
3319 | 3319 | # by the flagprocessor. |
|
3320 | 3320 | rev = self._addrevision( |
|
3321 | 3321 | node, |
|
3322 | 3322 | None, |
|
3323 | 3323 | transaction, |
|
3324 | 3324 | link, |
|
3325 | 3325 | p1, |
|
3326 | 3326 | p2, |
|
3327 | 3327 | flags, |
|
3328 | 3328 | (baserev, delta, delta_base_reuse_policy), |
|
3329 | 3329 | alwayscache=alwayscache, |
|
3330 | 3330 | deltacomputer=deltacomputer, |
|
3331 | 3331 | sidedata=sidedata, |
|
3332 | 3332 | ) |
|
3333 | 3333 | |
|
3334 | 3334 | if addrevisioncb: |
|
3335 | 3335 | addrevisioncb(self, rev) |
|
3336 | 3336 | empty = False |
|
3337 | 3337 | finally: |
|
3338 | 3338 | self._adding_group = False |
|
3339 | 3339 | return not empty |
|
3340 | 3340 | |
|
3341 | 3341 | def iscensored(self, rev): |
|
3342 | 3342 | """Check if a file revision is censored.""" |
|
3343 | 3343 | if not self.feature_config.censorable: |
|
3344 | 3344 | return False |
|
3345 | 3345 | |
|
3346 | 3346 | return self.flags(rev) & REVIDX_ISCENSORED |
|
3347 | 3347 | |
|
3348 | 3348 | def _peek_iscensored(self, baserev, delta): |
|
3349 | 3349 | """Quickly check if a delta produces a censored revision.""" |
|
3350 | 3350 | if not self.feature_config.censorable: |
|
3351 | 3351 | return False |
|
3352 | 3352 | |
|
3353 | 3353 | return storageutil.deltaiscensored(delta, baserev, self.rawsize) |
|
3354 | 3354 | |
|
3355 | 3355 | def getstrippoint(self, minlink): |
|
3356 | 3356 | """find the minimum rev that must be stripped to strip the linkrev |
|
3357 | 3357 | |
|
3358 | 3358 | Returns a tuple containing the minimum rev and a set of all revs that |
|
3359 | 3359 | have linkrevs that will be broken by this strip. |
|
3360 | 3360 | """ |
|
3361 | 3361 | return storageutil.resolvestripinfo( |
|
3362 | 3362 | minlink, |
|
3363 | 3363 | len(self) - 1, |
|
3364 | 3364 | self.headrevs(), |
|
3365 | 3365 | self.linkrev, |
|
3366 | 3366 | self.parentrevs, |
|
3367 | 3367 | ) |
|
3368 | 3368 | |
|
3369 | 3369 | def strip(self, minlink, transaction): |
|
3370 | 3370 | """truncate the revlog on the first revision with a linkrev >= minlink |
|
3371 | 3371 | |
|
3372 | 3372 | This function is called when we're stripping revision minlink and |
|
3373 | 3373 | its descendants from the repository. |
|
3374 | 3374 | |
|
3375 | 3375 | We have to remove all revisions with linkrev >= minlink, because |
|
3376 | 3376 | the equivalent changelog revisions will be renumbered after the |
|
3377 | 3377 | strip. |
|
3378 | 3378 | |
|
3379 | 3379 | So we truncate the revlog on the first of these revisions, and |
|
3380 | 3380 | trust that the caller has saved the revisions that shouldn't be |
|
3381 | 3381 | removed and that it'll re-add them after this truncation. |
|
3382 | 3382 | """ |
|
3383 | 3383 | if len(self) == 0: |
|
3384 | 3384 | return |
|
3385 | 3385 | |
|
3386 | 3386 | rev, _ = self.getstrippoint(minlink) |
|
3387 | 3387 | if rev == len(self): |
|
3388 | 3388 | return |
|
3389 | 3389 | |
|
3390 | 3390 | # first truncate the files on disk |
|
3391 | 3391 | data_end = self.start(rev) |
|
3392 | 3392 | if not self._inline: |
|
3393 | 3393 | transaction.add(self._datafile, data_end) |
|
3394 | 3394 | end = rev * self.index.entry_size |
|
3395 | 3395 | else: |
|
3396 | 3396 | end = data_end + (rev * self.index.entry_size) |
|
3397 | 3397 | |
|
3398 | 3398 | if self._sidedatafile: |
|
3399 | 3399 | sidedata_end = self.sidedata_cut_off(rev) |
|
3400 | 3400 | transaction.add(self._sidedatafile, sidedata_end) |
|
3401 | 3401 | |
|
3402 | 3402 | transaction.add(self._indexfile, end) |
|
3403 | 3403 | if self._docket is not None: |
|
3404 | 3404 | # XXX we could, leverage the docket while stripping. However it is |
|
3405 | 3405 | # not powerfull enough at the time of this comment |
|
3406 | 3406 | self._docket.index_end = end |
|
3407 | 3407 | self._docket.data_end = data_end |
|
3408 | 3408 | self._docket.sidedata_end = sidedata_end |
|
3409 | 3409 | self._docket.write(transaction, stripping=True) |
|
3410 | 3410 | |
|
3411 | 3411 | # then reset internal state in memory to forget those revisions |
|
3412 | 3412 | self._revisioncache = None |
|
3413 | 3413 | self._chaininfocache = util.lrucachedict(500) |
|
3414 | 3414 | self._inner._segmentfile.clear_cache() |
|
3415 | 3415 | self._inner._segmentfile_sidedata.clear_cache() |
|
3416 | 3416 | |
|
3417 | 3417 | del self.index[rev:-1] |
|
3418 | 3418 | |
|
3419 | 3419 | def checksize(self): |
|
3420 | 3420 | """Check size of index and data files |
|
3421 | 3421 | |
|
3422 | 3422 | return a (dd, di) tuple. |
|
3423 | 3423 | - dd: extra bytes for the "data" file |
|
3424 | 3424 | - di: extra bytes for the "index" file |
|
3425 | 3425 | |
|
3426 | 3426 | A healthy revlog will return (0, 0). |
|
3427 | 3427 | """ |
|
3428 | 3428 | expected = 0 |
|
3429 | 3429 | if len(self): |
|
3430 | 3430 | expected = max(0, self.end(len(self) - 1)) |
|
3431 | 3431 | |
|
3432 | 3432 | try: |
|
3433 | 3433 | with self._datafp() as f: |
|
3434 | 3434 | f.seek(0, io.SEEK_END) |
|
3435 | 3435 | actual = f.tell() |
|
3436 | 3436 | dd = actual - expected |
|
3437 | 3437 | except FileNotFoundError: |
|
3438 | 3438 | dd = 0 |
|
3439 | 3439 | |
|
3440 | 3440 | try: |
|
3441 | 3441 | f = self.opener(self._indexfile) |
|
3442 | 3442 | f.seek(0, io.SEEK_END) |
|
3443 | 3443 | actual = f.tell() |
|
3444 | 3444 | f.close() |
|
3445 | 3445 | s = self.index.entry_size |
|
3446 | 3446 | i = max(0, actual // s) |
|
3447 | 3447 | di = actual - (i * s) |
|
3448 | 3448 | if self._inline: |
|
3449 | 3449 | databytes = 0 |
|
3450 | 3450 | for r in self: |
|
3451 | 3451 | databytes += max(0, self.length(r)) |
|
3452 | 3452 | dd = 0 |
|
3453 | 3453 | di = actual - len(self) * s - databytes |
|
3454 | 3454 | except FileNotFoundError: |
|
3455 | 3455 | di = 0 |
|
3456 | 3456 | |
|
3457 | 3457 | return (dd, di) |
|
3458 | 3458 | |
|
3459 | 3459 | def files(self): |
|
3460 | 3460 | """return list of files that compose this revlog""" |
|
3461 | 3461 | res = [self._indexfile] |
|
3462 | 3462 | if self._docket_file is None: |
|
3463 | 3463 | if not self._inline: |
|
3464 | 3464 | res.append(self._datafile) |
|
3465 | 3465 | else: |
|
3466 | 3466 | res.append(self._docket_file) |
|
3467 | 3467 | res.extend(self._docket.old_index_filepaths(include_empty=False)) |
|
3468 | 3468 | if self._docket.data_end: |
|
3469 | 3469 | res.append(self._datafile) |
|
3470 | 3470 | res.extend(self._docket.old_data_filepaths(include_empty=False)) |
|
3471 | 3471 | if self._docket.sidedata_end: |
|
3472 | 3472 | res.append(self._sidedatafile) |
|
3473 | 3473 | res.extend(self._docket.old_sidedata_filepaths(include_empty=False)) |
|
3474 | 3474 | return res |
|
3475 | 3475 | |
|
3476 | 3476 | def emitrevisions( |
|
3477 | 3477 | self, |
|
3478 | 3478 | nodes, |
|
3479 | 3479 | nodesorder=None, |
|
3480 | 3480 | revisiondata=False, |
|
3481 | 3481 | assumehaveparentrevisions=False, |
|
3482 | 3482 | deltamode=repository.CG_DELTAMODE_STD, |
|
3483 | 3483 | sidedata_helpers=None, |
|
3484 | 3484 | debug_info=None, |
|
3485 | 3485 | ): |
|
3486 | 3486 | if nodesorder not in (b'nodes', b'storage', b'linear', None): |
|
3487 | 3487 | raise error.ProgrammingError( |
|
3488 | 3488 | b'unhandled value for nodesorder: %s' % nodesorder |
|
3489 | 3489 | ) |
|
3490 | 3490 | |
|
3491 | 3491 | if nodesorder is None and not self.delta_config.general_delta: |
|
3492 | 3492 | nodesorder = b'storage' |
|
3493 | 3493 | |
|
3494 | 3494 | if ( |
|
3495 | 3495 | not self._storedeltachains |
|
3496 | 3496 | and deltamode != repository.CG_DELTAMODE_PREV |
|
3497 | 3497 | ): |
|
3498 | 3498 | deltamode = repository.CG_DELTAMODE_FULL |
|
3499 | 3499 | |
|
3500 | 3500 | return storageutil.emitrevisions( |
|
3501 | 3501 | self, |
|
3502 | 3502 | nodes, |
|
3503 | 3503 | nodesorder, |
|
3504 | 3504 | revlogrevisiondelta, |
|
3505 | 3505 | deltaparentfn=self.deltaparent, |
|
3506 | 3506 | candeltafn=self._candelta, |
|
3507 | 3507 | rawsizefn=self.rawsize, |
|
3508 | 3508 | revdifffn=self.revdiff, |
|
3509 | 3509 | flagsfn=self.flags, |
|
3510 | 3510 | deltamode=deltamode, |
|
3511 | 3511 | revisiondata=revisiondata, |
|
3512 | 3512 | assumehaveparentrevisions=assumehaveparentrevisions, |
|
3513 | 3513 | sidedata_helpers=sidedata_helpers, |
|
3514 | 3514 | debug_info=debug_info, |
|
3515 | 3515 | ) |
|
3516 | 3516 | |
|
3517 | 3517 | DELTAREUSEALWAYS = b'always' |
|
3518 | 3518 | DELTAREUSESAMEREVS = b'samerevs' |
|
3519 | 3519 | DELTAREUSENEVER = b'never' |
|
3520 | 3520 | |
|
3521 | 3521 | DELTAREUSEFULLADD = b'fulladd' |
|
3522 | 3522 | |
|
3523 | 3523 | DELTAREUSEALL = {b'always', b'samerevs', b'never', b'fulladd'} |
|
3524 | 3524 | |
|
3525 | 3525 | def clone( |
|
3526 | 3526 | self, |
|
3527 | 3527 | tr, |
|
3528 | 3528 | destrevlog, |
|
3529 | 3529 | addrevisioncb=None, |
|
3530 | 3530 | deltareuse=DELTAREUSESAMEREVS, |
|
3531 | 3531 | forcedeltabothparents=None, |
|
3532 | 3532 | sidedata_helpers=None, |
|
3533 | 3533 | ): |
|
3534 | 3534 | """Copy this revlog to another, possibly with format changes. |
|
3535 | 3535 | |
|
3536 | 3536 | The destination revlog will contain the same revisions and nodes. |
|
3537 | 3537 | However, it may not be bit-for-bit identical due to e.g. delta encoding |
|
3538 | 3538 | differences. |
|
3539 | 3539 | |
|
3540 | 3540 | The ``deltareuse`` argument control how deltas from the existing revlog |
|
3541 | 3541 | are preserved in the destination revlog. The argument can have the |
|
3542 | 3542 | following values: |
|
3543 | 3543 | |
|
3544 | 3544 | DELTAREUSEALWAYS |
|
3545 | 3545 | Deltas will always be reused (if possible), even if the destination |
|
3546 | 3546 | revlog would not select the same revisions for the delta. This is the |
|
3547 | 3547 | fastest mode of operation. |
|
3548 | 3548 | DELTAREUSESAMEREVS |
|
3549 | 3549 | Deltas will be reused if the destination revlog would pick the same |
|
3550 | 3550 | revisions for the delta. This mode strikes a balance between speed |
|
3551 | 3551 | and optimization. |
|
3552 | 3552 | DELTAREUSENEVER |
|
3553 | 3553 | Deltas will never be reused. This is the slowest mode of execution. |
|
3554 | 3554 | This mode can be used to recompute deltas (e.g. if the diff/delta |
|
3555 | 3555 | algorithm changes). |
|
3556 | 3556 | DELTAREUSEFULLADD |
|
3557 | 3557 | Revision will be re-added as if their were new content. This is |
|
3558 | 3558 | slower than DELTAREUSEALWAYS but allow more mechanism to kicks in. |
|
3559 | 3559 | eg: large file detection and handling. |
|
3560 | 3560 | |
|
3561 | 3561 | Delta computation can be slow, so the choice of delta reuse policy can |
|
3562 | 3562 | significantly affect run time. |
|
3563 | 3563 | |
|
3564 | 3564 | The default policy (``DELTAREUSESAMEREVS``) strikes a balance between |
|
3565 | 3565 | two extremes. Deltas will be reused if they are appropriate. But if the |
|
3566 | 3566 | delta could choose a better revision, it will do so. This means if you |
|
3567 | 3567 | are converting a non-generaldelta revlog to a generaldelta revlog, |
|
3568 | 3568 | deltas will be recomputed if the delta's parent isn't a parent of the |
|
3569 | 3569 | revision. |
|
3570 | 3570 | |
|
3571 | 3571 | In addition to the delta policy, the ``forcedeltabothparents`` |
|
3572 | 3572 | argument controls whether to force compute deltas against both parents |
|
3573 | 3573 | for merges. By default, the current default is used. |
|
3574 | 3574 | |
|
3575 | 3575 | See `revlogutil.sidedata.get_sidedata_helpers` for the doc on |
|
3576 | 3576 | `sidedata_helpers`. |
|
3577 | 3577 | """ |
|
3578 | 3578 | if deltareuse not in self.DELTAREUSEALL: |
|
3579 | 3579 | raise ValueError( |
|
3580 | 3580 | _(b'value for deltareuse invalid: %s') % deltareuse |
|
3581 | 3581 | ) |
|
3582 | 3582 | |
|
3583 | 3583 | if len(destrevlog): |
|
3584 | 3584 | raise ValueError(_(b'destination revlog is not empty')) |
|
3585 | 3585 | |
|
3586 | 3586 | if getattr(self, 'filteredrevs', None): |
|
3587 | 3587 | raise ValueError(_(b'source revlog has filtered revisions')) |
|
3588 | 3588 | if getattr(destrevlog, 'filteredrevs', None): |
|
3589 | 3589 | raise ValueError(_(b'destination revlog has filtered revisions')) |
|
3590 | 3590 | |
|
3591 | 3591 | # lazydelta and lazydeltabase controls whether to reuse a cached delta, |
|
3592 | 3592 | # if possible. |
|
3593 | 3593 | old_delta_config = destrevlog.delta_config |
|
3594 | 3594 | destrevlog.delta_config = destrevlog.delta_config.copy() |
|
3595 | 3595 | |
|
3596 | 3596 | try: |
|
3597 | 3597 | if deltareuse == self.DELTAREUSEALWAYS: |
|
3598 | 3598 | destrevlog.delta_config.lazy_delta_base = True |
|
3599 | 3599 | destrevlog.delta_config.lazy_delta = True |
|
3600 | 3600 | elif deltareuse == self.DELTAREUSESAMEREVS: |
|
3601 | 3601 | destrevlog.delta_config.lazy_delta_base = False |
|
3602 | 3602 | destrevlog.delta_config.lazy_delta = True |
|
3603 | 3603 | elif deltareuse == self.DELTAREUSENEVER: |
|
3604 | 3604 | destrevlog.delta_config.lazy_delta_base = False |
|
3605 | 3605 | destrevlog.delta_config.lazy_delta = False |
|
3606 | 3606 | |
|
3607 | 3607 | delta_both_parents = ( |
|
3608 | 3608 | forcedeltabothparents or old_delta_config.delta_both_parents |
|
3609 | 3609 | ) |
|
3610 | 3610 | destrevlog.delta_config.delta_both_parents = delta_both_parents |
|
3611 | 3611 | |
|
3612 | 3612 | with self.reading(), destrevlog._writing(tr): |
|
3613 | 3613 | self._clone( |
|
3614 | 3614 | tr, |
|
3615 | 3615 | destrevlog, |
|
3616 | 3616 | addrevisioncb, |
|
3617 | 3617 | deltareuse, |
|
3618 | 3618 | forcedeltabothparents, |
|
3619 | 3619 | sidedata_helpers, |
|
3620 | 3620 | ) |
|
3621 | 3621 | |
|
3622 | 3622 | finally: |
|
3623 | 3623 | destrevlog.delta_config = old_delta_config |
|
3624 | 3624 | |
|
3625 | 3625 | def _clone( |
|
3626 | 3626 | self, |
|
3627 | 3627 | tr, |
|
3628 | 3628 | destrevlog, |
|
3629 | 3629 | addrevisioncb, |
|
3630 | 3630 | deltareuse, |
|
3631 | 3631 | forcedeltabothparents, |
|
3632 | 3632 | sidedata_helpers, |
|
3633 | 3633 | ): |
|
3634 | 3634 | """perform the core duty of `revlog.clone` after parameter processing""" |
|
3635 | 3635 | write_debug = None |
|
3636 | 3636 | if self.delta_config.debug_delta: |
|
3637 | 3637 | write_debug = tr._report |
|
3638 | 3638 | deltacomputer = deltautil.deltacomputer( |
|
3639 | 3639 | destrevlog, |
|
3640 | 3640 | write_debug=write_debug, |
|
3641 | 3641 | ) |
|
3642 | 3642 | index = self.index |
|
3643 | 3643 | for rev in self: |
|
3644 | 3644 | entry = index[rev] |
|
3645 | 3645 | |
|
3646 | 3646 | # Some classes override linkrev to take filtered revs into |
|
3647 | 3647 | # account. Use raw entry from index. |
|
3648 | 3648 | flags = entry[0] & 0xFFFF |
|
3649 | 3649 | linkrev = entry[4] |
|
3650 | 3650 | p1 = index[entry[5]][7] |
|
3651 | 3651 | p2 = index[entry[6]][7] |
|
3652 | 3652 | node = entry[7] |
|
3653 | 3653 | |
|
3654 | 3654 | # (Possibly) reuse the delta from the revlog if allowed and |
|
3655 | 3655 | # the revlog chunk is a delta. |
|
3656 | 3656 | cachedelta = None |
|
3657 | 3657 | rawtext = None |
|
3658 | 3658 | if deltareuse == self.DELTAREUSEFULLADD: |
|
3659 | 3659 | text = self._revisiondata(rev) |
|
3660 | 3660 | sidedata = self.sidedata(rev) |
|
3661 | 3661 | |
|
3662 | 3662 | if sidedata_helpers is not None: |
|
3663 | 3663 | (sidedata, new_flags) = sidedatautil.run_sidedata_helpers( |
|
3664 | 3664 | self, sidedata_helpers, sidedata, rev |
|
3665 | 3665 | ) |
|
3666 | 3666 | flags = flags | new_flags[0] & ~new_flags[1] |
|
3667 | 3667 | |
|
3668 | 3668 | destrevlog.addrevision( |
|
3669 | 3669 | text, |
|
3670 | 3670 | tr, |
|
3671 | 3671 | linkrev, |
|
3672 | 3672 | p1, |
|
3673 | 3673 | p2, |
|
3674 | 3674 | cachedelta=cachedelta, |
|
3675 | 3675 | node=node, |
|
3676 | 3676 | flags=flags, |
|
3677 | 3677 | deltacomputer=deltacomputer, |
|
3678 | 3678 | sidedata=sidedata, |
|
3679 | 3679 | ) |
|
3680 | 3680 | else: |
|
3681 | 3681 | if destrevlog.delta_config.lazy_delta: |
|
3682 | 3682 | dp = self.deltaparent(rev) |
|
3683 | 3683 | if dp != nullrev: |
|
3684 | cachedelta = (dp, bytes(self._chunk(rev))) | |
|
3684 | cachedelta = (dp, bytes(self._inner._chunk(rev))) | |
|
3685 | 3685 | |
|
3686 | 3686 | sidedata = None |
|
3687 | 3687 | if not cachedelta: |
|
3688 | 3688 | try: |
|
3689 | 3689 | rawtext = self._revisiondata(rev) |
|
3690 | 3690 | except error.CensoredNodeError as censored: |
|
3691 | 3691 | assert flags & REVIDX_ISCENSORED |
|
3692 | 3692 | rawtext = censored.tombstone |
|
3693 | 3693 | sidedata = self.sidedata(rev) |
|
3694 | 3694 | if sidedata is None: |
|
3695 | 3695 | sidedata = self.sidedata(rev) |
|
3696 | 3696 | |
|
3697 | 3697 | if sidedata_helpers is not None: |
|
3698 | 3698 | (sidedata, new_flags) = sidedatautil.run_sidedata_helpers( |
|
3699 | 3699 | self, sidedata_helpers, sidedata, rev |
|
3700 | 3700 | ) |
|
3701 | 3701 | flags = flags | new_flags[0] & ~new_flags[1] |
|
3702 | 3702 | |
|
3703 | 3703 | destrevlog._addrevision( |
|
3704 | 3704 | node, |
|
3705 | 3705 | rawtext, |
|
3706 | 3706 | tr, |
|
3707 | 3707 | linkrev, |
|
3708 | 3708 | p1, |
|
3709 | 3709 | p2, |
|
3710 | 3710 | flags, |
|
3711 | 3711 | cachedelta, |
|
3712 | 3712 | deltacomputer=deltacomputer, |
|
3713 | 3713 | sidedata=sidedata, |
|
3714 | 3714 | ) |
|
3715 | 3715 | |
|
3716 | 3716 | if addrevisioncb: |
|
3717 | 3717 | addrevisioncb(self, rev, node) |
|
3718 | 3718 | |
|
3719 | 3719 | def censorrevision(self, tr, censornode, tombstone=b''): |
|
3720 | 3720 | if self._format_version == REVLOGV0: |
|
3721 | 3721 | raise error.RevlogError( |
|
3722 | 3722 | _(b'cannot censor with version %d revlogs') |
|
3723 | 3723 | % self._format_version |
|
3724 | 3724 | ) |
|
3725 | 3725 | elif self._format_version == REVLOGV1: |
|
3726 | 3726 | rewrite.v1_censor(self, tr, censornode, tombstone) |
|
3727 | 3727 | else: |
|
3728 | 3728 | rewrite.v2_censor(self, tr, censornode, tombstone) |
|
3729 | 3729 | |
|
3730 | 3730 | def verifyintegrity(self, state): |
|
3731 | 3731 | """Verifies the integrity of the revlog. |
|
3732 | 3732 | |
|
3733 | 3733 | Yields ``revlogproblem`` instances describing problems that are |
|
3734 | 3734 | found. |
|
3735 | 3735 | """ |
|
3736 | 3736 | dd, di = self.checksize() |
|
3737 | 3737 | if dd: |
|
3738 | 3738 | yield revlogproblem(error=_(b'data length off by %d bytes') % dd) |
|
3739 | 3739 | if di: |
|
3740 | 3740 | yield revlogproblem(error=_(b'index contains %d extra bytes') % di) |
|
3741 | 3741 | |
|
3742 | 3742 | version = self._format_version |
|
3743 | 3743 | |
|
3744 | 3744 | # The verifier tells us what version revlog we should be. |
|
3745 | 3745 | if version != state[b'expectedversion']: |
|
3746 | 3746 | yield revlogproblem( |
|
3747 | 3747 | warning=_(b"warning: '%s' uses revlog format %d; expected %d") |
|
3748 | 3748 | % (self.display_id, version, state[b'expectedversion']) |
|
3749 | 3749 | ) |
|
3750 | 3750 | |
|
3751 | 3751 | state[b'skipread'] = set() |
|
3752 | 3752 | state[b'safe_renamed'] = set() |
|
3753 | 3753 | |
|
3754 | 3754 | for rev in self: |
|
3755 | 3755 | node = self.node(rev) |
|
3756 | 3756 | |
|
3757 | 3757 | # Verify contents. 4 cases to care about: |
|
3758 | 3758 | # |
|
3759 | 3759 | # common: the most common case |
|
3760 | 3760 | # rename: with a rename |
|
3761 | 3761 | # meta: file content starts with b'\1\n', the metadata |
|
3762 | 3762 | # header defined in filelog.py, but without a rename |
|
3763 | 3763 | # ext: content stored externally |
|
3764 | 3764 | # |
|
3765 | 3765 | # More formally, their differences are shown below: |
|
3766 | 3766 | # |
|
3767 | 3767 | # | common | rename | meta | ext |
|
3768 | 3768 | # ------------------------------------------------------- |
|
3769 | 3769 | # flags() | 0 | 0 | 0 | not 0 |
|
3770 | 3770 | # renamed() | False | True | False | ? |
|
3771 | 3771 | # rawtext[0:2]=='\1\n'| False | True | True | ? |
|
3772 | 3772 | # |
|
3773 | 3773 | # "rawtext" means the raw text stored in revlog data, which |
|
3774 | 3774 | # could be retrieved by "rawdata(rev)". "text" |
|
3775 | 3775 | # mentioned below is "revision(rev)". |
|
3776 | 3776 | # |
|
3777 | 3777 | # There are 3 different lengths stored physically: |
|
3778 | 3778 | # 1. L1: rawsize, stored in revlog index |
|
3779 | 3779 | # 2. L2: len(rawtext), stored in revlog data |
|
3780 | 3780 | # 3. L3: len(text), stored in revlog data if flags==0, or |
|
3781 | 3781 | # possibly somewhere else if flags!=0 |
|
3782 | 3782 | # |
|
3783 | 3783 | # L1 should be equal to L2. L3 could be different from them. |
|
3784 | 3784 | # "text" may or may not affect commit hash depending on flag |
|
3785 | 3785 | # processors (see flagutil.addflagprocessor). |
|
3786 | 3786 | # |
|
3787 | 3787 | # | common | rename | meta | ext |
|
3788 | 3788 | # ------------------------------------------------- |
|
3789 | 3789 | # rawsize() | L1 | L1 | L1 | L1 |
|
3790 | 3790 | # size() | L1 | L2-LM | L1(*) | L1 (?) |
|
3791 | 3791 | # len(rawtext) | L2 | L2 | L2 | L2 |
|
3792 | 3792 | # len(text) | L2 | L2 | L2 | L3 |
|
3793 | 3793 | # len(read()) | L2 | L2-LM | L2-LM | L3 (?) |
|
3794 | 3794 | # |
|
3795 | 3795 | # LM: length of metadata, depending on rawtext |
|
3796 | 3796 | # (*): not ideal, see comment in filelog.size |
|
3797 | 3797 | # (?): could be "- len(meta)" if the resolved content has |
|
3798 | 3798 | # rename metadata |
|
3799 | 3799 | # |
|
3800 | 3800 | # Checks needed to be done: |
|
3801 | 3801 | # 1. length check: L1 == L2, in all cases. |
|
3802 | 3802 | # 2. hash check: depending on flag processor, we may need to |
|
3803 | 3803 | # use either "text" (external), or "rawtext" (in revlog). |
|
3804 | 3804 | |
|
3805 | 3805 | try: |
|
3806 | 3806 | skipflags = state.get(b'skipflags', 0) |
|
3807 | 3807 | if skipflags: |
|
3808 | 3808 | skipflags &= self.flags(rev) |
|
3809 | 3809 | |
|
3810 | 3810 | _verify_revision(self, skipflags, state, node) |
|
3811 | 3811 | |
|
3812 | 3812 | l1 = self.rawsize(rev) |
|
3813 | 3813 | l2 = len(self.rawdata(node)) |
|
3814 | 3814 | |
|
3815 | 3815 | if l1 != l2: |
|
3816 | 3816 | yield revlogproblem( |
|
3817 | 3817 | error=_(b'unpacked size is %d, %d expected') % (l2, l1), |
|
3818 | 3818 | node=node, |
|
3819 | 3819 | ) |
|
3820 | 3820 | |
|
3821 | 3821 | except error.CensoredNodeError: |
|
3822 | 3822 | if state[b'erroroncensored']: |
|
3823 | 3823 | yield revlogproblem( |
|
3824 | 3824 | error=_(b'censored file data'), node=node |
|
3825 | 3825 | ) |
|
3826 | 3826 | state[b'skipread'].add(node) |
|
3827 | 3827 | except Exception as e: |
|
3828 | 3828 | yield revlogproblem( |
|
3829 | 3829 | error=_(b'unpacking %s: %s') |
|
3830 | 3830 | % (short(node), stringutil.forcebytestr(e)), |
|
3831 | 3831 | node=node, |
|
3832 | 3832 | ) |
|
3833 | 3833 | state[b'skipread'].add(node) |
|
3834 | 3834 | |
|
3835 | 3835 | def storageinfo( |
|
3836 | 3836 | self, |
|
3837 | 3837 | exclusivefiles=False, |
|
3838 | 3838 | sharedfiles=False, |
|
3839 | 3839 | revisionscount=False, |
|
3840 | 3840 | trackedsize=False, |
|
3841 | 3841 | storedsize=False, |
|
3842 | 3842 | ): |
|
3843 | 3843 | d = {} |
|
3844 | 3844 | |
|
3845 | 3845 | if exclusivefiles: |
|
3846 | 3846 | d[b'exclusivefiles'] = [(self.opener, self._indexfile)] |
|
3847 | 3847 | if not self._inline: |
|
3848 | 3848 | d[b'exclusivefiles'].append((self.opener, self._datafile)) |
|
3849 | 3849 | |
|
3850 | 3850 | if sharedfiles: |
|
3851 | 3851 | d[b'sharedfiles'] = [] |
|
3852 | 3852 | |
|
3853 | 3853 | if revisionscount: |
|
3854 | 3854 | d[b'revisionscount'] = len(self) |
|
3855 | 3855 | |
|
3856 | 3856 | if trackedsize: |
|
3857 | 3857 | d[b'trackedsize'] = sum(map(self.rawsize, iter(self))) |
|
3858 | 3858 | |
|
3859 | 3859 | if storedsize: |
|
3860 | 3860 | d[b'storedsize'] = sum( |
|
3861 | 3861 | self.opener.stat(path).st_size for path in self.files() |
|
3862 | 3862 | ) |
|
3863 | 3863 | |
|
3864 | 3864 | return d |
|
3865 | 3865 | |
|
3866 | 3866 | def rewrite_sidedata(self, transaction, helpers, startrev, endrev): |
|
3867 | 3867 | if not self.feature_config.has_side_data: |
|
3868 | 3868 | return |
|
3869 | 3869 | # revlog formats with sidedata support does not support inline |
|
3870 | 3870 | assert not self._inline |
|
3871 | 3871 | if not helpers[1] and not helpers[2]: |
|
3872 | 3872 | # Nothing to generate or remove |
|
3873 | 3873 | return |
|
3874 | 3874 | |
|
3875 | 3875 | new_entries = [] |
|
3876 | 3876 | # append the new sidedata |
|
3877 | 3877 | with self._writing(transaction): |
|
3878 | 3878 | ifh, dfh, sdfh = self._inner._writinghandles |
|
3879 | 3879 | dfh.seek(self._docket.sidedata_end, os.SEEK_SET) |
|
3880 | 3880 | |
|
3881 | 3881 | current_offset = sdfh.tell() |
|
3882 | 3882 | for rev in range(startrev, endrev + 1): |
|
3883 | 3883 | entry = self.index[rev] |
|
3884 | 3884 | new_sidedata, flags = sidedatautil.run_sidedata_helpers( |
|
3885 | 3885 | store=self, |
|
3886 | 3886 | sidedata_helpers=helpers, |
|
3887 | 3887 | sidedata={}, |
|
3888 | 3888 | rev=rev, |
|
3889 | 3889 | ) |
|
3890 | 3890 | |
|
3891 | 3891 | serialized_sidedata = sidedatautil.serialize_sidedata( |
|
3892 | 3892 | new_sidedata |
|
3893 | 3893 | ) |
|
3894 | 3894 | |
|
3895 | 3895 | sidedata_compression_mode = COMP_MODE_INLINE |
|
3896 | 3896 | if serialized_sidedata and self.feature_config.has_side_data: |
|
3897 | 3897 | sidedata_compression_mode = COMP_MODE_PLAIN |
|
3898 | 3898 | h, comp_sidedata = self._inner.compress(serialized_sidedata) |
|
3899 | 3899 | if ( |
|
3900 | 3900 | h != b'u' |
|
3901 | 3901 | and comp_sidedata[0] != b'\0' |
|
3902 | 3902 | and len(comp_sidedata) < len(serialized_sidedata) |
|
3903 | 3903 | ): |
|
3904 | 3904 | assert not h |
|
3905 | 3905 | if ( |
|
3906 | 3906 | comp_sidedata[0] |
|
3907 | 3907 | == self._docket.default_compression_header |
|
3908 | 3908 | ): |
|
3909 | 3909 | sidedata_compression_mode = COMP_MODE_DEFAULT |
|
3910 | 3910 | serialized_sidedata = comp_sidedata |
|
3911 | 3911 | else: |
|
3912 | 3912 | sidedata_compression_mode = COMP_MODE_INLINE |
|
3913 | 3913 | serialized_sidedata = comp_sidedata |
|
3914 | 3914 | if entry[8] != 0 or entry[9] != 0: |
|
3915 | 3915 | # rewriting entries that already have sidedata is not |
|
3916 | 3916 | # supported yet, because it introduces garbage data in the |
|
3917 | 3917 | # revlog. |
|
3918 | 3918 | msg = b"rewriting existing sidedata is not supported yet" |
|
3919 | 3919 | raise error.Abort(msg) |
|
3920 | 3920 | |
|
3921 | 3921 | # Apply (potential) flags to add and to remove after running |
|
3922 | 3922 | # the sidedata helpers |
|
3923 | 3923 | new_offset_flags = entry[0] | flags[0] & ~flags[1] |
|
3924 | 3924 | entry_update = ( |
|
3925 | 3925 | current_offset, |
|
3926 | 3926 | len(serialized_sidedata), |
|
3927 | 3927 | new_offset_flags, |
|
3928 | 3928 | sidedata_compression_mode, |
|
3929 | 3929 | ) |
|
3930 | 3930 | |
|
3931 | 3931 | # the sidedata computation might have move the file cursors around |
|
3932 | 3932 | sdfh.seek(current_offset, os.SEEK_SET) |
|
3933 | 3933 | sdfh.write(serialized_sidedata) |
|
3934 | 3934 | new_entries.append(entry_update) |
|
3935 | 3935 | current_offset += len(serialized_sidedata) |
|
3936 | 3936 | self._docket.sidedata_end = sdfh.tell() |
|
3937 | 3937 | |
|
3938 | 3938 | # rewrite the new index entries |
|
3939 | 3939 | ifh.seek(startrev * self.index.entry_size) |
|
3940 | 3940 | for i, e in enumerate(new_entries): |
|
3941 | 3941 | rev = startrev + i |
|
3942 | 3942 | self.index.replace_sidedata_info(rev, *e) |
|
3943 | 3943 | packed = self.index.entry_binary(rev) |
|
3944 | 3944 | if rev == 0 and self._docket is None: |
|
3945 | 3945 | header = self._format_flags | self._format_version |
|
3946 | 3946 | header = self.index.pack_header(header) |
|
3947 | 3947 | packed = header + packed |
|
3948 | 3948 | ifh.write(packed) |
@@ -1,876 +1,876 b'' | |||
|
1 | 1 | # censor code related to censoring revision |
|
2 | 2 | # coding: utf8 |
|
3 | 3 | # |
|
4 | 4 | # Copyright 2021 Pierre-Yves David <pierre-yves.david@octobus.net> |
|
5 | 5 | # Copyright 2015 Google, Inc <martinvonz@google.com> |
|
6 | 6 | # |
|
7 | 7 | # This software may be used and distributed according to the terms of the |
|
8 | 8 | # GNU General Public License version 2 or any later version. |
|
9 | 9 | |
|
10 | 10 | import binascii |
|
11 | 11 | import contextlib |
|
12 | 12 | import os |
|
13 | 13 | import struct |
|
14 | 14 | |
|
15 | 15 | from ..node import ( |
|
16 | 16 | nullrev, |
|
17 | 17 | ) |
|
18 | 18 | from .constants import ( |
|
19 | 19 | COMP_MODE_PLAIN, |
|
20 | 20 | ENTRY_DATA_COMPRESSED_LENGTH, |
|
21 | 21 | ENTRY_DATA_COMPRESSION_MODE, |
|
22 | 22 | ENTRY_DATA_OFFSET, |
|
23 | 23 | ENTRY_DATA_UNCOMPRESSED_LENGTH, |
|
24 | 24 | ENTRY_DELTA_BASE, |
|
25 | 25 | ENTRY_LINK_REV, |
|
26 | 26 | ENTRY_NODE_ID, |
|
27 | 27 | ENTRY_PARENT_1, |
|
28 | 28 | ENTRY_PARENT_2, |
|
29 | 29 | ENTRY_SIDEDATA_COMPRESSED_LENGTH, |
|
30 | 30 | ENTRY_SIDEDATA_COMPRESSION_MODE, |
|
31 | 31 | ENTRY_SIDEDATA_OFFSET, |
|
32 | 32 | REVIDX_ISCENSORED, |
|
33 | 33 | REVLOGV0, |
|
34 | 34 | REVLOGV1, |
|
35 | 35 | ) |
|
36 | 36 | from ..i18n import _ |
|
37 | 37 | |
|
38 | 38 | from .. import ( |
|
39 | 39 | error, |
|
40 | 40 | mdiff, |
|
41 | 41 | pycompat, |
|
42 | 42 | revlogutils, |
|
43 | 43 | util, |
|
44 | 44 | ) |
|
45 | 45 | from ..utils import ( |
|
46 | 46 | storageutil, |
|
47 | 47 | ) |
|
48 | 48 | from . import ( |
|
49 | 49 | constants, |
|
50 | 50 | deltas, |
|
51 | 51 | ) |
|
52 | 52 | |
|
53 | 53 | |
|
54 | 54 | def v1_censor(rl, tr, censornode, tombstone=b''): |
|
55 | 55 | """censors a revision in a "version 1" revlog""" |
|
56 | 56 | assert rl._format_version == constants.REVLOGV1, rl._format_version |
|
57 | 57 | |
|
58 | 58 | # avoid cycle |
|
59 | 59 | from .. import revlog |
|
60 | 60 | |
|
61 | 61 | censorrev = rl.rev(censornode) |
|
62 | 62 | tombstone = storageutil.packmeta({b'censored': tombstone}, b'') |
|
63 | 63 | |
|
64 | 64 | # Rewriting the revlog in place is hard. Our strategy for censoring is |
|
65 | 65 | # to create a new revlog, copy all revisions to it, then replace the |
|
66 | 66 | # revlogs on transaction close. |
|
67 | 67 | # |
|
68 | 68 | # This is a bit dangerous. We could easily have a mismatch of state. |
|
69 | 69 | newrl = revlog.revlog( |
|
70 | 70 | rl.opener, |
|
71 | 71 | target=rl.target, |
|
72 | 72 | radix=rl.radix, |
|
73 | 73 | postfix=b'tmpcensored', |
|
74 | 74 | censorable=True, |
|
75 | 75 | ) |
|
76 | 76 | newrl._format_version = rl._format_version |
|
77 | 77 | newrl._format_flags = rl._format_flags |
|
78 | 78 | newrl.delta_config.general_delta = rl.delta_config.general_delta |
|
79 | 79 | newrl._parse_index = rl._parse_index |
|
80 | 80 | |
|
81 | 81 | for rev in rl.revs(): |
|
82 | 82 | node = rl.node(rev) |
|
83 | 83 | p1, p2 = rl.parents(node) |
|
84 | 84 | |
|
85 | 85 | if rev == censorrev: |
|
86 | 86 | newrl.addrawrevision( |
|
87 | 87 | tombstone, |
|
88 | 88 | tr, |
|
89 | 89 | rl.linkrev(censorrev), |
|
90 | 90 | p1, |
|
91 | 91 | p2, |
|
92 | 92 | censornode, |
|
93 | 93 | constants.REVIDX_ISCENSORED, |
|
94 | 94 | ) |
|
95 | 95 | |
|
96 | 96 | if newrl.deltaparent(rev) != nullrev: |
|
97 | 97 | m = _(b'censored revision stored as delta; cannot censor') |
|
98 | 98 | h = _( |
|
99 | 99 | b'censoring of revlogs is not fully implemented;' |
|
100 | 100 | b' please report this bug' |
|
101 | 101 | ) |
|
102 | 102 | raise error.Abort(m, hint=h) |
|
103 | 103 | continue |
|
104 | 104 | |
|
105 | 105 | if rl.iscensored(rev): |
|
106 | 106 | if rl.deltaparent(rev) != nullrev: |
|
107 | 107 | m = _( |
|
108 | 108 | b'cannot censor due to censored ' |
|
109 | 109 | b'revision having delta stored' |
|
110 | 110 | ) |
|
111 | 111 | raise error.Abort(m) |
|
112 | rawtext = rl._chunk(rev) | |
|
112 | rawtext = rl._inner._chunk(rev) | |
|
113 | 113 | else: |
|
114 | 114 | rawtext = rl.rawdata(rev) |
|
115 | 115 | |
|
116 | 116 | newrl.addrawrevision( |
|
117 | 117 | rawtext, tr, rl.linkrev(rev), p1, p2, node, rl.flags(rev) |
|
118 | 118 | ) |
|
119 | 119 | |
|
120 | 120 | tr.addbackup(rl._indexfile, location=b'store') |
|
121 | 121 | if not rl._inline: |
|
122 | 122 | tr.addbackup(rl._datafile, location=b'store') |
|
123 | 123 | |
|
124 | 124 | rl.opener.rename(newrl._indexfile, rl._indexfile) |
|
125 | 125 | if not rl._inline: |
|
126 | 126 | rl.opener.rename(newrl._datafile, rl._datafile) |
|
127 | 127 | |
|
128 | 128 | rl.clearcaches() |
|
129 | 129 | chunk_cache = rl._loadindex() |
|
130 | 130 | rl._load_inner(chunk_cache) |
|
131 | 131 | |
|
132 | 132 | |
|
133 | 133 | def v2_censor(revlog, tr, censornode, tombstone=b''): |
|
134 | 134 | """censors a revision in a "version 2" revlog""" |
|
135 | 135 | assert revlog._format_version != REVLOGV0, revlog._format_version |
|
136 | 136 | assert revlog._format_version != REVLOGV1, revlog._format_version |
|
137 | 137 | |
|
138 | 138 | censor_revs = {revlog.rev(censornode)} |
|
139 | 139 | _rewrite_v2(revlog, tr, censor_revs, tombstone) |
|
140 | 140 | |
|
141 | 141 | |
|
142 | 142 | def _rewrite_v2(revlog, tr, censor_revs, tombstone=b''): |
|
143 | 143 | """rewrite a revlog to censor some of its content |
|
144 | 144 | |
|
145 | 145 | General principle |
|
146 | 146 | |
|
147 | 147 | We create new revlog files (index/data/sidedata) to copy the content of |
|
148 | 148 | the existing data without the censored data. |
|
149 | 149 | |
|
150 | 150 | We need to recompute new delta for any revision that used the censored |
|
151 | 151 | revision as delta base. As the cumulative size of the new delta may be |
|
152 | 152 | large, we store them in a temporary file until they are stored in their |
|
153 | 153 | final destination. |
|
154 | 154 | |
|
155 | 155 | All data before the censored data can be blindly copied. The rest needs |
|
156 | 156 | to be copied as we go and the associated index entry needs adjustement. |
|
157 | 157 | """ |
|
158 | 158 | assert revlog._format_version != REVLOGV0, revlog._format_version |
|
159 | 159 | assert revlog._format_version != REVLOGV1, revlog._format_version |
|
160 | 160 | |
|
161 | 161 | old_index = revlog.index |
|
162 | 162 | docket = revlog._docket |
|
163 | 163 | |
|
164 | 164 | tombstone = storageutil.packmeta({b'censored': tombstone}, b'') |
|
165 | 165 | |
|
166 | 166 | first_excl_rev = min(censor_revs) |
|
167 | 167 | |
|
168 | 168 | first_excl_entry = revlog.index[first_excl_rev] |
|
169 | 169 | index_cutoff = revlog.index.entry_size * first_excl_rev |
|
170 | 170 | data_cutoff = first_excl_entry[ENTRY_DATA_OFFSET] >> 16 |
|
171 | 171 | sidedata_cutoff = revlog.sidedata_cut_off(first_excl_rev) |
|
172 | 172 | |
|
173 | 173 | with pycompat.unnamedtempfile(mode=b"w+b") as tmp_storage: |
|
174 | 174 | # rev β (new_base, data_start, data_end, compression_mode) |
|
175 | 175 | rewritten_entries = _precompute_rewritten_delta( |
|
176 | 176 | revlog, |
|
177 | 177 | old_index, |
|
178 | 178 | censor_revs, |
|
179 | 179 | tmp_storage, |
|
180 | 180 | ) |
|
181 | 181 | |
|
182 | 182 | all_files = _setup_new_files( |
|
183 | 183 | revlog, |
|
184 | 184 | index_cutoff, |
|
185 | 185 | data_cutoff, |
|
186 | 186 | sidedata_cutoff, |
|
187 | 187 | ) |
|
188 | 188 | |
|
189 | 189 | # we dont need to open the old index file since its content already |
|
190 | 190 | # exist in a usable form in `old_index`. |
|
191 | 191 | with all_files() as open_files: |
|
192 | 192 | ( |
|
193 | 193 | old_data_file, |
|
194 | 194 | old_sidedata_file, |
|
195 | 195 | new_index_file, |
|
196 | 196 | new_data_file, |
|
197 | 197 | new_sidedata_file, |
|
198 | 198 | ) = open_files |
|
199 | 199 | |
|
200 | 200 | # writing the censored revision |
|
201 | 201 | |
|
202 | 202 | # Writing all subsequent revisions |
|
203 | 203 | for rev in range(first_excl_rev, len(old_index)): |
|
204 | 204 | if rev in censor_revs: |
|
205 | 205 | _rewrite_censor( |
|
206 | 206 | revlog, |
|
207 | 207 | old_index, |
|
208 | 208 | open_files, |
|
209 | 209 | rev, |
|
210 | 210 | tombstone, |
|
211 | 211 | ) |
|
212 | 212 | else: |
|
213 | 213 | _rewrite_simple( |
|
214 | 214 | revlog, |
|
215 | 215 | old_index, |
|
216 | 216 | open_files, |
|
217 | 217 | rev, |
|
218 | 218 | rewritten_entries, |
|
219 | 219 | tmp_storage, |
|
220 | 220 | ) |
|
221 | 221 | docket.write(transaction=None, stripping=True) |
|
222 | 222 | |
|
223 | 223 | |
|
224 | 224 | def _precompute_rewritten_delta( |
|
225 | 225 | revlog, |
|
226 | 226 | old_index, |
|
227 | 227 | excluded_revs, |
|
228 | 228 | tmp_storage, |
|
229 | 229 | ): |
|
230 | 230 | """Compute new delta for revisions whose delta is based on revision that |
|
231 | 231 | will not survive as is. |
|
232 | 232 | |
|
233 | 233 | Return a mapping: {rev β (new_base, data_start, data_end, compression_mode)} |
|
234 | 234 | """ |
|
235 | 235 | dc = deltas.deltacomputer(revlog) |
|
236 | 236 | rewritten_entries = {} |
|
237 | 237 | first_excl_rev = min(excluded_revs) |
|
238 | 238 | with revlog.reading(): |
|
239 | 239 | for rev in range(first_excl_rev, len(old_index)): |
|
240 | 240 | if rev in excluded_revs: |
|
241 | 241 | # this revision will be preserved as is, so we don't need to |
|
242 | 242 | # consider recomputing a delta. |
|
243 | 243 | continue |
|
244 | 244 | entry = old_index[rev] |
|
245 | 245 | if entry[ENTRY_DELTA_BASE] not in excluded_revs: |
|
246 | 246 | continue |
|
247 | 247 | # This is a revision that use the censored revision as the base |
|
248 | 248 | # for its delta. We need a need new deltas |
|
249 | 249 | if entry[ENTRY_DATA_UNCOMPRESSED_LENGTH] == 0: |
|
250 | 250 | # this revision is empty, we can delta against nullrev |
|
251 | 251 | rewritten_entries[rev] = (nullrev, 0, 0, COMP_MODE_PLAIN) |
|
252 | 252 | else: |
|
253 | 253 | |
|
254 | 254 | text = revlog.rawdata(rev) |
|
255 | 255 | info = revlogutils.revisioninfo( |
|
256 | 256 | node=entry[ENTRY_NODE_ID], |
|
257 | 257 | p1=revlog.node(entry[ENTRY_PARENT_1]), |
|
258 | 258 | p2=revlog.node(entry[ENTRY_PARENT_2]), |
|
259 | 259 | btext=[text], |
|
260 | 260 | textlen=len(text), |
|
261 | 261 | cachedelta=None, |
|
262 | 262 | flags=entry[ENTRY_DATA_OFFSET] & 0xFFFF, |
|
263 | 263 | ) |
|
264 | 264 | d = dc.finddeltainfo( |
|
265 | 265 | info, excluded_bases=excluded_revs, target_rev=rev |
|
266 | 266 | ) |
|
267 | 267 | default_comp = revlog._docket.default_compression_header |
|
268 | 268 | comp_mode, d = deltas.delta_compression(default_comp, d) |
|
269 | 269 | # using `tell` is a bit lazy, but we are not here for speed |
|
270 | 270 | start = tmp_storage.tell() |
|
271 | 271 | tmp_storage.write(d.data[1]) |
|
272 | 272 | end = tmp_storage.tell() |
|
273 | 273 | rewritten_entries[rev] = (d.base, start, end, comp_mode) |
|
274 | 274 | return rewritten_entries |
|
275 | 275 | |
|
276 | 276 | |
|
277 | 277 | def _setup_new_files( |
|
278 | 278 | revlog, |
|
279 | 279 | index_cutoff, |
|
280 | 280 | data_cutoff, |
|
281 | 281 | sidedata_cutoff, |
|
282 | 282 | ): |
|
283 | 283 | """ |
|
284 | 284 | |
|
285 | 285 | return a context manager to open all the relevant files: |
|
286 | 286 | - old_data_file, |
|
287 | 287 | - old_sidedata_file, |
|
288 | 288 | - new_index_file, |
|
289 | 289 | - new_data_file, |
|
290 | 290 | - new_sidedata_file, |
|
291 | 291 | |
|
292 | 292 | The old_index_file is not here because it is accessed through the |
|
293 | 293 | `old_index` object if the caller function. |
|
294 | 294 | """ |
|
295 | 295 | docket = revlog._docket |
|
296 | 296 | old_index_filepath = revlog.opener.join(docket.index_filepath()) |
|
297 | 297 | old_data_filepath = revlog.opener.join(docket.data_filepath()) |
|
298 | 298 | old_sidedata_filepath = revlog.opener.join(docket.sidedata_filepath()) |
|
299 | 299 | |
|
300 | 300 | new_index_filepath = revlog.opener.join(docket.new_index_file()) |
|
301 | 301 | new_data_filepath = revlog.opener.join(docket.new_data_file()) |
|
302 | 302 | new_sidedata_filepath = revlog.opener.join(docket.new_sidedata_file()) |
|
303 | 303 | |
|
304 | 304 | util.copyfile(old_index_filepath, new_index_filepath, nb_bytes=index_cutoff) |
|
305 | 305 | util.copyfile(old_data_filepath, new_data_filepath, nb_bytes=data_cutoff) |
|
306 | 306 | util.copyfile( |
|
307 | 307 | old_sidedata_filepath, |
|
308 | 308 | new_sidedata_filepath, |
|
309 | 309 | nb_bytes=sidedata_cutoff, |
|
310 | 310 | ) |
|
311 | 311 | revlog.opener.register_file(docket.index_filepath()) |
|
312 | 312 | revlog.opener.register_file(docket.data_filepath()) |
|
313 | 313 | revlog.opener.register_file(docket.sidedata_filepath()) |
|
314 | 314 | |
|
315 | 315 | docket.index_end = index_cutoff |
|
316 | 316 | docket.data_end = data_cutoff |
|
317 | 317 | docket.sidedata_end = sidedata_cutoff |
|
318 | 318 | |
|
319 | 319 | # reload the revlog internal information |
|
320 | 320 | revlog.clearcaches() |
|
321 | 321 | revlog._loadindex(docket=docket) |
|
322 | 322 | |
|
323 | 323 | @contextlib.contextmanager |
|
324 | 324 | def all_files_opener(): |
|
325 | 325 | # hide opening in an helper function to please check-code, black |
|
326 | 326 | # and various python version at the same time |
|
327 | 327 | with open(old_data_filepath, 'rb') as old_data_file: |
|
328 | 328 | with open(old_sidedata_filepath, 'rb') as old_sidedata_file: |
|
329 | 329 | with open(new_index_filepath, 'r+b') as new_index_file: |
|
330 | 330 | with open(new_data_filepath, 'r+b') as new_data_file: |
|
331 | 331 | with open( |
|
332 | 332 | new_sidedata_filepath, 'r+b' |
|
333 | 333 | ) as new_sidedata_file: |
|
334 | 334 | new_index_file.seek(0, os.SEEK_END) |
|
335 | 335 | assert new_index_file.tell() == index_cutoff |
|
336 | 336 | new_data_file.seek(0, os.SEEK_END) |
|
337 | 337 | assert new_data_file.tell() == data_cutoff |
|
338 | 338 | new_sidedata_file.seek(0, os.SEEK_END) |
|
339 | 339 | assert new_sidedata_file.tell() == sidedata_cutoff |
|
340 | 340 | yield ( |
|
341 | 341 | old_data_file, |
|
342 | 342 | old_sidedata_file, |
|
343 | 343 | new_index_file, |
|
344 | 344 | new_data_file, |
|
345 | 345 | new_sidedata_file, |
|
346 | 346 | ) |
|
347 | 347 | |
|
348 | 348 | return all_files_opener |
|
349 | 349 | |
|
350 | 350 | |
|
351 | 351 | def _rewrite_simple( |
|
352 | 352 | revlog, |
|
353 | 353 | old_index, |
|
354 | 354 | all_files, |
|
355 | 355 | rev, |
|
356 | 356 | rewritten_entries, |
|
357 | 357 | tmp_storage, |
|
358 | 358 | ): |
|
359 | 359 | """append a normal revision to the index after the rewritten one(s)""" |
|
360 | 360 | ( |
|
361 | 361 | old_data_file, |
|
362 | 362 | old_sidedata_file, |
|
363 | 363 | new_index_file, |
|
364 | 364 | new_data_file, |
|
365 | 365 | new_sidedata_file, |
|
366 | 366 | ) = all_files |
|
367 | 367 | entry = old_index[rev] |
|
368 | 368 | flags = entry[ENTRY_DATA_OFFSET] & 0xFFFF |
|
369 | 369 | old_data_offset = entry[ENTRY_DATA_OFFSET] >> 16 |
|
370 | 370 | |
|
371 | 371 | if rev not in rewritten_entries: |
|
372 | 372 | old_data_file.seek(old_data_offset) |
|
373 | 373 | new_data_size = entry[ENTRY_DATA_COMPRESSED_LENGTH] |
|
374 | 374 | new_data = old_data_file.read(new_data_size) |
|
375 | 375 | data_delta_base = entry[ENTRY_DELTA_BASE] |
|
376 | 376 | d_comp_mode = entry[ENTRY_DATA_COMPRESSION_MODE] |
|
377 | 377 | else: |
|
378 | 378 | ( |
|
379 | 379 | data_delta_base, |
|
380 | 380 | start, |
|
381 | 381 | end, |
|
382 | 382 | d_comp_mode, |
|
383 | 383 | ) = rewritten_entries[rev] |
|
384 | 384 | new_data_size = end - start |
|
385 | 385 | tmp_storage.seek(start) |
|
386 | 386 | new_data = tmp_storage.read(new_data_size) |
|
387 | 387 | |
|
388 | 388 | # It might be faster to group continuous read/write operation, |
|
389 | 389 | # however, this is censor, an operation that is not focussed |
|
390 | 390 | # around stellar performance. So I have not written this |
|
391 | 391 | # optimisation yet. |
|
392 | 392 | new_data_offset = new_data_file.tell() |
|
393 | 393 | new_data_file.write(new_data) |
|
394 | 394 | |
|
395 | 395 | sidedata_size = entry[ENTRY_SIDEDATA_COMPRESSED_LENGTH] |
|
396 | 396 | new_sidedata_offset = new_sidedata_file.tell() |
|
397 | 397 | if 0 < sidedata_size: |
|
398 | 398 | old_sidedata_offset = entry[ENTRY_SIDEDATA_OFFSET] |
|
399 | 399 | old_sidedata_file.seek(old_sidedata_offset) |
|
400 | 400 | new_sidedata = old_sidedata_file.read(sidedata_size) |
|
401 | 401 | new_sidedata_file.write(new_sidedata) |
|
402 | 402 | |
|
403 | 403 | data_uncompressed_length = entry[ENTRY_DATA_UNCOMPRESSED_LENGTH] |
|
404 | 404 | sd_com_mode = entry[ENTRY_SIDEDATA_COMPRESSION_MODE] |
|
405 | 405 | assert data_delta_base <= rev, (data_delta_base, rev) |
|
406 | 406 | |
|
407 | 407 | new_entry = revlogutils.entry( |
|
408 | 408 | flags=flags, |
|
409 | 409 | data_offset=new_data_offset, |
|
410 | 410 | data_compressed_length=new_data_size, |
|
411 | 411 | data_uncompressed_length=data_uncompressed_length, |
|
412 | 412 | data_delta_base=data_delta_base, |
|
413 | 413 | link_rev=entry[ENTRY_LINK_REV], |
|
414 | 414 | parent_rev_1=entry[ENTRY_PARENT_1], |
|
415 | 415 | parent_rev_2=entry[ENTRY_PARENT_2], |
|
416 | 416 | node_id=entry[ENTRY_NODE_ID], |
|
417 | 417 | sidedata_offset=new_sidedata_offset, |
|
418 | 418 | sidedata_compressed_length=sidedata_size, |
|
419 | 419 | data_compression_mode=d_comp_mode, |
|
420 | 420 | sidedata_compression_mode=sd_com_mode, |
|
421 | 421 | ) |
|
422 | 422 | revlog.index.append(new_entry) |
|
423 | 423 | entry_bin = revlog.index.entry_binary(rev) |
|
424 | 424 | new_index_file.write(entry_bin) |
|
425 | 425 | |
|
426 | 426 | revlog._docket.index_end = new_index_file.tell() |
|
427 | 427 | revlog._docket.data_end = new_data_file.tell() |
|
428 | 428 | revlog._docket.sidedata_end = new_sidedata_file.tell() |
|
429 | 429 | |
|
430 | 430 | |
|
431 | 431 | def _rewrite_censor( |
|
432 | 432 | revlog, |
|
433 | 433 | old_index, |
|
434 | 434 | all_files, |
|
435 | 435 | rev, |
|
436 | 436 | tombstone, |
|
437 | 437 | ): |
|
438 | 438 | """rewrite and append a censored revision""" |
|
439 | 439 | ( |
|
440 | 440 | old_data_file, |
|
441 | 441 | old_sidedata_file, |
|
442 | 442 | new_index_file, |
|
443 | 443 | new_data_file, |
|
444 | 444 | new_sidedata_file, |
|
445 | 445 | ) = all_files |
|
446 | 446 | entry = old_index[rev] |
|
447 | 447 | |
|
448 | 448 | # XXX consider trying the default compression too |
|
449 | 449 | new_data_size = len(tombstone) |
|
450 | 450 | new_data_offset = new_data_file.tell() |
|
451 | 451 | new_data_file.write(tombstone) |
|
452 | 452 | |
|
453 | 453 | # we are not adding any sidedata as they might leak info about the censored version |
|
454 | 454 | |
|
455 | 455 | link_rev = entry[ENTRY_LINK_REV] |
|
456 | 456 | |
|
457 | 457 | p1 = entry[ENTRY_PARENT_1] |
|
458 | 458 | p2 = entry[ENTRY_PARENT_2] |
|
459 | 459 | |
|
460 | 460 | new_entry = revlogutils.entry( |
|
461 | 461 | flags=constants.REVIDX_ISCENSORED, |
|
462 | 462 | data_offset=new_data_offset, |
|
463 | 463 | data_compressed_length=new_data_size, |
|
464 | 464 | data_uncompressed_length=new_data_size, |
|
465 | 465 | data_delta_base=rev, |
|
466 | 466 | link_rev=link_rev, |
|
467 | 467 | parent_rev_1=p1, |
|
468 | 468 | parent_rev_2=p2, |
|
469 | 469 | node_id=entry[ENTRY_NODE_ID], |
|
470 | 470 | sidedata_offset=0, |
|
471 | 471 | sidedata_compressed_length=0, |
|
472 | 472 | data_compression_mode=COMP_MODE_PLAIN, |
|
473 | 473 | sidedata_compression_mode=COMP_MODE_PLAIN, |
|
474 | 474 | ) |
|
475 | 475 | revlog.index.append(new_entry) |
|
476 | 476 | entry_bin = revlog.index.entry_binary(rev) |
|
477 | 477 | new_index_file.write(entry_bin) |
|
478 | 478 | revlog._docket.index_end = new_index_file.tell() |
|
479 | 479 | revlog._docket.data_end = new_data_file.tell() |
|
480 | 480 | |
|
481 | 481 | |
|
482 | 482 | def _get_filename_from_filelog_index(path): |
|
483 | 483 | # Drop the extension and the `data/` prefix |
|
484 | 484 | path_part = path.rsplit(b'.', 1)[0].split(b'/', 1) |
|
485 | 485 | if len(path_part) < 2: |
|
486 | 486 | msg = _(b"cannot recognize filelog from filename: '%s'") |
|
487 | 487 | msg %= path |
|
488 | 488 | raise error.Abort(msg) |
|
489 | 489 | |
|
490 | 490 | return path_part[1] |
|
491 | 491 | |
|
492 | 492 | |
|
493 | 493 | def _filelog_from_filename(repo, path): |
|
494 | 494 | """Returns the filelog for the given `path`. Stolen from `engine.py`""" |
|
495 | 495 | |
|
496 | 496 | from .. import filelog # avoid cycle |
|
497 | 497 | |
|
498 | 498 | fl = filelog.filelog(repo.svfs, path) |
|
499 | 499 | return fl |
|
500 | 500 | |
|
501 | 501 | |
|
502 | 502 | def _write_swapped_parents(repo, rl, rev, offset, fp): |
|
503 | 503 | """Swaps p1 and p2 and overwrites the revlog entry for `rev` in `fp`""" |
|
504 | 504 | from ..pure import parsers # avoid cycle |
|
505 | 505 | |
|
506 | 506 | if repo._currentlock(repo._lockref) is None: |
|
507 | 507 | # Let's be paranoid about it |
|
508 | 508 | msg = "repo needs to be locked to rewrite parents" |
|
509 | 509 | raise error.ProgrammingError(msg) |
|
510 | 510 | |
|
511 | 511 | index_format = parsers.IndexObject.index_format |
|
512 | 512 | entry = rl.index[rev] |
|
513 | 513 | new_entry = list(entry) |
|
514 | 514 | new_entry[5], new_entry[6] = entry[6], entry[5] |
|
515 | 515 | packed = index_format.pack(*new_entry[:8]) |
|
516 | 516 | fp.seek(offset) |
|
517 | 517 | fp.write(packed) |
|
518 | 518 | |
|
519 | 519 | |
|
520 | 520 | def _reorder_filelog_parents(repo, fl, to_fix): |
|
521 | 521 | """ |
|
522 | 522 | Swaps p1 and p2 for all `to_fix` revisions of filelog `fl` and writes the |
|
523 | 523 | new version to disk, overwriting the old one with a rename. |
|
524 | 524 | """ |
|
525 | 525 | from ..pure import parsers # avoid cycle |
|
526 | 526 | |
|
527 | 527 | ui = repo.ui |
|
528 | 528 | assert len(to_fix) > 0 |
|
529 | 529 | rl = fl._revlog |
|
530 | 530 | if rl._format_version != constants.REVLOGV1: |
|
531 | 531 | msg = "expected version 1 revlog, got version '%d'" % rl._format_version |
|
532 | 532 | raise error.ProgrammingError(msg) |
|
533 | 533 | |
|
534 | 534 | index_file = rl._indexfile |
|
535 | 535 | new_file_path = index_file + b'.tmp-parents-fix' |
|
536 | 536 | repaired_msg = _(b"repaired revision %d of 'filelog %s'\n") |
|
537 | 537 | |
|
538 | 538 | with ui.uninterruptible(): |
|
539 | 539 | try: |
|
540 | 540 | util.copyfile( |
|
541 | 541 | rl.opener.join(index_file), |
|
542 | 542 | rl.opener.join(new_file_path), |
|
543 | 543 | checkambig=rl.data_config.check_ambig, |
|
544 | 544 | ) |
|
545 | 545 | |
|
546 | 546 | with rl.opener(new_file_path, mode=b"r+") as fp: |
|
547 | 547 | if rl._inline: |
|
548 | 548 | index = parsers.InlinedIndexObject(fp.read()) |
|
549 | 549 | for rev in fl.revs(): |
|
550 | 550 | if rev in to_fix: |
|
551 | 551 | offset = index._calculate_index(rev) |
|
552 | 552 | _write_swapped_parents(repo, rl, rev, offset, fp) |
|
553 | 553 | ui.write(repaired_msg % (rev, index_file)) |
|
554 | 554 | else: |
|
555 | 555 | index_format = parsers.IndexObject.index_format |
|
556 | 556 | for rev in to_fix: |
|
557 | 557 | offset = rev * index_format.size |
|
558 | 558 | _write_swapped_parents(repo, rl, rev, offset, fp) |
|
559 | 559 | ui.write(repaired_msg % (rev, index_file)) |
|
560 | 560 | |
|
561 | 561 | rl.opener.rename(new_file_path, index_file) |
|
562 | 562 | rl.clearcaches() |
|
563 | 563 | rl._loadindex() |
|
564 | 564 | finally: |
|
565 | 565 | util.tryunlink(new_file_path) |
|
566 | 566 | |
|
567 | 567 | |
|
568 | 568 | def _is_revision_affected(fl, filerev, metadata_cache=None): |
|
569 | 569 | full_text = lambda: fl._revlog.rawdata(filerev) |
|
570 | 570 | parent_revs = lambda: fl._revlog.parentrevs(filerev) |
|
571 | 571 | return _is_revision_affected_inner( |
|
572 | 572 | full_text, parent_revs, filerev, metadata_cache |
|
573 | 573 | ) |
|
574 | 574 | |
|
575 | 575 | |
|
576 | 576 | def _is_revision_affected_inner( |
|
577 | 577 | full_text, |
|
578 | 578 | parents_revs, |
|
579 | 579 | filerev, |
|
580 | 580 | metadata_cache=None, |
|
581 | 581 | ): |
|
582 | 582 | """Mercurial currently (5.9rc0) uses `p1 == nullrev and p2 != nullrev` as a |
|
583 | 583 | special meaning compared to the reverse in the context of filelog-based |
|
584 | 584 | copytracing. issue6528 exists because new code assumed that parent ordering |
|
585 | 585 | didn't matter, so this detects if the revision contains metadata (since |
|
586 | 586 | it's only used for filelog-based copytracing) and its parents are in the |
|
587 | 587 | "wrong" order.""" |
|
588 | 588 | try: |
|
589 | 589 | raw_text = full_text() |
|
590 | 590 | except error.CensoredNodeError: |
|
591 | 591 | # We don't care about censored nodes as they never carry metadata |
|
592 | 592 | return False |
|
593 | 593 | |
|
594 | 594 | # raw text can be a `memoryview`, which doesn't implement `startswith` |
|
595 | 595 | has_meta = bytes(raw_text[:2]) == b'\x01\n' |
|
596 | 596 | if metadata_cache is not None: |
|
597 | 597 | metadata_cache[filerev] = has_meta |
|
598 | 598 | if has_meta: |
|
599 | 599 | (p1, p2) = parents_revs() |
|
600 | 600 | if p1 != nullrev and p2 == nullrev: |
|
601 | 601 | return True |
|
602 | 602 | return False |
|
603 | 603 | |
|
604 | 604 | |
|
605 | 605 | def _is_revision_affected_fast(repo, fl, filerev, metadata_cache): |
|
606 | 606 | rl = fl._revlog |
|
607 | 607 | is_censored = lambda: rl.iscensored(filerev) |
|
608 | 608 | delta_base = lambda: rl.deltaparent(filerev) |
|
609 | 609 | delta = lambda: rl._chunk(filerev) |
|
610 | 610 | full_text = lambda: rl.rawdata(filerev) |
|
611 | 611 | parent_revs = lambda: rl.parentrevs(filerev) |
|
612 | 612 | return _is_revision_affected_fast_inner( |
|
613 | 613 | is_censored, |
|
614 | 614 | delta_base, |
|
615 | 615 | delta, |
|
616 | 616 | full_text, |
|
617 | 617 | parent_revs, |
|
618 | 618 | filerev, |
|
619 | 619 | metadata_cache, |
|
620 | 620 | ) |
|
621 | 621 | |
|
622 | 622 | |
|
623 | 623 | def _is_revision_affected_fast_inner( |
|
624 | 624 | is_censored, |
|
625 | 625 | delta_base, |
|
626 | 626 | delta, |
|
627 | 627 | full_text, |
|
628 | 628 | parent_revs, |
|
629 | 629 | filerev, |
|
630 | 630 | metadata_cache, |
|
631 | 631 | ): |
|
632 | 632 | """Optimization fast-path for `_is_revision_affected`. |
|
633 | 633 | |
|
634 | 634 | `metadata_cache` is a dict of `{rev: has_metadata}` which allows any |
|
635 | 635 | revision to check if its base has metadata, saving computation of the full |
|
636 | 636 | text, instead looking at the current delta. |
|
637 | 637 | |
|
638 | 638 | This optimization only works if the revisions are looked at in order.""" |
|
639 | 639 | |
|
640 | 640 | if is_censored(): |
|
641 | 641 | # Censored revisions don't contain metadata, so they cannot be affected |
|
642 | 642 | metadata_cache[filerev] = False |
|
643 | 643 | return False |
|
644 | 644 | |
|
645 | 645 | p1, p2 = parent_revs() |
|
646 | 646 | if p1 == nullrev or p2 != nullrev: |
|
647 | 647 | return False |
|
648 | 648 | |
|
649 | 649 | delta_parent = delta_base() |
|
650 | 650 | parent_has_metadata = metadata_cache.get(delta_parent) |
|
651 | 651 | if parent_has_metadata is None: |
|
652 | 652 | return _is_revision_affected_inner( |
|
653 | 653 | full_text, |
|
654 | 654 | parent_revs, |
|
655 | 655 | filerev, |
|
656 | 656 | metadata_cache, |
|
657 | 657 | ) |
|
658 | 658 | |
|
659 | 659 | chunk = delta() |
|
660 | 660 | if not len(chunk): |
|
661 | 661 | # No diff for this revision |
|
662 | 662 | return parent_has_metadata |
|
663 | 663 | |
|
664 | 664 | header_length = 12 |
|
665 | 665 | if len(chunk) < header_length: |
|
666 | 666 | raise error.Abort(_(b"patch cannot be decoded")) |
|
667 | 667 | |
|
668 | 668 | start, _end, _length = struct.unpack(b">lll", chunk[:header_length]) |
|
669 | 669 | |
|
670 | 670 | if start < 2: # len(b'\x01\n') == 2 |
|
671 | 671 | # This delta does *something* to the metadata marker (if any). |
|
672 | 672 | # Check it the slow way |
|
673 | 673 | is_affected = _is_revision_affected_inner( |
|
674 | 674 | full_text, |
|
675 | 675 | parent_revs, |
|
676 | 676 | filerev, |
|
677 | 677 | metadata_cache, |
|
678 | 678 | ) |
|
679 | 679 | return is_affected |
|
680 | 680 | |
|
681 | 681 | # The diff did not remove or add the metadata header, it's then in the same |
|
682 | 682 | # situation as its parent |
|
683 | 683 | metadata_cache[filerev] = parent_has_metadata |
|
684 | 684 | return parent_has_metadata |
|
685 | 685 | |
|
686 | 686 | |
|
687 | 687 | def _from_report(ui, repo, context, from_report, dry_run): |
|
688 | 688 | """ |
|
689 | 689 | Fix the revisions given in the `from_report` file, but still checks if the |
|
690 | 690 | revisions are indeed affected to prevent an unfortunate cyclic situation |
|
691 | 691 | where we'd swap well-ordered parents again. |
|
692 | 692 | |
|
693 | 693 | See the doc for `debug_fix_issue6528` for the format documentation. |
|
694 | 694 | """ |
|
695 | 695 | ui.write(_(b"loading report file '%s'\n") % from_report) |
|
696 | 696 | |
|
697 | 697 | with context(), open(from_report, mode='rb') as f: |
|
698 | 698 | for line in f.read().split(b'\n'): |
|
699 | 699 | if not line: |
|
700 | 700 | continue |
|
701 | 701 | filenodes, filename = line.split(b' ', 1) |
|
702 | 702 | fl = _filelog_from_filename(repo, filename) |
|
703 | 703 | to_fix = set( |
|
704 | 704 | fl.rev(binascii.unhexlify(n)) for n in filenodes.split(b',') |
|
705 | 705 | ) |
|
706 | 706 | excluded = set() |
|
707 | 707 | |
|
708 | 708 | for filerev in to_fix: |
|
709 | 709 | if _is_revision_affected(fl, filerev): |
|
710 | 710 | msg = b"found affected revision %d for filelog '%s'\n" |
|
711 | 711 | ui.warn(msg % (filerev, filename)) |
|
712 | 712 | else: |
|
713 | 713 | msg = _(b"revision %s of file '%s' is not affected\n") |
|
714 | 714 | msg %= (binascii.hexlify(fl.node(filerev)), filename) |
|
715 | 715 | ui.warn(msg) |
|
716 | 716 | excluded.add(filerev) |
|
717 | 717 | |
|
718 | 718 | to_fix = to_fix - excluded |
|
719 | 719 | if not to_fix: |
|
720 | 720 | msg = _(b"no affected revisions were found for '%s'\n") |
|
721 | 721 | ui.write(msg % filename) |
|
722 | 722 | continue |
|
723 | 723 | if not dry_run: |
|
724 | 724 | _reorder_filelog_parents(repo, fl, sorted(to_fix)) |
|
725 | 725 | |
|
726 | 726 | |
|
727 | 727 | def filter_delta_issue6528(revlog, deltas_iter): |
|
728 | 728 | """filter incomind deltas to repaire issue 6528 on the fly""" |
|
729 | 729 | metadata_cache = {} |
|
730 | 730 | |
|
731 | 731 | deltacomputer = deltas.deltacomputer(revlog) |
|
732 | 732 | |
|
733 | 733 | for rev, d in enumerate(deltas_iter, len(revlog)): |
|
734 | 734 | ( |
|
735 | 735 | node, |
|
736 | 736 | p1_node, |
|
737 | 737 | p2_node, |
|
738 | 738 | linknode, |
|
739 | 739 | deltabase, |
|
740 | 740 | delta, |
|
741 | 741 | flags, |
|
742 | 742 | sidedata, |
|
743 | 743 | ) = d |
|
744 | 744 | |
|
745 | 745 | if not revlog.index.has_node(deltabase): |
|
746 | 746 | raise error.LookupError( |
|
747 | 747 | deltabase, revlog.radix, _(b'unknown parent') |
|
748 | 748 | ) |
|
749 | 749 | base_rev = revlog.rev(deltabase) |
|
750 | 750 | if not revlog.index.has_node(p1_node): |
|
751 | 751 | raise error.LookupError(p1_node, revlog.radix, _(b'unknown parent')) |
|
752 | 752 | p1_rev = revlog.rev(p1_node) |
|
753 | 753 | if not revlog.index.has_node(p2_node): |
|
754 | 754 | raise error.LookupError(p2_node, revlog.radix, _(b'unknown parent')) |
|
755 | 755 | p2_rev = revlog.rev(p2_node) |
|
756 | 756 | |
|
757 | 757 | is_censored = lambda: bool(flags & REVIDX_ISCENSORED) |
|
758 | 758 | delta_base = lambda: revlog.rev(delta_base) |
|
759 | 759 | delta_base = lambda: base_rev |
|
760 | 760 | parent_revs = lambda: (p1_rev, p2_rev) |
|
761 | 761 | |
|
762 | 762 | def full_text(): |
|
763 | 763 | # note: being able to reuse the full text computation in the |
|
764 | 764 | # underlying addrevision would be useful however this is a bit too |
|
765 | 765 | # intrusive the for the "quick" issue6528 we are writing before the |
|
766 | 766 | # 5.8 release |
|
767 | 767 | textlen = mdiff.patchedsize(revlog.size(base_rev), delta) |
|
768 | 768 | |
|
769 | 769 | revinfo = revlogutils.revisioninfo( |
|
770 | 770 | node, |
|
771 | 771 | p1_node, |
|
772 | 772 | p2_node, |
|
773 | 773 | [None], |
|
774 | 774 | textlen, |
|
775 | 775 | (base_rev, delta), |
|
776 | 776 | flags, |
|
777 | 777 | ) |
|
778 | 778 | return deltacomputer.buildtext(revinfo) |
|
779 | 779 | |
|
780 | 780 | is_affected = _is_revision_affected_fast_inner( |
|
781 | 781 | is_censored, |
|
782 | 782 | delta_base, |
|
783 | 783 | lambda: delta, |
|
784 | 784 | full_text, |
|
785 | 785 | parent_revs, |
|
786 | 786 | rev, |
|
787 | 787 | metadata_cache, |
|
788 | 788 | ) |
|
789 | 789 | if is_affected: |
|
790 | 790 | d = ( |
|
791 | 791 | node, |
|
792 | 792 | p2_node, |
|
793 | 793 | p1_node, |
|
794 | 794 | linknode, |
|
795 | 795 | deltabase, |
|
796 | 796 | delta, |
|
797 | 797 | flags, |
|
798 | 798 | sidedata, |
|
799 | 799 | ) |
|
800 | 800 | yield d |
|
801 | 801 | |
|
802 | 802 | |
|
803 | 803 | def repair_issue6528( |
|
804 | 804 | ui, repo, dry_run=False, to_report=None, from_report=None, paranoid=False |
|
805 | 805 | ): |
|
806 | 806 | @contextlib.contextmanager |
|
807 | 807 | def context(): |
|
808 | 808 | if dry_run or to_report: # No need for locking |
|
809 | 809 | yield |
|
810 | 810 | else: |
|
811 | 811 | with repo.wlock(), repo.lock(): |
|
812 | 812 | yield |
|
813 | 813 | |
|
814 | 814 | if from_report: |
|
815 | 815 | return _from_report(ui, repo, context, from_report, dry_run) |
|
816 | 816 | |
|
817 | 817 | report_entries = [] |
|
818 | 818 | |
|
819 | 819 | with context(): |
|
820 | 820 | files = list( |
|
821 | 821 | entry |
|
822 | 822 | for entry in repo.store.data_entries() |
|
823 | 823 | if entry.is_revlog and entry.is_filelog |
|
824 | 824 | ) |
|
825 | 825 | |
|
826 | 826 | progress = ui.makeprogress( |
|
827 | 827 | _(b"looking for affected revisions"), |
|
828 | 828 | unit=_(b"filelogs"), |
|
829 | 829 | total=len(files), |
|
830 | 830 | ) |
|
831 | 831 | found_nothing = True |
|
832 | 832 | |
|
833 | 833 | for entry in files: |
|
834 | 834 | progress.increment() |
|
835 | 835 | filename = entry.target_id |
|
836 | 836 | fl = _filelog_from_filename(repo, entry.target_id) |
|
837 | 837 | |
|
838 | 838 | # Set of filerevs (or hex filenodes if `to_report`) that need fixing |
|
839 | 839 | to_fix = set() |
|
840 | 840 | metadata_cache = {} |
|
841 | 841 | for filerev in fl.revs(): |
|
842 | 842 | affected = _is_revision_affected_fast( |
|
843 | 843 | repo, fl, filerev, metadata_cache |
|
844 | 844 | ) |
|
845 | 845 | if paranoid: |
|
846 | 846 | slow = _is_revision_affected(fl, filerev) |
|
847 | 847 | if slow != affected: |
|
848 | 848 | msg = _(b"paranoid check failed for '%s' at node %s") |
|
849 | 849 | node = binascii.hexlify(fl.node(filerev)) |
|
850 | 850 | raise error.Abort(msg % (filename, node)) |
|
851 | 851 | if affected: |
|
852 | 852 | msg = b"found affected revision %d for file '%s'\n" |
|
853 | 853 | ui.warn(msg % (filerev, filename)) |
|
854 | 854 | found_nothing = False |
|
855 | 855 | if not dry_run: |
|
856 | 856 | if to_report: |
|
857 | 857 | to_fix.add(binascii.hexlify(fl.node(filerev))) |
|
858 | 858 | else: |
|
859 | 859 | to_fix.add(filerev) |
|
860 | 860 | |
|
861 | 861 | if to_fix: |
|
862 | 862 | to_fix = sorted(to_fix) |
|
863 | 863 | if to_report: |
|
864 | 864 | report_entries.append((filename, to_fix)) |
|
865 | 865 | else: |
|
866 | 866 | _reorder_filelog_parents(repo, fl, to_fix) |
|
867 | 867 | |
|
868 | 868 | if found_nothing: |
|
869 | 869 | ui.write(_(b"no affected revisions were found\n")) |
|
870 | 870 | |
|
871 | 871 | if to_report and report_entries: |
|
872 | 872 | with open(to_report, mode="wb") as f: |
|
873 | 873 | for path, to_fix in report_entries: |
|
874 | 874 | f.write(b"%s %s\n" % (b",".join(to_fix), path)) |
|
875 | 875 | |
|
876 | 876 | progress.complete() |
General Comments 0
You need to be logged in to leave comments.
Login now