Show More
@@ -1,2643 +1,2559 b'' | |||||
1 | # revlog.py - storage back-end for mercurial |
|
1 | # revlog.py - storage back-end for mercurial | |
2 | # |
|
2 | # | |
3 | # Copyright 2005-2007 Matt Mackall <mpm@selenic.com> |
|
3 | # Copyright 2005-2007 Matt Mackall <mpm@selenic.com> | |
4 | # |
|
4 | # | |
5 | # This software may be used and distributed according to the terms of the |
|
5 | # This software may be used and distributed according to the terms of the | |
6 | # GNU General Public License version 2 or any later version. |
|
6 | # GNU General Public License version 2 or any later version. | |
7 |
|
7 | |||
8 | """Storage back-end for Mercurial. |
|
8 | """Storage back-end for Mercurial. | |
9 |
|
9 | |||
10 | This provides efficient delta storage with O(1) retrieve and append |
|
10 | This provides efficient delta storage with O(1) retrieve and append | |
11 | and O(changes) merge between branches. |
|
11 | and O(changes) merge between branches. | |
12 | """ |
|
12 | """ | |
13 |
|
13 | |||
14 | from __future__ import absolute_import |
|
14 | from __future__ import absolute_import | |
15 |
|
15 | |||
16 | import collections |
|
16 | import collections | |
17 | import contextlib |
|
17 | import contextlib | |
18 | import errno |
|
18 | import errno | |
19 | import os |
|
19 | import os | |
20 | import struct |
|
20 | import struct | |
21 | import zlib |
|
21 | import zlib | |
22 |
|
22 | |||
23 | # import stuff from node for others to import from revlog |
|
23 | # import stuff from node for others to import from revlog | |
24 | from .node import ( |
|
24 | from .node import ( | |
25 | bin, |
|
25 | bin, | |
26 | hex, |
|
26 | hex, | |
27 | nullhex, |
|
27 | nullhex, | |
28 | nullid, |
|
28 | nullid, | |
29 | nullrev, |
|
29 | nullrev, | |
30 | short, |
|
30 | short, | |
31 | wdirfilenodeids, |
|
31 | wdirfilenodeids, | |
32 | wdirhex, |
|
32 | wdirhex, | |
33 | wdirid, |
|
33 | wdirid, | |
34 | wdirrev, |
|
34 | wdirrev, | |
35 | ) |
|
35 | ) | |
36 | from .i18n import _ |
|
36 | from .i18n import _ | |
37 | from .revlogutils.constants import ( |
|
37 | from .revlogutils.constants import ( | |
38 | FLAG_GENERALDELTA, |
|
38 | FLAG_GENERALDELTA, | |
39 | FLAG_INLINE_DATA, |
|
39 | FLAG_INLINE_DATA, | |
40 | REVIDX_DEFAULT_FLAGS, |
|
40 | REVIDX_DEFAULT_FLAGS, | |
41 | REVIDX_ELLIPSIS, |
|
41 | REVIDX_ELLIPSIS, | |
42 | REVIDX_EXTSTORED, |
|
42 | REVIDX_EXTSTORED, | |
43 | REVIDX_FLAGS_ORDER, |
|
43 | REVIDX_FLAGS_ORDER, | |
44 | REVIDX_ISCENSORED, |
|
44 | REVIDX_ISCENSORED, | |
45 | REVIDX_KNOWN_FLAGS, |
|
45 | REVIDX_KNOWN_FLAGS, | |
46 | REVIDX_RAWTEXT_CHANGING_FLAGS, |
|
46 | REVIDX_RAWTEXT_CHANGING_FLAGS, | |
47 | REVLOGV0, |
|
47 | REVLOGV0, | |
48 | REVLOGV1, |
|
48 | REVLOGV1, | |
49 | REVLOGV1_FLAGS, |
|
49 | REVLOGV1_FLAGS, | |
50 | REVLOGV2, |
|
50 | REVLOGV2, | |
51 | REVLOGV2_FLAGS, |
|
51 | REVLOGV2_FLAGS, | |
52 | REVLOG_DEFAULT_FLAGS, |
|
52 | REVLOG_DEFAULT_FLAGS, | |
53 | REVLOG_DEFAULT_FORMAT, |
|
53 | REVLOG_DEFAULT_FORMAT, | |
54 | REVLOG_DEFAULT_VERSION, |
|
54 | REVLOG_DEFAULT_VERSION, | |
55 | ) |
|
55 | ) | |
56 | from .thirdparty import ( |
|
56 | from .thirdparty import ( | |
57 | attr, |
|
57 | attr, | |
58 | ) |
|
58 | ) | |
59 | from . import ( |
|
59 | from . import ( | |
60 | ancestor, |
|
60 | ancestor, | |
61 | dagop, |
|
61 | dagop, | |
62 | error, |
|
62 | error, | |
63 | mdiff, |
|
63 | mdiff, | |
64 | policy, |
|
64 | policy, | |
65 | pycompat, |
|
65 | pycompat, | |
66 | repository, |
|
66 | repository, | |
67 | templatefilters, |
|
67 | templatefilters, | |
68 | util, |
|
68 | util, | |
69 | ) |
|
69 | ) | |
70 | from .revlogutils import ( |
|
70 | from .revlogutils import ( | |
71 | deltas as deltautil, |
|
71 | deltas as deltautil, | |
72 | ) |
|
72 | ) | |
73 | from .utils import ( |
|
73 | from .utils import ( | |
74 | interfaceutil, |
|
74 | interfaceutil, | |
75 | storageutil, |
|
75 | storageutil, | |
76 | stringutil, |
|
76 | stringutil, | |
77 | ) |
|
77 | ) | |
78 |
|
78 | |||
79 | # blanked usage of all the name to prevent pyflakes constraints |
|
79 | # blanked usage of all the name to prevent pyflakes constraints | |
80 | # We need these name available in the module for extensions. |
|
80 | # We need these name available in the module for extensions. | |
81 | REVLOGV0 |
|
81 | REVLOGV0 | |
82 | REVLOGV1 |
|
82 | REVLOGV1 | |
83 | REVLOGV2 |
|
83 | REVLOGV2 | |
84 | FLAG_INLINE_DATA |
|
84 | FLAG_INLINE_DATA | |
85 | FLAG_GENERALDELTA |
|
85 | FLAG_GENERALDELTA | |
86 | REVLOG_DEFAULT_FLAGS |
|
86 | REVLOG_DEFAULT_FLAGS | |
87 | REVLOG_DEFAULT_FORMAT |
|
87 | REVLOG_DEFAULT_FORMAT | |
88 | REVLOG_DEFAULT_VERSION |
|
88 | REVLOG_DEFAULT_VERSION | |
89 | REVLOGV1_FLAGS |
|
89 | REVLOGV1_FLAGS | |
90 | REVLOGV2_FLAGS |
|
90 | REVLOGV2_FLAGS | |
91 | REVIDX_ISCENSORED |
|
91 | REVIDX_ISCENSORED | |
92 | REVIDX_ELLIPSIS |
|
92 | REVIDX_ELLIPSIS | |
93 | REVIDX_EXTSTORED |
|
93 | REVIDX_EXTSTORED | |
94 | REVIDX_DEFAULT_FLAGS |
|
94 | REVIDX_DEFAULT_FLAGS | |
95 | REVIDX_FLAGS_ORDER |
|
95 | REVIDX_FLAGS_ORDER | |
96 | REVIDX_KNOWN_FLAGS |
|
96 | REVIDX_KNOWN_FLAGS | |
97 | REVIDX_RAWTEXT_CHANGING_FLAGS |
|
97 | REVIDX_RAWTEXT_CHANGING_FLAGS | |
98 |
|
98 | |||
99 | parsers = policy.importmod(r'parsers') |
|
99 | parsers = policy.importmod(r'parsers') | |
100 |
|
100 | |||
101 | # Aliased for performance. |
|
101 | # Aliased for performance. | |
102 | _zlibdecompress = zlib.decompress |
|
102 | _zlibdecompress = zlib.decompress | |
103 |
|
103 | |||
104 | # max size of revlog with inline data |
|
104 | # max size of revlog with inline data | |
105 | _maxinline = 131072 |
|
105 | _maxinline = 131072 | |
106 | _chunksize = 1048576 |
|
106 | _chunksize = 1048576 | |
107 |
|
107 | |||
108 | # Store flag processors (cf. 'addflagprocessor()' to register) |
|
108 | # Store flag processors (cf. 'addflagprocessor()' to register) | |
109 | _flagprocessors = { |
|
109 | _flagprocessors = { | |
110 | REVIDX_ISCENSORED: None, |
|
110 | REVIDX_ISCENSORED: None, | |
111 | } |
|
111 | } | |
112 |
|
112 | |||
113 | # Flag processors for REVIDX_ELLIPSIS. |
|
113 | # Flag processors for REVIDX_ELLIPSIS. | |
114 | def ellipsisreadprocessor(rl, text): |
|
114 | def ellipsisreadprocessor(rl, text): | |
115 | return text, False |
|
115 | return text, False | |
116 |
|
116 | |||
117 | def ellipsiswriteprocessor(rl, text): |
|
117 | def ellipsiswriteprocessor(rl, text): | |
118 | return text, False |
|
118 | return text, False | |
119 |
|
119 | |||
120 | def ellipsisrawprocessor(rl, text): |
|
120 | def ellipsisrawprocessor(rl, text): | |
121 | return False |
|
121 | return False | |
122 |
|
122 | |||
123 | ellipsisprocessor = ( |
|
123 | ellipsisprocessor = ( | |
124 | ellipsisreadprocessor, |
|
124 | ellipsisreadprocessor, | |
125 | ellipsiswriteprocessor, |
|
125 | ellipsiswriteprocessor, | |
126 | ellipsisrawprocessor, |
|
126 | ellipsisrawprocessor, | |
127 | ) |
|
127 | ) | |
128 |
|
128 | |||
129 | def addflagprocessor(flag, processor): |
|
129 | def addflagprocessor(flag, processor): | |
130 | """Register a flag processor on a revision data flag. |
|
130 | """Register a flag processor on a revision data flag. | |
131 |
|
131 | |||
132 | Invariant: |
|
132 | Invariant: | |
133 | - Flags need to be defined in REVIDX_KNOWN_FLAGS and REVIDX_FLAGS_ORDER, |
|
133 | - Flags need to be defined in REVIDX_KNOWN_FLAGS and REVIDX_FLAGS_ORDER, | |
134 | and REVIDX_RAWTEXT_CHANGING_FLAGS if they can alter rawtext. |
|
134 | and REVIDX_RAWTEXT_CHANGING_FLAGS if they can alter rawtext. | |
135 | - Only one flag processor can be registered on a specific flag. |
|
135 | - Only one flag processor can be registered on a specific flag. | |
136 | - flagprocessors must be 3-tuples of functions (read, write, raw) with the |
|
136 | - flagprocessors must be 3-tuples of functions (read, write, raw) with the | |
137 | following signatures: |
|
137 | following signatures: | |
138 | - (read) f(self, rawtext) -> text, bool |
|
138 | - (read) f(self, rawtext) -> text, bool | |
139 | - (write) f(self, text) -> rawtext, bool |
|
139 | - (write) f(self, text) -> rawtext, bool | |
140 | - (raw) f(self, rawtext) -> bool |
|
140 | - (raw) f(self, rawtext) -> bool | |
141 | "text" is presented to the user. "rawtext" is stored in revlog data, not |
|
141 | "text" is presented to the user. "rawtext" is stored in revlog data, not | |
142 | directly visible to the user. |
|
142 | directly visible to the user. | |
143 | The boolean returned by these transforms is used to determine whether |
|
143 | The boolean returned by these transforms is used to determine whether | |
144 | the returned text can be used for hash integrity checking. For example, |
|
144 | the returned text can be used for hash integrity checking. For example, | |
145 | if "write" returns False, then "text" is used to generate hash. If |
|
145 | if "write" returns False, then "text" is used to generate hash. If | |
146 | "write" returns True, that basically means "rawtext" returned by "write" |
|
146 | "write" returns True, that basically means "rawtext" returned by "write" | |
147 | should be used to generate hash. Usually, "write" and "read" return |
|
147 | should be used to generate hash. Usually, "write" and "read" return | |
148 | different booleans. And "raw" returns a same boolean as "write". |
|
148 | different booleans. And "raw" returns a same boolean as "write". | |
149 |
|
149 | |||
150 | Note: The 'raw' transform is used for changegroup generation and in some |
|
150 | Note: The 'raw' transform is used for changegroup generation and in some | |
151 | debug commands. In this case the transform only indicates whether the |
|
151 | debug commands. In this case the transform only indicates whether the | |
152 | contents can be used for hash integrity checks. |
|
152 | contents can be used for hash integrity checks. | |
153 | """ |
|
153 | """ | |
154 | if not flag & REVIDX_KNOWN_FLAGS: |
|
154 | if not flag & REVIDX_KNOWN_FLAGS: | |
155 | msg = _("cannot register processor on unknown flag '%#x'.") % (flag) |
|
155 | msg = _("cannot register processor on unknown flag '%#x'.") % (flag) | |
156 | raise error.ProgrammingError(msg) |
|
156 | raise error.ProgrammingError(msg) | |
157 | if flag not in REVIDX_FLAGS_ORDER: |
|
157 | if flag not in REVIDX_FLAGS_ORDER: | |
158 | msg = _("flag '%#x' undefined in REVIDX_FLAGS_ORDER.") % (flag) |
|
158 | msg = _("flag '%#x' undefined in REVIDX_FLAGS_ORDER.") % (flag) | |
159 | raise error.ProgrammingError(msg) |
|
159 | raise error.ProgrammingError(msg) | |
160 | if flag in _flagprocessors: |
|
160 | if flag in _flagprocessors: | |
161 | msg = _("cannot register multiple processors on flag '%#x'.") % (flag) |
|
161 | msg = _("cannot register multiple processors on flag '%#x'.") % (flag) | |
162 | raise error.Abort(msg) |
|
162 | raise error.Abort(msg) | |
163 | _flagprocessors[flag] = processor |
|
163 | _flagprocessors[flag] = processor | |
164 |
|
164 | |||
165 | def getoffset(q): |
|
165 | def getoffset(q): | |
166 | return int(q >> 16) |
|
166 | return int(q >> 16) | |
167 |
|
167 | |||
168 | def gettype(q): |
|
168 | def gettype(q): | |
169 | return int(q & 0xFFFF) |
|
169 | return int(q & 0xFFFF) | |
170 |
|
170 | |||
171 | def offset_type(offset, type): |
|
171 | def offset_type(offset, type): | |
172 | if (type & ~REVIDX_KNOWN_FLAGS) != 0: |
|
172 | if (type & ~REVIDX_KNOWN_FLAGS) != 0: | |
173 | raise ValueError('unknown revlog index flags') |
|
173 | raise ValueError('unknown revlog index flags') | |
174 | return int(int(offset) << 16 | type) |
|
174 | return int(int(offset) << 16 | type) | |
175 |
|
175 | |||
176 | @attr.s(slots=True, frozen=True) |
|
176 | @attr.s(slots=True, frozen=True) | |
177 | class _revisioninfo(object): |
|
177 | class _revisioninfo(object): | |
178 | """Information about a revision that allows building its fulltext |
|
178 | """Information about a revision that allows building its fulltext | |
179 | node: expected hash of the revision |
|
179 | node: expected hash of the revision | |
180 | p1, p2: parent revs of the revision |
|
180 | p1, p2: parent revs of the revision | |
181 | btext: built text cache consisting of a one-element list |
|
181 | btext: built text cache consisting of a one-element list | |
182 | cachedelta: (baserev, uncompressed_delta) or None |
|
182 | cachedelta: (baserev, uncompressed_delta) or None | |
183 | flags: flags associated to the revision storage |
|
183 | flags: flags associated to the revision storage | |
184 |
|
184 | |||
185 | One of btext[0] or cachedelta must be set. |
|
185 | One of btext[0] or cachedelta must be set. | |
186 | """ |
|
186 | """ | |
187 | node = attr.ib() |
|
187 | node = attr.ib() | |
188 | p1 = attr.ib() |
|
188 | p1 = attr.ib() | |
189 | p2 = attr.ib() |
|
189 | p2 = attr.ib() | |
190 | btext = attr.ib() |
|
190 | btext = attr.ib() | |
191 | textlen = attr.ib() |
|
191 | textlen = attr.ib() | |
192 | cachedelta = attr.ib() |
|
192 | cachedelta = attr.ib() | |
193 | flags = attr.ib() |
|
193 | flags = attr.ib() | |
194 |
|
194 | |||
195 | @interfaceutil.implementer(repository.irevisiondelta) |
|
195 | @interfaceutil.implementer(repository.irevisiondelta) | |
196 | @attr.s(slots=True) |
|
196 | @attr.s(slots=True) | |
197 | class revlogrevisiondelta(object): |
|
197 | class revlogrevisiondelta(object): | |
198 | node = attr.ib() |
|
198 | node = attr.ib() | |
199 | p1node = attr.ib() |
|
199 | p1node = attr.ib() | |
200 | p2node = attr.ib() |
|
200 | p2node = attr.ib() | |
201 | basenode = attr.ib() |
|
201 | basenode = attr.ib() | |
202 | flags = attr.ib() |
|
202 | flags = attr.ib() | |
203 | baserevisionsize = attr.ib() |
|
203 | baserevisionsize = attr.ib() | |
204 | revision = attr.ib() |
|
204 | revision = attr.ib() | |
205 | delta = attr.ib() |
|
205 | delta = attr.ib() | |
206 | linknode = attr.ib(default=None) |
|
206 | linknode = attr.ib(default=None) | |
207 |
|
207 | |||
208 | @interfaceutil.implementer(repository.iverifyproblem) |
|
208 | @interfaceutil.implementer(repository.iverifyproblem) | |
209 | @attr.s(frozen=True) |
|
209 | @attr.s(frozen=True) | |
210 | class revlogproblem(object): |
|
210 | class revlogproblem(object): | |
211 | warning = attr.ib(default=None) |
|
211 | warning = attr.ib(default=None) | |
212 | error = attr.ib(default=None) |
|
212 | error = attr.ib(default=None) | |
213 | node = attr.ib(default=None) |
|
213 | node = attr.ib(default=None) | |
214 |
|
214 | |||
215 | # index v0: |
|
215 | # index v0: | |
216 | # 4 bytes: offset |
|
216 | # 4 bytes: offset | |
217 | # 4 bytes: compressed length |
|
217 | # 4 bytes: compressed length | |
218 | # 4 bytes: base rev |
|
218 | # 4 bytes: base rev | |
219 | # 4 bytes: link rev |
|
219 | # 4 bytes: link rev | |
220 | # 20 bytes: parent 1 nodeid |
|
220 | # 20 bytes: parent 1 nodeid | |
221 | # 20 bytes: parent 2 nodeid |
|
221 | # 20 bytes: parent 2 nodeid | |
222 | # 20 bytes: nodeid |
|
222 | # 20 bytes: nodeid | |
223 | indexformatv0 = struct.Struct(">4l20s20s20s") |
|
223 | indexformatv0 = struct.Struct(">4l20s20s20s") | |
224 | indexformatv0_pack = indexformatv0.pack |
|
224 | indexformatv0_pack = indexformatv0.pack | |
225 | indexformatv0_unpack = indexformatv0.unpack |
|
225 | indexformatv0_unpack = indexformatv0.unpack | |
226 |
|
226 | |||
227 | class revlogoldindex(list): |
|
227 | class revlogoldindex(list): | |
228 | def __getitem__(self, i): |
|
228 | def __getitem__(self, i): | |
229 | if i == -1: |
|
229 | if i == -1: | |
230 | return (0, 0, 0, -1, -1, -1, -1, nullid) |
|
230 | return (0, 0, 0, -1, -1, -1, -1, nullid) | |
231 | return list.__getitem__(self, i) |
|
231 | return list.__getitem__(self, i) | |
232 |
|
232 | |||
233 | class revlogoldio(object): |
|
233 | class revlogoldio(object): | |
234 | def __init__(self): |
|
234 | def __init__(self): | |
235 | self.size = indexformatv0.size |
|
235 | self.size = indexformatv0.size | |
236 |
|
236 | |||
237 | def parseindex(self, data, inline): |
|
237 | def parseindex(self, data, inline): | |
238 | s = self.size |
|
238 | s = self.size | |
239 | index = [] |
|
239 | index = [] | |
240 | nodemap = {nullid: nullrev} |
|
240 | nodemap = {nullid: nullrev} | |
241 | n = off = 0 |
|
241 | n = off = 0 | |
242 | l = len(data) |
|
242 | l = len(data) | |
243 | while off + s <= l: |
|
243 | while off + s <= l: | |
244 | cur = data[off:off + s] |
|
244 | cur = data[off:off + s] | |
245 | off += s |
|
245 | off += s | |
246 | e = indexformatv0_unpack(cur) |
|
246 | e = indexformatv0_unpack(cur) | |
247 | # transform to revlogv1 format |
|
247 | # transform to revlogv1 format | |
248 | e2 = (offset_type(e[0], 0), e[1], -1, e[2], e[3], |
|
248 | e2 = (offset_type(e[0], 0), e[1], -1, e[2], e[3], | |
249 | nodemap.get(e[4], nullrev), nodemap.get(e[5], nullrev), e[6]) |
|
249 | nodemap.get(e[4], nullrev), nodemap.get(e[5], nullrev), e[6]) | |
250 | index.append(e2) |
|
250 | index.append(e2) | |
251 | nodemap[e[6]] = n |
|
251 | nodemap[e[6]] = n | |
252 | n += 1 |
|
252 | n += 1 | |
253 |
|
253 | |||
254 | return revlogoldindex(index), nodemap, None |
|
254 | return revlogoldindex(index), nodemap, None | |
255 |
|
255 | |||
256 | def packentry(self, entry, node, version, rev): |
|
256 | def packentry(self, entry, node, version, rev): | |
257 | if gettype(entry[0]): |
|
257 | if gettype(entry[0]): | |
258 | raise error.RevlogError(_('index entry flags need revlog ' |
|
258 | raise error.RevlogError(_('index entry flags need revlog ' | |
259 | 'version 1')) |
|
259 | 'version 1')) | |
260 | e2 = (getoffset(entry[0]), entry[1], entry[3], entry[4], |
|
260 | e2 = (getoffset(entry[0]), entry[1], entry[3], entry[4], | |
261 | node(entry[5]), node(entry[6]), entry[7]) |
|
261 | node(entry[5]), node(entry[6]), entry[7]) | |
262 | return indexformatv0_pack(*e2) |
|
262 | return indexformatv0_pack(*e2) | |
263 |
|
263 | |||
264 | # index ng: |
|
264 | # index ng: | |
265 | # 6 bytes: offset |
|
265 | # 6 bytes: offset | |
266 | # 2 bytes: flags |
|
266 | # 2 bytes: flags | |
267 | # 4 bytes: compressed length |
|
267 | # 4 bytes: compressed length | |
268 | # 4 bytes: uncompressed length |
|
268 | # 4 bytes: uncompressed length | |
269 | # 4 bytes: base rev |
|
269 | # 4 bytes: base rev | |
270 | # 4 bytes: link rev |
|
270 | # 4 bytes: link rev | |
271 | # 4 bytes: parent 1 rev |
|
271 | # 4 bytes: parent 1 rev | |
272 | # 4 bytes: parent 2 rev |
|
272 | # 4 bytes: parent 2 rev | |
273 | # 32 bytes: nodeid |
|
273 | # 32 bytes: nodeid | |
274 | indexformatng = struct.Struct(">Qiiiiii20s12x") |
|
274 | indexformatng = struct.Struct(">Qiiiiii20s12x") | |
275 | indexformatng_pack = indexformatng.pack |
|
275 | indexformatng_pack = indexformatng.pack | |
276 | versionformat = struct.Struct(">I") |
|
276 | versionformat = struct.Struct(">I") | |
277 | versionformat_pack = versionformat.pack |
|
277 | versionformat_pack = versionformat.pack | |
278 | versionformat_unpack = versionformat.unpack |
|
278 | versionformat_unpack = versionformat.unpack | |
279 |
|
279 | |||
280 | # corresponds to uncompressed length of indexformatng (2 gigs, 4-byte |
|
280 | # corresponds to uncompressed length of indexformatng (2 gigs, 4-byte | |
281 | # signed integer) |
|
281 | # signed integer) | |
282 | _maxentrysize = 0x7fffffff |
|
282 | _maxentrysize = 0x7fffffff | |
283 |
|
283 | |||
284 | class revlogio(object): |
|
284 | class revlogio(object): | |
285 | def __init__(self): |
|
285 | def __init__(self): | |
286 | self.size = indexformatng.size |
|
286 | self.size = indexformatng.size | |
287 |
|
287 | |||
288 | def parseindex(self, data, inline): |
|
288 | def parseindex(self, data, inline): | |
289 | # call the C implementation to parse the index data |
|
289 | # call the C implementation to parse the index data | |
290 | index, cache = parsers.parse_index2(data, inline) |
|
290 | index, cache = parsers.parse_index2(data, inline) | |
291 | return index, getattr(index, 'nodemap', None), cache |
|
291 | return index, getattr(index, 'nodemap', None), cache | |
292 |
|
292 | |||
293 | def packentry(self, entry, node, version, rev): |
|
293 | def packentry(self, entry, node, version, rev): | |
294 | p = indexformatng_pack(*entry) |
|
294 | p = indexformatng_pack(*entry) | |
295 | if rev == 0: |
|
295 | if rev == 0: | |
296 | p = versionformat_pack(version) + p[4:] |
|
296 | p = versionformat_pack(version) + p[4:] | |
297 | return p |
|
297 | return p | |
298 |
|
298 | |||
299 | class revlog(object): |
|
299 | class revlog(object): | |
300 | """ |
|
300 | """ | |
301 | the underlying revision storage object |
|
301 | the underlying revision storage object | |
302 |
|
302 | |||
303 | A revlog consists of two parts, an index and the revision data. |
|
303 | A revlog consists of two parts, an index and the revision data. | |
304 |
|
304 | |||
305 | The index is a file with a fixed record size containing |
|
305 | The index is a file with a fixed record size containing | |
306 | information on each revision, including its nodeid (hash), the |
|
306 | information on each revision, including its nodeid (hash), the | |
307 | nodeids of its parents, the position and offset of its data within |
|
307 | nodeids of its parents, the position and offset of its data within | |
308 | the data file, and the revision it's based on. Finally, each entry |
|
308 | the data file, and the revision it's based on. Finally, each entry | |
309 | contains a linkrev entry that can serve as a pointer to external |
|
309 | contains a linkrev entry that can serve as a pointer to external | |
310 | data. |
|
310 | data. | |
311 |
|
311 | |||
312 | The revision data itself is a linear collection of data chunks. |
|
312 | The revision data itself is a linear collection of data chunks. | |
313 | Each chunk represents a revision and is usually represented as a |
|
313 | Each chunk represents a revision and is usually represented as a | |
314 | delta against the previous chunk. To bound lookup time, runs of |
|
314 | delta against the previous chunk. To bound lookup time, runs of | |
315 | deltas are limited to about 2 times the length of the original |
|
315 | deltas are limited to about 2 times the length of the original | |
316 | version data. This makes retrieval of a version proportional to |
|
316 | version data. This makes retrieval of a version proportional to | |
317 | its size, or O(1) relative to the number of revisions. |
|
317 | its size, or O(1) relative to the number of revisions. | |
318 |
|
318 | |||
319 | Both pieces of the revlog are written to in an append-only |
|
319 | Both pieces of the revlog are written to in an append-only | |
320 | fashion, which means we never need to rewrite a file to insert or |
|
320 | fashion, which means we never need to rewrite a file to insert or | |
321 | remove data, and can use some simple techniques to avoid the need |
|
321 | remove data, and can use some simple techniques to avoid the need | |
322 | for locking while reading. |
|
322 | for locking while reading. | |
323 |
|
323 | |||
324 | If checkambig, indexfile is opened with checkambig=True at |
|
324 | If checkambig, indexfile is opened with checkambig=True at | |
325 | writing, to avoid file stat ambiguity. |
|
325 | writing, to avoid file stat ambiguity. | |
326 |
|
326 | |||
327 | If mmaplargeindex is True, and an mmapindexthreshold is set, the |
|
327 | If mmaplargeindex is True, and an mmapindexthreshold is set, the | |
328 | index will be mmapped rather than read if it is larger than the |
|
328 | index will be mmapped rather than read if it is larger than the | |
329 | configured threshold. |
|
329 | configured threshold. | |
330 |
|
330 | |||
331 | If censorable is True, the revlog can have censored revisions. |
|
331 | If censorable is True, the revlog can have censored revisions. | |
332 | """ |
|
332 | """ | |
333 | def __init__(self, opener, indexfile, datafile=None, checkambig=False, |
|
333 | def __init__(self, opener, indexfile, datafile=None, checkambig=False, | |
334 | mmaplargeindex=False, censorable=False): |
|
334 | mmaplargeindex=False, censorable=False): | |
335 | """ |
|
335 | """ | |
336 | create a revlog object |
|
336 | create a revlog object | |
337 |
|
337 | |||
338 | opener is a function that abstracts the file opening operation |
|
338 | opener is a function that abstracts the file opening operation | |
339 | and can be used to implement COW semantics or the like. |
|
339 | and can be used to implement COW semantics or the like. | |
340 | """ |
|
340 | """ | |
341 | self.indexfile = indexfile |
|
341 | self.indexfile = indexfile | |
342 | self.datafile = datafile or (indexfile[:-2] + ".d") |
|
342 | self.datafile = datafile or (indexfile[:-2] + ".d") | |
343 | self.opener = opener |
|
343 | self.opener = opener | |
344 | # When True, indexfile is opened with checkambig=True at writing, to |
|
344 | # When True, indexfile is opened with checkambig=True at writing, to | |
345 | # avoid file stat ambiguity. |
|
345 | # avoid file stat ambiguity. | |
346 | self._checkambig = checkambig |
|
346 | self._checkambig = checkambig | |
347 | self._censorable = censorable |
|
347 | self._censorable = censorable | |
348 | # 3-tuple of (node, rev, text) for a raw revision. |
|
348 | # 3-tuple of (node, rev, text) for a raw revision. | |
349 | self._cache = None |
|
349 | self._cache = None | |
350 | # Maps rev to chain base rev. |
|
350 | # Maps rev to chain base rev. | |
351 | self._chainbasecache = util.lrucachedict(100) |
|
351 | self._chainbasecache = util.lrucachedict(100) | |
352 | # 2-tuple of (offset, data) of raw data from the revlog at an offset. |
|
352 | # 2-tuple of (offset, data) of raw data from the revlog at an offset. | |
353 | self._chunkcache = (0, '') |
|
353 | self._chunkcache = (0, '') | |
354 | # How much data to read and cache into the raw revlog data cache. |
|
354 | # How much data to read and cache into the raw revlog data cache. | |
355 | self._chunkcachesize = 65536 |
|
355 | self._chunkcachesize = 65536 | |
356 | self._maxchainlen = None |
|
356 | self._maxchainlen = None | |
357 | self._deltabothparents = True |
|
357 | self._deltabothparents = True | |
358 | self.index = [] |
|
358 | self.index = [] | |
359 | # Mapping of partial identifiers to full nodes. |
|
359 | # Mapping of partial identifiers to full nodes. | |
360 | self._pcache = {} |
|
360 | self._pcache = {} | |
361 | # Mapping of revision integer to full node. |
|
361 | # Mapping of revision integer to full node. | |
362 | self._nodecache = {nullid: nullrev} |
|
362 | self._nodecache = {nullid: nullrev} | |
363 | self._nodepos = None |
|
363 | self._nodepos = None | |
364 | self._compengine = 'zlib' |
|
364 | self._compengine = 'zlib' | |
365 | self._maxdeltachainspan = -1 |
|
365 | self._maxdeltachainspan = -1 | |
366 | self._withsparseread = False |
|
366 | self._withsparseread = False | |
367 | self._sparserevlog = False |
|
367 | self._sparserevlog = False | |
368 | self._srdensitythreshold = 0.50 |
|
368 | self._srdensitythreshold = 0.50 | |
369 | self._srmingapsize = 262144 |
|
369 | self._srmingapsize = 262144 | |
370 |
|
370 | |||
371 | # Make copy of flag processors so each revlog instance can support |
|
371 | # Make copy of flag processors so each revlog instance can support | |
372 | # custom flags. |
|
372 | # custom flags. | |
373 | self._flagprocessors = dict(_flagprocessors) |
|
373 | self._flagprocessors = dict(_flagprocessors) | |
374 |
|
374 | |||
375 | mmapindexthreshold = None |
|
375 | mmapindexthreshold = None | |
376 | v = REVLOG_DEFAULT_VERSION |
|
376 | v = REVLOG_DEFAULT_VERSION | |
377 | opts = getattr(opener, 'options', None) |
|
377 | opts = getattr(opener, 'options', None) | |
378 | if opts is not None: |
|
378 | if opts is not None: | |
379 | if 'revlogv2' in opts: |
|
379 | if 'revlogv2' in opts: | |
380 | # version 2 revlogs always use generaldelta. |
|
380 | # version 2 revlogs always use generaldelta. | |
381 | v = REVLOGV2 | FLAG_GENERALDELTA | FLAG_INLINE_DATA |
|
381 | v = REVLOGV2 | FLAG_GENERALDELTA | FLAG_INLINE_DATA | |
382 | elif 'revlogv1' in opts: |
|
382 | elif 'revlogv1' in opts: | |
383 | if 'generaldelta' in opts: |
|
383 | if 'generaldelta' in opts: | |
384 | v |= FLAG_GENERALDELTA |
|
384 | v |= FLAG_GENERALDELTA | |
385 | else: |
|
385 | else: | |
386 | v = 0 |
|
386 | v = 0 | |
387 | if 'chunkcachesize' in opts: |
|
387 | if 'chunkcachesize' in opts: | |
388 | self._chunkcachesize = opts['chunkcachesize'] |
|
388 | self._chunkcachesize = opts['chunkcachesize'] | |
389 | if 'maxchainlen' in opts: |
|
389 | if 'maxchainlen' in opts: | |
390 | self._maxchainlen = opts['maxchainlen'] |
|
390 | self._maxchainlen = opts['maxchainlen'] | |
391 | if 'deltabothparents' in opts: |
|
391 | if 'deltabothparents' in opts: | |
392 | self._deltabothparents = opts['deltabothparents'] |
|
392 | self._deltabothparents = opts['deltabothparents'] | |
393 | self._lazydeltabase = bool(opts.get('lazydeltabase', False)) |
|
393 | self._lazydeltabase = bool(opts.get('lazydeltabase', False)) | |
394 | if 'compengine' in opts: |
|
394 | if 'compengine' in opts: | |
395 | self._compengine = opts['compengine'] |
|
395 | self._compengine = opts['compengine'] | |
396 | if 'maxdeltachainspan' in opts: |
|
396 | if 'maxdeltachainspan' in opts: | |
397 | self._maxdeltachainspan = opts['maxdeltachainspan'] |
|
397 | self._maxdeltachainspan = opts['maxdeltachainspan'] | |
398 | if mmaplargeindex and 'mmapindexthreshold' in opts: |
|
398 | if mmaplargeindex and 'mmapindexthreshold' in opts: | |
399 | mmapindexthreshold = opts['mmapindexthreshold'] |
|
399 | mmapindexthreshold = opts['mmapindexthreshold'] | |
400 | self._sparserevlog = bool(opts.get('sparse-revlog', False)) |
|
400 | self._sparserevlog = bool(opts.get('sparse-revlog', False)) | |
401 | withsparseread = bool(opts.get('with-sparse-read', False)) |
|
401 | withsparseread = bool(opts.get('with-sparse-read', False)) | |
402 | # sparse-revlog forces sparse-read |
|
402 | # sparse-revlog forces sparse-read | |
403 | self._withsparseread = self._sparserevlog or withsparseread |
|
403 | self._withsparseread = self._sparserevlog or withsparseread | |
404 | if 'sparse-read-density-threshold' in opts: |
|
404 | if 'sparse-read-density-threshold' in opts: | |
405 | self._srdensitythreshold = opts['sparse-read-density-threshold'] |
|
405 | self._srdensitythreshold = opts['sparse-read-density-threshold'] | |
406 | if 'sparse-read-min-gap-size' in opts: |
|
406 | if 'sparse-read-min-gap-size' in opts: | |
407 | self._srmingapsize = opts['sparse-read-min-gap-size'] |
|
407 | self._srmingapsize = opts['sparse-read-min-gap-size'] | |
408 | if opts.get('enableellipsis'): |
|
408 | if opts.get('enableellipsis'): | |
409 | self._flagprocessors[REVIDX_ELLIPSIS] = ellipsisprocessor |
|
409 | self._flagprocessors[REVIDX_ELLIPSIS] = ellipsisprocessor | |
410 |
|
410 | |||
411 | if self._chunkcachesize <= 0: |
|
411 | if self._chunkcachesize <= 0: | |
412 | raise error.RevlogError(_('revlog chunk cache size %r is not ' |
|
412 | raise error.RevlogError(_('revlog chunk cache size %r is not ' | |
413 | 'greater than 0') % self._chunkcachesize) |
|
413 | 'greater than 0') % self._chunkcachesize) | |
414 | elif self._chunkcachesize & (self._chunkcachesize - 1): |
|
414 | elif self._chunkcachesize & (self._chunkcachesize - 1): | |
415 | raise error.RevlogError(_('revlog chunk cache size %r is not a ' |
|
415 | raise error.RevlogError(_('revlog chunk cache size %r is not a ' | |
416 | 'power of 2') % self._chunkcachesize) |
|
416 | 'power of 2') % self._chunkcachesize) | |
417 |
|
417 | |||
418 | indexdata = '' |
|
418 | indexdata = '' | |
419 | self._initempty = True |
|
419 | self._initempty = True | |
420 | try: |
|
420 | try: | |
421 | with self._indexfp() as f: |
|
421 | with self._indexfp() as f: | |
422 | if (mmapindexthreshold is not None and |
|
422 | if (mmapindexthreshold is not None and | |
423 | self.opener.fstat(f).st_size >= mmapindexthreshold): |
|
423 | self.opener.fstat(f).st_size >= mmapindexthreshold): | |
424 | indexdata = util.buffer(util.mmapread(f)) |
|
424 | indexdata = util.buffer(util.mmapread(f)) | |
425 | else: |
|
425 | else: | |
426 | indexdata = f.read() |
|
426 | indexdata = f.read() | |
427 | if len(indexdata) > 0: |
|
427 | if len(indexdata) > 0: | |
428 | v = versionformat_unpack(indexdata[:4])[0] |
|
428 | v = versionformat_unpack(indexdata[:4])[0] | |
429 | self._initempty = False |
|
429 | self._initempty = False | |
430 | except IOError as inst: |
|
430 | except IOError as inst: | |
431 | if inst.errno != errno.ENOENT: |
|
431 | if inst.errno != errno.ENOENT: | |
432 | raise |
|
432 | raise | |
433 |
|
433 | |||
434 | self.version = v |
|
434 | self.version = v | |
435 | self._inline = v & FLAG_INLINE_DATA |
|
435 | self._inline = v & FLAG_INLINE_DATA | |
436 | self._generaldelta = v & FLAG_GENERALDELTA |
|
436 | self._generaldelta = v & FLAG_GENERALDELTA | |
437 | flags = v & ~0xFFFF |
|
437 | flags = v & ~0xFFFF | |
438 | fmt = v & 0xFFFF |
|
438 | fmt = v & 0xFFFF | |
439 | if fmt == REVLOGV0: |
|
439 | if fmt == REVLOGV0: | |
440 | if flags: |
|
440 | if flags: | |
441 | raise error.RevlogError(_('unknown flags (%#04x) in version %d ' |
|
441 | raise error.RevlogError(_('unknown flags (%#04x) in version %d ' | |
442 | 'revlog %s') % |
|
442 | 'revlog %s') % | |
443 | (flags >> 16, fmt, self.indexfile)) |
|
443 | (flags >> 16, fmt, self.indexfile)) | |
444 | elif fmt == REVLOGV1: |
|
444 | elif fmt == REVLOGV1: | |
445 | if flags & ~REVLOGV1_FLAGS: |
|
445 | if flags & ~REVLOGV1_FLAGS: | |
446 | raise error.RevlogError(_('unknown flags (%#04x) in version %d ' |
|
446 | raise error.RevlogError(_('unknown flags (%#04x) in version %d ' | |
447 | 'revlog %s') % |
|
447 | 'revlog %s') % | |
448 | (flags >> 16, fmt, self.indexfile)) |
|
448 | (flags >> 16, fmt, self.indexfile)) | |
449 | elif fmt == REVLOGV2: |
|
449 | elif fmt == REVLOGV2: | |
450 | if flags & ~REVLOGV2_FLAGS: |
|
450 | if flags & ~REVLOGV2_FLAGS: | |
451 | raise error.RevlogError(_('unknown flags (%#04x) in version %d ' |
|
451 | raise error.RevlogError(_('unknown flags (%#04x) in version %d ' | |
452 | 'revlog %s') % |
|
452 | 'revlog %s') % | |
453 | (flags >> 16, fmt, self.indexfile)) |
|
453 | (flags >> 16, fmt, self.indexfile)) | |
454 | else: |
|
454 | else: | |
455 | raise error.RevlogError(_('unknown version (%d) in revlog %s') % |
|
455 | raise error.RevlogError(_('unknown version (%d) in revlog %s') % | |
456 | (fmt, self.indexfile)) |
|
456 | (fmt, self.indexfile)) | |
457 |
|
457 | |||
458 | self._storedeltachains = True |
|
458 | self._storedeltachains = True | |
459 |
|
459 | |||
460 | self._io = revlogio() |
|
460 | self._io = revlogio() | |
461 | if self.version == REVLOGV0: |
|
461 | if self.version == REVLOGV0: | |
462 | self._io = revlogoldio() |
|
462 | self._io = revlogoldio() | |
463 | try: |
|
463 | try: | |
464 | d = self._io.parseindex(indexdata, self._inline) |
|
464 | d = self._io.parseindex(indexdata, self._inline) | |
465 | except (ValueError, IndexError): |
|
465 | except (ValueError, IndexError): | |
466 | raise error.RevlogError(_("index %s is corrupted") % |
|
466 | raise error.RevlogError(_("index %s is corrupted") % | |
467 | self.indexfile) |
|
467 | self.indexfile) | |
468 | self.index, nodemap, self._chunkcache = d |
|
468 | self.index, nodemap, self._chunkcache = d | |
469 | if nodemap is not None: |
|
469 | if nodemap is not None: | |
470 | self.nodemap = self._nodecache = nodemap |
|
470 | self.nodemap = self._nodecache = nodemap | |
471 | if not self._chunkcache: |
|
471 | if not self._chunkcache: | |
472 | self._chunkclear() |
|
472 | self._chunkclear() | |
473 | # revnum -> (chain-length, sum-delta-length) |
|
473 | # revnum -> (chain-length, sum-delta-length) | |
474 | self._chaininfocache = {} |
|
474 | self._chaininfocache = {} | |
475 | # revlog header -> revlog compressor |
|
475 | # revlog header -> revlog compressor | |
476 | self._decompressors = {} |
|
476 | self._decompressors = {} | |
477 |
|
477 | |||
478 | @util.propertycache |
|
478 | @util.propertycache | |
479 | def _compressor(self): |
|
479 | def _compressor(self): | |
480 | return util.compengines[self._compengine].revlogcompressor() |
|
480 | return util.compengines[self._compengine].revlogcompressor() | |
481 |
|
481 | |||
482 | def _indexfp(self, mode='r'): |
|
482 | def _indexfp(self, mode='r'): | |
483 | """file object for the revlog's index file""" |
|
483 | """file object for the revlog's index file""" | |
484 | args = {r'mode': mode} |
|
484 | args = {r'mode': mode} | |
485 | if mode != 'r': |
|
485 | if mode != 'r': | |
486 | args[r'checkambig'] = self._checkambig |
|
486 | args[r'checkambig'] = self._checkambig | |
487 | if mode == 'w': |
|
487 | if mode == 'w': | |
488 | args[r'atomictemp'] = True |
|
488 | args[r'atomictemp'] = True | |
489 | return self.opener(self.indexfile, **args) |
|
489 | return self.opener(self.indexfile, **args) | |
490 |
|
490 | |||
491 | def _datafp(self, mode='r'): |
|
491 | def _datafp(self, mode='r'): | |
492 | """file object for the revlog's data file""" |
|
492 | """file object for the revlog's data file""" | |
493 | return self.opener(self.datafile, mode=mode) |
|
493 | return self.opener(self.datafile, mode=mode) | |
494 |
|
494 | |||
495 | @contextlib.contextmanager |
|
495 | @contextlib.contextmanager | |
496 | def _datareadfp(self, existingfp=None): |
|
496 | def _datareadfp(self, existingfp=None): | |
497 | """file object suitable to read data""" |
|
497 | """file object suitable to read data""" | |
498 | if existingfp is not None: |
|
498 | if existingfp is not None: | |
499 | yield existingfp |
|
499 | yield existingfp | |
500 | else: |
|
500 | else: | |
501 | if self._inline: |
|
501 | if self._inline: | |
502 | func = self._indexfp |
|
502 | func = self._indexfp | |
503 | else: |
|
503 | else: | |
504 | func = self._datafp |
|
504 | func = self._datafp | |
505 | with func() as fp: |
|
505 | with func() as fp: | |
506 | yield fp |
|
506 | yield fp | |
507 |
|
507 | |||
508 | def tip(self): |
|
508 | def tip(self): | |
509 | return self.node(len(self.index) - 1) |
|
509 | return self.node(len(self.index) - 1) | |
510 | def __contains__(self, rev): |
|
510 | def __contains__(self, rev): | |
511 | return 0 <= rev < len(self) |
|
511 | return 0 <= rev < len(self) | |
512 | def __len__(self): |
|
512 | def __len__(self): | |
513 | return len(self.index) |
|
513 | return len(self.index) | |
514 | def __iter__(self): |
|
514 | def __iter__(self): | |
515 | return iter(pycompat.xrange(len(self))) |
|
515 | return iter(pycompat.xrange(len(self))) | |
516 | def revs(self, start=0, stop=None): |
|
516 | def revs(self, start=0, stop=None): | |
517 | """iterate over all rev in this revlog (from start to stop)""" |
|
517 | """iterate over all rev in this revlog (from start to stop)""" | |
518 | return storageutil.iterrevs(len(self), start=start, stop=stop) |
|
518 | return storageutil.iterrevs(len(self), start=start, stop=stop) | |
519 |
|
519 | |||
520 | @util.propertycache |
|
520 | @util.propertycache | |
521 | def nodemap(self): |
|
521 | def nodemap(self): | |
522 | if self.index: |
|
522 | if self.index: | |
523 | # populate mapping down to the initial node |
|
523 | # populate mapping down to the initial node | |
524 | node0 = self.index[0][7] # get around changelog filtering |
|
524 | node0 = self.index[0][7] # get around changelog filtering | |
525 | self.rev(node0) |
|
525 | self.rev(node0) | |
526 | return self._nodecache |
|
526 | return self._nodecache | |
527 |
|
527 | |||
528 | def hasnode(self, node): |
|
528 | def hasnode(self, node): | |
529 | try: |
|
529 | try: | |
530 | self.rev(node) |
|
530 | self.rev(node) | |
531 | return True |
|
531 | return True | |
532 | except KeyError: |
|
532 | except KeyError: | |
533 | return False |
|
533 | return False | |
534 |
|
534 | |||
535 | def candelta(self, baserev, rev): |
|
535 | def candelta(self, baserev, rev): | |
536 | """whether two revisions (baserev, rev) can be delta-ed or not""" |
|
536 | """whether two revisions (baserev, rev) can be delta-ed or not""" | |
537 | # Disable delta if either rev requires a content-changing flag |
|
537 | # Disable delta if either rev requires a content-changing flag | |
538 | # processor (ex. LFS). This is because such flag processor can alter |
|
538 | # processor (ex. LFS). This is because such flag processor can alter | |
539 | # the rawtext content that the delta will be based on, and two clients |
|
539 | # the rawtext content that the delta will be based on, and two clients | |
540 | # could have a same revlog node with different flags (i.e. different |
|
540 | # could have a same revlog node with different flags (i.e. different | |
541 | # rawtext contents) and the delta could be incompatible. |
|
541 | # rawtext contents) and the delta could be incompatible. | |
542 | if ((self.flags(baserev) & REVIDX_RAWTEXT_CHANGING_FLAGS) |
|
542 | if ((self.flags(baserev) & REVIDX_RAWTEXT_CHANGING_FLAGS) | |
543 | or (self.flags(rev) & REVIDX_RAWTEXT_CHANGING_FLAGS)): |
|
543 | or (self.flags(rev) & REVIDX_RAWTEXT_CHANGING_FLAGS)): | |
544 | return False |
|
544 | return False | |
545 | return True |
|
545 | return True | |
546 |
|
546 | |||
547 | def clearcaches(self): |
|
547 | def clearcaches(self): | |
548 | self._cache = None |
|
548 | self._cache = None | |
549 | self._chainbasecache.clear() |
|
549 | self._chainbasecache.clear() | |
550 | self._chunkcache = (0, '') |
|
550 | self._chunkcache = (0, '') | |
551 | self._pcache = {} |
|
551 | self._pcache = {} | |
552 |
|
552 | |||
553 | try: |
|
553 | try: | |
554 | self._nodecache.clearcaches() |
|
554 | self._nodecache.clearcaches() | |
555 | except AttributeError: |
|
555 | except AttributeError: | |
556 | self._nodecache = {nullid: nullrev} |
|
556 | self._nodecache = {nullid: nullrev} | |
557 | self._nodepos = None |
|
557 | self._nodepos = None | |
558 |
|
558 | |||
559 | def rev(self, node): |
|
559 | def rev(self, node): | |
560 | try: |
|
560 | try: | |
561 | return self._nodecache[node] |
|
561 | return self._nodecache[node] | |
562 | except TypeError: |
|
562 | except TypeError: | |
563 | raise |
|
563 | raise | |
564 | except error.RevlogError: |
|
564 | except error.RevlogError: | |
565 | # parsers.c radix tree lookup failed |
|
565 | # parsers.c radix tree lookup failed | |
566 | if node == wdirid or node in wdirfilenodeids: |
|
566 | if node == wdirid or node in wdirfilenodeids: | |
567 | raise error.WdirUnsupported |
|
567 | raise error.WdirUnsupported | |
568 | raise error.LookupError(node, self.indexfile, _('no node')) |
|
568 | raise error.LookupError(node, self.indexfile, _('no node')) | |
569 | except KeyError: |
|
569 | except KeyError: | |
570 | # pure python cache lookup failed |
|
570 | # pure python cache lookup failed | |
571 | n = self._nodecache |
|
571 | n = self._nodecache | |
572 | i = self.index |
|
572 | i = self.index | |
573 | p = self._nodepos |
|
573 | p = self._nodepos | |
574 | if p is None: |
|
574 | if p is None: | |
575 | p = len(i) - 1 |
|
575 | p = len(i) - 1 | |
576 | else: |
|
576 | else: | |
577 | assert p < len(i) |
|
577 | assert p < len(i) | |
578 | for r in pycompat.xrange(p, -1, -1): |
|
578 | for r in pycompat.xrange(p, -1, -1): | |
579 | v = i[r][7] |
|
579 | v = i[r][7] | |
580 | n[v] = r |
|
580 | n[v] = r | |
581 | if v == node: |
|
581 | if v == node: | |
582 | self._nodepos = r - 1 |
|
582 | self._nodepos = r - 1 | |
583 | return r |
|
583 | return r | |
584 | if node == wdirid or node in wdirfilenodeids: |
|
584 | if node == wdirid or node in wdirfilenodeids: | |
585 | raise error.WdirUnsupported |
|
585 | raise error.WdirUnsupported | |
586 | raise error.LookupError(node, self.indexfile, _('no node')) |
|
586 | raise error.LookupError(node, self.indexfile, _('no node')) | |
587 |
|
587 | |||
588 | # Accessors for index entries. |
|
588 | # Accessors for index entries. | |
589 |
|
589 | |||
590 | # First tuple entry is 8 bytes. First 6 bytes are offset. Last 2 bytes |
|
590 | # First tuple entry is 8 bytes. First 6 bytes are offset. Last 2 bytes | |
591 | # are flags. |
|
591 | # are flags. | |
592 | def start(self, rev): |
|
592 | def start(self, rev): | |
593 | return int(self.index[rev][0] >> 16) |
|
593 | return int(self.index[rev][0] >> 16) | |
594 |
|
594 | |||
595 | def flags(self, rev): |
|
595 | def flags(self, rev): | |
596 | return self.index[rev][0] & 0xFFFF |
|
596 | return self.index[rev][0] & 0xFFFF | |
597 |
|
597 | |||
598 | def length(self, rev): |
|
598 | def length(self, rev): | |
599 | return self.index[rev][1] |
|
599 | return self.index[rev][1] | |
600 |
|
600 | |||
601 | def rawsize(self, rev): |
|
601 | def rawsize(self, rev): | |
602 | """return the length of the uncompressed text for a given revision""" |
|
602 | """return the length of the uncompressed text for a given revision""" | |
603 | l = self.index[rev][2] |
|
603 | l = self.index[rev][2] | |
604 | if l >= 0: |
|
604 | if l >= 0: | |
605 | return l |
|
605 | return l | |
606 |
|
606 | |||
607 | t = self.revision(rev, raw=True) |
|
607 | t = self.revision(rev, raw=True) | |
608 | return len(t) |
|
608 | return len(t) | |
609 |
|
609 | |||
610 | def size(self, rev): |
|
610 | def size(self, rev): | |
611 | """length of non-raw text (processed by a "read" flag processor)""" |
|
611 | """length of non-raw text (processed by a "read" flag processor)""" | |
612 | # fast path: if no "read" flag processor could change the content, |
|
612 | # fast path: if no "read" flag processor could change the content, | |
613 | # size is rawsize. note: ELLIPSIS is known to not change the content. |
|
613 | # size is rawsize. note: ELLIPSIS is known to not change the content. | |
614 | flags = self.flags(rev) |
|
614 | flags = self.flags(rev) | |
615 | if flags & (REVIDX_KNOWN_FLAGS ^ REVIDX_ELLIPSIS) == 0: |
|
615 | if flags & (REVIDX_KNOWN_FLAGS ^ REVIDX_ELLIPSIS) == 0: | |
616 | return self.rawsize(rev) |
|
616 | return self.rawsize(rev) | |
617 |
|
617 | |||
618 | return len(self.revision(rev, raw=False)) |
|
618 | return len(self.revision(rev, raw=False)) | |
619 |
|
619 | |||
620 | def chainbase(self, rev): |
|
620 | def chainbase(self, rev): | |
621 | base = self._chainbasecache.get(rev) |
|
621 | base = self._chainbasecache.get(rev) | |
622 | if base is not None: |
|
622 | if base is not None: | |
623 | return base |
|
623 | return base | |
624 |
|
624 | |||
625 | index = self.index |
|
625 | index = self.index | |
626 | iterrev = rev |
|
626 | iterrev = rev | |
627 | base = index[iterrev][3] |
|
627 | base = index[iterrev][3] | |
628 | while base != iterrev: |
|
628 | while base != iterrev: | |
629 | iterrev = base |
|
629 | iterrev = base | |
630 | base = index[iterrev][3] |
|
630 | base = index[iterrev][3] | |
631 |
|
631 | |||
632 | self._chainbasecache[rev] = base |
|
632 | self._chainbasecache[rev] = base | |
633 | return base |
|
633 | return base | |
634 |
|
634 | |||
635 | def linkrev(self, rev): |
|
635 | def linkrev(self, rev): | |
636 | return self.index[rev][4] |
|
636 | return self.index[rev][4] | |
637 |
|
637 | |||
638 | def parentrevs(self, rev): |
|
638 | def parentrevs(self, rev): | |
639 | try: |
|
639 | try: | |
640 | entry = self.index[rev] |
|
640 | entry = self.index[rev] | |
641 | except IndexError: |
|
641 | except IndexError: | |
642 | if rev == wdirrev: |
|
642 | if rev == wdirrev: | |
643 | raise error.WdirUnsupported |
|
643 | raise error.WdirUnsupported | |
644 | raise |
|
644 | raise | |
645 |
|
645 | |||
646 | return entry[5], entry[6] |
|
646 | return entry[5], entry[6] | |
647 |
|
647 | |||
648 | def node(self, rev): |
|
648 | def node(self, rev): | |
649 | try: |
|
649 | try: | |
650 | return self.index[rev][7] |
|
650 | return self.index[rev][7] | |
651 | except IndexError: |
|
651 | except IndexError: | |
652 | if rev == wdirrev: |
|
652 | if rev == wdirrev: | |
653 | raise error.WdirUnsupported |
|
653 | raise error.WdirUnsupported | |
654 | raise |
|
654 | raise | |
655 |
|
655 | |||
656 | # Derived from index values. |
|
656 | # Derived from index values. | |
657 |
|
657 | |||
658 | def end(self, rev): |
|
658 | def end(self, rev): | |
659 | return self.start(rev) + self.length(rev) |
|
659 | return self.start(rev) + self.length(rev) | |
660 |
|
660 | |||
661 | def parents(self, node): |
|
661 | def parents(self, node): | |
662 | i = self.index |
|
662 | i = self.index | |
663 | d = i[self.rev(node)] |
|
663 | d = i[self.rev(node)] | |
664 | return i[d[5]][7], i[d[6]][7] # map revisions to nodes inline |
|
664 | return i[d[5]][7], i[d[6]][7] # map revisions to nodes inline | |
665 |
|
665 | |||
666 | def chainlen(self, rev): |
|
666 | def chainlen(self, rev): | |
667 | return self._chaininfo(rev)[0] |
|
667 | return self._chaininfo(rev)[0] | |
668 |
|
668 | |||
669 | def _chaininfo(self, rev): |
|
669 | def _chaininfo(self, rev): | |
670 | chaininfocache = self._chaininfocache |
|
670 | chaininfocache = self._chaininfocache | |
671 | if rev in chaininfocache: |
|
671 | if rev in chaininfocache: | |
672 | return chaininfocache[rev] |
|
672 | return chaininfocache[rev] | |
673 | index = self.index |
|
673 | index = self.index | |
674 | generaldelta = self._generaldelta |
|
674 | generaldelta = self._generaldelta | |
675 | iterrev = rev |
|
675 | iterrev = rev | |
676 | e = index[iterrev] |
|
676 | e = index[iterrev] | |
677 | clen = 0 |
|
677 | clen = 0 | |
678 | compresseddeltalen = 0 |
|
678 | compresseddeltalen = 0 | |
679 | while iterrev != e[3]: |
|
679 | while iterrev != e[3]: | |
680 | clen += 1 |
|
680 | clen += 1 | |
681 | compresseddeltalen += e[1] |
|
681 | compresseddeltalen += e[1] | |
682 | if generaldelta: |
|
682 | if generaldelta: | |
683 | iterrev = e[3] |
|
683 | iterrev = e[3] | |
684 | else: |
|
684 | else: | |
685 | iterrev -= 1 |
|
685 | iterrev -= 1 | |
686 | if iterrev in chaininfocache: |
|
686 | if iterrev in chaininfocache: | |
687 | t = chaininfocache[iterrev] |
|
687 | t = chaininfocache[iterrev] | |
688 | clen += t[0] |
|
688 | clen += t[0] | |
689 | compresseddeltalen += t[1] |
|
689 | compresseddeltalen += t[1] | |
690 | break |
|
690 | break | |
691 | e = index[iterrev] |
|
691 | e = index[iterrev] | |
692 | else: |
|
692 | else: | |
693 | # Add text length of base since decompressing that also takes |
|
693 | # Add text length of base since decompressing that also takes | |
694 | # work. For cache hits the length is already included. |
|
694 | # work. For cache hits the length is already included. | |
695 | compresseddeltalen += e[1] |
|
695 | compresseddeltalen += e[1] | |
696 | r = (clen, compresseddeltalen) |
|
696 | r = (clen, compresseddeltalen) | |
697 | chaininfocache[rev] = r |
|
697 | chaininfocache[rev] = r | |
698 | return r |
|
698 | return r | |
699 |
|
699 | |||
700 | def _deltachain(self, rev, stoprev=None): |
|
700 | def _deltachain(self, rev, stoprev=None): | |
701 | """Obtain the delta chain for a revision. |
|
701 | """Obtain the delta chain for a revision. | |
702 |
|
702 | |||
703 | ``stoprev`` specifies a revision to stop at. If not specified, we |
|
703 | ``stoprev`` specifies a revision to stop at. If not specified, we | |
704 | stop at the base of the chain. |
|
704 | stop at the base of the chain. | |
705 |
|
705 | |||
706 | Returns a 2-tuple of (chain, stopped) where ``chain`` is a list of |
|
706 | Returns a 2-tuple of (chain, stopped) where ``chain`` is a list of | |
707 | revs in ascending order and ``stopped`` is a bool indicating whether |
|
707 | revs in ascending order and ``stopped`` is a bool indicating whether | |
708 | ``stoprev`` was hit. |
|
708 | ``stoprev`` was hit. | |
709 | """ |
|
709 | """ | |
710 | # Try C implementation. |
|
710 | # Try C implementation. | |
711 | try: |
|
711 | try: | |
712 | return self.index.deltachain(rev, stoprev, self._generaldelta) |
|
712 | return self.index.deltachain(rev, stoprev, self._generaldelta) | |
713 | except AttributeError: |
|
713 | except AttributeError: | |
714 | pass |
|
714 | pass | |
715 |
|
715 | |||
716 | chain = [] |
|
716 | chain = [] | |
717 |
|
717 | |||
718 | # Alias to prevent attribute lookup in tight loop. |
|
718 | # Alias to prevent attribute lookup in tight loop. | |
719 | index = self.index |
|
719 | index = self.index | |
720 | generaldelta = self._generaldelta |
|
720 | generaldelta = self._generaldelta | |
721 |
|
721 | |||
722 | iterrev = rev |
|
722 | iterrev = rev | |
723 | e = index[iterrev] |
|
723 | e = index[iterrev] | |
724 | while iterrev != e[3] and iterrev != stoprev: |
|
724 | while iterrev != e[3] and iterrev != stoprev: | |
725 | chain.append(iterrev) |
|
725 | chain.append(iterrev) | |
726 | if generaldelta: |
|
726 | if generaldelta: | |
727 | iterrev = e[3] |
|
727 | iterrev = e[3] | |
728 | else: |
|
728 | else: | |
729 | iterrev -= 1 |
|
729 | iterrev -= 1 | |
730 | e = index[iterrev] |
|
730 | e = index[iterrev] | |
731 |
|
731 | |||
732 | if iterrev == stoprev: |
|
732 | if iterrev == stoprev: | |
733 | stopped = True |
|
733 | stopped = True | |
734 | else: |
|
734 | else: | |
735 | chain.append(iterrev) |
|
735 | chain.append(iterrev) | |
736 | stopped = False |
|
736 | stopped = False | |
737 |
|
737 | |||
738 | chain.reverse() |
|
738 | chain.reverse() | |
739 | return chain, stopped |
|
739 | return chain, stopped | |
740 |
|
740 | |||
741 | def ancestors(self, revs, stoprev=0, inclusive=False): |
|
741 | def ancestors(self, revs, stoprev=0, inclusive=False): | |
742 | """Generate the ancestors of 'revs' in reverse topological order. |
|
742 | """Generate the ancestors of 'revs' in reverse topological order. | |
743 | Does not generate revs lower than stoprev. |
|
743 | Does not generate revs lower than stoprev. | |
744 |
|
744 | |||
745 | See the documentation for ancestor.lazyancestors for more details.""" |
|
745 | See the documentation for ancestor.lazyancestors for more details.""" | |
746 |
|
746 | |||
747 | return ancestor.lazyancestors(self.parentrevs, revs, stoprev=stoprev, |
|
747 | return ancestor.lazyancestors(self.parentrevs, revs, stoprev=stoprev, | |
748 | inclusive=inclusive) |
|
748 | inclusive=inclusive) | |
749 |
|
749 | |||
750 | def descendants(self, revs): |
|
750 | def descendants(self, revs): | |
751 | return dagop.descendantrevs(revs, self.revs, self.parentrevs) |
|
751 | return dagop.descendantrevs(revs, self.revs, self.parentrevs) | |
752 |
|
752 | |||
753 | def findcommonmissing(self, common=None, heads=None): |
|
753 | def findcommonmissing(self, common=None, heads=None): | |
754 | """Return a tuple of the ancestors of common and the ancestors of heads |
|
754 | """Return a tuple of the ancestors of common and the ancestors of heads | |
755 | that are not ancestors of common. In revset terminology, we return the |
|
755 | that are not ancestors of common. In revset terminology, we return the | |
756 | tuple: |
|
756 | tuple: | |
757 |
|
757 | |||
758 | ::common, (::heads) - (::common) |
|
758 | ::common, (::heads) - (::common) | |
759 |
|
759 | |||
760 | The list is sorted by revision number, meaning it is |
|
760 | The list is sorted by revision number, meaning it is | |
761 | topologically sorted. |
|
761 | topologically sorted. | |
762 |
|
762 | |||
763 | 'heads' and 'common' are both lists of node IDs. If heads is |
|
763 | 'heads' and 'common' are both lists of node IDs. If heads is | |
764 | not supplied, uses all of the revlog's heads. If common is not |
|
764 | not supplied, uses all of the revlog's heads. If common is not | |
765 | supplied, uses nullid.""" |
|
765 | supplied, uses nullid.""" | |
766 | if common is None: |
|
766 | if common is None: | |
767 | common = [nullid] |
|
767 | common = [nullid] | |
768 | if heads is None: |
|
768 | if heads is None: | |
769 | heads = self.heads() |
|
769 | heads = self.heads() | |
770 |
|
770 | |||
771 | common = [self.rev(n) for n in common] |
|
771 | common = [self.rev(n) for n in common] | |
772 | heads = [self.rev(n) for n in heads] |
|
772 | heads = [self.rev(n) for n in heads] | |
773 |
|
773 | |||
774 | # we want the ancestors, but inclusive |
|
774 | # we want the ancestors, but inclusive | |
775 | class lazyset(object): |
|
775 | class lazyset(object): | |
776 | def __init__(self, lazyvalues): |
|
776 | def __init__(self, lazyvalues): | |
777 | self.addedvalues = set() |
|
777 | self.addedvalues = set() | |
778 | self.lazyvalues = lazyvalues |
|
778 | self.lazyvalues = lazyvalues | |
779 |
|
779 | |||
780 | def __contains__(self, value): |
|
780 | def __contains__(self, value): | |
781 | return value in self.addedvalues or value in self.lazyvalues |
|
781 | return value in self.addedvalues or value in self.lazyvalues | |
782 |
|
782 | |||
783 | def __iter__(self): |
|
783 | def __iter__(self): | |
784 | added = self.addedvalues |
|
784 | added = self.addedvalues | |
785 | for r in added: |
|
785 | for r in added: | |
786 | yield r |
|
786 | yield r | |
787 | for r in self.lazyvalues: |
|
787 | for r in self.lazyvalues: | |
788 | if not r in added: |
|
788 | if not r in added: | |
789 | yield r |
|
789 | yield r | |
790 |
|
790 | |||
791 | def add(self, value): |
|
791 | def add(self, value): | |
792 | self.addedvalues.add(value) |
|
792 | self.addedvalues.add(value) | |
793 |
|
793 | |||
794 | def update(self, values): |
|
794 | def update(self, values): | |
795 | self.addedvalues.update(values) |
|
795 | self.addedvalues.update(values) | |
796 |
|
796 | |||
797 | has = lazyset(self.ancestors(common)) |
|
797 | has = lazyset(self.ancestors(common)) | |
798 | has.add(nullrev) |
|
798 | has.add(nullrev) | |
799 | has.update(common) |
|
799 | has.update(common) | |
800 |
|
800 | |||
801 | # take all ancestors from heads that aren't in has |
|
801 | # take all ancestors from heads that aren't in has | |
802 | missing = set() |
|
802 | missing = set() | |
803 | visit = collections.deque(r for r in heads if r not in has) |
|
803 | visit = collections.deque(r for r in heads if r not in has) | |
804 | while visit: |
|
804 | while visit: | |
805 | r = visit.popleft() |
|
805 | r = visit.popleft() | |
806 | if r in missing: |
|
806 | if r in missing: | |
807 | continue |
|
807 | continue | |
808 | else: |
|
808 | else: | |
809 | missing.add(r) |
|
809 | missing.add(r) | |
810 | for p in self.parentrevs(r): |
|
810 | for p in self.parentrevs(r): | |
811 | if p not in has: |
|
811 | if p not in has: | |
812 | visit.append(p) |
|
812 | visit.append(p) | |
813 | missing = list(missing) |
|
813 | missing = list(missing) | |
814 | missing.sort() |
|
814 | missing.sort() | |
815 | return has, [self.node(miss) for miss in missing] |
|
815 | return has, [self.node(miss) for miss in missing] | |
816 |
|
816 | |||
817 | def incrementalmissingrevs(self, common=None): |
|
817 | def incrementalmissingrevs(self, common=None): | |
818 | """Return an object that can be used to incrementally compute the |
|
818 | """Return an object that can be used to incrementally compute the | |
819 | revision numbers of the ancestors of arbitrary sets that are not |
|
819 | revision numbers of the ancestors of arbitrary sets that are not | |
820 | ancestors of common. This is an ancestor.incrementalmissingancestors |
|
820 | ancestors of common. This is an ancestor.incrementalmissingancestors | |
821 | object. |
|
821 | object. | |
822 |
|
822 | |||
823 | 'common' is a list of revision numbers. If common is not supplied, uses |
|
823 | 'common' is a list of revision numbers. If common is not supplied, uses | |
824 | nullrev. |
|
824 | nullrev. | |
825 | """ |
|
825 | """ | |
826 | if common is None: |
|
826 | if common is None: | |
827 | common = [nullrev] |
|
827 | common = [nullrev] | |
828 |
|
828 | |||
829 | return ancestor.incrementalmissingancestors(self.parentrevs, common) |
|
829 | return ancestor.incrementalmissingancestors(self.parentrevs, common) | |
830 |
|
830 | |||
831 | def findmissingrevs(self, common=None, heads=None): |
|
831 | def findmissingrevs(self, common=None, heads=None): | |
832 | """Return the revision numbers of the ancestors of heads that |
|
832 | """Return the revision numbers of the ancestors of heads that | |
833 | are not ancestors of common. |
|
833 | are not ancestors of common. | |
834 |
|
834 | |||
835 | More specifically, return a list of revision numbers corresponding to |
|
835 | More specifically, return a list of revision numbers corresponding to | |
836 | nodes N such that every N satisfies the following constraints: |
|
836 | nodes N such that every N satisfies the following constraints: | |
837 |
|
837 | |||
838 | 1. N is an ancestor of some node in 'heads' |
|
838 | 1. N is an ancestor of some node in 'heads' | |
839 | 2. N is not an ancestor of any node in 'common' |
|
839 | 2. N is not an ancestor of any node in 'common' | |
840 |
|
840 | |||
841 | The list is sorted by revision number, meaning it is |
|
841 | The list is sorted by revision number, meaning it is | |
842 | topologically sorted. |
|
842 | topologically sorted. | |
843 |
|
843 | |||
844 | 'heads' and 'common' are both lists of revision numbers. If heads is |
|
844 | 'heads' and 'common' are both lists of revision numbers. If heads is | |
845 | not supplied, uses all of the revlog's heads. If common is not |
|
845 | not supplied, uses all of the revlog's heads. If common is not | |
846 | supplied, uses nullid.""" |
|
846 | supplied, uses nullid.""" | |
847 | if common is None: |
|
847 | if common is None: | |
848 | common = [nullrev] |
|
848 | common = [nullrev] | |
849 | if heads is None: |
|
849 | if heads is None: | |
850 | heads = self.headrevs() |
|
850 | heads = self.headrevs() | |
851 |
|
851 | |||
852 | inc = self.incrementalmissingrevs(common=common) |
|
852 | inc = self.incrementalmissingrevs(common=common) | |
853 | return inc.missingancestors(heads) |
|
853 | return inc.missingancestors(heads) | |
854 |
|
854 | |||
855 | def findmissing(self, common=None, heads=None): |
|
855 | def findmissing(self, common=None, heads=None): | |
856 | """Return the ancestors of heads that are not ancestors of common. |
|
856 | """Return the ancestors of heads that are not ancestors of common. | |
857 |
|
857 | |||
858 | More specifically, return a list of nodes N such that every N |
|
858 | More specifically, return a list of nodes N such that every N | |
859 | satisfies the following constraints: |
|
859 | satisfies the following constraints: | |
860 |
|
860 | |||
861 | 1. N is an ancestor of some node in 'heads' |
|
861 | 1. N is an ancestor of some node in 'heads' | |
862 | 2. N is not an ancestor of any node in 'common' |
|
862 | 2. N is not an ancestor of any node in 'common' | |
863 |
|
863 | |||
864 | The list is sorted by revision number, meaning it is |
|
864 | The list is sorted by revision number, meaning it is | |
865 | topologically sorted. |
|
865 | topologically sorted. | |
866 |
|
866 | |||
867 | 'heads' and 'common' are both lists of node IDs. If heads is |
|
867 | 'heads' and 'common' are both lists of node IDs. If heads is | |
868 | not supplied, uses all of the revlog's heads. If common is not |
|
868 | not supplied, uses all of the revlog's heads. If common is not | |
869 | supplied, uses nullid.""" |
|
869 | supplied, uses nullid.""" | |
870 | if common is None: |
|
870 | if common is None: | |
871 | common = [nullid] |
|
871 | common = [nullid] | |
872 | if heads is None: |
|
872 | if heads is None: | |
873 | heads = self.heads() |
|
873 | heads = self.heads() | |
874 |
|
874 | |||
875 | common = [self.rev(n) for n in common] |
|
875 | common = [self.rev(n) for n in common] | |
876 | heads = [self.rev(n) for n in heads] |
|
876 | heads = [self.rev(n) for n in heads] | |
877 |
|
877 | |||
878 | inc = self.incrementalmissingrevs(common=common) |
|
878 | inc = self.incrementalmissingrevs(common=common) | |
879 | return [self.node(r) for r in inc.missingancestors(heads)] |
|
879 | return [self.node(r) for r in inc.missingancestors(heads)] | |
880 |
|
880 | |||
881 | def nodesbetween(self, roots=None, heads=None): |
|
881 | def nodesbetween(self, roots=None, heads=None): | |
882 | """Return a topological path from 'roots' to 'heads'. |
|
882 | """Return a topological path from 'roots' to 'heads'. | |
883 |
|
883 | |||
884 | Return a tuple (nodes, outroots, outheads) where 'nodes' is a |
|
884 | Return a tuple (nodes, outroots, outheads) where 'nodes' is a | |
885 | topologically sorted list of all nodes N that satisfy both of |
|
885 | topologically sorted list of all nodes N that satisfy both of | |
886 | these constraints: |
|
886 | these constraints: | |
887 |
|
887 | |||
888 | 1. N is a descendant of some node in 'roots' |
|
888 | 1. N is a descendant of some node in 'roots' | |
889 | 2. N is an ancestor of some node in 'heads' |
|
889 | 2. N is an ancestor of some node in 'heads' | |
890 |
|
890 | |||
891 | Every node is considered to be both a descendant and an ancestor |
|
891 | Every node is considered to be both a descendant and an ancestor | |
892 | of itself, so every reachable node in 'roots' and 'heads' will be |
|
892 | of itself, so every reachable node in 'roots' and 'heads' will be | |
893 | included in 'nodes'. |
|
893 | included in 'nodes'. | |
894 |
|
894 | |||
895 | 'outroots' is the list of reachable nodes in 'roots', i.e., the |
|
895 | 'outroots' is the list of reachable nodes in 'roots', i.e., the | |
896 | subset of 'roots' that is returned in 'nodes'. Likewise, |
|
896 | subset of 'roots' that is returned in 'nodes'. Likewise, | |
897 | 'outheads' is the subset of 'heads' that is also in 'nodes'. |
|
897 | 'outheads' is the subset of 'heads' that is also in 'nodes'. | |
898 |
|
898 | |||
899 | 'roots' and 'heads' are both lists of node IDs. If 'roots' is |
|
899 | 'roots' and 'heads' are both lists of node IDs. If 'roots' is | |
900 | unspecified, uses nullid as the only root. If 'heads' is |
|
900 | unspecified, uses nullid as the only root. If 'heads' is | |
901 | unspecified, uses list of all of the revlog's heads.""" |
|
901 | unspecified, uses list of all of the revlog's heads.""" | |
902 | nonodes = ([], [], []) |
|
902 | nonodes = ([], [], []) | |
903 | if roots is not None: |
|
903 | if roots is not None: | |
904 | roots = list(roots) |
|
904 | roots = list(roots) | |
905 | if not roots: |
|
905 | if not roots: | |
906 | return nonodes |
|
906 | return nonodes | |
907 | lowestrev = min([self.rev(n) for n in roots]) |
|
907 | lowestrev = min([self.rev(n) for n in roots]) | |
908 | else: |
|
908 | else: | |
909 | roots = [nullid] # Everybody's a descendant of nullid |
|
909 | roots = [nullid] # Everybody's a descendant of nullid | |
910 | lowestrev = nullrev |
|
910 | lowestrev = nullrev | |
911 | if (lowestrev == nullrev) and (heads is None): |
|
911 | if (lowestrev == nullrev) and (heads is None): | |
912 | # We want _all_ the nodes! |
|
912 | # We want _all_ the nodes! | |
913 | return ([self.node(r) for r in self], [nullid], list(self.heads())) |
|
913 | return ([self.node(r) for r in self], [nullid], list(self.heads())) | |
914 | if heads is None: |
|
914 | if heads is None: | |
915 | # All nodes are ancestors, so the latest ancestor is the last |
|
915 | # All nodes are ancestors, so the latest ancestor is the last | |
916 | # node. |
|
916 | # node. | |
917 | highestrev = len(self) - 1 |
|
917 | highestrev = len(self) - 1 | |
918 | # Set ancestors to None to signal that every node is an ancestor. |
|
918 | # Set ancestors to None to signal that every node is an ancestor. | |
919 | ancestors = None |
|
919 | ancestors = None | |
920 | # Set heads to an empty dictionary for later discovery of heads |
|
920 | # Set heads to an empty dictionary for later discovery of heads | |
921 | heads = {} |
|
921 | heads = {} | |
922 | else: |
|
922 | else: | |
923 | heads = list(heads) |
|
923 | heads = list(heads) | |
924 | if not heads: |
|
924 | if not heads: | |
925 | return nonodes |
|
925 | return nonodes | |
926 | ancestors = set() |
|
926 | ancestors = set() | |
927 | # Turn heads into a dictionary so we can remove 'fake' heads. |
|
927 | # Turn heads into a dictionary so we can remove 'fake' heads. | |
928 | # Also, later we will be using it to filter out the heads we can't |
|
928 | # Also, later we will be using it to filter out the heads we can't | |
929 | # find from roots. |
|
929 | # find from roots. | |
930 | heads = dict.fromkeys(heads, False) |
|
930 | heads = dict.fromkeys(heads, False) | |
931 | # Start at the top and keep marking parents until we're done. |
|
931 | # Start at the top and keep marking parents until we're done. | |
932 | nodestotag = set(heads) |
|
932 | nodestotag = set(heads) | |
933 | # Remember where the top was so we can use it as a limit later. |
|
933 | # Remember where the top was so we can use it as a limit later. | |
934 | highestrev = max([self.rev(n) for n in nodestotag]) |
|
934 | highestrev = max([self.rev(n) for n in nodestotag]) | |
935 | while nodestotag: |
|
935 | while nodestotag: | |
936 | # grab a node to tag |
|
936 | # grab a node to tag | |
937 | n = nodestotag.pop() |
|
937 | n = nodestotag.pop() | |
938 | # Never tag nullid |
|
938 | # Never tag nullid | |
939 | if n == nullid: |
|
939 | if n == nullid: | |
940 | continue |
|
940 | continue | |
941 | # A node's revision number represents its place in a |
|
941 | # A node's revision number represents its place in a | |
942 | # topologically sorted list of nodes. |
|
942 | # topologically sorted list of nodes. | |
943 | r = self.rev(n) |
|
943 | r = self.rev(n) | |
944 | if r >= lowestrev: |
|
944 | if r >= lowestrev: | |
945 | if n not in ancestors: |
|
945 | if n not in ancestors: | |
946 | # If we are possibly a descendant of one of the roots |
|
946 | # If we are possibly a descendant of one of the roots | |
947 | # and we haven't already been marked as an ancestor |
|
947 | # and we haven't already been marked as an ancestor | |
948 | ancestors.add(n) # Mark as ancestor |
|
948 | ancestors.add(n) # Mark as ancestor | |
949 | # Add non-nullid parents to list of nodes to tag. |
|
949 | # Add non-nullid parents to list of nodes to tag. | |
950 | nodestotag.update([p for p in self.parents(n) if |
|
950 | nodestotag.update([p for p in self.parents(n) if | |
951 | p != nullid]) |
|
951 | p != nullid]) | |
952 | elif n in heads: # We've seen it before, is it a fake head? |
|
952 | elif n in heads: # We've seen it before, is it a fake head? | |
953 | # So it is, real heads should not be the ancestors of |
|
953 | # So it is, real heads should not be the ancestors of | |
954 | # any other heads. |
|
954 | # any other heads. | |
955 | heads.pop(n) |
|
955 | heads.pop(n) | |
956 | if not ancestors: |
|
956 | if not ancestors: | |
957 | return nonodes |
|
957 | return nonodes | |
958 | # Now that we have our set of ancestors, we want to remove any |
|
958 | # Now that we have our set of ancestors, we want to remove any | |
959 | # roots that are not ancestors. |
|
959 | # roots that are not ancestors. | |
960 |
|
960 | |||
961 | # If one of the roots was nullid, everything is included anyway. |
|
961 | # If one of the roots was nullid, everything is included anyway. | |
962 | if lowestrev > nullrev: |
|
962 | if lowestrev > nullrev: | |
963 | # But, since we weren't, let's recompute the lowest rev to not |
|
963 | # But, since we weren't, let's recompute the lowest rev to not | |
964 | # include roots that aren't ancestors. |
|
964 | # include roots that aren't ancestors. | |
965 |
|
965 | |||
966 | # Filter out roots that aren't ancestors of heads |
|
966 | # Filter out roots that aren't ancestors of heads | |
967 | roots = [root for root in roots if root in ancestors] |
|
967 | roots = [root for root in roots if root in ancestors] | |
968 | # Recompute the lowest revision |
|
968 | # Recompute the lowest revision | |
969 | if roots: |
|
969 | if roots: | |
970 | lowestrev = min([self.rev(root) for root in roots]) |
|
970 | lowestrev = min([self.rev(root) for root in roots]) | |
971 | else: |
|
971 | else: | |
972 | # No more roots? Return empty list |
|
972 | # No more roots? Return empty list | |
973 | return nonodes |
|
973 | return nonodes | |
974 | else: |
|
974 | else: | |
975 | # We are descending from nullid, and don't need to care about |
|
975 | # We are descending from nullid, and don't need to care about | |
976 | # any other roots. |
|
976 | # any other roots. | |
977 | lowestrev = nullrev |
|
977 | lowestrev = nullrev | |
978 | roots = [nullid] |
|
978 | roots = [nullid] | |
979 | # Transform our roots list into a set. |
|
979 | # Transform our roots list into a set. | |
980 | descendants = set(roots) |
|
980 | descendants = set(roots) | |
981 | # Also, keep the original roots so we can filter out roots that aren't |
|
981 | # Also, keep the original roots so we can filter out roots that aren't | |
982 | # 'real' roots (i.e. are descended from other roots). |
|
982 | # 'real' roots (i.e. are descended from other roots). | |
983 | roots = descendants.copy() |
|
983 | roots = descendants.copy() | |
984 | # Our topologically sorted list of output nodes. |
|
984 | # Our topologically sorted list of output nodes. | |
985 | orderedout = [] |
|
985 | orderedout = [] | |
986 | # Don't start at nullid since we don't want nullid in our output list, |
|
986 | # Don't start at nullid since we don't want nullid in our output list, | |
987 | # and if nullid shows up in descendants, empty parents will look like |
|
987 | # and if nullid shows up in descendants, empty parents will look like | |
988 | # they're descendants. |
|
988 | # they're descendants. | |
989 | for r in self.revs(start=max(lowestrev, 0), stop=highestrev + 1): |
|
989 | for r in self.revs(start=max(lowestrev, 0), stop=highestrev + 1): | |
990 | n = self.node(r) |
|
990 | n = self.node(r) | |
991 | isdescendant = False |
|
991 | isdescendant = False | |
992 | if lowestrev == nullrev: # Everybody is a descendant of nullid |
|
992 | if lowestrev == nullrev: # Everybody is a descendant of nullid | |
993 | isdescendant = True |
|
993 | isdescendant = True | |
994 | elif n in descendants: |
|
994 | elif n in descendants: | |
995 | # n is already a descendant |
|
995 | # n is already a descendant | |
996 | isdescendant = True |
|
996 | isdescendant = True | |
997 | # This check only needs to be done here because all the roots |
|
997 | # This check only needs to be done here because all the roots | |
998 | # will start being marked is descendants before the loop. |
|
998 | # will start being marked is descendants before the loop. | |
999 | if n in roots: |
|
999 | if n in roots: | |
1000 | # If n was a root, check if it's a 'real' root. |
|
1000 | # If n was a root, check if it's a 'real' root. | |
1001 | p = tuple(self.parents(n)) |
|
1001 | p = tuple(self.parents(n)) | |
1002 | # If any of its parents are descendants, it's not a root. |
|
1002 | # If any of its parents are descendants, it's not a root. | |
1003 | if (p[0] in descendants) or (p[1] in descendants): |
|
1003 | if (p[0] in descendants) or (p[1] in descendants): | |
1004 | roots.remove(n) |
|
1004 | roots.remove(n) | |
1005 | else: |
|
1005 | else: | |
1006 | p = tuple(self.parents(n)) |
|
1006 | p = tuple(self.parents(n)) | |
1007 | # A node is a descendant if either of its parents are |
|
1007 | # A node is a descendant if either of its parents are | |
1008 | # descendants. (We seeded the dependents list with the roots |
|
1008 | # descendants. (We seeded the dependents list with the roots | |
1009 | # up there, remember?) |
|
1009 | # up there, remember?) | |
1010 | if (p[0] in descendants) or (p[1] in descendants): |
|
1010 | if (p[0] in descendants) or (p[1] in descendants): | |
1011 | descendants.add(n) |
|
1011 | descendants.add(n) | |
1012 | isdescendant = True |
|
1012 | isdescendant = True | |
1013 | if isdescendant and ((ancestors is None) or (n in ancestors)): |
|
1013 | if isdescendant and ((ancestors is None) or (n in ancestors)): | |
1014 | # Only include nodes that are both descendants and ancestors. |
|
1014 | # Only include nodes that are both descendants and ancestors. | |
1015 | orderedout.append(n) |
|
1015 | orderedout.append(n) | |
1016 | if (ancestors is not None) and (n in heads): |
|
1016 | if (ancestors is not None) and (n in heads): | |
1017 | # We're trying to figure out which heads are reachable |
|
1017 | # We're trying to figure out which heads are reachable | |
1018 | # from roots. |
|
1018 | # from roots. | |
1019 | # Mark this head as having been reached |
|
1019 | # Mark this head as having been reached | |
1020 | heads[n] = True |
|
1020 | heads[n] = True | |
1021 | elif ancestors is None: |
|
1021 | elif ancestors is None: | |
1022 | # Otherwise, we're trying to discover the heads. |
|
1022 | # Otherwise, we're trying to discover the heads. | |
1023 | # Assume this is a head because if it isn't, the next step |
|
1023 | # Assume this is a head because if it isn't, the next step | |
1024 | # will eventually remove it. |
|
1024 | # will eventually remove it. | |
1025 | heads[n] = True |
|
1025 | heads[n] = True | |
1026 | # But, obviously its parents aren't. |
|
1026 | # But, obviously its parents aren't. | |
1027 | for p in self.parents(n): |
|
1027 | for p in self.parents(n): | |
1028 | heads.pop(p, None) |
|
1028 | heads.pop(p, None) | |
1029 | heads = [head for head, flag in heads.iteritems() if flag] |
|
1029 | heads = [head for head, flag in heads.iteritems() if flag] | |
1030 | roots = list(roots) |
|
1030 | roots = list(roots) | |
1031 | assert orderedout |
|
1031 | assert orderedout | |
1032 | assert roots |
|
1032 | assert roots | |
1033 | assert heads |
|
1033 | assert heads | |
1034 | return (orderedout, roots, heads) |
|
1034 | return (orderedout, roots, heads) | |
1035 |
|
1035 | |||
1036 | def headrevs(self): |
|
1036 | def headrevs(self): | |
1037 | try: |
|
1037 | try: | |
1038 | return self.index.headrevs() |
|
1038 | return self.index.headrevs() | |
1039 | except AttributeError: |
|
1039 | except AttributeError: | |
1040 | return self._headrevs() |
|
1040 | return self._headrevs() | |
1041 |
|
1041 | |||
1042 | def computephases(self, roots): |
|
1042 | def computephases(self, roots): | |
1043 | return self.index.computephasesmapsets(roots) |
|
1043 | return self.index.computephasesmapsets(roots) | |
1044 |
|
1044 | |||
1045 | def _headrevs(self): |
|
1045 | def _headrevs(self): | |
1046 | count = len(self) |
|
1046 | count = len(self) | |
1047 | if not count: |
|
1047 | if not count: | |
1048 | return [nullrev] |
|
1048 | return [nullrev] | |
1049 | # we won't iter over filtered rev so nobody is a head at start |
|
1049 | # we won't iter over filtered rev so nobody is a head at start | |
1050 | ishead = [0] * (count + 1) |
|
1050 | ishead = [0] * (count + 1) | |
1051 | index = self.index |
|
1051 | index = self.index | |
1052 | for r in self: |
|
1052 | for r in self: | |
1053 | ishead[r] = 1 # I may be an head |
|
1053 | ishead[r] = 1 # I may be an head | |
1054 | e = index[r] |
|
1054 | e = index[r] | |
1055 | ishead[e[5]] = ishead[e[6]] = 0 # my parent are not |
|
1055 | ishead[e[5]] = ishead[e[6]] = 0 # my parent are not | |
1056 | return [r for r, val in enumerate(ishead) if val] |
|
1056 | return [r for r, val in enumerate(ishead) if val] | |
1057 |
|
1057 | |||
1058 | def heads(self, start=None, stop=None): |
|
1058 | def heads(self, start=None, stop=None): | |
1059 | """return the list of all nodes that have no children |
|
1059 | """return the list of all nodes that have no children | |
1060 |
|
1060 | |||
1061 | if start is specified, only heads that are descendants of |
|
1061 | if start is specified, only heads that are descendants of | |
1062 | start will be returned |
|
1062 | start will be returned | |
1063 | if stop is specified, it will consider all the revs from stop |
|
1063 | if stop is specified, it will consider all the revs from stop | |
1064 | as if they had no children |
|
1064 | as if they had no children | |
1065 | """ |
|
1065 | """ | |
1066 | if start is None and stop is None: |
|
1066 | if start is None and stop is None: | |
1067 | if not len(self): |
|
1067 | if not len(self): | |
1068 | return [nullid] |
|
1068 | return [nullid] | |
1069 | return [self.node(r) for r in self.headrevs()] |
|
1069 | return [self.node(r) for r in self.headrevs()] | |
1070 |
|
1070 | |||
1071 | if start is None: |
|
1071 | if start is None: | |
1072 | start = nullrev |
|
1072 | start = nullrev | |
1073 | else: |
|
1073 | else: | |
1074 | start = self.rev(start) |
|
1074 | start = self.rev(start) | |
1075 |
|
1075 | |||
1076 | stoprevs = set(self.rev(n) for n in stop or []) |
|
1076 | stoprevs = set(self.rev(n) for n in stop or []) | |
1077 |
|
1077 | |||
1078 | revs = dagop.headrevssubset(self.revs, self.parentrevs, startrev=start, |
|
1078 | revs = dagop.headrevssubset(self.revs, self.parentrevs, startrev=start, | |
1079 | stoprevs=stoprevs) |
|
1079 | stoprevs=stoprevs) | |
1080 |
|
1080 | |||
1081 | return [self.node(rev) for rev in revs] |
|
1081 | return [self.node(rev) for rev in revs] | |
1082 |
|
1082 | |||
1083 | def children(self, node): |
|
1083 | def children(self, node): | |
1084 | """find the children of a given node""" |
|
1084 | """find the children of a given node""" | |
1085 | c = [] |
|
1085 | c = [] | |
1086 | p = self.rev(node) |
|
1086 | p = self.rev(node) | |
1087 | for r in self.revs(start=p + 1): |
|
1087 | for r in self.revs(start=p + 1): | |
1088 | prevs = [pr for pr in self.parentrevs(r) if pr != nullrev] |
|
1088 | prevs = [pr for pr in self.parentrevs(r) if pr != nullrev] | |
1089 | if prevs: |
|
1089 | if prevs: | |
1090 | for pr in prevs: |
|
1090 | for pr in prevs: | |
1091 | if pr == p: |
|
1091 | if pr == p: | |
1092 | c.append(self.node(r)) |
|
1092 | c.append(self.node(r)) | |
1093 | elif p == nullrev: |
|
1093 | elif p == nullrev: | |
1094 | c.append(self.node(r)) |
|
1094 | c.append(self.node(r)) | |
1095 | return c |
|
1095 | return c | |
1096 |
|
1096 | |||
1097 | def commonancestorsheads(self, a, b): |
|
1097 | def commonancestorsheads(self, a, b): | |
1098 | """calculate all the heads of the common ancestors of nodes a and b""" |
|
1098 | """calculate all the heads of the common ancestors of nodes a and b""" | |
1099 | a, b = self.rev(a), self.rev(b) |
|
1099 | a, b = self.rev(a), self.rev(b) | |
1100 | ancs = self._commonancestorsheads(a, b) |
|
1100 | ancs = self._commonancestorsheads(a, b) | |
1101 | return pycompat.maplist(self.node, ancs) |
|
1101 | return pycompat.maplist(self.node, ancs) | |
1102 |
|
1102 | |||
1103 | def _commonancestorsheads(self, *revs): |
|
1103 | def _commonancestorsheads(self, *revs): | |
1104 | """calculate all the heads of the common ancestors of revs""" |
|
1104 | """calculate all the heads of the common ancestors of revs""" | |
1105 | try: |
|
1105 | try: | |
1106 | ancs = self.index.commonancestorsheads(*revs) |
|
1106 | ancs = self.index.commonancestorsheads(*revs) | |
1107 | except (AttributeError, OverflowError): # C implementation failed |
|
1107 | except (AttributeError, OverflowError): # C implementation failed | |
1108 | ancs = ancestor.commonancestorsheads(self.parentrevs, *revs) |
|
1108 | ancs = ancestor.commonancestorsheads(self.parentrevs, *revs) | |
1109 | return ancs |
|
1109 | return ancs | |
1110 |
|
1110 | |||
1111 | def isancestor(self, a, b): |
|
1111 | def isancestor(self, a, b): | |
1112 | """return True if node a is an ancestor of node b |
|
1112 | """return True if node a is an ancestor of node b | |
1113 |
|
1113 | |||
1114 | A revision is considered an ancestor of itself.""" |
|
1114 | A revision is considered an ancestor of itself.""" | |
1115 | a, b = self.rev(a), self.rev(b) |
|
1115 | a, b = self.rev(a), self.rev(b) | |
1116 | return self.isancestorrev(a, b) |
|
1116 | return self.isancestorrev(a, b) | |
1117 |
|
1117 | |||
1118 | def isancestorrev(self, a, b): |
|
1118 | def isancestorrev(self, a, b): | |
1119 | """return True if revision a is an ancestor of revision b |
|
1119 | """return True if revision a is an ancestor of revision b | |
1120 |
|
1120 | |||
1121 | A revision is considered an ancestor of itself. |
|
1121 | A revision is considered an ancestor of itself. | |
1122 |
|
1122 | |||
1123 | The implementation of this is trivial but the use of |
|
1123 | The implementation of this is trivial but the use of | |
1124 | commonancestorsheads is not.""" |
|
1124 | commonancestorsheads is not.""" | |
1125 | if a == nullrev: |
|
1125 | if a == nullrev: | |
1126 | return True |
|
1126 | return True | |
1127 | elif a == b: |
|
1127 | elif a == b: | |
1128 | return True |
|
1128 | return True | |
1129 | elif a > b: |
|
1129 | elif a > b: | |
1130 | return False |
|
1130 | return False | |
1131 | return a in self._commonancestorsheads(a, b) |
|
1131 | return a in self._commonancestorsheads(a, b) | |
1132 |
|
1132 | |||
1133 | def ancestor(self, a, b): |
|
1133 | def ancestor(self, a, b): | |
1134 | """calculate the "best" common ancestor of nodes a and b""" |
|
1134 | """calculate the "best" common ancestor of nodes a and b""" | |
1135 |
|
1135 | |||
1136 | a, b = self.rev(a), self.rev(b) |
|
1136 | a, b = self.rev(a), self.rev(b) | |
1137 | try: |
|
1137 | try: | |
1138 | ancs = self.index.ancestors(a, b) |
|
1138 | ancs = self.index.ancestors(a, b) | |
1139 | except (AttributeError, OverflowError): |
|
1139 | except (AttributeError, OverflowError): | |
1140 | ancs = ancestor.ancestors(self.parentrevs, a, b) |
|
1140 | ancs = ancestor.ancestors(self.parentrevs, a, b) | |
1141 | if ancs: |
|
1141 | if ancs: | |
1142 | # choose a consistent winner when there's a tie |
|
1142 | # choose a consistent winner when there's a tie | |
1143 | return min(map(self.node, ancs)) |
|
1143 | return min(map(self.node, ancs)) | |
1144 | return nullid |
|
1144 | return nullid | |
1145 |
|
1145 | |||
1146 | def _match(self, id): |
|
1146 | def _match(self, id): | |
1147 | if isinstance(id, int): |
|
1147 | if isinstance(id, int): | |
1148 | # rev |
|
1148 | # rev | |
1149 | return self.node(id) |
|
1149 | return self.node(id) | |
1150 | if len(id) == 20: |
|
1150 | if len(id) == 20: | |
1151 | # possibly a binary node |
|
1151 | # possibly a binary node | |
1152 | # odds of a binary node being all hex in ASCII are 1 in 10**25 |
|
1152 | # odds of a binary node being all hex in ASCII are 1 in 10**25 | |
1153 | try: |
|
1153 | try: | |
1154 | node = id |
|
1154 | node = id | |
1155 | self.rev(node) # quick search the index |
|
1155 | self.rev(node) # quick search the index | |
1156 | return node |
|
1156 | return node | |
1157 | except error.LookupError: |
|
1157 | except error.LookupError: | |
1158 | pass # may be partial hex id |
|
1158 | pass # may be partial hex id | |
1159 | try: |
|
1159 | try: | |
1160 | # str(rev) |
|
1160 | # str(rev) | |
1161 | rev = int(id) |
|
1161 | rev = int(id) | |
1162 | if "%d" % rev != id: |
|
1162 | if "%d" % rev != id: | |
1163 | raise ValueError |
|
1163 | raise ValueError | |
1164 | if rev < 0: |
|
1164 | if rev < 0: | |
1165 | rev = len(self) + rev |
|
1165 | rev = len(self) + rev | |
1166 | if rev < 0 or rev >= len(self): |
|
1166 | if rev < 0 or rev >= len(self): | |
1167 | raise ValueError |
|
1167 | raise ValueError | |
1168 | return self.node(rev) |
|
1168 | return self.node(rev) | |
1169 | except (ValueError, OverflowError): |
|
1169 | except (ValueError, OverflowError): | |
1170 | pass |
|
1170 | pass | |
1171 | if len(id) == 40: |
|
1171 | if len(id) == 40: | |
1172 | try: |
|
1172 | try: | |
1173 | # a full hex nodeid? |
|
1173 | # a full hex nodeid? | |
1174 | node = bin(id) |
|
1174 | node = bin(id) | |
1175 | self.rev(node) |
|
1175 | self.rev(node) | |
1176 | return node |
|
1176 | return node | |
1177 | except (TypeError, error.LookupError): |
|
1177 | except (TypeError, error.LookupError): | |
1178 | pass |
|
1178 | pass | |
1179 |
|
1179 | |||
1180 | def _partialmatch(self, id): |
|
1180 | def _partialmatch(self, id): | |
1181 | # we don't care wdirfilenodeids as they should be always full hash |
|
1181 | # we don't care wdirfilenodeids as they should be always full hash | |
1182 | maybewdir = wdirhex.startswith(id) |
|
1182 | maybewdir = wdirhex.startswith(id) | |
1183 | try: |
|
1183 | try: | |
1184 | partial = self.index.partialmatch(id) |
|
1184 | partial = self.index.partialmatch(id) | |
1185 | if partial and self.hasnode(partial): |
|
1185 | if partial and self.hasnode(partial): | |
1186 | if maybewdir: |
|
1186 | if maybewdir: | |
1187 | # single 'ff...' match in radix tree, ambiguous with wdir |
|
1187 | # single 'ff...' match in radix tree, ambiguous with wdir | |
1188 | raise error.RevlogError |
|
1188 | raise error.RevlogError | |
1189 | return partial |
|
1189 | return partial | |
1190 | if maybewdir: |
|
1190 | if maybewdir: | |
1191 | # no 'ff...' match in radix tree, wdir identified |
|
1191 | # no 'ff...' match in radix tree, wdir identified | |
1192 | raise error.WdirUnsupported |
|
1192 | raise error.WdirUnsupported | |
1193 | return None |
|
1193 | return None | |
1194 | except error.RevlogError: |
|
1194 | except error.RevlogError: | |
1195 | # parsers.c radix tree lookup gave multiple matches |
|
1195 | # parsers.c radix tree lookup gave multiple matches | |
1196 | # fast path: for unfiltered changelog, radix tree is accurate |
|
1196 | # fast path: for unfiltered changelog, radix tree is accurate | |
1197 | if not getattr(self, 'filteredrevs', None): |
|
1197 | if not getattr(self, 'filteredrevs', None): | |
1198 | raise error.AmbiguousPrefixLookupError( |
|
1198 | raise error.AmbiguousPrefixLookupError( | |
1199 | id, self.indexfile, _('ambiguous identifier')) |
|
1199 | id, self.indexfile, _('ambiguous identifier')) | |
1200 | # fall through to slow path that filters hidden revisions |
|
1200 | # fall through to slow path that filters hidden revisions | |
1201 | except (AttributeError, ValueError): |
|
1201 | except (AttributeError, ValueError): | |
1202 | # we are pure python, or key was too short to search radix tree |
|
1202 | # we are pure python, or key was too short to search radix tree | |
1203 | pass |
|
1203 | pass | |
1204 |
|
1204 | |||
1205 | if id in self._pcache: |
|
1205 | if id in self._pcache: | |
1206 | return self._pcache[id] |
|
1206 | return self._pcache[id] | |
1207 |
|
1207 | |||
1208 | if len(id) <= 40: |
|
1208 | if len(id) <= 40: | |
1209 | try: |
|
1209 | try: | |
1210 | # hex(node)[:...] |
|
1210 | # hex(node)[:...] | |
1211 | l = len(id) // 2 # grab an even number of digits |
|
1211 | l = len(id) // 2 # grab an even number of digits | |
1212 | prefix = bin(id[:l * 2]) |
|
1212 | prefix = bin(id[:l * 2]) | |
1213 | nl = [e[7] for e in self.index if e[7].startswith(prefix)] |
|
1213 | nl = [e[7] for e in self.index if e[7].startswith(prefix)] | |
1214 | nl = [n for n in nl if hex(n).startswith(id) and |
|
1214 | nl = [n for n in nl if hex(n).startswith(id) and | |
1215 | self.hasnode(n)] |
|
1215 | self.hasnode(n)] | |
1216 | if nullhex.startswith(id): |
|
1216 | if nullhex.startswith(id): | |
1217 | nl.append(nullid) |
|
1217 | nl.append(nullid) | |
1218 | if len(nl) > 0: |
|
1218 | if len(nl) > 0: | |
1219 | if len(nl) == 1 and not maybewdir: |
|
1219 | if len(nl) == 1 and not maybewdir: | |
1220 | self._pcache[id] = nl[0] |
|
1220 | self._pcache[id] = nl[0] | |
1221 | return nl[0] |
|
1221 | return nl[0] | |
1222 | raise error.AmbiguousPrefixLookupError( |
|
1222 | raise error.AmbiguousPrefixLookupError( | |
1223 | id, self.indexfile, _('ambiguous identifier')) |
|
1223 | id, self.indexfile, _('ambiguous identifier')) | |
1224 | if maybewdir: |
|
1224 | if maybewdir: | |
1225 | raise error.WdirUnsupported |
|
1225 | raise error.WdirUnsupported | |
1226 | return None |
|
1226 | return None | |
1227 | except TypeError: |
|
1227 | except TypeError: | |
1228 | pass |
|
1228 | pass | |
1229 |
|
1229 | |||
1230 | def lookup(self, id): |
|
1230 | def lookup(self, id): | |
1231 | """locate a node based on: |
|
1231 | """locate a node based on: | |
1232 | - revision number or str(revision number) |
|
1232 | - revision number or str(revision number) | |
1233 | - nodeid or subset of hex nodeid |
|
1233 | - nodeid or subset of hex nodeid | |
1234 | """ |
|
1234 | """ | |
1235 | n = self._match(id) |
|
1235 | n = self._match(id) | |
1236 | if n is not None: |
|
1236 | if n is not None: | |
1237 | return n |
|
1237 | return n | |
1238 | n = self._partialmatch(id) |
|
1238 | n = self._partialmatch(id) | |
1239 | if n: |
|
1239 | if n: | |
1240 | return n |
|
1240 | return n | |
1241 |
|
1241 | |||
1242 | raise error.LookupError(id, self.indexfile, _('no match found')) |
|
1242 | raise error.LookupError(id, self.indexfile, _('no match found')) | |
1243 |
|
1243 | |||
1244 | def shortest(self, node, minlength=1): |
|
1244 | def shortest(self, node, minlength=1): | |
1245 | """Find the shortest unambiguous prefix that matches node.""" |
|
1245 | """Find the shortest unambiguous prefix that matches node.""" | |
1246 | def isvalid(prefix): |
|
1246 | def isvalid(prefix): | |
1247 | try: |
|
1247 | try: | |
1248 | node = self._partialmatch(prefix) |
|
1248 | node = self._partialmatch(prefix) | |
1249 | except error.AmbiguousPrefixLookupError: |
|
1249 | except error.AmbiguousPrefixLookupError: | |
1250 | return False |
|
1250 | return False | |
1251 | except error.WdirUnsupported: |
|
1251 | except error.WdirUnsupported: | |
1252 | # single 'ff...' match |
|
1252 | # single 'ff...' match | |
1253 | return True |
|
1253 | return True | |
1254 | if node is None: |
|
1254 | if node is None: | |
1255 | raise error.LookupError(node, self.indexfile, _('no node')) |
|
1255 | raise error.LookupError(node, self.indexfile, _('no node')) | |
1256 | return True |
|
1256 | return True | |
1257 |
|
1257 | |||
1258 | def maybewdir(prefix): |
|
1258 | def maybewdir(prefix): | |
1259 | return all(c == 'f' for c in prefix) |
|
1259 | return all(c == 'f' for c in prefix) | |
1260 |
|
1260 | |||
1261 | hexnode = hex(node) |
|
1261 | hexnode = hex(node) | |
1262 |
|
1262 | |||
1263 | def disambiguate(hexnode, minlength): |
|
1263 | def disambiguate(hexnode, minlength): | |
1264 | """Disambiguate against wdirid.""" |
|
1264 | """Disambiguate against wdirid.""" | |
1265 | for length in range(minlength, 41): |
|
1265 | for length in range(minlength, 41): | |
1266 | prefix = hexnode[:length] |
|
1266 | prefix = hexnode[:length] | |
1267 | if not maybewdir(prefix): |
|
1267 | if not maybewdir(prefix): | |
1268 | return prefix |
|
1268 | return prefix | |
1269 |
|
1269 | |||
1270 | if not getattr(self, 'filteredrevs', None): |
|
1270 | if not getattr(self, 'filteredrevs', None): | |
1271 | try: |
|
1271 | try: | |
1272 | length = max(self.index.shortest(node), minlength) |
|
1272 | length = max(self.index.shortest(node), minlength) | |
1273 | return disambiguate(hexnode, length) |
|
1273 | return disambiguate(hexnode, length) | |
1274 | except error.RevlogError: |
|
1274 | except error.RevlogError: | |
1275 | if node != wdirid: |
|
1275 | if node != wdirid: | |
1276 | raise error.LookupError(node, self.indexfile, _('no node')) |
|
1276 | raise error.LookupError(node, self.indexfile, _('no node')) | |
1277 | except AttributeError: |
|
1277 | except AttributeError: | |
1278 | # Fall through to pure code |
|
1278 | # Fall through to pure code | |
1279 | pass |
|
1279 | pass | |
1280 |
|
1280 | |||
1281 | if node == wdirid: |
|
1281 | if node == wdirid: | |
1282 | for length in range(minlength, 41): |
|
1282 | for length in range(minlength, 41): | |
1283 | prefix = hexnode[:length] |
|
1283 | prefix = hexnode[:length] | |
1284 | if isvalid(prefix): |
|
1284 | if isvalid(prefix): | |
1285 | return prefix |
|
1285 | return prefix | |
1286 |
|
1286 | |||
1287 | for length in range(minlength, 41): |
|
1287 | for length in range(minlength, 41): | |
1288 | prefix = hexnode[:length] |
|
1288 | prefix = hexnode[:length] | |
1289 | if isvalid(prefix): |
|
1289 | if isvalid(prefix): | |
1290 | return disambiguate(hexnode, length) |
|
1290 | return disambiguate(hexnode, length) | |
1291 |
|
1291 | |||
1292 | def cmp(self, node, text): |
|
1292 | def cmp(self, node, text): | |
1293 | """compare text with a given file revision |
|
1293 | """compare text with a given file revision | |
1294 |
|
1294 | |||
1295 | returns True if text is different than what is stored. |
|
1295 | returns True if text is different than what is stored. | |
1296 | """ |
|
1296 | """ | |
1297 | p1, p2 = self.parents(node) |
|
1297 | p1, p2 = self.parents(node) | |
1298 | return storageutil.hashrevisionsha1(text, p1, p2) != node |
|
1298 | return storageutil.hashrevisionsha1(text, p1, p2) != node | |
1299 |
|
1299 | |||
1300 | def _cachesegment(self, offset, data): |
|
1300 | def _cachesegment(self, offset, data): | |
1301 | """Add a segment to the revlog cache. |
|
1301 | """Add a segment to the revlog cache. | |
1302 |
|
1302 | |||
1303 | Accepts an absolute offset and the data that is at that location. |
|
1303 | Accepts an absolute offset and the data that is at that location. | |
1304 | """ |
|
1304 | """ | |
1305 | o, d = self._chunkcache |
|
1305 | o, d = self._chunkcache | |
1306 | # try to add to existing cache |
|
1306 | # try to add to existing cache | |
1307 | if o + len(d) == offset and len(d) + len(data) < _chunksize: |
|
1307 | if o + len(d) == offset and len(d) + len(data) < _chunksize: | |
1308 | self._chunkcache = o, d + data |
|
1308 | self._chunkcache = o, d + data | |
1309 | else: |
|
1309 | else: | |
1310 | self._chunkcache = offset, data |
|
1310 | self._chunkcache = offset, data | |
1311 |
|
1311 | |||
1312 | def _readsegment(self, offset, length, df=None): |
|
1312 | def _readsegment(self, offset, length, df=None): | |
1313 | """Load a segment of raw data from the revlog. |
|
1313 | """Load a segment of raw data from the revlog. | |
1314 |
|
1314 | |||
1315 | Accepts an absolute offset, length to read, and an optional existing |
|
1315 | Accepts an absolute offset, length to read, and an optional existing | |
1316 | file handle to read from. |
|
1316 | file handle to read from. | |
1317 |
|
1317 | |||
1318 | If an existing file handle is passed, it will be seeked and the |
|
1318 | If an existing file handle is passed, it will be seeked and the | |
1319 | original seek position will NOT be restored. |
|
1319 | original seek position will NOT be restored. | |
1320 |
|
1320 | |||
1321 | Returns a str or buffer of raw byte data. |
|
1321 | Returns a str or buffer of raw byte data. | |
1322 | """ |
|
1322 | """ | |
1323 | # Cache data both forward and backward around the requested |
|
1323 | # Cache data both forward and backward around the requested | |
1324 | # data, in a fixed size window. This helps speed up operations |
|
1324 | # data, in a fixed size window. This helps speed up operations | |
1325 | # involving reading the revlog backwards. |
|
1325 | # involving reading the revlog backwards. | |
1326 | cachesize = self._chunkcachesize |
|
1326 | cachesize = self._chunkcachesize | |
1327 | realoffset = offset & ~(cachesize - 1) |
|
1327 | realoffset = offset & ~(cachesize - 1) | |
1328 | reallength = (((offset + length + cachesize) & ~(cachesize - 1)) |
|
1328 | reallength = (((offset + length + cachesize) & ~(cachesize - 1)) | |
1329 | - realoffset) |
|
1329 | - realoffset) | |
1330 | with self._datareadfp(df) as df: |
|
1330 | with self._datareadfp(df) as df: | |
1331 | df.seek(realoffset) |
|
1331 | df.seek(realoffset) | |
1332 | d = df.read(reallength) |
|
1332 | d = df.read(reallength) | |
1333 | self._cachesegment(realoffset, d) |
|
1333 | self._cachesegment(realoffset, d) | |
1334 | if offset != realoffset or reallength != length: |
|
1334 | if offset != realoffset or reallength != length: | |
1335 | return util.buffer(d, offset - realoffset, length) |
|
1335 | return util.buffer(d, offset - realoffset, length) | |
1336 | return d |
|
1336 | return d | |
1337 |
|
1337 | |||
1338 | def _getsegment(self, offset, length, df=None): |
|
1338 | def _getsegment(self, offset, length, df=None): | |
1339 | """Obtain a segment of raw data from the revlog. |
|
1339 | """Obtain a segment of raw data from the revlog. | |
1340 |
|
1340 | |||
1341 | Accepts an absolute offset, length of bytes to obtain, and an |
|
1341 | Accepts an absolute offset, length of bytes to obtain, and an | |
1342 | optional file handle to the already-opened revlog. If the file |
|
1342 | optional file handle to the already-opened revlog. If the file | |
1343 | handle is used, it's original seek position will not be preserved. |
|
1343 | handle is used, it's original seek position will not be preserved. | |
1344 |
|
1344 | |||
1345 | Requests for data may be returned from a cache. |
|
1345 | Requests for data may be returned from a cache. | |
1346 |
|
1346 | |||
1347 | Returns a str or a buffer instance of raw byte data. |
|
1347 | Returns a str or a buffer instance of raw byte data. | |
1348 | """ |
|
1348 | """ | |
1349 | o, d = self._chunkcache |
|
1349 | o, d = self._chunkcache | |
1350 | l = len(d) |
|
1350 | l = len(d) | |
1351 |
|
1351 | |||
1352 | # is it in the cache? |
|
1352 | # is it in the cache? | |
1353 | cachestart = offset - o |
|
1353 | cachestart = offset - o | |
1354 | cacheend = cachestart + length |
|
1354 | cacheend = cachestart + length | |
1355 | if cachestart >= 0 and cacheend <= l: |
|
1355 | if cachestart >= 0 and cacheend <= l: | |
1356 | if cachestart == 0 and cacheend == l: |
|
1356 | if cachestart == 0 and cacheend == l: | |
1357 | return d # avoid a copy |
|
1357 | return d # avoid a copy | |
1358 | return util.buffer(d, cachestart, cacheend - cachestart) |
|
1358 | return util.buffer(d, cachestart, cacheend - cachestart) | |
1359 |
|
1359 | |||
1360 | return self._readsegment(offset, length, df=df) |
|
1360 | return self._readsegment(offset, length, df=df) | |
1361 |
|
1361 | |||
1362 | def _getsegmentforrevs(self, startrev, endrev, df=None): |
|
1362 | def _getsegmentforrevs(self, startrev, endrev, df=None): | |
1363 | """Obtain a segment of raw data corresponding to a range of revisions. |
|
1363 | """Obtain a segment of raw data corresponding to a range of revisions. | |
1364 |
|
1364 | |||
1365 | Accepts the start and end revisions and an optional already-open |
|
1365 | Accepts the start and end revisions and an optional already-open | |
1366 | file handle to be used for reading. If the file handle is read, its |
|
1366 | file handle to be used for reading. If the file handle is read, its | |
1367 | seek position will not be preserved. |
|
1367 | seek position will not be preserved. | |
1368 |
|
1368 | |||
1369 | Requests for data may be satisfied by a cache. |
|
1369 | Requests for data may be satisfied by a cache. | |
1370 |
|
1370 | |||
1371 | Returns a 2-tuple of (offset, data) for the requested range of |
|
1371 | Returns a 2-tuple of (offset, data) for the requested range of | |
1372 | revisions. Offset is the integer offset from the beginning of the |
|
1372 | revisions. Offset is the integer offset from the beginning of the | |
1373 | revlog and data is a str or buffer of the raw byte data. |
|
1373 | revlog and data is a str or buffer of the raw byte data. | |
1374 |
|
1374 | |||
1375 | Callers will need to call ``self.start(rev)`` and ``self.length(rev)`` |
|
1375 | Callers will need to call ``self.start(rev)`` and ``self.length(rev)`` | |
1376 | to determine where each revision's data begins and ends. |
|
1376 | to determine where each revision's data begins and ends. | |
1377 | """ |
|
1377 | """ | |
1378 | # Inlined self.start(startrev) & self.end(endrev) for perf reasons |
|
1378 | # Inlined self.start(startrev) & self.end(endrev) for perf reasons | |
1379 | # (functions are expensive). |
|
1379 | # (functions are expensive). | |
1380 | index = self.index |
|
1380 | index = self.index | |
1381 | istart = index[startrev] |
|
1381 | istart = index[startrev] | |
1382 | start = int(istart[0] >> 16) |
|
1382 | start = int(istart[0] >> 16) | |
1383 | if startrev == endrev: |
|
1383 | if startrev == endrev: | |
1384 | end = start + istart[1] |
|
1384 | end = start + istart[1] | |
1385 | else: |
|
1385 | else: | |
1386 | iend = index[endrev] |
|
1386 | iend = index[endrev] | |
1387 | end = int(iend[0] >> 16) + iend[1] |
|
1387 | end = int(iend[0] >> 16) + iend[1] | |
1388 |
|
1388 | |||
1389 | if self._inline: |
|
1389 | if self._inline: | |
1390 | start += (startrev + 1) * self._io.size |
|
1390 | start += (startrev + 1) * self._io.size | |
1391 | end += (endrev + 1) * self._io.size |
|
1391 | end += (endrev + 1) * self._io.size | |
1392 | length = end - start |
|
1392 | length = end - start | |
1393 |
|
1393 | |||
1394 | return start, self._getsegment(start, length, df=df) |
|
1394 | return start, self._getsegment(start, length, df=df) | |
1395 |
|
1395 | |||
1396 | def _chunk(self, rev, df=None): |
|
1396 | def _chunk(self, rev, df=None): | |
1397 | """Obtain a single decompressed chunk for a revision. |
|
1397 | """Obtain a single decompressed chunk for a revision. | |
1398 |
|
1398 | |||
1399 | Accepts an integer revision and an optional already-open file handle |
|
1399 | Accepts an integer revision and an optional already-open file handle | |
1400 | to be used for reading. If used, the seek position of the file will not |
|
1400 | to be used for reading. If used, the seek position of the file will not | |
1401 | be preserved. |
|
1401 | be preserved. | |
1402 |
|
1402 | |||
1403 | Returns a str holding uncompressed data for the requested revision. |
|
1403 | Returns a str holding uncompressed data for the requested revision. | |
1404 | """ |
|
1404 | """ | |
1405 | return self.decompress(self._getsegmentforrevs(rev, rev, df=df)[1]) |
|
1405 | return self.decompress(self._getsegmentforrevs(rev, rev, df=df)[1]) | |
1406 |
|
1406 | |||
1407 | def _chunks(self, revs, df=None, targetsize=None): |
|
1407 | def _chunks(self, revs, df=None, targetsize=None): | |
1408 | """Obtain decompressed chunks for the specified revisions. |
|
1408 | """Obtain decompressed chunks for the specified revisions. | |
1409 |
|
1409 | |||
1410 | Accepts an iterable of numeric revisions that are assumed to be in |
|
1410 | Accepts an iterable of numeric revisions that are assumed to be in | |
1411 | ascending order. Also accepts an optional already-open file handle |
|
1411 | ascending order. Also accepts an optional already-open file handle | |
1412 | to be used for reading. If used, the seek position of the file will |
|
1412 | to be used for reading. If used, the seek position of the file will | |
1413 | not be preserved. |
|
1413 | not be preserved. | |
1414 |
|
1414 | |||
1415 | This function is similar to calling ``self._chunk()`` multiple times, |
|
1415 | This function is similar to calling ``self._chunk()`` multiple times, | |
1416 | but is faster. |
|
1416 | but is faster. | |
1417 |
|
1417 | |||
1418 | Returns a list with decompressed data for each requested revision. |
|
1418 | Returns a list with decompressed data for each requested revision. | |
1419 | """ |
|
1419 | """ | |
1420 | if not revs: |
|
1420 | if not revs: | |
1421 | return [] |
|
1421 | return [] | |
1422 | start = self.start |
|
1422 | start = self.start | |
1423 | length = self.length |
|
1423 | length = self.length | |
1424 | inline = self._inline |
|
1424 | inline = self._inline | |
1425 | iosize = self._io.size |
|
1425 | iosize = self._io.size | |
1426 | buffer = util.buffer |
|
1426 | buffer = util.buffer | |
1427 |
|
1427 | |||
1428 | l = [] |
|
1428 | l = [] | |
1429 | ladd = l.append |
|
1429 | ladd = l.append | |
1430 |
|
1430 | |||
1431 | if not self._withsparseread: |
|
1431 | if not self._withsparseread: | |
1432 | slicedchunks = (revs,) |
|
1432 | slicedchunks = (revs,) | |
1433 | else: |
|
1433 | else: | |
1434 | slicedchunks = deltautil.slicechunk(self, revs, |
|
1434 | slicedchunks = deltautil.slicechunk(self, revs, | |
1435 | targetsize=targetsize) |
|
1435 | targetsize=targetsize) | |
1436 |
|
1436 | |||
1437 | for revschunk in slicedchunks: |
|
1437 | for revschunk in slicedchunks: | |
1438 | firstrev = revschunk[0] |
|
1438 | firstrev = revschunk[0] | |
1439 | # Skip trailing revisions with empty diff |
|
1439 | # Skip trailing revisions with empty diff | |
1440 | for lastrev in revschunk[::-1]: |
|
1440 | for lastrev in revschunk[::-1]: | |
1441 | if length(lastrev) != 0: |
|
1441 | if length(lastrev) != 0: | |
1442 | break |
|
1442 | break | |
1443 |
|
1443 | |||
1444 | try: |
|
1444 | try: | |
1445 | offset, data = self._getsegmentforrevs(firstrev, lastrev, df=df) |
|
1445 | offset, data = self._getsegmentforrevs(firstrev, lastrev, df=df) | |
1446 | except OverflowError: |
|
1446 | except OverflowError: | |
1447 | # issue4215 - we can't cache a run of chunks greater than |
|
1447 | # issue4215 - we can't cache a run of chunks greater than | |
1448 | # 2G on Windows |
|
1448 | # 2G on Windows | |
1449 | return [self._chunk(rev, df=df) for rev in revschunk] |
|
1449 | return [self._chunk(rev, df=df) for rev in revschunk] | |
1450 |
|
1450 | |||
1451 | decomp = self.decompress |
|
1451 | decomp = self.decompress | |
1452 | for rev in revschunk: |
|
1452 | for rev in revschunk: | |
1453 | chunkstart = start(rev) |
|
1453 | chunkstart = start(rev) | |
1454 | if inline: |
|
1454 | if inline: | |
1455 | chunkstart += (rev + 1) * iosize |
|
1455 | chunkstart += (rev + 1) * iosize | |
1456 | chunklength = length(rev) |
|
1456 | chunklength = length(rev) | |
1457 | ladd(decomp(buffer(data, chunkstart - offset, chunklength))) |
|
1457 | ladd(decomp(buffer(data, chunkstart - offset, chunklength))) | |
1458 |
|
1458 | |||
1459 | return l |
|
1459 | return l | |
1460 |
|
1460 | |||
1461 | def _chunkclear(self): |
|
1461 | def _chunkclear(self): | |
1462 | """Clear the raw chunk cache.""" |
|
1462 | """Clear the raw chunk cache.""" | |
1463 | self._chunkcache = (0, '') |
|
1463 | self._chunkcache = (0, '') | |
1464 |
|
1464 | |||
1465 | def deltaparent(self, rev): |
|
1465 | def deltaparent(self, rev): | |
1466 | """return deltaparent of the given revision""" |
|
1466 | """return deltaparent of the given revision""" | |
1467 | base = self.index[rev][3] |
|
1467 | base = self.index[rev][3] | |
1468 | if base == rev: |
|
1468 | if base == rev: | |
1469 | return nullrev |
|
1469 | return nullrev | |
1470 | elif self._generaldelta: |
|
1470 | elif self._generaldelta: | |
1471 | return base |
|
1471 | return base | |
1472 | else: |
|
1472 | else: | |
1473 | return rev - 1 |
|
1473 | return rev - 1 | |
1474 |
|
1474 | |||
1475 | def issnapshot(self, rev): |
|
1475 | def issnapshot(self, rev): | |
1476 | """tells whether rev is a snapshot |
|
1476 | """tells whether rev is a snapshot | |
1477 | """ |
|
1477 | """ | |
1478 | if rev == nullrev: |
|
1478 | if rev == nullrev: | |
1479 | return True |
|
1479 | return True | |
1480 | deltap = self.deltaparent(rev) |
|
1480 | deltap = self.deltaparent(rev) | |
1481 | if deltap == nullrev: |
|
1481 | if deltap == nullrev: | |
1482 | return True |
|
1482 | return True | |
1483 | p1, p2 = self.parentrevs(rev) |
|
1483 | p1, p2 = self.parentrevs(rev) | |
1484 | if deltap in (p1, p2): |
|
1484 | if deltap in (p1, p2): | |
1485 | return False |
|
1485 | return False | |
1486 | return self.issnapshot(deltap) |
|
1486 | return self.issnapshot(deltap) | |
1487 |
|
1487 | |||
1488 | def snapshotdepth(self, rev): |
|
1488 | def snapshotdepth(self, rev): | |
1489 | """number of snapshot in the chain before this one""" |
|
1489 | """number of snapshot in the chain before this one""" | |
1490 | if not self.issnapshot(rev): |
|
1490 | if not self.issnapshot(rev): | |
1491 | raise error.ProgrammingError('revision %d not a snapshot') |
|
1491 | raise error.ProgrammingError('revision %d not a snapshot') | |
1492 | return len(self._deltachain(rev)[0]) - 1 |
|
1492 | return len(self._deltachain(rev)[0]) - 1 | |
1493 |
|
1493 | |||
1494 | def revdiff(self, rev1, rev2): |
|
1494 | def revdiff(self, rev1, rev2): | |
1495 | """return or calculate a delta between two revisions |
|
1495 | """return or calculate a delta between two revisions | |
1496 |
|
1496 | |||
1497 | The delta calculated is in binary form and is intended to be written to |
|
1497 | The delta calculated is in binary form and is intended to be written to | |
1498 | revlog data directly. So this function needs raw revision data. |
|
1498 | revlog data directly. So this function needs raw revision data. | |
1499 | """ |
|
1499 | """ | |
1500 | if rev1 != nullrev and self.deltaparent(rev2) == rev1: |
|
1500 | if rev1 != nullrev and self.deltaparent(rev2) == rev1: | |
1501 | return bytes(self._chunk(rev2)) |
|
1501 | return bytes(self._chunk(rev2)) | |
1502 |
|
1502 | |||
1503 | return mdiff.textdiff(self.revision(rev1, raw=True), |
|
1503 | return mdiff.textdiff(self.revision(rev1, raw=True), | |
1504 | self.revision(rev2, raw=True)) |
|
1504 | self.revision(rev2, raw=True)) | |
1505 |
|
1505 | |||
1506 | def revision(self, nodeorrev, _df=None, raw=False): |
|
1506 | def revision(self, nodeorrev, _df=None, raw=False): | |
1507 | """return an uncompressed revision of a given node or revision |
|
1507 | """return an uncompressed revision of a given node or revision | |
1508 | number. |
|
1508 | number. | |
1509 |
|
1509 | |||
1510 | _df - an existing file handle to read from. (internal-only) |
|
1510 | _df - an existing file handle to read from. (internal-only) | |
1511 | raw - an optional argument specifying if the revision data is to be |
|
1511 | raw - an optional argument specifying if the revision data is to be | |
1512 | treated as raw data when applying flag transforms. 'raw' should be set |
|
1512 | treated as raw data when applying flag transforms. 'raw' should be set | |
1513 | to True when generating changegroups or in debug commands. |
|
1513 | to True when generating changegroups or in debug commands. | |
1514 | """ |
|
1514 | """ | |
1515 | if isinstance(nodeorrev, int): |
|
1515 | if isinstance(nodeorrev, int): | |
1516 | rev = nodeorrev |
|
1516 | rev = nodeorrev | |
1517 | node = self.node(rev) |
|
1517 | node = self.node(rev) | |
1518 | else: |
|
1518 | else: | |
1519 | node = nodeorrev |
|
1519 | node = nodeorrev | |
1520 | rev = None |
|
1520 | rev = None | |
1521 |
|
1521 | |||
1522 | cachedrev = None |
|
1522 | cachedrev = None | |
1523 | flags = None |
|
1523 | flags = None | |
1524 | rawtext = None |
|
1524 | rawtext = None | |
1525 | if node == nullid: |
|
1525 | if node == nullid: | |
1526 | return "" |
|
1526 | return "" | |
1527 | if self._cache: |
|
1527 | if self._cache: | |
1528 | if self._cache[0] == node: |
|
1528 | if self._cache[0] == node: | |
1529 | # _cache only stores rawtext |
|
1529 | # _cache only stores rawtext | |
1530 | if raw: |
|
1530 | if raw: | |
1531 | return self._cache[2] |
|
1531 | return self._cache[2] | |
1532 | # duplicated, but good for perf |
|
1532 | # duplicated, but good for perf | |
1533 | if rev is None: |
|
1533 | if rev is None: | |
1534 | rev = self.rev(node) |
|
1534 | rev = self.rev(node) | |
1535 | if flags is None: |
|
1535 | if flags is None: | |
1536 | flags = self.flags(rev) |
|
1536 | flags = self.flags(rev) | |
1537 | # no extra flags set, no flag processor runs, text = rawtext |
|
1537 | # no extra flags set, no flag processor runs, text = rawtext | |
1538 | if flags == REVIDX_DEFAULT_FLAGS: |
|
1538 | if flags == REVIDX_DEFAULT_FLAGS: | |
1539 | return self._cache[2] |
|
1539 | return self._cache[2] | |
1540 | # rawtext is reusable. need to run flag processor |
|
1540 | # rawtext is reusable. need to run flag processor | |
1541 | rawtext = self._cache[2] |
|
1541 | rawtext = self._cache[2] | |
1542 |
|
1542 | |||
1543 | cachedrev = self._cache[1] |
|
1543 | cachedrev = self._cache[1] | |
1544 |
|
1544 | |||
1545 | # look up what we need to read |
|
1545 | # look up what we need to read | |
1546 | if rawtext is None: |
|
1546 | if rawtext is None: | |
1547 | if rev is None: |
|
1547 | if rev is None: | |
1548 | rev = self.rev(node) |
|
1548 | rev = self.rev(node) | |
1549 |
|
1549 | |||
1550 | chain, stopped = self._deltachain(rev, stoprev=cachedrev) |
|
1550 | chain, stopped = self._deltachain(rev, stoprev=cachedrev) | |
1551 | if stopped: |
|
1551 | if stopped: | |
1552 | rawtext = self._cache[2] |
|
1552 | rawtext = self._cache[2] | |
1553 |
|
1553 | |||
1554 | # drop cache to save memory |
|
1554 | # drop cache to save memory | |
1555 | self._cache = None |
|
1555 | self._cache = None | |
1556 |
|
1556 | |||
1557 | targetsize = None |
|
1557 | targetsize = None | |
1558 | rawsize = self.index[rev][2] |
|
1558 | rawsize = self.index[rev][2] | |
1559 | if 0 <= rawsize: |
|
1559 | if 0 <= rawsize: | |
1560 | targetsize = 4 * rawsize |
|
1560 | targetsize = 4 * rawsize | |
1561 |
|
1561 | |||
1562 | bins = self._chunks(chain, df=_df, targetsize=targetsize) |
|
1562 | bins = self._chunks(chain, df=_df, targetsize=targetsize) | |
1563 | if rawtext is None: |
|
1563 | if rawtext is None: | |
1564 | rawtext = bytes(bins[0]) |
|
1564 | rawtext = bytes(bins[0]) | |
1565 | bins = bins[1:] |
|
1565 | bins = bins[1:] | |
1566 |
|
1566 | |||
1567 | rawtext = mdiff.patches(rawtext, bins) |
|
1567 | rawtext = mdiff.patches(rawtext, bins) | |
1568 | self._cache = (node, rev, rawtext) |
|
1568 | self._cache = (node, rev, rawtext) | |
1569 |
|
1569 | |||
1570 | if flags is None: |
|
1570 | if flags is None: | |
1571 | if rev is None: |
|
1571 | if rev is None: | |
1572 | rev = self.rev(node) |
|
1572 | rev = self.rev(node) | |
1573 | flags = self.flags(rev) |
|
1573 | flags = self.flags(rev) | |
1574 |
|
1574 | |||
1575 | text, validatehash = self._processflags(rawtext, flags, 'read', raw=raw) |
|
1575 | text, validatehash = self._processflags(rawtext, flags, 'read', raw=raw) | |
1576 | if validatehash: |
|
1576 | if validatehash: | |
1577 | self.checkhash(text, node, rev=rev) |
|
1577 | self.checkhash(text, node, rev=rev) | |
1578 |
|
1578 | |||
1579 | return text |
|
1579 | return text | |
1580 |
|
1580 | |||
1581 | def hash(self, text, p1, p2): |
|
1581 | def hash(self, text, p1, p2): | |
1582 | """Compute a node hash. |
|
1582 | """Compute a node hash. | |
1583 |
|
1583 | |||
1584 | Available as a function so that subclasses can replace the hash |
|
1584 | Available as a function so that subclasses can replace the hash | |
1585 | as needed. |
|
1585 | as needed. | |
1586 | """ |
|
1586 | """ | |
1587 | return storageutil.hashrevisionsha1(text, p1, p2) |
|
1587 | return storageutil.hashrevisionsha1(text, p1, p2) | |
1588 |
|
1588 | |||
1589 | def _processflags(self, text, flags, operation, raw=False): |
|
1589 | def _processflags(self, text, flags, operation, raw=False): | |
1590 | """Inspect revision data flags and applies transforms defined by |
|
1590 | """Inspect revision data flags and applies transforms defined by | |
1591 | registered flag processors. |
|
1591 | registered flag processors. | |
1592 |
|
1592 | |||
1593 | ``text`` - the revision data to process |
|
1593 | ``text`` - the revision data to process | |
1594 | ``flags`` - the revision flags |
|
1594 | ``flags`` - the revision flags | |
1595 | ``operation`` - the operation being performed (read or write) |
|
1595 | ``operation`` - the operation being performed (read or write) | |
1596 | ``raw`` - an optional argument describing if the raw transform should be |
|
1596 | ``raw`` - an optional argument describing if the raw transform should be | |
1597 | applied. |
|
1597 | applied. | |
1598 |
|
1598 | |||
1599 | This method processes the flags in the order (or reverse order if |
|
1599 | This method processes the flags in the order (or reverse order if | |
1600 | ``operation`` is 'write') defined by REVIDX_FLAGS_ORDER, applying the |
|
1600 | ``operation`` is 'write') defined by REVIDX_FLAGS_ORDER, applying the | |
1601 | flag processors registered for present flags. The order of flags defined |
|
1601 | flag processors registered for present flags. The order of flags defined | |
1602 | in REVIDX_FLAGS_ORDER needs to be stable to allow non-commutativity. |
|
1602 | in REVIDX_FLAGS_ORDER needs to be stable to allow non-commutativity. | |
1603 |
|
1603 | |||
1604 | Returns a 2-tuple of ``(text, validatehash)`` where ``text`` is the |
|
1604 | Returns a 2-tuple of ``(text, validatehash)`` where ``text`` is the | |
1605 | processed text and ``validatehash`` is a bool indicating whether the |
|
1605 | processed text and ``validatehash`` is a bool indicating whether the | |
1606 | returned text should be checked for hash integrity. |
|
1606 | returned text should be checked for hash integrity. | |
1607 |
|
1607 | |||
1608 | Note: If the ``raw`` argument is set, it has precedence over the |
|
1608 | Note: If the ``raw`` argument is set, it has precedence over the | |
1609 | operation and will only update the value of ``validatehash``. |
|
1609 | operation and will only update the value of ``validatehash``. | |
1610 | """ |
|
1610 | """ | |
1611 | # fast path: no flag processors will run |
|
1611 | # fast path: no flag processors will run | |
1612 | if flags == 0: |
|
1612 | if flags == 0: | |
1613 | return text, True |
|
1613 | return text, True | |
1614 | if not operation in ('read', 'write'): |
|
1614 | if not operation in ('read', 'write'): | |
1615 | raise error.ProgrammingError(_("invalid '%s' operation") % |
|
1615 | raise error.ProgrammingError(_("invalid '%s' operation") % | |
1616 | operation) |
|
1616 | operation) | |
1617 | # Check all flags are known. |
|
1617 | # Check all flags are known. | |
1618 | if flags & ~REVIDX_KNOWN_FLAGS: |
|
1618 | if flags & ~REVIDX_KNOWN_FLAGS: | |
1619 | raise error.RevlogError(_("incompatible revision flag '%#x'") % |
|
1619 | raise error.RevlogError(_("incompatible revision flag '%#x'") % | |
1620 | (flags & ~REVIDX_KNOWN_FLAGS)) |
|
1620 | (flags & ~REVIDX_KNOWN_FLAGS)) | |
1621 | validatehash = True |
|
1621 | validatehash = True | |
1622 | # Depending on the operation (read or write), the order might be |
|
1622 | # Depending on the operation (read or write), the order might be | |
1623 | # reversed due to non-commutative transforms. |
|
1623 | # reversed due to non-commutative transforms. | |
1624 | orderedflags = REVIDX_FLAGS_ORDER |
|
1624 | orderedflags = REVIDX_FLAGS_ORDER | |
1625 | if operation == 'write': |
|
1625 | if operation == 'write': | |
1626 | orderedflags = reversed(orderedflags) |
|
1626 | orderedflags = reversed(orderedflags) | |
1627 |
|
1627 | |||
1628 | for flag in orderedflags: |
|
1628 | for flag in orderedflags: | |
1629 | # If a flagprocessor has been registered for a known flag, apply the |
|
1629 | # If a flagprocessor has been registered for a known flag, apply the | |
1630 | # related operation transform and update result tuple. |
|
1630 | # related operation transform and update result tuple. | |
1631 | if flag & flags: |
|
1631 | if flag & flags: | |
1632 | vhash = True |
|
1632 | vhash = True | |
1633 |
|
1633 | |||
1634 | if flag not in self._flagprocessors: |
|
1634 | if flag not in self._flagprocessors: | |
1635 | message = _("missing processor for flag '%#x'") % (flag) |
|
1635 | message = _("missing processor for flag '%#x'") % (flag) | |
1636 | raise error.RevlogError(message) |
|
1636 | raise error.RevlogError(message) | |
1637 |
|
1637 | |||
1638 | processor = self._flagprocessors[flag] |
|
1638 | processor = self._flagprocessors[flag] | |
1639 | if processor is not None: |
|
1639 | if processor is not None: | |
1640 | readtransform, writetransform, rawtransform = processor |
|
1640 | readtransform, writetransform, rawtransform = processor | |
1641 |
|
1641 | |||
1642 | if raw: |
|
1642 | if raw: | |
1643 | vhash = rawtransform(self, text) |
|
1643 | vhash = rawtransform(self, text) | |
1644 | elif operation == 'read': |
|
1644 | elif operation == 'read': | |
1645 | text, vhash = readtransform(self, text) |
|
1645 | text, vhash = readtransform(self, text) | |
1646 | else: # write operation |
|
1646 | else: # write operation | |
1647 | text, vhash = writetransform(self, text) |
|
1647 | text, vhash = writetransform(self, text) | |
1648 | validatehash = validatehash and vhash |
|
1648 | validatehash = validatehash and vhash | |
1649 |
|
1649 | |||
1650 | return text, validatehash |
|
1650 | return text, validatehash | |
1651 |
|
1651 | |||
1652 | def checkhash(self, text, node, p1=None, p2=None, rev=None): |
|
1652 | def checkhash(self, text, node, p1=None, p2=None, rev=None): | |
1653 | """Check node hash integrity. |
|
1653 | """Check node hash integrity. | |
1654 |
|
1654 | |||
1655 | Available as a function so that subclasses can extend hash mismatch |
|
1655 | Available as a function so that subclasses can extend hash mismatch | |
1656 | behaviors as needed. |
|
1656 | behaviors as needed. | |
1657 | """ |
|
1657 | """ | |
1658 | try: |
|
1658 | try: | |
1659 | if p1 is None and p2 is None: |
|
1659 | if p1 is None and p2 is None: | |
1660 | p1, p2 = self.parents(node) |
|
1660 | p1, p2 = self.parents(node) | |
1661 | if node != self.hash(text, p1, p2): |
|
1661 | if node != self.hash(text, p1, p2): | |
1662 | revornode = rev |
|
1662 | revornode = rev | |
1663 | if revornode is None: |
|
1663 | if revornode is None: | |
1664 | revornode = templatefilters.short(hex(node)) |
|
1664 | revornode = templatefilters.short(hex(node)) | |
1665 | raise error.RevlogError(_("integrity check failed on %s:%s") |
|
1665 | raise error.RevlogError(_("integrity check failed on %s:%s") | |
1666 | % (self.indexfile, pycompat.bytestr(revornode))) |
|
1666 | % (self.indexfile, pycompat.bytestr(revornode))) | |
1667 | except error.RevlogError: |
|
1667 | except error.RevlogError: | |
1668 | if self._censorable and storageutil.iscensoredtext(text): |
|
1668 | if self._censorable and storageutil.iscensoredtext(text): | |
1669 | raise error.CensoredNodeError(self.indexfile, node, text) |
|
1669 | raise error.CensoredNodeError(self.indexfile, node, text) | |
1670 | raise |
|
1670 | raise | |
1671 |
|
1671 | |||
1672 | def _enforceinlinesize(self, tr, fp=None): |
|
1672 | def _enforceinlinesize(self, tr, fp=None): | |
1673 | """Check if the revlog is too big for inline and convert if so. |
|
1673 | """Check if the revlog is too big for inline and convert if so. | |
1674 |
|
1674 | |||
1675 | This should be called after revisions are added to the revlog. If the |
|
1675 | This should be called after revisions are added to the revlog. If the | |
1676 | revlog has grown too large to be an inline revlog, it will convert it |
|
1676 | revlog has grown too large to be an inline revlog, it will convert it | |
1677 | to use multiple index and data files. |
|
1677 | to use multiple index and data files. | |
1678 | """ |
|
1678 | """ | |
1679 | tiprev = len(self) - 1 |
|
1679 | tiprev = len(self) - 1 | |
1680 | if (not self._inline or |
|
1680 | if (not self._inline or | |
1681 | (self.start(tiprev) + self.length(tiprev)) < _maxinline): |
|
1681 | (self.start(tiprev) + self.length(tiprev)) < _maxinline): | |
1682 | return |
|
1682 | return | |
1683 |
|
1683 | |||
1684 | trinfo = tr.find(self.indexfile) |
|
1684 | trinfo = tr.find(self.indexfile) | |
1685 | if trinfo is None: |
|
1685 | if trinfo is None: | |
1686 | raise error.RevlogError(_("%s not found in the transaction") |
|
1686 | raise error.RevlogError(_("%s not found in the transaction") | |
1687 | % self.indexfile) |
|
1687 | % self.indexfile) | |
1688 |
|
1688 | |||
1689 | trindex = trinfo[2] |
|
1689 | trindex = trinfo[2] | |
1690 | if trindex is not None: |
|
1690 | if trindex is not None: | |
1691 | dataoff = self.start(trindex) |
|
1691 | dataoff = self.start(trindex) | |
1692 | else: |
|
1692 | else: | |
1693 | # revlog was stripped at start of transaction, use all leftover data |
|
1693 | # revlog was stripped at start of transaction, use all leftover data | |
1694 | trindex = len(self) - 1 |
|
1694 | trindex = len(self) - 1 | |
1695 | dataoff = self.end(tiprev) |
|
1695 | dataoff = self.end(tiprev) | |
1696 |
|
1696 | |||
1697 | tr.add(self.datafile, dataoff) |
|
1697 | tr.add(self.datafile, dataoff) | |
1698 |
|
1698 | |||
1699 | if fp: |
|
1699 | if fp: | |
1700 | fp.flush() |
|
1700 | fp.flush() | |
1701 | fp.close() |
|
1701 | fp.close() | |
1702 |
|
1702 | |||
1703 | with self._datafp('w') as df: |
|
1703 | with self._datafp('w') as df: | |
1704 | for r in self: |
|
1704 | for r in self: | |
1705 | df.write(self._getsegmentforrevs(r, r)[1]) |
|
1705 | df.write(self._getsegmentforrevs(r, r)[1]) | |
1706 |
|
1706 | |||
1707 | with self._indexfp('w') as fp: |
|
1707 | with self._indexfp('w') as fp: | |
1708 | self.version &= ~FLAG_INLINE_DATA |
|
1708 | self.version &= ~FLAG_INLINE_DATA | |
1709 | self._inline = False |
|
1709 | self._inline = False | |
1710 | io = self._io |
|
1710 | io = self._io | |
1711 | for i in self: |
|
1711 | for i in self: | |
1712 | e = io.packentry(self.index[i], self.node, self.version, i) |
|
1712 | e = io.packentry(self.index[i], self.node, self.version, i) | |
1713 | fp.write(e) |
|
1713 | fp.write(e) | |
1714 |
|
1714 | |||
1715 | # the temp file replace the real index when we exit the context |
|
1715 | # the temp file replace the real index when we exit the context | |
1716 | # manager |
|
1716 | # manager | |
1717 |
|
1717 | |||
1718 | tr.replace(self.indexfile, trindex * self._io.size) |
|
1718 | tr.replace(self.indexfile, trindex * self._io.size) | |
1719 | self._chunkclear() |
|
1719 | self._chunkclear() | |
1720 |
|
1720 | |||
1721 | def _nodeduplicatecallback(self, transaction, node): |
|
1721 | def _nodeduplicatecallback(self, transaction, node): | |
1722 | """called when trying to add a node already stored. |
|
1722 | """called when trying to add a node already stored. | |
1723 | """ |
|
1723 | """ | |
1724 |
|
1724 | |||
1725 | def addrevision(self, text, transaction, link, p1, p2, cachedelta=None, |
|
1725 | def addrevision(self, text, transaction, link, p1, p2, cachedelta=None, | |
1726 | node=None, flags=REVIDX_DEFAULT_FLAGS, deltacomputer=None): |
|
1726 | node=None, flags=REVIDX_DEFAULT_FLAGS, deltacomputer=None): | |
1727 | """add a revision to the log |
|
1727 | """add a revision to the log | |
1728 |
|
1728 | |||
1729 | text - the revision data to add |
|
1729 | text - the revision data to add | |
1730 | transaction - the transaction object used for rollback |
|
1730 | transaction - the transaction object used for rollback | |
1731 | link - the linkrev data to add |
|
1731 | link - the linkrev data to add | |
1732 | p1, p2 - the parent nodeids of the revision |
|
1732 | p1, p2 - the parent nodeids of the revision | |
1733 | cachedelta - an optional precomputed delta |
|
1733 | cachedelta - an optional precomputed delta | |
1734 | node - nodeid of revision; typically node is not specified, and it is |
|
1734 | node - nodeid of revision; typically node is not specified, and it is | |
1735 | computed by default as hash(text, p1, p2), however subclasses might |
|
1735 | computed by default as hash(text, p1, p2), however subclasses might | |
1736 | use different hashing method (and override checkhash() in such case) |
|
1736 | use different hashing method (and override checkhash() in such case) | |
1737 | flags - the known flags to set on the revision |
|
1737 | flags - the known flags to set on the revision | |
1738 | deltacomputer - an optional deltacomputer instance shared between |
|
1738 | deltacomputer - an optional deltacomputer instance shared between | |
1739 | multiple calls |
|
1739 | multiple calls | |
1740 | """ |
|
1740 | """ | |
1741 | if link == nullrev: |
|
1741 | if link == nullrev: | |
1742 | raise error.RevlogError(_("attempted to add linkrev -1 to %s") |
|
1742 | raise error.RevlogError(_("attempted to add linkrev -1 to %s") | |
1743 | % self.indexfile) |
|
1743 | % self.indexfile) | |
1744 |
|
1744 | |||
1745 | if flags: |
|
1745 | if flags: | |
1746 | node = node or self.hash(text, p1, p2) |
|
1746 | node = node or self.hash(text, p1, p2) | |
1747 |
|
1747 | |||
1748 | rawtext, validatehash = self._processflags(text, flags, 'write') |
|
1748 | rawtext, validatehash = self._processflags(text, flags, 'write') | |
1749 |
|
1749 | |||
1750 | # If the flag processor modifies the revision data, ignore any provided |
|
1750 | # If the flag processor modifies the revision data, ignore any provided | |
1751 | # cachedelta. |
|
1751 | # cachedelta. | |
1752 | if rawtext != text: |
|
1752 | if rawtext != text: | |
1753 | cachedelta = None |
|
1753 | cachedelta = None | |
1754 |
|
1754 | |||
1755 | if len(rawtext) > _maxentrysize: |
|
1755 | if len(rawtext) > _maxentrysize: | |
1756 | raise error.RevlogError( |
|
1756 | raise error.RevlogError( | |
1757 | _("%s: size of %d bytes exceeds maximum revlog storage of 2GiB") |
|
1757 | _("%s: size of %d bytes exceeds maximum revlog storage of 2GiB") | |
1758 | % (self.indexfile, len(rawtext))) |
|
1758 | % (self.indexfile, len(rawtext))) | |
1759 |
|
1759 | |||
1760 | node = node or self.hash(rawtext, p1, p2) |
|
1760 | node = node or self.hash(rawtext, p1, p2) | |
1761 | if node in self.nodemap: |
|
1761 | if node in self.nodemap: | |
1762 | return node |
|
1762 | return node | |
1763 |
|
1763 | |||
1764 | if validatehash: |
|
1764 | if validatehash: | |
1765 | self.checkhash(rawtext, node, p1=p1, p2=p2) |
|
1765 | self.checkhash(rawtext, node, p1=p1, p2=p2) | |
1766 |
|
1766 | |||
1767 | return self.addrawrevision(rawtext, transaction, link, p1, p2, node, |
|
1767 | return self.addrawrevision(rawtext, transaction, link, p1, p2, node, | |
1768 | flags, cachedelta=cachedelta, |
|
1768 | flags, cachedelta=cachedelta, | |
1769 | deltacomputer=deltacomputer) |
|
1769 | deltacomputer=deltacomputer) | |
1770 |
|
1770 | |||
1771 | def addrawrevision(self, rawtext, transaction, link, p1, p2, node, flags, |
|
1771 | def addrawrevision(self, rawtext, transaction, link, p1, p2, node, flags, | |
1772 | cachedelta=None, deltacomputer=None): |
|
1772 | cachedelta=None, deltacomputer=None): | |
1773 | """add a raw revision with known flags, node and parents |
|
1773 | """add a raw revision with known flags, node and parents | |
1774 | useful when reusing a revision not stored in this revlog (ex: received |
|
1774 | useful when reusing a revision not stored in this revlog (ex: received | |
1775 | over wire, or read from an external bundle). |
|
1775 | over wire, or read from an external bundle). | |
1776 | """ |
|
1776 | """ | |
1777 | dfh = None |
|
1777 | dfh = None | |
1778 | if not self._inline: |
|
1778 | if not self._inline: | |
1779 | dfh = self._datafp("a+") |
|
1779 | dfh = self._datafp("a+") | |
1780 | ifh = self._indexfp("a+") |
|
1780 | ifh = self._indexfp("a+") | |
1781 | try: |
|
1781 | try: | |
1782 | return self._addrevision(node, rawtext, transaction, link, p1, p2, |
|
1782 | return self._addrevision(node, rawtext, transaction, link, p1, p2, | |
1783 | flags, cachedelta, ifh, dfh, |
|
1783 | flags, cachedelta, ifh, dfh, | |
1784 | deltacomputer=deltacomputer) |
|
1784 | deltacomputer=deltacomputer) | |
1785 | finally: |
|
1785 | finally: | |
1786 | if dfh: |
|
1786 | if dfh: | |
1787 | dfh.close() |
|
1787 | dfh.close() | |
1788 | ifh.close() |
|
1788 | ifh.close() | |
1789 |
|
1789 | |||
1790 | def compress(self, data): |
|
1790 | def compress(self, data): | |
1791 | """Generate a possibly-compressed representation of data.""" |
|
1791 | """Generate a possibly-compressed representation of data.""" | |
1792 | if not data: |
|
1792 | if not data: | |
1793 | return '', data |
|
1793 | return '', data | |
1794 |
|
1794 | |||
1795 | compressed = self._compressor.compress(data) |
|
1795 | compressed = self._compressor.compress(data) | |
1796 |
|
1796 | |||
1797 | if compressed: |
|
1797 | if compressed: | |
1798 | # The revlog compressor added the header in the returned data. |
|
1798 | # The revlog compressor added the header in the returned data. | |
1799 | return '', compressed |
|
1799 | return '', compressed | |
1800 |
|
1800 | |||
1801 | if data[0:1] == '\0': |
|
1801 | if data[0:1] == '\0': | |
1802 | return '', data |
|
1802 | return '', data | |
1803 | return 'u', data |
|
1803 | return 'u', data | |
1804 |
|
1804 | |||
1805 | def decompress(self, data): |
|
1805 | def decompress(self, data): | |
1806 | """Decompress a revlog chunk. |
|
1806 | """Decompress a revlog chunk. | |
1807 |
|
1807 | |||
1808 | The chunk is expected to begin with a header identifying the |
|
1808 | The chunk is expected to begin with a header identifying the | |
1809 | format type so it can be routed to an appropriate decompressor. |
|
1809 | format type so it can be routed to an appropriate decompressor. | |
1810 | """ |
|
1810 | """ | |
1811 | if not data: |
|
1811 | if not data: | |
1812 | return data |
|
1812 | return data | |
1813 |
|
1813 | |||
1814 | # Revlogs are read much more frequently than they are written and many |
|
1814 | # Revlogs are read much more frequently than they are written and many | |
1815 | # chunks only take microseconds to decompress, so performance is |
|
1815 | # chunks only take microseconds to decompress, so performance is | |
1816 | # important here. |
|
1816 | # important here. | |
1817 | # |
|
1817 | # | |
1818 | # We can make a few assumptions about revlogs: |
|
1818 | # We can make a few assumptions about revlogs: | |
1819 | # |
|
1819 | # | |
1820 | # 1) the majority of chunks will be compressed (as opposed to inline |
|
1820 | # 1) the majority of chunks will be compressed (as opposed to inline | |
1821 | # raw data). |
|
1821 | # raw data). | |
1822 | # 2) decompressing *any* data will likely by at least 10x slower than |
|
1822 | # 2) decompressing *any* data will likely by at least 10x slower than | |
1823 | # returning raw inline data. |
|
1823 | # returning raw inline data. | |
1824 | # 3) we want to prioritize common and officially supported compression |
|
1824 | # 3) we want to prioritize common and officially supported compression | |
1825 | # engines |
|
1825 | # engines | |
1826 | # |
|
1826 | # | |
1827 | # It follows that we want to optimize for "decompress compressed data |
|
1827 | # It follows that we want to optimize for "decompress compressed data | |
1828 | # when encoded with common and officially supported compression engines" |
|
1828 | # when encoded with common and officially supported compression engines" | |
1829 | # case over "raw data" and "data encoded by less common or non-official |
|
1829 | # case over "raw data" and "data encoded by less common or non-official | |
1830 | # compression engines." That is why we have the inline lookup first |
|
1830 | # compression engines." That is why we have the inline lookup first | |
1831 | # followed by the compengines lookup. |
|
1831 | # followed by the compengines lookup. | |
1832 | # |
|
1832 | # | |
1833 | # According to `hg perfrevlogchunks`, this is ~0.5% faster for zlib |
|
1833 | # According to `hg perfrevlogchunks`, this is ~0.5% faster for zlib | |
1834 | # compressed chunks. And this matters for changelog and manifest reads. |
|
1834 | # compressed chunks. And this matters for changelog and manifest reads. | |
1835 | t = data[0:1] |
|
1835 | t = data[0:1] | |
1836 |
|
1836 | |||
1837 | if t == 'x': |
|
1837 | if t == 'x': | |
1838 | try: |
|
1838 | try: | |
1839 | return _zlibdecompress(data) |
|
1839 | return _zlibdecompress(data) | |
1840 | except zlib.error as e: |
|
1840 | except zlib.error as e: | |
1841 | raise error.RevlogError(_('revlog decompress error: %s') % |
|
1841 | raise error.RevlogError(_('revlog decompress error: %s') % | |
1842 | stringutil.forcebytestr(e)) |
|
1842 | stringutil.forcebytestr(e)) | |
1843 | # '\0' is more common than 'u' so it goes first. |
|
1843 | # '\0' is more common than 'u' so it goes first. | |
1844 | elif t == '\0': |
|
1844 | elif t == '\0': | |
1845 | return data |
|
1845 | return data | |
1846 | elif t == 'u': |
|
1846 | elif t == 'u': | |
1847 | return util.buffer(data, 1) |
|
1847 | return util.buffer(data, 1) | |
1848 |
|
1848 | |||
1849 | try: |
|
1849 | try: | |
1850 | compressor = self._decompressors[t] |
|
1850 | compressor = self._decompressors[t] | |
1851 | except KeyError: |
|
1851 | except KeyError: | |
1852 | try: |
|
1852 | try: | |
1853 | engine = util.compengines.forrevlogheader(t) |
|
1853 | engine = util.compengines.forrevlogheader(t) | |
1854 | compressor = engine.revlogcompressor() |
|
1854 | compressor = engine.revlogcompressor() | |
1855 | self._decompressors[t] = compressor |
|
1855 | self._decompressors[t] = compressor | |
1856 | except KeyError: |
|
1856 | except KeyError: | |
1857 | raise error.RevlogError(_('unknown compression type %r') % t) |
|
1857 | raise error.RevlogError(_('unknown compression type %r') % t) | |
1858 |
|
1858 | |||
1859 | return compressor.decompress(data) |
|
1859 | return compressor.decompress(data) | |
1860 |
|
1860 | |||
1861 | def _addrevision(self, node, rawtext, transaction, link, p1, p2, flags, |
|
1861 | def _addrevision(self, node, rawtext, transaction, link, p1, p2, flags, | |
1862 | cachedelta, ifh, dfh, alwayscache=False, |
|
1862 | cachedelta, ifh, dfh, alwayscache=False, | |
1863 | deltacomputer=None): |
|
1863 | deltacomputer=None): | |
1864 | """internal function to add revisions to the log |
|
1864 | """internal function to add revisions to the log | |
1865 |
|
1865 | |||
1866 | see addrevision for argument descriptions. |
|
1866 | see addrevision for argument descriptions. | |
1867 |
|
1867 | |||
1868 | note: "addrevision" takes non-raw text, "_addrevision" takes raw text. |
|
1868 | note: "addrevision" takes non-raw text, "_addrevision" takes raw text. | |
1869 |
|
1869 | |||
1870 | if "deltacomputer" is not provided or None, a defaultdeltacomputer will |
|
1870 | if "deltacomputer" is not provided or None, a defaultdeltacomputer will | |
1871 | be used. |
|
1871 | be used. | |
1872 |
|
1872 | |||
1873 | invariants: |
|
1873 | invariants: | |
1874 | - rawtext is optional (can be None); if not set, cachedelta must be set. |
|
1874 | - rawtext is optional (can be None); if not set, cachedelta must be set. | |
1875 | if both are set, they must correspond to each other. |
|
1875 | if both are set, they must correspond to each other. | |
1876 | """ |
|
1876 | """ | |
1877 | if node == nullid: |
|
1877 | if node == nullid: | |
1878 | raise error.RevlogError(_("%s: attempt to add null revision") % |
|
1878 | raise error.RevlogError(_("%s: attempt to add null revision") % | |
1879 | self.indexfile) |
|
1879 | self.indexfile) | |
1880 | if node == wdirid or node in wdirfilenodeids: |
|
1880 | if node == wdirid or node in wdirfilenodeids: | |
1881 | raise error.RevlogError(_("%s: attempt to add wdir revision") % |
|
1881 | raise error.RevlogError(_("%s: attempt to add wdir revision") % | |
1882 | self.indexfile) |
|
1882 | self.indexfile) | |
1883 |
|
1883 | |||
1884 | if self._inline: |
|
1884 | if self._inline: | |
1885 | fh = ifh |
|
1885 | fh = ifh | |
1886 | else: |
|
1886 | else: | |
1887 | fh = dfh |
|
1887 | fh = dfh | |
1888 |
|
1888 | |||
1889 | btext = [rawtext] |
|
1889 | btext = [rawtext] | |
1890 |
|
1890 | |||
1891 | curr = len(self) |
|
1891 | curr = len(self) | |
1892 | prev = curr - 1 |
|
1892 | prev = curr - 1 | |
1893 | offset = self.end(prev) |
|
1893 | offset = self.end(prev) | |
1894 | p1r, p2r = self.rev(p1), self.rev(p2) |
|
1894 | p1r, p2r = self.rev(p1), self.rev(p2) | |
1895 |
|
1895 | |||
1896 | # full versions are inserted when the needed deltas |
|
1896 | # full versions are inserted when the needed deltas | |
1897 | # become comparable to the uncompressed text |
|
1897 | # become comparable to the uncompressed text | |
1898 | if rawtext is None: |
|
1898 | if rawtext is None: | |
1899 | # need rawtext size, before changed by flag processors, which is |
|
1899 | # need rawtext size, before changed by flag processors, which is | |
1900 | # the non-raw size. use revlog explicitly to avoid filelog's extra |
|
1900 | # the non-raw size. use revlog explicitly to avoid filelog's extra | |
1901 | # logic that might remove metadata size. |
|
1901 | # logic that might remove metadata size. | |
1902 | textlen = mdiff.patchedsize(revlog.size(self, cachedelta[0]), |
|
1902 | textlen = mdiff.patchedsize(revlog.size(self, cachedelta[0]), | |
1903 | cachedelta[1]) |
|
1903 | cachedelta[1]) | |
1904 | else: |
|
1904 | else: | |
1905 | textlen = len(rawtext) |
|
1905 | textlen = len(rawtext) | |
1906 |
|
1906 | |||
1907 | if deltacomputer is None: |
|
1907 | if deltacomputer is None: | |
1908 | deltacomputer = deltautil.deltacomputer(self) |
|
1908 | deltacomputer = deltautil.deltacomputer(self) | |
1909 |
|
1909 | |||
1910 | revinfo = _revisioninfo(node, p1, p2, btext, textlen, cachedelta, flags) |
|
1910 | revinfo = _revisioninfo(node, p1, p2, btext, textlen, cachedelta, flags) | |
1911 |
|
1911 | |||
1912 | deltainfo = deltacomputer.finddeltainfo(revinfo, fh) |
|
1912 | deltainfo = deltacomputer.finddeltainfo(revinfo, fh) | |
1913 |
|
1913 | |||
1914 | e = (offset_type(offset, flags), deltainfo.deltalen, textlen, |
|
1914 | e = (offset_type(offset, flags), deltainfo.deltalen, textlen, | |
1915 | deltainfo.base, link, p1r, p2r, node) |
|
1915 | deltainfo.base, link, p1r, p2r, node) | |
1916 | self.index.append(e) |
|
1916 | self.index.append(e) | |
1917 | self.nodemap[node] = curr |
|
1917 | self.nodemap[node] = curr | |
1918 |
|
1918 | |||
1919 | entry = self._io.packentry(e, self.node, self.version, curr) |
|
1919 | entry = self._io.packentry(e, self.node, self.version, curr) | |
1920 | self._writeentry(transaction, ifh, dfh, entry, deltainfo.data, |
|
1920 | self._writeentry(transaction, ifh, dfh, entry, deltainfo.data, | |
1921 | link, offset) |
|
1921 | link, offset) | |
1922 |
|
1922 | |||
1923 | rawtext = btext[0] |
|
1923 | rawtext = btext[0] | |
1924 |
|
1924 | |||
1925 | if alwayscache and rawtext is None: |
|
1925 | if alwayscache and rawtext is None: | |
1926 | rawtext = deltacomputer.buildtext(revinfo, fh) |
|
1926 | rawtext = deltacomputer.buildtext(revinfo, fh) | |
1927 |
|
1927 | |||
1928 | if type(rawtext) == bytes: # only accept immutable objects |
|
1928 | if type(rawtext) == bytes: # only accept immutable objects | |
1929 | self._cache = (node, curr, rawtext) |
|
1929 | self._cache = (node, curr, rawtext) | |
1930 | self._chainbasecache[curr] = deltainfo.chainbase |
|
1930 | self._chainbasecache[curr] = deltainfo.chainbase | |
1931 | return node |
|
1931 | return node | |
1932 |
|
1932 | |||
1933 | def _writeentry(self, transaction, ifh, dfh, entry, data, link, offset): |
|
1933 | def _writeentry(self, transaction, ifh, dfh, entry, data, link, offset): | |
1934 | # Files opened in a+ mode have inconsistent behavior on various |
|
1934 | # Files opened in a+ mode have inconsistent behavior on various | |
1935 | # platforms. Windows requires that a file positioning call be made |
|
1935 | # platforms. Windows requires that a file positioning call be made | |
1936 | # when the file handle transitions between reads and writes. See |
|
1936 | # when the file handle transitions between reads and writes. See | |
1937 | # 3686fa2b8eee and the mixedfilemodewrapper in windows.py. On other |
|
1937 | # 3686fa2b8eee and the mixedfilemodewrapper in windows.py. On other | |
1938 | # platforms, Python or the platform itself can be buggy. Some versions |
|
1938 | # platforms, Python or the platform itself can be buggy. Some versions | |
1939 | # of Solaris have been observed to not append at the end of the file |
|
1939 | # of Solaris have been observed to not append at the end of the file | |
1940 | # if the file was seeked to before the end. See issue4943 for more. |
|
1940 | # if the file was seeked to before the end. See issue4943 for more. | |
1941 | # |
|
1941 | # | |
1942 | # We work around this issue by inserting a seek() before writing. |
|
1942 | # We work around this issue by inserting a seek() before writing. | |
1943 | # Note: This is likely not necessary on Python 3. |
|
1943 | # Note: This is likely not necessary on Python 3. | |
1944 | ifh.seek(0, os.SEEK_END) |
|
1944 | ifh.seek(0, os.SEEK_END) | |
1945 | if dfh: |
|
1945 | if dfh: | |
1946 | dfh.seek(0, os.SEEK_END) |
|
1946 | dfh.seek(0, os.SEEK_END) | |
1947 |
|
1947 | |||
1948 | curr = len(self) - 1 |
|
1948 | curr = len(self) - 1 | |
1949 | if not self._inline: |
|
1949 | if not self._inline: | |
1950 | transaction.add(self.datafile, offset) |
|
1950 | transaction.add(self.datafile, offset) | |
1951 | transaction.add(self.indexfile, curr * len(entry)) |
|
1951 | transaction.add(self.indexfile, curr * len(entry)) | |
1952 | if data[0]: |
|
1952 | if data[0]: | |
1953 | dfh.write(data[0]) |
|
1953 | dfh.write(data[0]) | |
1954 | dfh.write(data[1]) |
|
1954 | dfh.write(data[1]) | |
1955 | ifh.write(entry) |
|
1955 | ifh.write(entry) | |
1956 | else: |
|
1956 | else: | |
1957 | offset += curr * self._io.size |
|
1957 | offset += curr * self._io.size | |
1958 | transaction.add(self.indexfile, offset, curr) |
|
1958 | transaction.add(self.indexfile, offset, curr) | |
1959 | ifh.write(entry) |
|
1959 | ifh.write(entry) | |
1960 | ifh.write(data[0]) |
|
1960 | ifh.write(data[0]) | |
1961 | ifh.write(data[1]) |
|
1961 | ifh.write(data[1]) | |
1962 | self._enforceinlinesize(transaction, ifh) |
|
1962 | self._enforceinlinesize(transaction, ifh) | |
1963 |
|
1963 | |||
1964 | def addgroup(self, deltas, linkmapper, transaction, addrevisioncb=None): |
|
1964 | def addgroup(self, deltas, linkmapper, transaction, addrevisioncb=None): | |
1965 | """ |
|
1965 | """ | |
1966 | add a delta group |
|
1966 | add a delta group | |
1967 |
|
1967 | |||
1968 | given a set of deltas, add them to the revision log. the |
|
1968 | given a set of deltas, add them to the revision log. the | |
1969 | first delta is against its parent, which should be in our |
|
1969 | first delta is against its parent, which should be in our | |
1970 | log, the rest are against the previous delta. |
|
1970 | log, the rest are against the previous delta. | |
1971 |
|
1971 | |||
1972 | If ``addrevisioncb`` is defined, it will be called with arguments of |
|
1972 | If ``addrevisioncb`` is defined, it will be called with arguments of | |
1973 | this revlog and the node that was added. |
|
1973 | this revlog and the node that was added. | |
1974 | """ |
|
1974 | """ | |
1975 |
|
1975 | |||
1976 | nodes = [] |
|
1976 | nodes = [] | |
1977 |
|
1977 | |||
1978 | r = len(self) |
|
1978 | r = len(self) | |
1979 | end = 0 |
|
1979 | end = 0 | |
1980 | if r: |
|
1980 | if r: | |
1981 | end = self.end(r - 1) |
|
1981 | end = self.end(r - 1) | |
1982 | ifh = self._indexfp("a+") |
|
1982 | ifh = self._indexfp("a+") | |
1983 | isize = r * self._io.size |
|
1983 | isize = r * self._io.size | |
1984 | if self._inline: |
|
1984 | if self._inline: | |
1985 | transaction.add(self.indexfile, end + isize, r) |
|
1985 | transaction.add(self.indexfile, end + isize, r) | |
1986 | dfh = None |
|
1986 | dfh = None | |
1987 | else: |
|
1987 | else: | |
1988 | transaction.add(self.indexfile, isize, r) |
|
1988 | transaction.add(self.indexfile, isize, r) | |
1989 | transaction.add(self.datafile, end) |
|
1989 | transaction.add(self.datafile, end) | |
1990 | dfh = self._datafp("a+") |
|
1990 | dfh = self._datafp("a+") | |
1991 | def flush(): |
|
1991 | def flush(): | |
1992 | if dfh: |
|
1992 | if dfh: | |
1993 | dfh.flush() |
|
1993 | dfh.flush() | |
1994 | ifh.flush() |
|
1994 | ifh.flush() | |
1995 | try: |
|
1995 | try: | |
1996 | deltacomputer = deltautil.deltacomputer(self) |
|
1996 | deltacomputer = deltautil.deltacomputer(self) | |
1997 | # loop through our set of deltas |
|
1997 | # loop through our set of deltas | |
1998 | for data in deltas: |
|
1998 | for data in deltas: | |
1999 | node, p1, p2, linknode, deltabase, delta, flags = data |
|
1999 | node, p1, p2, linknode, deltabase, delta, flags = data | |
2000 | link = linkmapper(linknode) |
|
2000 | link = linkmapper(linknode) | |
2001 | flags = flags or REVIDX_DEFAULT_FLAGS |
|
2001 | flags = flags or REVIDX_DEFAULT_FLAGS | |
2002 |
|
2002 | |||
2003 | nodes.append(node) |
|
2003 | nodes.append(node) | |
2004 |
|
2004 | |||
2005 | if node in self.nodemap: |
|
2005 | if node in self.nodemap: | |
2006 | self._nodeduplicatecallback(transaction, node) |
|
2006 | self._nodeduplicatecallback(transaction, node) | |
2007 | # this can happen if two branches make the same change |
|
2007 | # this can happen if two branches make the same change | |
2008 | continue |
|
2008 | continue | |
2009 |
|
2009 | |||
2010 | for p in (p1, p2): |
|
2010 | for p in (p1, p2): | |
2011 | if p not in self.nodemap: |
|
2011 | if p not in self.nodemap: | |
2012 | raise error.LookupError(p, self.indexfile, |
|
2012 | raise error.LookupError(p, self.indexfile, | |
2013 | _('unknown parent')) |
|
2013 | _('unknown parent')) | |
2014 |
|
2014 | |||
2015 | if deltabase not in self.nodemap: |
|
2015 | if deltabase not in self.nodemap: | |
2016 | raise error.LookupError(deltabase, self.indexfile, |
|
2016 | raise error.LookupError(deltabase, self.indexfile, | |
2017 | _('unknown delta base')) |
|
2017 | _('unknown delta base')) | |
2018 |
|
2018 | |||
2019 | baserev = self.rev(deltabase) |
|
2019 | baserev = self.rev(deltabase) | |
2020 |
|
2020 | |||
2021 | if baserev != nullrev and self.iscensored(baserev): |
|
2021 | if baserev != nullrev and self.iscensored(baserev): | |
2022 | # if base is censored, delta must be full replacement in a |
|
2022 | # if base is censored, delta must be full replacement in a | |
2023 | # single patch operation |
|
2023 | # single patch operation | |
2024 | hlen = struct.calcsize(">lll") |
|
2024 | hlen = struct.calcsize(">lll") | |
2025 | oldlen = self.rawsize(baserev) |
|
2025 | oldlen = self.rawsize(baserev) | |
2026 | newlen = len(delta) - hlen |
|
2026 | newlen = len(delta) - hlen | |
2027 | if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen): |
|
2027 | if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen): | |
2028 | raise error.CensoredBaseError(self.indexfile, |
|
2028 | raise error.CensoredBaseError(self.indexfile, | |
2029 | self.node(baserev)) |
|
2029 | self.node(baserev)) | |
2030 |
|
2030 | |||
2031 | if not flags and self._peek_iscensored(baserev, delta, flush): |
|
2031 | if not flags and self._peek_iscensored(baserev, delta, flush): | |
2032 | flags |= REVIDX_ISCENSORED |
|
2032 | flags |= REVIDX_ISCENSORED | |
2033 |
|
2033 | |||
2034 | # We assume consumers of addrevisioncb will want to retrieve |
|
2034 | # We assume consumers of addrevisioncb will want to retrieve | |
2035 | # the added revision, which will require a call to |
|
2035 | # the added revision, which will require a call to | |
2036 | # revision(). revision() will fast path if there is a cache |
|
2036 | # revision(). revision() will fast path if there is a cache | |
2037 | # hit. So, we tell _addrevision() to always cache in this case. |
|
2037 | # hit. So, we tell _addrevision() to always cache in this case. | |
2038 | # We're only using addgroup() in the context of changegroup |
|
2038 | # We're only using addgroup() in the context of changegroup | |
2039 | # generation so the revision data can always be handled as raw |
|
2039 | # generation so the revision data can always be handled as raw | |
2040 | # by the flagprocessor. |
|
2040 | # by the flagprocessor. | |
2041 | self._addrevision(node, None, transaction, link, |
|
2041 | self._addrevision(node, None, transaction, link, | |
2042 | p1, p2, flags, (baserev, delta), |
|
2042 | p1, p2, flags, (baserev, delta), | |
2043 | ifh, dfh, |
|
2043 | ifh, dfh, | |
2044 | alwayscache=bool(addrevisioncb), |
|
2044 | alwayscache=bool(addrevisioncb), | |
2045 | deltacomputer=deltacomputer) |
|
2045 | deltacomputer=deltacomputer) | |
2046 |
|
2046 | |||
2047 | if addrevisioncb: |
|
2047 | if addrevisioncb: | |
2048 | addrevisioncb(self, node) |
|
2048 | addrevisioncb(self, node) | |
2049 |
|
2049 | |||
2050 | if not dfh and not self._inline: |
|
2050 | if not dfh and not self._inline: | |
2051 | # addrevision switched from inline to conventional |
|
2051 | # addrevision switched from inline to conventional | |
2052 | # reopen the index |
|
2052 | # reopen the index | |
2053 | ifh.close() |
|
2053 | ifh.close() | |
2054 | dfh = self._datafp("a+") |
|
2054 | dfh = self._datafp("a+") | |
2055 | ifh = self._indexfp("a+") |
|
2055 | ifh = self._indexfp("a+") | |
2056 | finally: |
|
2056 | finally: | |
2057 | if dfh: |
|
2057 | if dfh: | |
2058 | dfh.close() |
|
2058 | dfh.close() | |
2059 | ifh.close() |
|
2059 | ifh.close() | |
2060 |
|
2060 | |||
2061 | return nodes |
|
2061 | return nodes | |
2062 |
|
2062 | |||
2063 | def iscensored(self, rev): |
|
2063 | def iscensored(self, rev): | |
2064 | """Check if a file revision is censored.""" |
|
2064 | """Check if a file revision is censored.""" | |
2065 | if not self._censorable: |
|
2065 | if not self._censorable: | |
2066 | return False |
|
2066 | return False | |
2067 |
|
2067 | |||
2068 | return self.flags(rev) & REVIDX_ISCENSORED |
|
2068 | return self.flags(rev) & REVIDX_ISCENSORED | |
2069 |
|
2069 | |||
2070 | def _peek_iscensored(self, baserev, delta, flush): |
|
2070 | def _peek_iscensored(self, baserev, delta, flush): | |
2071 | """Quickly check if a delta produces a censored revision.""" |
|
2071 | """Quickly check if a delta produces a censored revision.""" | |
2072 | if not self._censorable: |
|
2072 | if not self._censorable: | |
2073 | return False |
|
2073 | return False | |
2074 |
|
2074 | |||
2075 | # Fragile heuristic: unless new file meta keys are added alphabetically |
|
2075 | # Fragile heuristic: unless new file meta keys are added alphabetically | |
2076 | # preceding "censored", all censored revisions are prefixed by |
|
2076 | # preceding "censored", all censored revisions are prefixed by | |
2077 | # "\1\ncensored:". A delta producing such a censored revision must be a |
|
2077 | # "\1\ncensored:". A delta producing such a censored revision must be a | |
2078 | # full-replacement delta, so we inspect the first and only patch in the |
|
2078 | # full-replacement delta, so we inspect the first and only patch in the | |
2079 | # delta for this prefix. |
|
2079 | # delta for this prefix. | |
2080 | hlen = struct.calcsize(">lll") |
|
2080 | hlen = struct.calcsize(">lll") | |
2081 | if len(delta) <= hlen: |
|
2081 | if len(delta) <= hlen: | |
2082 | return False |
|
2082 | return False | |
2083 |
|
2083 | |||
2084 | oldlen = self.rawsize(baserev) |
|
2084 | oldlen = self.rawsize(baserev) | |
2085 | newlen = len(delta) - hlen |
|
2085 | newlen = len(delta) - hlen | |
2086 | if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen): |
|
2086 | if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen): | |
2087 | return False |
|
2087 | return False | |
2088 |
|
2088 | |||
2089 | add = "\1\ncensored:" |
|
2089 | add = "\1\ncensored:" | |
2090 | addlen = len(add) |
|
2090 | addlen = len(add) | |
2091 | return newlen >= addlen and delta[hlen:hlen + addlen] == add |
|
2091 | return newlen >= addlen and delta[hlen:hlen + addlen] == add | |
2092 |
|
2092 | |||
2093 | def getstrippoint(self, minlink): |
|
2093 | def getstrippoint(self, minlink): | |
2094 | """find the minimum rev that must be stripped to strip the linkrev |
|
2094 | """find the minimum rev that must be stripped to strip the linkrev | |
2095 |
|
2095 | |||
2096 | Returns a tuple containing the minimum rev and a set of all revs that |
|
2096 | Returns a tuple containing the minimum rev and a set of all revs that | |
2097 | have linkrevs that will be broken by this strip. |
|
2097 | have linkrevs that will be broken by this strip. | |
2098 | """ |
|
2098 | """ | |
2099 | return storageutil.resolvestripinfo(minlink, len(self) - 1, |
|
2099 | return storageutil.resolvestripinfo(minlink, len(self) - 1, | |
2100 | self.headrevs(), |
|
2100 | self.headrevs(), | |
2101 | self.linkrev, self.parentrevs) |
|
2101 | self.linkrev, self.parentrevs) | |
2102 |
|
2102 | |||
2103 | def strip(self, minlink, transaction): |
|
2103 | def strip(self, minlink, transaction): | |
2104 | """truncate the revlog on the first revision with a linkrev >= minlink |
|
2104 | """truncate the revlog on the first revision with a linkrev >= minlink | |
2105 |
|
2105 | |||
2106 | This function is called when we're stripping revision minlink and |
|
2106 | This function is called when we're stripping revision minlink and | |
2107 | its descendants from the repository. |
|
2107 | its descendants from the repository. | |
2108 |
|
2108 | |||
2109 | We have to remove all revisions with linkrev >= minlink, because |
|
2109 | We have to remove all revisions with linkrev >= minlink, because | |
2110 | the equivalent changelog revisions will be renumbered after the |
|
2110 | the equivalent changelog revisions will be renumbered after the | |
2111 | strip. |
|
2111 | strip. | |
2112 |
|
2112 | |||
2113 | So we truncate the revlog on the first of these revisions, and |
|
2113 | So we truncate the revlog on the first of these revisions, and | |
2114 | trust that the caller has saved the revisions that shouldn't be |
|
2114 | trust that the caller has saved the revisions that shouldn't be | |
2115 | removed and that it'll re-add them after this truncation. |
|
2115 | removed and that it'll re-add them after this truncation. | |
2116 | """ |
|
2116 | """ | |
2117 | if len(self) == 0: |
|
2117 | if len(self) == 0: | |
2118 | return |
|
2118 | return | |
2119 |
|
2119 | |||
2120 | rev, _ = self.getstrippoint(minlink) |
|
2120 | rev, _ = self.getstrippoint(minlink) | |
2121 | if rev == len(self): |
|
2121 | if rev == len(self): | |
2122 | return |
|
2122 | return | |
2123 |
|
2123 | |||
2124 | # first truncate the files on disk |
|
2124 | # first truncate the files on disk | |
2125 | end = self.start(rev) |
|
2125 | end = self.start(rev) | |
2126 | if not self._inline: |
|
2126 | if not self._inline: | |
2127 | transaction.add(self.datafile, end) |
|
2127 | transaction.add(self.datafile, end) | |
2128 | end = rev * self._io.size |
|
2128 | end = rev * self._io.size | |
2129 | else: |
|
2129 | else: | |
2130 | end += rev * self._io.size |
|
2130 | end += rev * self._io.size | |
2131 |
|
2131 | |||
2132 | transaction.add(self.indexfile, end) |
|
2132 | transaction.add(self.indexfile, end) | |
2133 |
|
2133 | |||
2134 | # then reset internal state in memory to forget those revisions |
|
2134 | # then reset internal state in memory to forget those revisions | |
2135 | self._cache = None |
|
2135 | self._cache = None | |
2136 | self._chaininfocache = {} |
|
2136 | self._chaininfocache = {} | |
2137 | self._chunkclear() |
|
2137 | self._chunkclear() | |
2138 | for x in pycompat.xrange(rev, len(self)): |
|
2138 | for x in pycompat.xrange(rev, len(self)): | |
2139 | del self.nodemap[self.node(x)] |
|
2139 | del self.nodemap[self.node(x)] | |
2140 |
|
2140 | |||
2141 | del self.index[rev:-1] |
|
2141 | del self.index[rev:-1] | |
2142 | self._nodepos = None |
|
2142 | self._nodepos = None | |
2143 |
|
2143 | |||
2144 | def checksize(self): |
|
2144 | def checksize(self): | |
2145 | expected = 0 |
|
2145 | expected = 0 | |
2146 | if len(self): |
|
2146 | if len(self): | |
2147 | expected = max(0, self.end(len(self) - 1)) |
|
2147 | expected = max(0, self.end(len(self) - 1)) | |
2148 |
|
2148 | |||
2149 | try: |
|
2149 | try: | |
2150 | with self._datafp() as f: |
|
2150 | with self._datafp() as f: | |
2151 | f.seek(0, 2) |
|
2151 | f.seek(0, 2) | |
2152 | actual = f.tell() |
|
2152 | actual = f.tell() | |
2153 | dd = actual - expected |
|
2153 | dd = actual - expected | |
2154 | except IOError as inst: |
|
2154 | except IOError as inst: | |
2155 | if inst.errno != errno.ENOENT: |
|
2155 | if inst.errno != errno.ENOENT: | |
2156 | raise |
|
2156 | raise | |
2157 | dd = 0 |
|
2157 | dd = 0 | |
2158 |
|
2158 | |||
2159 | try: |
|
2159 | try: | |
2160 | f = self.opener(self.indexfile) |
|
2160 | f = self.opener(self.indexfile) | |
2161 | f.seek(0, 2) |
|
2161 | f.seek(0, 2) | |
2162 | actual = f.tell() |
|
2162 | actual = f.tell() | |
2163 | f.close() |
|
2163 | f.close() | |
2164 | s = self._io.size |
|
2164 | s = self._io.size | |
2165 | i = max(0, actual // s) |
|
2165 | i = max(0, actual // s) | |
2166 | di = actual - (i * s) |
|
2166 | di = actual - (i * s) | |
2167 | if self._inline: |
|
2167 | if self._inline: | |
2168 | databytes = 0 |
|
2168 | databytes = 0 | |
2169 | for r in self: |
|
2169 | for r in self: | |
2170 | databytes += max(0, self.length(r)) |
|
2170 | databytes += max(0, self.length(r)) | |
2171 | dd = 0 |
|
2171 | dd = 0 | |
2172 | di = actual - len(self) * s - databytes |
|
2172 | di = actual - len(self) * s - databytes | |
2173 | except IOError as inst: |
|
2173 | except IOError as inst: | |
2174 | if inst.errno != errno.ENOENT: |
|
2174 | if inst.errno != errno.ENOENT: | |
2175 | raise |
|
2175 | raise | |
2176 | di = 0 |
|
2176 | di = 0 | |
2177 |
|
2177 | |||
2178 | return (dd, di) |
|
2178 | return (dd, di) | |
2179 |
|
2179 | |||
2180 | def files(self): |
|
2180 | def files(self): | |
2181 | res = [self.indexfile] |
|
2181 | res = [self.indexfile] | |
2182 | if not self._inline: |
|
2182 | if not self._inline: | |
2183 | res.append(self.datafile) |
|
2183 | res.append(self.datafile) | |
2184 | return res |
|
2184 | return res | |
2185 |
|
2185 | |||
2186 | def emitrevisions(self, nodes, nodesorder=None, revisiondata=False, |
|
2186 | def emitrevisions(self, nodes, nodesorder=None, revisiondata=False, | |
2187 | assumehaveparentrevisions=False, deltaprevious=False): |
|
2187 | assumehaveparentrevisions=False, deltaprevious=False): | |
2188 | if nodesorder not in ('nodes', 'storage', None): |
|
2188 | if nodesorder not in ('nodes', 'storage', None): | |
2189 | raise error.ProgrammingError('unhandled value for nodesorder: %s' % |
|
2189 | raise error.ProgrammingError('unhandled value for nodesorder: %s' % | |
2190 | nodesorder) |
|
2190 | nodesorder) | |
2191 |
|
2191 | |||
2192 | if nodesorder is None and not self._generaldelta: |
|
2192 | if nodesorder is None and not self._generaldelta: | |
2193 | nodesorder = 'storage' |
|
2193 | nodesorder = 'storage' | |
2194 |
|
2194 | |||
2195 | frev = self.rev |
|
2195 | frev = self.rev | |
2196 | fnode = self.node |
|
|||
2197 |
|
2196 | |||
2198 | if nodesorder == 'nodes': |
|
2197 | if nodesorder == 'nodes': | |
2199 | revs = [frev(n) for n in nodes] |
|
2198 | revs = [frev(n) for n in nodes] | |
2200 | elif nodesorder == 'storage': |
|
2199 | elif nodesorder == 'storage': | |
2201 | revs = sorted(frev(n) for n in nodes) |
|
2200 | revs = sorted(frev(n) for n in nodes) | |
2202 | else: |
|
2201 | else: | |
2203 | assert self._generaldelta |
|
2202 | assert self._generaldelta | |
2204 | revs = set(frev(n) for n in nodes) |
|
2203 | revs = set(frev(n) for n in nodes) | |
2205 | revs = dagop.linearize(revs, self.parentrevs) |
|
2204 | revs = dagop.linearize(revs, self.parentrevs) | |
2206 |
|
2205 | |||
2207 | prevrev = None |
|
2206 | return storageutil.emitrevisions( | |
2208 |
|
2207 | self, revs, revlogrevisiondelta, | ||
2209 | if deltaprevious or assumehaveparentrevisions: |
|
2208 | deltaparentfn=self.deltaparent, | |
2210 | prevrev = self.parentrevs(revs[0])[0] |
|
2209 | candeltafn=self.candelta, | |
2211 |
|
2210 | rawsizefn=self.rawsize, | ||
2212 | # Set of revs available to delta against. |
|
2211 | revdifffn=self.revdiff, | |
2213 | available = set() |
|
2212 | flagsfn=self.flags, | |
2214 |
|
2213 | sendfulltext=not self._storedeltachains, | ||
2215 | for rev in revs: |
|
2214 | revisiondata=revisiondata, | |
2216 | if rev == nullrev: |
|
2215 | assumehaveparentrevisions=assumehaveparentrevisions, | |
2217 | continue |
|
2216 | deltaprevious=deltaprevious) | |
2218 |
|
||||
2219 | node = fnode(rev) |
|
|||
2220 | deltaparentrev = self.deltaparent(rev) |
|
|||
2221 | p1rev, p2rev = self.parentrevs(rev) |
|
|||
2222 |
|
||||
2223 | # Forced delta against previous mode. |
|
|||
2224 | if deltaprevious: |
|
|||
2225 | baserev = prevrev |
|
|||
2226 |
|
||||
2227 | # Revlog is configured to use full snapshots. Stick to that. |
|
|||
2228 | elif not self._storedeltachains: |
|
|||
2229 | baserev = nullrev |
|
|||
2230 |
|
||||
2231 | # There is a delta in storage. We try to use that because it |
|
|||
2232 | # amounts to effectively copying data from storage and is |
|
|||
2233 | # therefore the fastest. |
|
|||
2234 | elif deltaparentrev != nullrev: |
|
|||
2235 | # Base revision was already emitted in this group. We can |
|
|||
2236 | # always safely use the delta. |
|
|||
2237 | if deltaparentrev in available: |
|
|||
2238 | baserev = deltaparentrev |
|
|||
2239 |
|
||||
2240 | # Base revision is a parent that hasn't been emitted already. |
|
|||
2241 | # Use it if we can assume the receiver has the parent revision. |
|
|||
2242 | elif (assumehaveparentrevisions |
|
|||
2243 | and deltaparentrev in (p1rev, p2rev)): |
|
|||
2244 | baserev = deltaparentrev |
|
|||
2245 |
|
||||
2246 | # No guarantee the receiver has the delta parent. Send delta |
|
|||
2247 | # against last revision (if possible), which in the common case |
|
|||
2248 | # should be similar enough to this revision that the delta is |
|
|||
2249 | # reasonable. |
|
|||
2250 | elif prevrev is not None: |
|
|||
2251 | baserev = prevrev |
|
|||
2252 | else: |
|
|||
2253 | baserev = nullrev |
|
|||
2254 |
|
||||
2255 | # Storage has a fulltext revision. |
|
|||
2256 |
|
||||
2257 | # Let's use the previous revision, which is as good a guess as any. |
|
|||
2258 | # There is definitely room to improve this logic. |
|
|||
2259 | elif prevrev is not None: |
|
|||
2260 | baserev = prevrev |
|
|||
2261 | else: |
|
|||
2262 | baserev = nullrev |
|
|||
2263 |
|
||||
2264 | # But we can't actually use our chosen delta base for whatever |
|
|||
2265 | # reason. Reset to fulltext. |
|
|||
2266 | if baserev != nullrev and not self.candelta(baserev, rev): |
|
|||
2267 | baserev = nullrev |
|
|||
2268 |
|
||||
2269 | revision = None |
|
|||
2270 | delta = None |
|
|||
2271 | baserevisionsize = None |
|
|||
2272 |
|
||||
2273 | if revisiondata: |
|
|||
2274 | if self.iscensored(baserev) or self.iscensored(rev): |
|
|||
2275 | try: |
|
|||
2276 | revision = self.revision(node, raw=True) |
|
|||
2277 | except error.CensoredNodeError as e: |
|
|||
2278 | revision = e.tombstone |
|
|||
2279 |
|
||||
2280 | if baserev != nullrev: |
|
|||
2281 | baserevisionsize = self.rawsize(baserev) |
|
|||
2282 |
|
||||
2283 | elif baserev == nullrev and not deltaprevious: |
|
|||
2284 | revision = self.revision(node, raw=True) |
|
|||
2285 | available.add(rev) |
|
|||
2286 | else: |
|
|||
2287 | delta = self.revdiff(baserev, rev) |
|
|||
2288 | available.add(rev) |
|
|||
2289 |
|
||||
2290 | yield revlogrevisiondelta( |
|
|||
2291 | node=node, |
|
|||
2292 | p1node=fnode(p1rev), |
|
|||
2293 | p2node=fnode(p2rev), |
|
|||
2294 | basenode=fnode(baserev), |
|
|||
2295 | flags=self.flags(rev), |
|
|||
2296 | baserevisionsize=baserevisionsize, |
|
|||
2297 | revision=revision, |
|
|||
2298 | delta=delta) |
|
|||
2299 |
|
||||
2300 | prevrev = rev |
|
|||
2301 |
|
2217 | |||
2302 | DELTAREUSEALWAYS = 'always' |
|
2218 | DELTAREUSEALWAYS = 'always' | |
2303 | DELTAREUSESAMEREVS = 'samerevs' |
|
2219 | DELTAREUSESAMEREVS = 'samerevs' | |
2304 | DELTAREUSENEVER = 'never' |
|
2220 | DELTAREUSENEVER = 'never' | |
2305 |
|
2221 | |||
2306 | DELTAREUSEFULLADD = 'fulladd' |
|
2222 | DELTAREUSEFULLADD = 'fulladd' | |
2307 |
|
2223 | |||
2308 | DELTAREUSEALL = {'always', 'samerevs', 'never', 'fulladd'} |
|
2224 | DELTAREUSEALL = {'always', 'samerevs', 'never', 'fulladd'} | |
2309 |
|
2225 | |||
2310 | def clone(self, tr, destrevlog, addrevisioncb=None, |
|
2226 | def clone(self, tr, destrevlog, addrevisioncb=None, | |
2311 | deltareuse=DELTAREUSESAMEREVS, deltabothparents=None): |
|
2227 | deltareuse=DELTAREUSESAMEREVS, deltabothparents=None): | |
2312 | """Copy this revlog to another, possibly with format changes. |
|
2228 | """Copy this revlog to another, possibly with format changes. | |
2313 |
|
2229 | |||
2314 | The destination revlog will contain the same revisions and nodes. |
|
2230 | The destination revlog will contain the same revisions and nodes. | |
2315 | However, it may not be bit-for-bit identical due to e.g. delta encoding |
|
2231 | However, it may not be bit-for-bit identical due to e.g. delta encoding | |
2316 | differences. |
|
2232 | differences. | |
2317 |
|
2233 | |||
2318 | The ``deltareuse`` argument control how deltas from the existing revlog |
|
2234 | The ``deltareuse`` argument control how deltas from the existing revlog | |
2319 | are preserved in the destination revlog. The argument can have the |
|
2235 | are preserved in the destination revlog. The argument can have the | |
2320 | following values: |
|
2236 | following values: | |
2321 |
|
2237 | |||
2322 | DELTAREUSEALWAYS |
|
2238 | DELTAREUSEALWAYS | |
2323 | Deltas will always be reused (if possible), even if the destination |
|
2239 | Deltas will always be reused (if possible), even if the destination | |
2324 | revlog would not select the same revisions for the delta. This is the |
|
2240 | revlog would not select the same revisions for the delta. This is the | |
2325 | fastest mode of operation. |
|
2241 | fastest mode of operation. | |
2326 | DELTAREUSESAMEREVS |
|
2242 | DELTAREUSESAMEREVS | |
2327 | Deltas will be reused if the destination revlog would pick the same |
|
2243 | Deltas will be reused if the destination revlog would pick the same | |
2328 | revisions for the delta. This mode strikes a balance between speed |
|
2244 | revisions for the delta. This mode strikes a balance between speed | |
2329 | and optimization. |
|
2245 | and optimization. | |
2330 | DELTAREUSENEVER |
|
2246 | DELTAREUSENEVER | |
2331 | Deltas will never be reused. This is the slowest mode of execution. |
|
2247 | Deltas will never be reused. This is the slowest mode of execution. | |
2332 | This mode can be used to recompute deltas (e.g. if the diff/delta |
|
2248 | This mode can be used to recompute deltas (e.g. if the diff/delta | |
2333 | algorithm changes). |
|
2249 | algorithm changes). | |
2334 |
|
2250 | |||
2335 | Delta computation can be slow, so the choice of delta reuse policy can |
|
2251 | Delta computation can be slow, so the choice of delta reuse policy can | |
2336 | significantly affect run time. |
|
2252 | significantly affect run time. | |
2337 |
|
2253 | |||
2338 | The default policy (``DELTAREUSESAMEREVS``) strikes a balance between |
|
2254 | The default policy (``DELTAREUSESAMEREVS``) strikes a balance between | |
2339 | two extremes. Deltas will be reused if they are appropriate. But if the |
|
2255 | two extremes. Deltas will be reused if they are appropriate. But if the | |
2340 | delta could choose a better revision, it will do so. This means if you |
|
2256 | delta could choose a better revision, it will do so. This means if you | |
2341 | are converting a non-generaldelta revlog to a generaldelta revlog, |
|
2257 | are converting a non-generaldelta revlog to a generaldelta revlog, | |
2342 | deltas will be recomputed if the delta's parent isn't a parent of the |
|
2258 | deltas will be recomputed if the delta's parent isn't a parent of the | |
2343 | revision. |
|
2259 | revision. | |
2344 |
|
2260 | |||
2345 | In addition to the delta policy, the ``deltabothparents`` argument |
|
2261 | In addition to the delta policy, the ``deltabothparents`` argument | |
2346 | controls whether to compute deltas against both parents for merges. |
|
2262 | controls whether to compute deltas against both parents for merges. | |
2347 | By default, the current default is used. |
|
2263 | By default, the current default is used. | |
2348 | """ |
|
2264 | """ | |
2349 | if deltareuse not in self.DELTAREUSEALL: |
|
2265 | if deltareuse not in self.DELTAREUSEALL: | |
2350 | raise ValueError(_('value for deltareuse invalid: %s') % deltareuse) |
|
2266 | raise ValueError(_('value for deltareuse invalid: %s') % deltareuse) | |
2351 |
|
2267 | |||
2352 | if len(destrevlog): |
|
2268 | if len(destrevlog): | |
2353 | raise ValueError(_('destination revlog is not empty')) |
|
2269 | raise ValueError(_('destination revlog is not empty')) | |
2354 |
|
2270 | |||
2355 | if getattr(self, 'filteredrevs', None): |
|
2271 | if getattr(self, 'filteredrevs', None): | |
2356 | raise ValueError(_('source revlog has filtered revisions')) |
|
2272 | raise ValueError(_('source revlog has filtered revisions')) | |
2357 | if getattr(destrevlog, 'filteredrevs', None): |
|
2273 | if getattr(destrevlog, 'filteredrevs', None): | |
2358 | raise ValueError(_('destination revlog has filtered revisions')) |
|
2274 | raise ValueError(_('destination revlog has filtered revisions')) | |
2359 |
|
2275 | |||
2360 | # lazydeltabase controls whether to reuse a cached delta, if possible. |
|
2276 | # lazydeltabase controls whether to reuse a cached delta, if possible. | |
2361 | oldlazydeltabase = destrevlog._lazydeltabase |
|
2277 | oldlazydeltabase = destrevlog._lazydeltabase | |
2362 | oldamd = destrevlog._deltabothparents |
|
2278 | oldamd = destrevlog._deltabothparents | |
2363 |
|
2279 | |||
2364 | try: |
|
2280 | try: | |
2365 | if deltareuse == self.DELTAREUSEALWAYS: |
|
2281 | if deltareuse == self.DELTAREUSEALWAYS: | |
2366 | destrevlog._lazydeltabase = True |
|
2282 | destrevlog._lazydeltabase = True | |
2367 | elif deltareuse == self.DELTAREUSESAMEREVS: |
|
2283 | elif deltareuse == self.DELTAREUSESAMEREVS: | |
2368 | destrevlog._lazydeltabase = False |
|
2284 | destrevlog._lazydeltabase = False | |
2369 |
|
2285 | |||
2370 | destrevlog._deltabothparents = deltabothparents or oldamd |
|
2286 | destrevlog._deltabothparents = deltabothparents or oldamd | |
2371 |
|
2287 | |||
2372 | populatecachedelta = deltareuse in (self.DELTAREUSEALWAYS, |
|
2288 | populatecachedelta = deltareuse in (self.DELTAREUSEALWAYS, | |
2373 | self.DELTAREUSESAMEREVS) |
|
2289 | self.DELTAREUSESAMEREVS) | |
2374 |
|
2290 | |||
2375 | deltacomputer = deltautil.deltacomputer(destrevlog) |
|
2291 | deltacomputer = deltautil.deltacomputer(destrevlog) | |
2376 | index = self.index |
|
2292 | index = self.index | |
2377 | for rev in self: |
|
2293 | for rev in self: | |
2378 | entry = index[rev] |
|
2294 | entry = index[rev] | |
2379 |
|
2295 | |||
2380 | # Some classes override linkrev to take filtered revs into |
|
2296 | # Some classes override linkrev to take filtered revs into | |
2381 | # account. Use raw entry from index. |
|
2297 | # account. Use raw entry from index. | |
2382 | flags = entry[0] & 0xffff |
|
2298 | flags = entry[0] & 0xffff | |
2383 | linkrev = entry[4] |
|
2299 | linkrev = entry[4] | |
2384 | p1 = index[entry[5]][7] |
|
2300 | p1 = index[entry[5]][7] | |
2385 | p2 = index[entry[6]][7] |
|
2301 | p2 = index[entry[6]][7] | |
2386 | node = entry[7] |
|
2302 | node = entry[7] | |
2387 |
|
2303 | |||
2388 | # (Possibly) reuse the delta from the revlog if allowed and |
|
2304 | # (Possibly) reuse the delta from the revlog if allowed and | |
2389 | # the revlog chunk is a delta. |
|
2305 | # the revlog chunk is a delta. | |
2390 | cachedelta = None |
|
2306 | cachedelta = None | |
2391 | rawtext = None |
|
2307 | rawtext = None | |
2392 | if populatecachedelta: |
|
2308 | if populatecachedelta: | |
2393 | dp = self.deltaparent(rev) |
|
2309 | dp = self.deltaparent(rev) | |
2394 | if dp != nullrev: |
|
2310 | if dp != nullrev: | |
2395 | cachedelta = (dp, bytes(self._chunk(rev))) |
|
2311 | cachedelta = (dp, bytes(self._chunk(rev))) | |
2396 |
|
2312 | |||
2397 | if not cachedelta: |
|
2313 | if not cachedelta: | |
2398 | rawtext = self.revision(rev, raw=True) |
|
2314 | rawtext = self.revision(rev, raw=True) | |
2399 |
|
2315 | |||
2400 |
|
2316 | |||
2401 | if deltareuse == self.DELTAREUSEFULLADD: |
|
2317 | if deltareuse == self.DELTAREUSEFULLADD: | |
2402 | destrevlog.addrevision(rawtext, tr, linkrev, p1, p2, |
|
2318 | destrevlog.addrevision(rawtext, tr, linkrev, p1, p2, | |
2403 | cachedelta=cachedelta, |
|
2319 | cachedelta=cachedelta, | |
2404 | node=node, flags=flags, |
|
2320 | node=node, flags=flags, | |
2405 | deltacomputer=deltacomputer) |
|
2321 | deltacomputer=deltacomputer) | |
2406 | else: |
|
2322 | else: | |
2407 | ifh = destrevlog.opener(destrevlog.indexfile, 'a+', |
|
2323 | ifh = destrevlog.opener(destrevlog.indexfile, 'a+', | |
2408 | checkambig=False) |
|
2324 | checkambig=False) | |
2409 | dfh = None |
|
2325 | dfh = None | |
2410 | if not destrevlog._inline: |
|
2326 | if not destrevlog._inline: | |
2411 | dfh = destrevlog.opener(destrevlog.datafile, 'a+') |
|
2327 | dfh = destrevlog.opener(destrevlog.datafile, 'a+') | |
2412 | try: |
|
2328 | try: | |
2413 | destrevlog._addrevision(node, rawtext, tr, linkrev, p1, |
|
2329 | destrevlog._addrevision(node, rawtext, tr, linkrev, p1, | |
2414 | p2, flags, cachedelta, ifh, dfh, |
|
2330 | p2, flags, cachedelta, ifh, dfh, | |
2415 | deltacomputer=deltacomputer) |
|
2331 | deltacomputer=deltacomputer) | |
2416 | finally: |
|
2332 | finally: | |
2417 | if dfh: |
|
2333 | if dfh: | |
2418 | dfh.close() |
|
2334 | dfh.close() | |
2419 | ifh.close() |
|
2335 | ifh.close() | |
2420 |
|
2336 | |||
2421 | if addrevisioncb: |
|
2337 | if addrevisioncb: | |
2422 | addrevisioncb(self, rev, node) |
|
2338 | addrevisioncb(self, rev, node) | |
2423 | finally: |
|
2339 | finally: | |
2424 | destrevlog._lazydeltabase = oldlazydeltabase |
|
2340 | destrevlog._lazydeltabase = oldlazydeltabase | |
2425 | destrevlog._deltabothparents = oldamd |
|
2341 | destrevlog._deltabothparents = oldamd | |
2426 |
|
2342 | |||
2427 | def censorrevision(self, node, tombstone=b''): |
|
2343 | def censorrevision(self, node, tombstone=b''): | |
2428 | if (self.version & 0xFFFF) == REVLOGV0: |
|
2344 | if (self.version & 0xFFFF) == REVLOGV0: | |
2429 | raise error.RevlogError(_('cannot censor with version %d revlogs') % |
|
2345 | raise error.RevlogError(_('cannot censor with version %d revlogs') % | |
2430 | self.version) |
|
2346 | self.version) | |
2431 |
|
2347 | |||
2432 | rev = self.rev(node) |
|
2348 | rev = self.rev(node) | |
2433 | tombstone = storageutil.packmeta({b'censored': tombstone}, b'') |
|
2349 | tombstone = storageutil.packmeta({b'censored': tombstone}, b'') | |
2434 |
|
2350 | |||
2435 | if len(tombstone) > self.rawsize(rev): |
|
2351 | if len(tombstone) > self.rawsize(rev): | |
2436 | raise error.Abort(_('censor tombstone must be no longer than ' |
|
2352 | raise error.Abort(_('censor tombstone must be no longer than ' | |
2437 | 'censored data')) |
|
2353 | 'censored data')) | |
2438 |
|
2354 | |||
2439 | # Using two files instead of one makes it easy to rewrite entry-by-entry |
|
2355 | # Using two files instead of one makes it easy to rewrite entry-by-entry | |
2440 | idxread = self.opener(self.indexfile, 'r') |
|
2356 | idxread = self.opener(self.indexfile, 'r') | |
2441 | idxwrite = self.opener(self.indexfile, 'wb', atomictemp=True) |
|
2357 | idxwrite = self.opener(self.indexfile, 'wb', atomictemp=True) | |
2442 | if self.version & FLAG_INLINE_DATA: |
|
2358 | if self.version & FLAG_INLINE_DATA: | |
2443 | dataread, datawrite = idxread, idxwrite |
|
2359 | dataread, datawrite = idxread, idxwrite | |
2444 | else: |
|
2360 | else: | |
2445 | dataread = self.opener(self.datafile, 'r') |
|
2361 | dataread = self.opener(self.datafile, 'r') | |
2446 | datawrite = self.opener(self.datafile, 'wb', atomictemp=True) |
|
2362 | datawrite = self.opener(self.datafile, 'wb', atomictemp=True) | |
2447 |
|
2363 | |||
2448 | # Copy all revlog data up to the entry to be censored. |
|
2364 | # Copy all revlog data up to the entry to be censored. | |
2449 | offset = self.start(rev) |
|
2365 | offset = self.start(rev) | |
2450 |
|
2366 | |||
2451 | for chunk in util.filechunkiter(idxread, limit=rev * self._io.size): |
|
2367 | for chunk in util.filechunkiter(idxread, limit=rev * self._io.size): | |
2452 | idxwrite.write(chunk) |
|
2368 | idxwrite.write(chunk) | |
2453 | for chunk in util.filechunkiter(dataread, limit=offset): |
|
2369 | for chunk in util.filechunkiter(dataread, limit=offset): | |
2454 | datawrite.write(chunk) |
|
2370 | datawrite.write(chunk) | |
2455 |
|
2371 | |||
2456 | def rewriteindex(r, newoffs, newdata=None): |
|
2372 | def rewriteindex(r, newoffs, newdata=None): | |
2457 | """Rewrite the index entry with a new data offset and new data. |
|
2373 | """Rewrite the index entry with a new data offset and new data. | |
2458 |
|
2374 | |||
2459 | The newdata argument, if given, is a tuple of three positive |
|
2375 | The newdata argument, if given, is a tuple of three positive | |
2460 | integers: (new compressed, new uncompressed, added flag bits). |
|
2376 | integers: (new compressed, new uncompressed, added flag bits). | |
2461 | """ |
|
2377 | """ | |
2462 | offlags, comp, uncomp, base, link, p1, p2, nodeid = self.index[r] |
|
2378 | offlags, comp, uncomp, base, link, p1, p2, nodeid = self.index[r] | |
2463 | flags = gettype(offlags) |
|
2379 | flags = gettype(offlags) | |
2464 | if newdata: |
|
2380 | if newdata: | |
2465 | comp, uncomp, nflags = newdata |
|
2381 | comp, uncomp, nflags = newdata | |
2466 | flags |= nflags |
|
2382 | flags |= nflags | |
2467 | offlags = offset_type(newoffs, flags) |
|
2383 | offlags = offset_type(newoffs, flags) | |
2468 | e = (offlags, comp, uncomp, r, link, p1, p2, nodeid) |
|
2384 | e = (offlags, comp, uncomp, r, link, p1, p2, nodeid) | |
2469 | idxwrite.write(self._io.packentry(e, None, self.version, r)) |
|
2385 | idxwrite.write(self._io.packentry(e, None, self.version, r)) | |
2470 | idxread.seek(self._io.size, 1) |
|
2386 | idxread.seek(self._io.size, 1) | |
2471 |
|
2387 | |||
2472 | def rewrite(r, offs, data, nflags=REVIDX_DEFAULT_FLAGS): |
|
2388 | def rewrite(r, offs, data, nflags=REVIDX_DEFAULT_FLAGS): | |
2473 | """Write the given fulltext with the given data offset. |
|
2389 | """Write the given fulltext with the given data offset. | |
2474 |
|
2390 | |||
2475 | Returns: |
|
2391 | Returns: | |
2476 | The integer number of data bytes written, for tracking data |
|
2392 | The integer number of data bytes written, for tracking data | |
2477 | offsets. |
|
2393 | offsets. | |
2478 | """ |
|
2394 | """ | |
2479 | flag, compdata = self.compress(data) |
|
2395 | flag, compdata = self.compress(data) | |
2480 | newcomp = len(flag) + len(compdata) |
|
2396 | newcomp = len(flag) + len(compdata) | |
2481 | rewriteindex(r, offs, (newcomp, len(data), nflags)) |
|
2397 | rewriteindex(r, offs, (newcomp, len(data), nflags)) | |
2482 | datawrite.write(flag) |
|
2398 | datawrite.write(flag) | |
2483 | datawrite.write(compdata) |
|
2399 | datawrite.write(compdata) | |
2484 | dataread.seek(self.length(r), 1) |
|
2400 | dataread.seek(self.length(r), 1) | |
2485 | return newcomp |
|
2401 | return newcomp | |
2486 |
|
2402 | |||
2487 | # Rewrite censored entry with (padded) tombstone data. |
|
2403 | # Rewrite censored entry with (padded) tombstone data. | |
2488 | pad = ' ' * (self.rawsize(rev) - len(tombstone)) |
|
2404 | pad = ' ' * (self.rawsize(rev) - len(tombstone)) | |
2489 | offset += rewrite(rev, offset, tombstone + pad, REVIDX_ISCENSORED) |
|
2405 | offset += rewrite(rev, offset, tombstone + pad, REVIDX_ISCENSORED) | |
2490 |
|
2406 | |||
2491 | # Rewrite all following filelog revisions fixing up offsets and deltas. |
|
2407 | # Rewrite all following filelog revisions fixing up offsets and deltas. | |
2492 | for srev in pycompat.xrange(rev + 1, len(self)): |
|
2408 | for srev in pycompat.xrange(rev + 1, len(self)): | |
2493 | if rev in self.parentrevs(srev): |
|
2409 | if rev in self.parentrevs(srev): | |
2494 | # Immediate children of censored node must be re-added as |
|
2410 | # Immediate children of censored node must be re-added as | |
2495 | # fulltext. |
|
2411 | # fulltext. | |
2496 | try: |
|
2412 | try: | |
2497 | revdata = self.revision(srev) |
|
2413 | revdata = self.revision(srev) | |
2498 | except error.CensoredNodeError as e: |
|
2414 | except error.CensoredNodeError as e: | |
2499 | revdata = e.tombstone |
|
2415 | revdata = e.tombstone | |
2500 | dlen = rewrite(srev, offset, revdata) |
|
2416 | dlen = rewrite(srev, offset, revdata) | |
2501 | else: |
|
2417 | else: | |
2502 | # Copy any other revision data verbatim after fixing up the |
|
2418 | # Copy any other revision data verbatim after fixing up the | |
2503 | # offset. |
|
2419 | # offset. | |
2504 | rewriteindex(srev, offset) |
|
2420 | rewriteindex(srev, offset) | |
2505 | dlen = self.length(srev) |
|
2421 | dlen = self.length(srev) | |
2506 | for chunk in util.filechunkiter(dataread, limit=dlen): |
|
2422 | for chunk in util.filechunkiter(dataread, limit=dlen): | |
2507 | datawrite.write(chunk) |
|
2423 | datawrite.write(chunk) | |
2508 | offset += dlen |
|
2424 | offset += dlen | |
2509 |
|
2425 | |||
2510 | idxread.close() |
|
2426 | idxread.close() | |
2511 | idxwrite.close() |
|
2427 | idxwrite.close() | |
2512 | if dataread is not idxread: |
|
2428 | if dataread is not idxread: | |
2513 | dataread.close() |
|
2429 | dataread.close() | |
2514 | datawrite.close() |
|
2430 | datawrite.close() | |
2515 |
|
2431 | |||
2516 | def verifyintegrity(self, state): |
|
2432 | def verifyintegrity(self, state): | |
2517 | """Verifies the integrity of the revlog. |
|
2433 | """Verifies the integrity of the revlog. | |
2518 |
|
2434 | |||
2519 | Yields ``revlogproblem`` instances describing problems that are |
|
2435 | Yields ``revlogproblem`` instances describing problems that are | |
2520 | found. |
|
2436 | found. | |
2521 | """ |
|
2437 | """ | |
2522 | dd, di = self.checksize() |
|
2438 | dd, di = self.checksize() | |
2523 | if dd: |
|
2439 | if dd: | |
2524 | yield revlogproblem(error=_('data length off by %d bytes') % dd) |
|
2440 | yield revlogproblem(error=_('data length off by %d bytes') % dd) | |
2525 | if di: |
|
2441 | if di: | |
2526 | yield revlogproblem(error=_('index contains %d extra bytes') % di) |
|
2442 | yield revlogproblem(error=_('index contains %d extra bytes') % di) | |
2527 |
|
2443 | |||
2528 | version = self.version & 0xFFFF |
|
2444 | version = self.version & 0xFFFF | |
2529 |
|
2445 | |||
2530 | # The verifier tells us what version revlog we should be. |
|
2446 | # The verifier tells us what version revlog we should be. | |
2531 | if version != state['expectedversion']: |
|
2447 | if version != state['expectedversion']: | |
2532 | yield revlogproblem( |
|
2448 | yield revlogproblem( | |
2533 | warning=_("warning: '%s' uses revlog format %d; expected %d") % |
|
2449 | warning=_("warning: '%s' uses revlog format %d; expected %d") % | |
2534 | (self.indexfile, version, state['expectedversion'])) |
|
2450 | (self.indexfile, version, state['expectedversion'])) | |
2535 |
|
2451 | |||
2536 | state['skipread'] = set() |
|
2452 | state['skipread'] = set() | |
2537 |
|
2453 | |||
2538 | for rev in self: |
|
2454 | for rev in self: | |
2539 | node = self.node(rev) |
|
2455 | node = self.node(rev) | |
2540 |
|
2456 | |||
2541 | # Verify contents. 4 cases to care about: |
|
2457 | # Verify contents. 4 cases to care about: | |
2542 | # |
|
2458 | # | |
2543 | # common: the most common case |
|
2459 | # common: the most common case | |
2544 | # rename: with a rename |
|
2460 | # rename: with a rename | |
2545 | # meta: file content starts with b'\1\n', the metadata |
|
2461 | # meta: file content starts with b'\1\n', the metadata | |
2546 | # header defined in filelog.py, but without a rename |
|
2462 | # header defined in filelog.py, but without a rename | |
2547 | # ext: content stored externally |
|
2463 | # ext: content stored externally | |
2548 | # |
|
2464 | # | |
2549 | # More formally, their differences are shown below: |
|
2465 | # More formally, their differences are shown below: | |
2550 | # |
|
2466 | # | |
2551 | # | common | rename | meta | ext |
|
2467 | # | common | rename | meta | ext | |
2552 | # ------------------------------------------------------- |
|
2468 | # ------------------------------------------------------- | |
2553 | # flags() | 0 | 0 | 0 | not 0 |
|
2469 | # flags() | 0 | 0 | 0 | not 0 | |
2554 | # renamed() | False | True | False | ? |
|
2470 | # renamed() | False | True | False | ? | |
2555 | # rawtext[0:2]=='\1\n'| False | True | True | ? |
|
2471 | # rawtext[0:2]=='\1\n'| False | True | True | ? | |
2556 | # |
|
2472 | # | |
2557 | # "rawtext" means the raw text stored in revlog data, which |
|
2473 | # "rawtext" means the raw text stored in revlog data, which | |
2558 | # could be retrieved by "revision(rev, raw=True)". "text" |
|
2474 | # could be retrieved by "revision(rev, raw=True)". "text" | |
2559 | # mentioned below is "revision(rev, raw=False)". |
|
2475 | # mentioned below is "revision(rev, raw=False)". | |
2560 | # |
|
2476 | # | |
2561 | # There are 3 different lengths stored physically: |
|
2477 | # There are 3 different lengths stored physically: | |
2562 | # 1. L1: rawsize, stored in revlog index |
|
2478 | # 1. L1: rawsize, stored in revlog index | |
2563 | # 2. L2: len(rawtext), stored in revlog data |
|
2479 | # 2. L2: len(rawtext), stored in revlog data | |
2564 | # 3. L3: len(text), stored in revlog data if flags==0, or |
|
2480 | # 3. L3: len(text), stored in revlog data if flags==0, or | |
2565 | # possibly somewhere else if flags!=0 |
|
2481 | # possibly somewhere else if flags!=0 | |
2566 | # |
|
2482 | # | |
2567 | # L1 should be equal to L2. L3 could be different from them. |
|
2483 | # L1 should be equal to L2. L3 could be different from them. | |
2568 | # "text" may or may not affect commit hash depending on flag |
|
2484 | # "text" may or may not affect commit hash depending on flag | |
2569 | # processors (see revlog.addflagprocessor). |
|
2485 | # processors (see revlog.addflagprocessor). | |
2570 | # |
|
2486 | # | |
2571 | # | common | rename | meta | ext |
|
2487 | # | common | rename | meta | ext | |
2572 | # ------------------------------------------------- |
|
2488 | # ------------------------------------------------- | |
2573 | # rawsize() | L1 | L1 | L1 | L1 |
|
2489 | # rawsize() | L1 | L1 | L1 | L1 | |
2574 | # size() | L1 | L2-LM | L1(*) | L1 (?) |
|
2490 | # size() | L1 | L2-LM | L1(*) | L1 (?) | |
2575 | # len(rawtext) | L2 | L2 | L2 | L2 |
|
2491 | # len(rawtext) | L2 | L2 | L2 | L2 | |
2576 | # len(text) | L2 | L2 | L2 | L3 |
|
2492 | # len(text) | L2 | L2 | L2 | L3 | |
2577 | # len(read()) | L2 | L2-LM | L2-LM | L3 (?) |
|
2493 | # len(read()) | L2 | L2-LM | L2-LM | L3 (?) | |
2578 | # |
|
2494 | # | |
2579 | # LM: length of metadata, depending on rawtext |
|
2495 | # LM: length of metadata, depending on rawtext | |
2580 | # (*): not ideal, see comment in filelog.size |
|
2496 | # (*): not ideal, see comment in filelog.size | |
2581 | # (?): could be "- len(meta)" if the resolved content has |
|
2497 | # (?): could be "- len(meta)" if the resolved content has | |
2582 | # rename metadata |
|
2498 | # rename metadata | |
2583 | # |
|
2499 | # | |
2584 | # Checks needed to be done: |
|
2500 | # Checks needed to be done: | |
2585 | # 1. length check: L1 == L2, in all cases. |
|
2501 | # 1. length check: L1 == L2, in all cases. | |
2586 | # 2. hash check: depending on flag processor, we may need to |
|
2502 | # 2. hash check: depending on flag processor, we may need to | |
2587 | # use either "text" (external), or "rawtext" (in revlog). |
|
2503 | # use either "text" (external), or "rawtext" (in revlog). | |
2588 |
|
2504 | |||
2589 | try: |
|
2505 | try: | |
2590 | skipflags = state.get('skipflags', 0) |
|
2506 | skipflags = state.get('skipflags', 0) | |
2591 | if skipflags: |
|
2507 | if skipflags: | |
2592 | skipflags &= self.flags(rev) |
|
2508 | skipflags &= self.flags(rev) | |
2593 |
|
2509 | |||
2594 | if skipflags: |
|
2510 | if skipflags: | |
2595 | state['skipread'].add(node) |
|
2511 | state['skipread'].add(node) | |
2596 | else: |
|
2512 | else: | |
2597 | # Side-effect: read content and verify hash. |
|
2513 | # Side-effect: read content and verify hash. | |
2598 | self.revision(node) |
|
2514 | self.revision(node) | |
2599 |
|
2515 | |||
2600 | l1 = self.rawsize(rev) |
|
2516 | l1 = self.rawsize(rev) | |
2601 | l2 = len(self.revision(node, raw=True)) |
|
2517 | l2 = len(self.revision(node, raw=True)) | |
2602 |
|
2518 | |||
2603 | if l1 != l2: |
|
2519 | if l1 != l2: | |
2604 | yield revlogproblem( |
|
2520 | yield revlogproblem( | |
2605 | error=_('unpacked size is %d, %d expected') % (l2, l1), |
|
2521 | error=_('unpacked size is %d, %d expected') % (l2, l1), | |
2606 | node=node) |
|
2522 | node=node) | |
2607 |
|
2523 | |||
2608 | except error.CensoredNodeError: |
|
2524 | except error.CensoredNodeError: | |
2609 | if state['erroroncensored']: |
|
2525 | if state['erroroncensored']: | |
2610 | yield revlogproblem(error=_('censored file data'), |
|
2526 | yield revlogproblem(error=_('censored file data'), | |
2611 | node=node) |
|
2527 | node=node) | |
2612 | state['skipread'].add(node) |
|
2528 | state['skipread'].add(node) | |
2613 | except Exception as e: |
|
2529 | except Exception as e: | |
2614 | yield revlogproblem( |
|
2530 | yield revlogproblem( | |
2615 | error=_('unpacking %s: %s') % (short(node), |
|
2531 | error=_('unpacking %s: %s') % (short(node), | |
2616 | stringutil.forcebytestr(e)), |
|
2532 | stringutil.forcebytestr(e)), | |
2617 | node=node) |
|
2533 | node=node) | |
2618 | state['skipread'].add(node) |
|
2534 | state['skipread'].add(node) | |
2619 |
|
2535 | |||
2620 | def storageinfo(self, exclusivefiles=False, sharedfiles=False, |
|
2536 | def storageinfo(self, exclusivefiles=False, sharedfiles=False, | |
2621 | revisionscount=False, trackedsize=False, |
|
2537 | revisionscount=False, trackedsize=False, | |
2622 | storedsize=False): |
|
2538 | storedsize=False): | |
2623 | d = {} |
|
2539 | d = {} | |
2624 |
|
2540 | |||
2625 | if exclusivefiles: |
|
2541 | if exclusivefiles: | |
2626 | d['exclusivefiles'] = [(self.opener, self.indexfile)] |
|
2542 | d['exclusivefiles'] = [(self.opener, self.indexfile)] | |
2627 | if not self._inline: |
|
2543 | if not self._inline: | |
2628 | d['exclusivefiles'].append((self.opener, self.datafile)) |
|
2544 | d['exclusivefiles'].append((self.opener, self.datafile)) | |
2629 |
|
2545 | |||
2630 | if sharedfiles: |
|
2546 | if sharedfiles: | |
2631 | d['sharedfiles'] = [] |
|
2547 | d['sharedfiles'] = [] | |
2632 |
|
2548 | |||
2633 | if revisionscount: |
|
2549 | if revisionscount: | |
2634 | d['revisionscount'] = len(self) |
|
2550 | d['revisionscount'] = len(self) | |
2635 |
|
2551 | |||
2636 | if trackedsize: |
|
2552 | if trackedsize: | |
2637 | d['trackedsize'] = sum(map(self.rawsize, iter(self))) |
|
2553 | d['trackedsize'] = sum(map(self.rawsize, iter(self))) | |
2638 |
|
2554 | |||
2639 | if storedsize: |
|
2555 | if storedsize: | |
2640 | d['storedsize'] = sum(self.opener.stat(path).st_size |
|
2556 | d['storedsize'] = sum(self.opener.stat(path).st_size | |
2641 | for path in self.files()) |
|
2557 | for path in self.files()) | |
2642 |
|
2558 | |||
2643 | return d |
|
2559 | return d |
@@ -1,264 +1,410 b'' | |||||
1 | # storageutil.py - Storage functionality agnostic of backend implementation. |
|
1 | # storageutil.py - Storage functionality agnostic of backend implementation. | |
2 | # |
|
2 | # | |
3 | # Copyright 2018 Gregory Szorc <gregory.szorc@gmail.com> |
|
3 | # Copyright 2018 Gregory Szorc <gregory.szorc@gmail.com> | |
4 | # |
|
4 | # | |
5 | # This software may be used and distributed according to the terms of the |
|
5 | # This software may be used and distributed according to the terms of the | |
6 | # GNU General Public License version 2 or any later version. |
|
6 | # GNU General Public License version 2 or any later version. | |
7 |
|
7 | |||
8 | from __future__ import absolute_import |
|
8 | from __future__ import absolute_import | |
9 |
|
9 | |||
10 | import hashlib |
|
10 | import hashlib | |
11 | import re |
|
11 | import re | |
12 |
|
12 | |||
13 | from ..i18n import _ |
|
13 | from ..i18n import _ | |
14 | from ..node import ( |
|
14 | from ..node import ( | |
15 | bin, |
|
15 | bin, | |
16 | nullid, |
|
16 | nullid, | |
17 | nullrev, |
|
17 | nullrev, | |
18 | ) |
|
18 | ) | |
19 | from .. import ( |
|
19 | from .. import ( | |
20 | error, |
|
20 | error, | |
21 | pycompat, |
|
21 | pycompat, | |
22 | ) |
|
22 | ) | |
23 |
|
23 | |||
24 | _nullhash = hashlib.sha1(nullid) |
|
24 | _nullhash = hashlib.sha1(nullid) | |
25 |
|
25 | |||
26 | def hashrevisionsha1(text, p1, p2): |
|
26 | def hashrevisionsha1(text, p1, p2): | |
27 | """Compute the SHA-1 for revision data and its parents. |
|
27 | """Compute the SHA-1 for revision data and its parents. | |
28 |
|
28 | |||
29 | This hash combines both the current file contents and its history |
|
29 | This hash combines both the current file contents and its history | |
30 | in a manner that makes it easy to distinguish nodes with the same |
|
30 | in a manner that makes it easy to distinguish nodes with the same | |
31 | content in the revision graph. |
|
31 | content in the revision graph. | |
32 | """ |
|
32 | """ | |
33 | # As of now, if one of the parent node is null, p2 is null |
|
33 | # As of now, if one of the parent node is null, p2 is null | |
34 | if p2 == nullid: |
|
34 | if p2 == nullid: | |
35 | # deep copy of a hash is faster than creating one |
|
35 | # deep copy of a hash is faster than creating one | |
36 | s = _nullhash.copy() |
|
36 | s = _nullhash.copy() | |
37 | s.update(p1) |
|
37 | s.update(p1) | |
38 | else: |
|
38 | else: | |
39 | # none of the parent nodes are nullid |
|
39 | # none of the parent nodes are nullid | |
40 | if p1 < p2: |
|
40 | if p1 < p2: | |
41 | a = p1 |
|
41 | a = p1 | |
42 | b = p2 |
|
42 | b = p2 | |
43 | else: |
|
43 | else: | |
44 | a = p2 |
|
44 | a = p2 | |
45 | b = p1 |
|
45 | b = p1 | |
46 | s = hashlib.sha1(a) |
|
46 | s = hashlib.sha1(a) | |
47 | s.update(b) |
|
47 | s.update(b) | |
48 | s.update(text) |
|
48 | s.update(text) | |
49 | return s.digest() |
|
49 | return s.digest() | |
50 |
|
50 | |||
51 | METADATA_RE = re.compile(b'\x01\n') |
|
51 | METADATA_RE = re.compile(b'\x01\n') | |
52 |
|
52 | |||
53 | def parsemeta(text): |
|
53 | def parsemeta(text): | |
54 | """Parse metadata header from revision data. |
|
54 | """Parse metadata header from revision data. | |
55 |
|
55 | |||
56 | Returns a 2-tuple of (metadata, offset), where both can be None if there |
|
56 | Returns a 2-tuple of (metadata, offset), where both can be None if there | |
57 | is no metadata. |
|
57 | is no metadata. | |
58 | """ |
|
58 | """ | |
59 | # text can be buffer, so we can't use .startswith or .index |
|
59 | # text can be buffer, so we can't use .startswith or .index | |
60 | if text[:2] != b'\x01\n': |
|
60 | if text[:2] != b'\x01\n': | |
61 | return None, None |
|
61 | return None, None | |
62 | s = METADATA_RE.search(text, 2).start() |
|
62 | s = METADATA_RE.search(text, 2).start() | |
63 | mtext = text[2:s] |
|
63 | mtext = text[2:s] | |
64 | meta = {} |
|
64 | meta = {} | |
65 | for l in mtext.splitlines(): |
|
65 | for l in mtext.splitlines(): | |
66 | k, v = l.split(b': ', 1) |
|
66 | k, v = l.split(b': ', 1) | |
67 | meta[k] = v |
|
67 | meta[k] = v | |
68 | return meta, s + 2 |
|
68 | return meta, s + 2 | |
69 |
|
69 | |||
70 | def packmeta(meta, text): |
|
70 | def packmeta(meta, text): | |
71 | """Add metadata to fulltext to produce revision text.""" |
|
71 | """Add metadata to fulltext to produce revision text.""" | |
72 | keys = sorted(meta) |
|
72 | keys = sorted(meta) | |
73 | metatext = b''.join(b'%s: %s\n' % (k, meta[k]) for k in keys) |
|
73 | metatext = b''.join(b'%s: %s\n' % (k, meta[k]) for k in keys) | |
74 | return b'\x01\n%s\x01\n%s' % (metatext, text) |
|
74 | return b'\x01\n%s\x01\n%s' % (metatext, text) | |
75 |
|
75 | |||
76 | def iscensoredtext(text): |
|
76 | def iscensoredtext(text): | |
77 | meta = parsemeta(text)[0] |
|
77 | meta = parsemeta(text)[0] | |
78 | return meta and b'censored' in meta |
|
78 | return meta and b'censored' in meta | |
79 |
|
79 | |||
80 | def filtermetadata(text): |
|
80 | def filtermetadata(text): | |
81 | """Extract just the revision data from source text. |
|
81 | """Extract just the revision data from source text. | |
82 |
|
82 | |||
83 | Returns ``text`` unless it has a metadata header, in which case we return |
|
83 | Returns ``text`` unless it has a metadata header, in which case we return | |
84 | a new buffer without hte metadata. |
|
84 | a new buffer without hte metadata. | |
85 | """ |
|
85 | """ | |
86 | if not text.startswith(b'\x01\n'): |
|
86 | if not text.startswith(b'\x01\n'): | |
87 | return text |
|
87 | return text | |
88 |
|
88 | |||
89 | offset = text.index(b'\x01\n', 2) |
|
89 | offset = text.index(b'\x01\n', 2) | |
90 | return text[offset + 2:] |
|
90 | return text[offset + 2:] | |
91 |
|
91 | |||
92 | def filerevisioncopied(store, node): |
|
92 | def filerevisioncopied(store, node): | |
93 | """Resolve file revision copy metadata. |
|
93 | """Resolve file revision copy metadata. | |
94 |
|
94 | |||
95 | Returns ``False`` if the file has no copy metadata. Otherwise a |
|
95 | Returns ``False`` if the file has no copy metadata. Otherwise a | |
96 | 2-tuple of the source filename and node. |
|
96 | 2-tuple of the source filename and node. | |
97 | """ |
|
97 | """ | |
98 | if store.parents(node)[0] != nullid: |
|
98 | if store.parents(node)[0] != nullid: | |
99 | return False |
|
99 | return False | |
100 |
|
100 | |||
101 | meta = parsemeta(store.revision(node))[0] |
|
101 | meta = parsemeta(store.revision(node))[0] | |
102 |
|
102 | |||
103 | # copy and copyrev occur in pairs. In rare cases due to old bugs, |
|
103 | # copy and copyrev occur in pairs. In rare cases due to old bugs, | |
104 | # one can occur without the other. So ensure both are present to flag |
|
104 | # one can occur without the other. So ensure both are present to flag | |
105 | # as a copy. |
|
105 | # as a copy. | |
106 | if meta and b'copy' in meta and b'copyrev' in meta: |
|
106 | if meta and b'copy' in meta and b'copyrev' in meta: | |
107 | return meta[b'copy'], bin(meta[b'copyrev']) |
|
107 | return meta[b'copy'], bin(meta[b'copyrev']) | |
108 |
|
108 | |||
109 | return False |
|
109 | return False | |
110 |
|
110 | |||
111 | def filedataequivalent(store, node, filedata): |
|
111 | def filedataequivalent(store, node, filedata): | |
112 | """Determines whether file data is equivalent to a stored node. |
|
112 | """Determines whether file data is equivalent to a stored node. | |
113 |
|
113 | |||
114 | Returns True if the passed file data would hash to the same value |
|
114 | Returns True if the passed file data would hash to the same value | |
115 | as a stored revision and False otherwise. |
|
115 | as a stored revision and False otherwise. | |
116 |
|
116 | |||
117 | When a stored revision is censored, filedata must be empty to have |
|
117 | When a stored revision is censored, filedata must be empty to have | |
118 | equivalence. |
|
118 | equivalence. | |
119 |
|
119 | |||
120 | When a stored revision has copy metadata, it is ignored as part |
|
120 | When a stored revision has copy metadata, it is ignored as part | |
121 | of the compare. |
|
121 | of the compare. | |
122 | """ |
|
122 | """ | |
123 |
|
123 | |||
124 | if filedata.startswith(b'\x01\n'): |
|
124 | if filedata.startswith(b'\x01\n'): | |
125 | revisiontext = b'\x01\n\x01\n' + filedata |
|
125 | revisiontext = b'\x01\n\x01\n' + filedata | |
126 | else: |
|
126 | else: | |
127 | revisiontext = filedata |
|
127 | revisiontext = filedata | |
128 |
|
128 | |||
129 | p1, p2 = store.parents(node) |
|
129 | p1, p2 = store.parents(node) | |
130 |
|
130 | |||
131 | computednode = hashrevisionsha1(revisiontext, p1, p2) |
|
131 | computednode = hashrevisionsha1(revisiontext, p1, p2) | |
132 |
|
132 | |||
133 | if computednode == node: |
|
133 | if computednode == node: | |
134 | return True |
|
134 | return True | |
135 |
|
135 | |||
136 | # Censored files compare against the empty file. |
|
136 | # Censored files compare against the empty file. | |
137 | if store.iscensored(store.rev(node)): |
|
137 | if store.iscensored(store.rev(node)): | |
138 | return filedata == b'' |
|
138 | return filedata == b'' | |
139 |
|
139 | |||
140 | # Renaming a file produces a different hash, even if the data |
|
140 | # Renaming a file produces a different hash, even if the data | |
141 | # remains unchanged. Check if that's the case. |
|
141 | # remains unchanged. Check if that's the case. | |
142 | if store.renamed(node): |
|
142 | if store.renamed(node): | |
143 | return store.read(node) == filedata |
|
143 | return store.read(node) == filedata | |
144 |
|
144 | |||
145 | return False |
|
145 | return False | |
146 |
|
146 | |||
147 | def iterrevs(storelen, start=0, stop=None): |
|
147 | def iterrevs(storelen, start=0, stop=None): | |
148 | """Iterate over revision numbers in a store.""" |
|
148 | """Iterate over revision numbers in a store.""" | |
149 | step = 1 |
|
149 | step = 1 | |
150 |
|
150 | |||
151 | if stop is not None: |
|
151 | if stop is not None: | |
152 | if start > stop: |
|
152 | if start > stop: | |
153 | step = -1 |
|
153 | step = -1 | |
154 | stop += step |
|
154 | stop += step | |
155 | if stop > storelen: |
|
155 | if stop > storelen: | |
156 | stop = storelen |
|
156 | stop = storelen | |
157 | else: |
|
157 | else: | |
158 | stop = storelen |
|
158 | stop = storelen | |
159 |
|
159 | |||
160 | return pycompat.xrange(start, stop, step) |
|
160 | return pycompat.xrange(start, stop, step) | |
161 |
|
161 | |||
162 | def fileidlookup(store, fileid, identifier): |
|
162 | def fileidlookup(store, fileid, identifier): | |
163 | """Resolve the file node for a value. |
|
163 | """Resolve the file node for a value. | |
164 |
|
164 | |||
165 | ``store`` is an object implementing the ``ifileindex`` interface. |
|
165 | ``store`` is an object implementing the ``ifileindex`` interface. | |
166 |
|
166 | |||
167 | ``fileid`` can be: |
|
167 | ``fileid`` can be: | |
168 |
|
168 | |||
169 | * A 20 byte binary node. |
|
169 | * A 20 byte binary node. | |
170 | * An integer revision number |
|
170 | * An integer revision number | |
171 | * A 40 byte hex node. |
|
171 | * A 40 byte hex node. | |
172 | * A bytes that can be parsed as an integer representing a revision number. |
|
172 | * A bytes that can be parsed as an integer representing a revision number. | |
173 |
|
173 | |||
174 | ``identifier`` is used to populate ``error.LookupError`` with an identifier |
|
174 | ``identifier`` is used to populate ``error.LookupError`` with an identifier | |
175 | for the store. |
|
175 | for the store. | |
176 |
|
176 | |||
177 | Raises ``error.LookupError`` on failure. |
|
177 | Raises ``error.LookupError`` on failure. | |
178 | """ |
|
178 | """ | |
179 | if isinstance(fileid, int): |
|
179 | if isinstance(fileid, int): | |
180 | try: |
|
180 | try: | |
181 | return store.node(fileid) |
|
181 | return store.node(fileid) | |
182 | except IndexError: |
|
182 | except IndexError: | |
183 | raise error.LookupError(fileid, identifier, _('no match found')) |
|
183 | raise error.LookupError(fileid, identifier, _('no match found')) | |
184 |
|
184 | |||
185 | if len(fileid) == 20: |
|
185 | if len(fileid) == 20: | |
186 | try: |
|
186 | try: | |
187 | store.rev(fileid) |
|
187 | store.rev(fileid) | |
188 | return fileid |
|
188 | return fileid | |
189 | except error.LookupError: |
|
189 | except error.LookupError: | |
190 | pass |
|
190 | pass | |
191 |
|
191 | |||
192 | if len(fileid) == 40: |
|
192 | if len(fileid) == 40: | |
193 | try: |
|
193 | try: | |
194 | rawnode = bin(fileid) |
|
194 | rawnode = bin(fileid) | |
195 | store.rev(rawnode) |
|
195 | store.rev(rawnode) | |
196 | return rawnode |
|
196 | return rawnode | |
197 | except TypeError: |
|
197 | except TypeError: | |
198 | pass |
|
198 | pass | |
199 |
|
199 | |||
200 | try: |
|
200 | try: | |
201 | rev = int(fileid) |
|
201 | rev = int(fileid) | |
202 |
|
202 | |||
203 | if b'%d' % rev != fileid: |
|
203 | if b'%d' % rev != fileid: | |
204 | raise ValueError |
|
204 | raise ValueError | |
205 |
|
205 | |||
206 | try: |
|
206 | try: | |
207 | return store.node(rev) |
|
207 | return store.node(rev) | |
208 | except (IndexError, TypeError): |
|
208 | except (IndexError, TypeError): | |
209 | pass |
|
209 | pass | |
210 | except (ValueError, OverflowError): |
|
210 | except (ValueError, OverflowError): | |
211 | pass |
|
211 | pass | |
212 |
|
212 | |||
213 | raise error.LookupError(fileid, identifier, _('no match found')) |
|
213 | raise error.LookupError(fileid, identifier, _('no match found')) | |
214 |
|
214 | |||
215 | def resolvestripinfo(minlinkrev, tiprev, headrevs, linkrevfn, parentrevsfn): |
|
215 | def resolvestripinfo(minlinkrev, tiprev, headrevs, linkrevfn, parentrevsfn): | |
216 | """Resolve information needed to strip revisions. |
|
216 | """Resolve information needed to strip revisions. | |
217 |
|
217 | |||
218 | Finds the minimum revision number that must be stripped in order to |
|
218 | Finds the minimum revision number that must be stripped in order to | |
219 | strip ``minlinkrev``. |
|
219 | strip ``minlinkrev``. | |
220 |
|
220 | |||
221 | Returns a 2-tuple of the minimum revision number to do that and a set |
|
221 | Returns a 2-tuple of the minimum revision number to do that and a set | |
222 | of all revision numbers that have linkrevs that would be broken |
|
222 | of all revision numbers that have linkrevs that would be broken | |
223 | by that strip. |
|
223 | by that strip. | |
224 |
|
224 | |||
225 | ``tiprev`` is the current tip-most revision. It is ``len(store) - 1``. |
|
225 | ``tiprev`` is the current tip-most revision. It is ``len(store) - 1``. | |
226 | ``headrevs`` is an iterable of head revisions. |
|
226 | ``headrevs`` is an iterable of head revisions. | |
227 | ``linkrevfn`` is a callable that receives a revision and returns a linked |
|
227 | ``linkrevfn`` is a callable that receives a revision and returns a linked | |
228 | revision. |
|
228 | revision. | |
229 | ``parentrevsfn`` is a callable that receives a revision number and returns |
|
229 | ``parentrevsfn`` is a callable that receives a revision number and returns | |
230 | an iterable of its parent revision numbers. |
|
230 | an iterable of its parent revision numbers. | |
231 | """ |
|
231 | """ | |
232 | brokenrevs = set() |
|
232 | brokenrevs = set() | |
233 | strippoint = tiprev + 1 |
|
233 | strippoint = tiprev + 1 | |
234 |
|
234 | |||
235 | heads = {} |
|
235 | heads = {} | |
236 | futurelargelinkrevs = set() |
|
236 | futurelargelinkrevs = set() | |
237 | for head in headrevs: |
|
237 | for head in headrevs: | |
238 | headlinkrev = linkrevfn(head) |
|
238 | headlinkrev = linkrevfn(head) | |
239 | heads[head] = headlinkrev |
|
239 | heads[head] = headlinkrev | |
240 | if headlinkrev >= minlinkrev: |
|
240 | if headlinkrev >= minlinkrev: | |
241 | futurelargelinkrevs.add(headlinkrev) |
|
241 | futurelargelinkrevs.add(headlinkrev) | |
242 |
|
242 | |||
243 | # This algorithm involves walking down the rev graph, starting at the |
|
243 | # This algorithm involves walking down the rev graph, starting at the | |
244 | # heads. Since the revs are topologically sorted according to linkrev, |
|
244 | # heads. Since the revs are topologically sorted according to linkrev, | |
245 | # once all head linkrevs are below the minlink, we know there are |
|
245 | # once all head linkrevs are below the minlink, we know there are | |
246 | # no more revs that could have a linkrev greater than minlink. |
|
246 | # no more revs that could have a linkrev greater than minlink. | |
247 | # So we can stop walking. |
|
247 | # So we can stop walking. | |
248 | while futurelargelinkrevs: |
|
248 | while futurelargelinkrevs: | |
249 | strippoint -= 1 |
|
249 | strippoint -= 1 | |
250 | linkrev = heads.pop(strippoint) |
|
250 | linkrev = heads.pop(strippoint) | |
251 |
|
251 | |||
252 | if linkrev < minlinkrev: |
|
252 | if linkrev < minlinkrev: | |
253 | brokenrevs.add(strippoint) |
|
253 | brokenrevs.add(strippoint) | |
254 | else: |
|
254 | else: | |
255 | futurelargelinkrevs.remove(linkrev) |
|
255 | futurelargelinkrevs.remove(linkrev) | |
256 |
|
256 | |||
257 | for p in parentrevsfn(strippoint): |
|
257 | for p in parentrevsfn(strippoint): | |
258 | if p != nullrev: |
|
258 | if p != nullrev: | |
259 | plinkrev = linkrevfn(p) |
|
259 | plinkrev = linkrevfn(p) | |
260 | heads[p] = plinkrev |
|
260 | heads[p] = plinkrev | |
261 | if plinkrev >= minlinkrev: |
|
261 | if plinkrev >= minlinkrev: | |
262 | futurelargelinkrevs.add(plinkrev) |
|
262 | futurelargelinkrevs.add(plinkrev) | |
263 |
|
263 | |||
264 | return strippoint, brokenrevs |
|
264 | return strippoint, brokenrevs | |
|
265 | ||||
|
266 | def emitrevisions(store, revs, resultcls, deltaparentfn, candeltafn, | |||
|
267 | rawsizefn, revdifffn, flagsfn, sendfulltext=False, | |||
|
268 | revisiondata=False, assumehaveparentrevisions=False, | |||
|
269 | deltaprevious=False): | |||
|
270 | """Generic implementation of ifiledata.emitrevisions(). | |||
|
271 | ||||
|
272 | Emitting revision data is subtly complex. This function attempts to | |||
|
273 | encapsulate all the logic for doing so in a backend-agnostic way. | |||
|
274 | ||||
|
275 | ``store`` | |||
|
276 | Object conforming to ``ifilestorage`` interface. | |||
|
277 | ||||
|
278 | ``revs`` | |||
|
279 | List of integer revision numbers whose data to emit. | |||
|
280 | ||||
|
281 | ``resultcls`` | |||
|
282 | A type implementing the ``irevisiondelta`` interface that will be | |||
|
283 | constructed and returned. | |||
|
284 | ||||
|
285 | ``deltaparentfn`` | |||
|
286 | Callable receiving a revision number and returning the revision number | |||
|
287 | of a revision that the internal delta is stored against. This delta | |||
|
288 | will be preferred over computing a new arbitrary delta. | |||
|
289 | ||||
|
290 | ``candeltafn`` | |||
|
291 | Callable receiving a pair of revision numbers that returns a bool | |||
|
292 | indicating whether a delta between them can be produced. | |||
|
293 | ||||
|
294 | ``rawsizefn`` | |||
|
295 | Callable receiving a revision number and returning the length of the | |||
|
296 | ``store.revision(rev, raw=True)``. | |||
|
297 | ||||
|
298 | ``revdifffn`` | |||
|
299 | Callable receiving a pair of revision numbers that returns a delta | |||
|
300 | between them. | |||
|
301 | ||||
|
302 | ``flagsfn`` | |||
|
303 | Callable receiving a revision number and returns the integer flags | |||
|
304 | value for it. | |||
|
305 | ||||
|
306 | ``sendfulltext`` | |||
|
307 | Whether to send fulltext revisions instead of deltas, if allowed. | |||
|
308 | ||||
|
309 | ``revisiondata`` | |||
|
310 | ``assumehaveparentrevisions`` | |||
|
311 | ``deltaprevious`` | |||
|
312 | See ``ifiledata.emitrevisions()`` interface documentation. | |||
|
313 | """ | |||
|
314 | ||||
|
315 | fnode = store.node | |||
|
316 | ||||
|
317 | prevrev = None | |||
|
318 | ||||
|
319 | if deltaprevious or assumehaveparentrevisions: | |||
|
320 | prevrev = store.parentrevs(revs[0])[0] | |||
|
321 | ||||
|
322 | # Set of revs available to delta against. | |||
|
323 | available = set() | |||
|
324 | ||||
|
325 | for rev in revs: | |||
|
326 | if rev == nullrev: | |||
|
327 | continue | |||
|
328 | ||||
|
329 | node = fnode(rev) | |||
|
330 | deltaparentrev = deltaparentfn(rev) | |||
|
331 | p1rev, p2rev = store.parentrevs(rev) | |||
|
332 | ||||
|
333 | # Forced delta against previous mode. | |||
|
334 | if deltaprevious: | |||
|
335 | baserev = prevrev | |||
|
336 | ||||
|
337 | # We're instructed to send fulltext. Honor that. | |||
|
338 | elif sendfulltext: | |||
|
339 | baserev = nullrev | |||
|
340 | ||||
|
341 | # There is a delta in storage. We try to use that because it | |||
|
342 | # amounts to effectively copying data from storage and is | |||
|
343 | # therefore the fastest. | |||
|
344 | elif deltaparentrev != nullrev: | |||
|
345 | # Base revision was already emitted in this group. We can | |||
|
346 | # always safely use the delta. | |||
|
347 | if deltaparentrev in available: | |||
|
348 | baserev = deltaparentrev | |||
|
349 | ||||
|
350 | # Base revision is a parent that hasn't been emitted already. | |||
|
351 | # Use it if we can assume the receiver has the parent revision. | |||
|
352 | elif (assumehaveparentrevisions | |||
|
353 | and deltaparentrev in (p1rev, p2rev)): | |||
|
354 | baserev = deltaparentrev | |||
|
355 | ||||
|
356 | # No guarantee the receiver has the delta parent. Send delta | |||
|
357 | # against last revision (if possible), which in the common case | |||
|
358 | # should be similar enough to this revision that the delta is | |||
|
359 | # reasonable. | |||
|
360 | elif prevrev is not None: | |||
|
361 | baserev = prevrev | |||
|
362 | else: | |||
|
363 | baserev = nullrev | |||
|
364 | ||||
|
365 | # Storage has a fulltext revision. | |||
|
366 | ||||
|
367 | # Let's use the previous revision, which is as good a guess as any. | |||
|
368 | # There is definitely room to improve this logic. | |||
|
369 | elif prevrev is not None: | |||
|
370 | baserev = prevrev | |||
|
371 | else: | |||
|
372 | baserev = nullrev | |||
|
373 | ||||
|
374 | # But we can't actually use our chosen delta base for whatever | |||
|
375 | # reason. Reset to fulltext. | |||
|
376 | if baserev != nullrev and not candeltafn(baserev, rev): | |||
|
377 | baserev = nullrev | |||
|
378 | ||||
|
379 | revision = None | |||
|
380 | delta = None | |||
|
381 | baserevisionsize = None | |||
|
382 | ||||
|
383 | if revisiondata: | |||
|
384 | if store.iscensored(baserev) or store.iscensored(rev): | |||
|
385 | try: | |||
|
386 | revision = store.revision(node, raw=True) | |||
|
387 | except error.CensoredNodeError as e: | |||
|
388 | revision = e.tombstone | |||
|
389 | ||||
|
390 | if baserev != nullrev: | |||
|
391 | baserevisionsize = rawsizefn(baserev) | |||
|
392 | ||||
|
393 | elif baserev == nullrev and not deltaprevious: | |||
|
394 | revision = store.revision(node, raw=True) | |||
|
395 | available.add(rev) | |||
|
396 | else: | |||
|
397 | delta = revdifffn(baserev, rev) | |||
|
398 | available.add(rev) | |||
|
399 | ||||
|
400 | yield resultcls( | |||
|
401 | node=node, | |||
|
402 | p1node=fnode(p1rev), | |||
|
403 | p2node=fnode(p2rev), | |||
|
404 | basenode=fnode(baserev), | |||
|
405 | flags=flagsfn(rev), | |||
|
406 | baserevisionsize=baserevisionsize, | |||
|
407 | revision=revision, | |||
|
408 | delta=delta) | |||
|
409 | ||||
|
410 | prevrev = rev |
General Comments 0
You need to be logged in to leave comments.
Login now