Show More
@@ -1,605 +1,585 b'' | |||
|
1 | 1 | # changelog.py - changelog class for mercurial |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2005-2007 Matt Mackall <mpm@selenic.com> |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | from .i18n import _ |
|
11 | 11 | from .node import ( |
|
12 | 12 | bin, |
|
13 | 13 | hex, |
|
14 | 14 | nullid, |
|
15 | 15 | ) |
|
16 | 16 | from .thirdparty import attr |
|
17 | 17 | |
|
18 | 18 | from . import ( |
|
19 | 19 | encoding, |
|
20 | 20 | error, |
|
21 | 21 | metadata, |
|
22 | 22 | pycompat, |
|
23 | 23 | revlog, |
|
24 | 24 | ) |
|
25 | 25 | from .utils import ( |
|
26 | 26 | dateutil, |
|
27 | 27 | stringutil, |
|
28 | 28 | ) |
|
29 | 29 | |
|
30 | 30 | from .revlogutils import sidedata as sidedatamod |
|
31 | 31 | |
|
32 | 32 | _defaultextra = {b'branch': b'default'} |
|
33 | 33 | |
|
34 | 34 | |
|
35 | 35 | def _string_escape(text): |
|
36 | 36 | """ |
|
37 | 37 | >>> from .pycompat import bytechr as chr |
|
38 | 38 | >>> d = {b'nl': chr(10), b'bs': chr(92), b'cr': chr(13), b'nul': chr(0)} |
|
39 | 39 | >>> s = b"ab%(nl)scd%(bs)s%(bs)sn%(nul)s12ab%(cr)scd%(bs)s%(nl)s" % d |
|
40 | 40 | >>> s |
|
41 | 41 | 'ab\\ncd\\\\\\\\n\\x0012ab\\rcd\\\\\\n' |
|
42 | 42 | >>> res = _string_escape(s) |
|
43 | 43 | >>> s == _string_unescape(res) |
|
44 | 44 | True |
|
45 | 45 | """ |
|
46 | 46 | # subset of the string_escape codec |
|
47 | 47 | text = ( |
|
48 | 48 | text.replace(b'\\', b'\\\\') |
|
49 | 49 | .replace(b'\n', b'\\n') |
|
50 | 50 | .replace(b'\r', b'\\r') |
|
51 | 51 | ) |
|
52 | 52 | return text.replace(b'\0', b'\\0') |
|
53 | 53 | |
|
54 | 54 | |
|
55 | 55 | def _string_unescape(text): |
|
56 | 56 | if b'\\0' in text: |
|
57 | 57 | # fix up \0 without getting into trouble with \\0 |
|
58 | 58 | text = text.replace(b'\\\\', b'\\\\\n') |
|
59 | 59 | text = text.replace(b'\\0', b'\0') |
|
60 | 60 | text = text.replace(b'\n', b'') |
|
61 | 61 | return stringutil.unescapestr(text) |
|
62 | 62 | |
|
63 | 63 | |
|
64 | 64 | def decodeextra(text): |
|
65 | 65 | """ |
|
66 | 66 | >>> from .pycompat import bytechr as chr |
|
67 | 67 | >>> sorted(decodeextra(encodeextra({b'foo': b'bar', b'baz': chr(0) + b'2'}) |
|
68 | 68 | ... ).items()) |
|
69 | 69 | [('baz', '\\x002'), ('branch', 'default'), ('foo', 'bar')] |
|
70 | 70 | >>> sorted(decodeextra(encodeextra({b'foo': b'bar', |
|
71 | 71 | ... b'baz': chr(92) + chr(0) + b'2'}) |
|
72 | 72 | ... ).items()) |
|
73 | 73 | [('baz', '\\\\\\x002'), ('branch', 'default'), ('foo', 'bar')] |
|
74 | 74 | """ |
|
75 | 75 | extra = _defaultextra.copy() |
|
76 | 76 | for l in text.split(b'\0'): |
|
77 | 77 | if l: |
|
78 | 78 | k, v = _string_unescape(l).split(b':', 1) |
|
79 | 79 | extra[k] = v |
|
80 | 80 | return extra |
|
81 | 81 | |
|
82 | 82 | |
|
83 | 83 | def encodeextra(d): |
|
84 | 84 | # keys must be sorted to produce a deterministic changelog entry |
|
85 | 85 | items = [_string_escape(b'%s:%s' % (k, d[k])) for k in sorted(d)] |
|
86 | 86 | return b"\0".join(items) |
|
87 | 87 | |
|
88 | 88 | |
|
89 | 89 | def stripdesc(desc): |
|
90 | 90 | """strip trailing whitespace and leading and trailing empty lines""" |
|
91 | 91 | return b'\n'.join([l.rstrip() for l in desc.splitlines()]).strip(b'\n') |
|
92 | 92 | |
|
93 | 93 | |
|
94 | 94 | class appender(object): |
|
95 | 95 | '''the changelog index must be updated last on disk, so we use this class |
|
96 | 96 | to delay writes to it''' |
|
97 | 97 | |
|
98 | 98 | def __init__(self, vfs, name, mode, buf): |
|
99 | 99 | self.data = buf |
|
100 | 100 | fp = vfs(name, mode) |
|
101 | 101 | self.fp = fp |
|
102 | 102 | self.offset = fp.tell() |
|
103 | 103 | self.size = vfs.fstat(fp).st_size |
|
104 | 104 | self._end = self.size |
|
105 | 105 | |
|
106 | 106 | def end(self): |
|
107 | 107 | return self._end |
|
108 | 108 | |
|
109 | 109 | def tell(self): |
|
110 | 110 | return self.offset |
|
111 | 111 | |
|
112 | 112 | def flush(self): |
|
113 | 113 | pass |
|
114 | 114 | |
|
115 | 115 | @property |
|
116 | 116 | def closed(self): |
|
117 | 117 | return self.fp.closed |
|
118 | 118 | |
|
119 | 119 | def close(self): |
|
120 | 120 | self.fp.close() |
|
121 | 121 | |
|
122 | 122 | def seek(self, offset, whence=0): |
|
123 | 123 | '''virtual file offset spans real file and data''' |
|
124 | 124 | if whence == 0: |
|
125 | 125 | self.offset = offset |
|
126 | 126 | elif whence == 1: |
|
127 | 127 | self.offset += offset |
|
128 | 128 | elif whence == 2: |
|
129 | 129 | self.offset = self.end() + offset |
|
130 | 130 | if self.offset < self.size: |
|
131 | 131 | self.fp.seek(self.offset) |
|
132 | 132 | |
|
133 | 133 | def read(self, count=-1): |
|
134 | 134 | '''only trick here is reads that span real file and data''' |
|
135 | 135 | ret = b"" |
|
136 | 136 | if self.offset < self.size: |
|
137 | 137 | s = self.fp.read(count) |
|
138 | 138 | ret = s |
|
139 | 139 | self.offset += len(s) |
|
140 | 140 | if count > 0: |
|
141 | 141 | count -= len(s) |
|
142 | 142 | if count != 0: |
|
143 | 143 | doff = self.offset - self.size |
|
144 | 144 | self.data.insert(0, b"".join(self.data)) |
|
145 | 145 | del self.data[1:] |
|
146 | 146 | s = self.data[0][doff : doff + count] |
|
147 | 147 | self.offset += len(s) |
|
148 | 148 | ret += s |
|
149 | 149 | return ret |
|
150 | 150 | |
|
151 | 151 | def write(self, s): |
|
152 | 152 | self.data.append(bytes(s)) |
|
153 | 153 | self.offset += len(s) |
|
154 | 154 | self._end += len(s) |
|
155 | 155 | |
|
156 | 156 | def __enter__(self): |
|
157 | 157 | self.fp.__enter__() |
|
158 | 158 | return self |
|
159 | 159 | |
|
160 | 160 | def __exit__(self, *args): |
|
161 | 161 | return self.fp.__exit__(*args) |
|
162 | 162 | |
|
163 | 163 | |
|
164 | 164 | class _divertopener(object): |
|
165 | 165 | def __init__(self, opener, target): |
|
166 | 166 | self._opener = opener |
|
167 | 167 | self._target = target |
|
168 | 168 | |
|
169 | 169 | def __call__(self, name, mode=b'r', checkambig=False, **kwargs): |
|
170 | 170 | if name != self._target: |
|
171 | 171 | return self._opener(name, mode, **kwargs) |
|
172 | 172 | return self._opener(name + b".a", mode, **kwargs) |
|
173 | 173 | |
|
174 | 174 | def __getattr__(self, attr): |
|
175 | 175 | return getattr(self._opener, attr) |
|
176 | 176 | |
|
177 | 177 | |
|
178 | 178 | def _delayopener(opener, target, buf): |
|
179 | 179 | """build an opener that stores chunks in 'buf' instead of 'target'""" |
|
180 | 180 | |
|
181 | 181 | def _delay(name, mode=b'r', checkambig=False, **kwargs): |
|
182 | 182 | if name != target: |
|
183 | 183 | return opener(name, mode, **kwargs) |
|
184 | 184 | assert not kwargs |
|
185 | 185 | return appender(opener, name, mode, buf) |
|
186 | 186 | |
|
187 | 187 | return _delay |
|
188 | 188 | |
|
189 | 189 | |
|
190 | 190 | @attr.s |
|
191 | 191 | class _changelogrevision(object): |
|
192 | 192 | # Extensions might modify _defaultextra, so let the constructor below pass |
|
193 | 193 | # it in |
|
194 | 194 | extra = attr.ib() |
|
195 | 195 | manifest = attr.ib(default=nullid) |
|
196 | 196 | user = attr.ib(default=b'') |
|
197 | 197 | date = attr.ib(default=(0, 0)) |
|
198 | 198 | files = attr.ib(default=attr.Factory(list)) |
|
199 | 199 | filesadded = attr.ib(default=None) |
|
200 | 200 | filesremoved = attr.ib(default=None) |
|
201 | 201 | p1copies = attr.ib(default=None) |
|
202 | 202 | p2copies = attr.ib(default=None) |
|
203 | 203 | description = attr.ib(default=b'') |
|
204 | 204 | |
|
205 | 205 | |
|
206 | 206 | class changelogrevision(object): |
|
207 | 207 | """Holds results of a parsed changelog revision. |
|
208 | 208 | |
|
209 | 209 | Changelog revisions consist of multiple pieces of data, including |
|
210 | 210 | the manifest node, user, and date. This object exposes a view into |
|
211 | 211 | the parsed object. |
|
212 | 212 | """ |
|
213 | 213 | |
|
214 | 214 | __slots__ = ( |
|
215 | 215 | '_offsets', |
|
216 | 216 | '_text', |
|
217 | 217 | '_sidedata', |
|
218 | 218 | '_cpsd', |
|
219 | 219 | ) |
|
220 | 220 | |
|
221 | 221 | def __new__(cls, text, sidedata, cpsd): |
|
222 | 222 | if not text: |
|
223 | 223 | return _changelogrevision(extra=_defaultextra) |
|
224 | 224 | |
|
225 | 225 | self = super(changelogrevision, cls).__new__(cls) |
|
226 | 226 | # We could return here and implement the following as an __init__. |
|
227 | 227 | # But doing it here is equivalent and saves an extra function call. |
|
228 | 228 | |
|
229 | 229 | # format used: |
|
230 | 230 | # nodeid\n : manifest node in ascii |
|
231 | 231 | # user\n : user, no \n or \r allowed |
|
232 | 232 | # time tz extra\n : date (time is int or float, timezone is int) |
|
233 | 233 | # : extra is metadata, encoded and separated by '\0' |
|
234 | 234 | # : older versions ignore it |
|
235 | 235 | # files\n\n : files modified by the cset, no \n or \r allowed |
|
236 | 236 | # (.*) : comment (free text, ideally utf-8) |
|
237 | 237 | # |
|
238 | 238 | # changelog v0 doesn't use extra |
|
239 | 239 | |
|
240 | 240 | nl1 = text.index(b'\n') |
|
241 | 241 | nl2 = text.index(b'\n', nl1 + 1) |
|
242 | 242 | nl3 = text.index(b'\n', nl2 + 1) |
|
243 | 243 | |
|
244 | 244 | # The list of files may be empty. Which means nl3 is the first of the |
|
245 | 245 | # double newline that precedes the description. |
|
246 | 246 | if text[nl3 + 1 : nl3 + 2] == b'\n': |
|
247 | 247 | doublenl = nl3 |
|
248 | 248 | else: |
|
249 | 249 | doublenl = text.index(b'\n\n', nl3 + 1) |
|
250 | 250 | |
|
251 | 251 | self._offsets = (nl1, nl2, nl3, doublenl) |
|
252 | 252 | self._text = text |
|
253 | 253 | self._sidedata = sidedata |
|
254 | 254 | self._cpsd = cpsd |
|
255 | 255 | |
|
256 | 256 | return self |
|
257 | 257 | |
|
258 | 258 | @property |
|
259 | 259 | def manifest(self): |
|
260 | 260 | return bin(self._text[0 : self._offsets[0]]) |
|
261 | 261 | |
|
262 | 262 | @property |
|
263 | 263 | def user(self): |
|
264 | 264 | off = self._offsets |
|
265 | 265 | return encoding.tolocal(self._text[off[0] + 1 : off[1]]) |
|
266 | 266 | |
|
267 | 267 | @property |
|
268 | 268 | def _rawdate(self): |
|
269 | 269 | off = self._offsets |
|
270 | 270 | dateextra = self._text[off[1] + 1 : off[2]] |
|
271 | 271 | return dateextra.split(b' ', 2)[0:2] |
|
272 | 272 | |
|
273 | 273 | @property |
|
274 | 274 | def _rawextra(self): |
|
275 | 275 | off = self._offsets |
|
276 | 276 | dateextra = self._text[off[1] + 1 : off[2]] |
|
277 | 277 | fields = dateextra.split(b' ', 2) |
|
278 | 278 | if len(fields) != 3: |
|
279 | 279 | return None |
|
280 | 280 | |
|
281 | 281 | return fields[2] |
|
282 | 282 | |
|
283 | 283 | @property |
|
284 | 284 | def date(self): |
|
285 | 285 | raw = self._rawdate |
|
286 | 286 | time = float(raw[0]) |
|
287 | 287 | # Various tools did silly things with the timezone. |
|
288 | 288 | try: |
|
289 | 289 | timezone = int(raw[1]) |
|
290 | 290 | except ValueError: |
|
291 | 291 | timezone = 0 |
|
292 | 292 | |
|
293 | 293 | return time, timezone |
|
294 | 294 | |
|
295 | 295 | @property |
|
296 | 296 | def extra(self): |
|
297 | 297 | raw = self._rawextra |
|
298 | 298 | if raw is None: |
|
299 | 299 | return _defaultextra |
|
300 | 300 | |
|
301 | 301 | return decodeextra(raw) |
|
302 | 302 | |
|
303 | 303 | @property |
|
304 | 304 | def files(self): |
|
305 | 305 | off = self._offsets |
|
306 | 306 | if off[2] == off[3]: |
|
307 | 307 | return [] |
|
308 | 308 | |
|
309 | 309 | return self._text[off[2] + 1 : off[3]].split(b'\n') |
|
310 | 310 | |
|
311 | 311 | @property |
|
312 | 312 | def filesadded(self): |
|
313 | 313 | if self._cpsd: |
|
314 | 314 | rawindices = self._sidedata.get(sidedatamod.SD_FILESADDED) |
|
315 | 315 | if not rawindices: |
|
316 | 316 | return [] |
|
317 | 317 | else: |
|
318 | 318 | rawindices = self.extra.get(b'filesadded') |
|
319 | 319 | if rawindices is None: |
|
320 | 320 | return None |
|
321 | 321 | return metadata.decodefileindices(self.files, rawindices) |
|
322 | 322 | |
|
323 | 323 | @property |
|
324 | 324 | def filesremoved(self): |
|
325 | 325 | if self._cpsd: |
|
326 | 326 | rawindices = self._sidedata.get(sidedatamod.SD_FILESREMOVED) |
|
327 | 327 | if not rawindices: |
|
328 | 328 | return [] |
|
329 | 329 | else: |
|
330 | 330 | rawindices = self.extra.get(b'filesremoved') |
|
331 | 331 | if rawindices is None: |
|
332 | 332 | return None |
|
333 | 333 | return metadata.decodefileindices(self.files, rawindices) |
|
334 | 334 | |
|
335 | 335 | @property |
|
336 | 336 | def p1copies(self): |
|
337 | 337 | if self._cpsd: |
|
338 | 338 | rawcopies = self._sidedata.get(sidedatamod.SD_P1COPIES) |
|
339 | 339 | if not rawcopies: |
|
340 | 340 | return {} |
|
341 | 341 | else: |
|
342 | 342 | rawcopies = self.extra.get(b'p1copies') |
|
343 | 343 | if rawcopies is None: |
|
344 | 344 | return None |
|
345 | 345 | return metadata.decodecopies(self.files, rawcopies) |
|
346 | 346 | |
|
347 | 347 | @property |
|
348 | 348 | def p2copies(self): |
|
349 | 349 | if self._cpsd: |
|
350 | 350 | rawcopies = self._sidedata.get(sidedatamod.SD_P2COPIES) |
|
351 | 351 | if not rawcopies: |
|
352 | 352 | return {} |
|
353 | 353 | else: |
|
354 | 354 | rawcopies = self.extra.get(b'p2copies') |
|
355 | 355 | if rawcopies is None: |
|
356 | 356 | return None |
|
357 | 357 | return metadata.decodecopies(self.files, rawcopies) |
|
358 | 358 | |
|
359 | 359 | @property |
|
360 | 360 | def description(self): |
|
361 | 361 | return encoding.tolocal(self._text[self._offsets[3] + 2 :]) |
|
362 | 362 | |
|
363 | 363 | |
|
364 | 364 | class changelog(revlog.revlog): |
|
365 | 365 | def __init__(self, opener, trypending=False): |
|
366 | 366 | """Load a changelog revlog using an opener. |
|
367 | 367 | |
|
368 | 368 | If ``trypending`` is true, we attempt to load the index from a |
|
369 | 369 | ``00changelog.i.a`` file instead of the default ``00changelog.i``. |
|
370 | 370 | The ``00changelog.i.a`` file contains index (and possibly inline |
|
371 | 371 | revision) data for a transaction that hasn't been finalized yet. |
|
372 | 372 | It exists in a separate file to facilitate readers (such as |
|
373 | 373 | hooks processes) accessing data before a transaction is finalized. |
|
374 | 374 | """ |
|
375 | 375 | if trypending and opener.exists(b'00changelog.i.a'): |
|
376 | 376 | indexfile = b'00changelog.i.a' |
|
377 | 377 | else: |
|
378 | 378 | indexfile = b'00changelog.i' |
|
379 | 379 | |
|
380 | 380 | datafile = b'00changelog.d' |
|
381 | 381 | revlog.revlog.__init__( |
|
382 | 382 | self, |
|
383 | 383 | opener, |
|
384 | 384 | indexfile, |
|
385 | 385 | datafile=datafile, |
|
386 | 386 | checkambig=True, |
|
387 | 387 | mmaplargeindex=True, |
|
388 | 388 | persistentnodemap=opener.options.get(b'persistent-nodemap', False), |
|
389 | 389 | ) |
|
390 | 390 | |
|
391 | 391 | if self._initempty and (self.version & 0xFFFF == revlog.REVLOGV1): |
|
392 | 392 | # changelogs don't benefit from generaldelta. |
|
393 | 393 | |
|
394 | 394 | self.version &= ~revlog.FLAG_GENERALDELTA |
|
395 | 395 | self._generaldelta = False |
|
396 | 396 | |
|
397 | 397 | # Delta chains for changelogs tend to be very small because entries |
|
398 | 398 | # tend to be small and don't delta well with each. So disable delta |
|
399 | 399 | # chains. |
|
400 | 400 | self._storedeltachains = False |
|
401 | 401 | |
|
402 | 402 | self._realopener = opener |
|
403 | 403 | self._delayed = False |
|
404 | 404 | self._delaybuf = None |
|
405 | 405 | self._divert = False |
|
406 | 406 | self.filteredrevs = frozenset() |
|
407 | 407 | self._copiesstorage = opener.options.get(b'copies-storage') |
|
408 | 408 | |
|
409 | 409 | def delayupdate(self, tr): |
|
410 | 410 | """delay visibility of index updates to other readers""" |
|
411 | 411 | |
|
412 | 412 | if not self._delayed: |
|
413 | 413 | if len(self) == 0: |
|
414 | 414 | self._divert = True |
|
415 | 415 | if self._realopener.exists(self.indexfile + b'.a'): |
|
416 | 416 | self._realopener.unlink(self.indexfile + b'.a') |
|
417 | 417 | self.opener = _divertopener(self._realopener, self.indexfile) |
|
418 | 418 | else: |
|
419 | 419 | self._delaybuf = [] |
|
420 | 420 | self.opener = _delayopener( |
|
421 | 421 | self._realopener, self.indexfile, self._delaybuf |
|
422 | 422 | ) |
|
423 | 423 | self._delayed = True |
|
424 | 424 | tr.addpending(b'cl-%i' % id(self), self._writepending) |
|
425 | 425 | tr.addfinalize(b'cl-%i' % id(self), self._finalize) |
|
426 | 426 | |
|
427 | 427 | def _finalize(self, tr): |
|
428 | 428 | """finalize index updates""" |
|
429 | 429 | self._delayed = False |
|
430 | 430 | self.opener = self._realopener |
|
431 | 431 | # move redirected index data back into place |
|
432 | 432 | if self._divert: |
|
433 | 433 | assert not self._delaybuf |
|
434 | 434 | tmpname = self.indexfile + b".a" |
|
435 | 435 | nfile = self.opener.open(tmpname) |
|
436 | 436 | nfile.close() |
|
437 | 437 | self.opener.rename(tmpname, self.indexfile, checkambig=True) |
|
438 | 438 | elif self._delaybuf: |
|
439 | 439 | fp = self.opener(self.indexfile, b'a', checkambig=True) |
|
440 | 440 | fp.write(b"".join(self._delaybuf)) |
|
441 | 441 | fp.close() |
|
442 | 442 | self._delaybuf = None |
|
443 | 443 | self._divert = False |
|
444 | 444 | # split when we're done |
|
445 | 445 | self._enforceinlinesize(tr) |
|
446 | 446 | |
|
447 | 447 | def _writepending(self, tr): |
|
448 | 448 | """create a file containing the unfinalized state for |
|
449 | 449 | pretxnchangegroup""" |
|
450 | 450 | if self._delaybuf: |
|
451 | 451 | # make a temporary copy of the index |
|
452 | 452 | fp1 = self._realopener(self.indexfile) |
|
453 | 453 | pendingfilename = self.indexfile + b".a" |
|
454 | 454 | # register as a temp file to ensure cleanup on failure |
|
455 | 455 | tr.registertmp(pendingfilename) |
|
456 | 456 | # write existing data |
|
457 | 457 | fp2 = self._realopener(pendingfilename, b"w") |
|
458 | 458 | fp2.write(fp1.read()) |
|
459 | 459 | # add pending data |
|
460 | 460 | fp2.write(b"".join(self._delaybuf)) |
|
461 | 461 | fp2.close() |
|
462 | 462 | # switch modes so finalize can simply rename |
|
463 | 463 | self._delaybuf = None |
|
464 | 464 | self._divert = True |
|
465 | 465 | self.opener = _divertopener(self._realopener, self.indexfile) |
|
466 | 466 | |
|
467 | 467 | if self._divert: |
|
468 | 468 | return True |
|
469 | 469 | |
|
470 | 470 | return False |
|
471 | 471 | |
|
472 | 472 | def _enforceinlinesize(self, tr, fp=None): |
|
473 | 473 | if not self._delayed: |
|
474 | 474 | revlog.revlog._enforceinlinesize(self, tr, fp) |
|
475 | 475 | |
|
476 | 476 | def read(self, node): |
|
477 | 477 | """Obtain data from a parsed changelog revision. |
|
478 | 478 | |
|
479 | 479 | Returns a 6-tuple of: |
|
480 | 480 | |
|
481 | 481 | - manifest node in binary |
|
482 | 482 | - author/user as a localstr |
|
483 | 483 | - date as a 2-tuple of (time, timezone) |
|
484 | 484 | - list of files |
|
485 | 485 | - commit message as a localstr |
|
486 | 486 | - dict of extra metadata |
|
487 | 487 | |
|
488 | 488 | Unless you need to access all fields, consider calling |
|
489 | 489 | ``changelogrevision`` instead, as it is faster for partial object |
|
490 | 490 | access. |
|
491 | 491 | """ |
|
492 | 492 | d, s = self._revisiondata(node) |
|
493 | 493 | c = changelogrevision( |
|
494 | 494 | d, s, self._copiesstorage == b'changeset-sidedata' |
|
495 | 495 | ) |
|
496 | 496 | return (c.manifest, c.user, c.date, c.files, c.description, c.extra) |
|
497 | 497 | |
|
498 | 498 | def changelogrevision(self, nodeorrev): |
|
499 | 499 | """Obtain a ``changelogrevision`` for a node or revision.""" |
|
500 | 500 | text, sidedata = self._revisiondata(nodeorrev) |
|
501 | 501 | return changelogrevision( |
|
502 | 502 | text, sidedata, self._copiesstorage == b'changeset-sidedata' |
|
503 | 503 | ) |
|
504 | 504 | |
|
505 | 505 | def readfiles(self, node): |
|
506 | 506 | """ |
|
507 | 507 | short version of read that only returns the files modified by the cset |
|
508 | 508 | """ |
|
509 | 509 | text = self.revision(node) |
|
510 | 510 | if not text: |
|
511 | 511 | return [] |
|
512 | 512 | last = text.index(b"\n\n") |
|
513 | 513 | l = text[:last].split(b'\n') |
|
514 | 514 | return l[3:] |
|
515 | 515 | |
|
516 | 516 | def add( |
|
517 | 517 | self, |
|
518 | 518 | manifest, |
|
519 | 519 | files, |
|
520 | 520 | desc, |
|
521 | 521 | transaction, |
|
522 | 522 | p1, |
|
523 | 523 | p2, |
|
524 | 524 | user, |
|
525 | 525 | date=None, |
|
526 | 526 | extra=None, |
|
527 | 527 | ): |
|
528 | 528 | # Convert to UTF-8 encoded bytestrings as the very first |
|
529 | 529 | # thing: calling any method on a localstr object will turn it |
|
530 | 530 | # into a str object and the cached UTF-8 string is thus lost. |
|
531 | 531 | user, desc = encoding.fromlocal(user), encoding.fromlocal(desc) |
|
532 | 532 | |
|
533 | 533 | user = user.strip() |
|
534 | 534 | # An empty username or a username with a "\n" will make the |
|
535 | 535 | # revision text contain two "\n\n" sequences -> corrupt |
|
536 | 536 | # repository since read cannot unpack the revision. |
|
537 | 537 | if not user: |
|
538 | 538 | raise error.StorageError(_(b"empty username")) |
|
539 | 539 | if b"\n" in user: |
|
540 | 540 | raise error.StorageError( |
|
541 | 541 | _(b"username %r contains a newline") % pycompat.bytestr(user) |
|
542 | 542 | ) |
|
543 | 543 | |
|
544 | 544 | desc = stripdesc(desc) |
|
545 | 545 | |
|
546 | 546 | if date: |
|
547 | 547 | parseddate = b"%d %d" % dateutil.parsedate(date) |
|
548 | 548 | else: |
|
549 | 549 | parseddate = b"%d %d" % dateutil.makedate() |
|
550 | 550 | if extra: |
|
551 | 551 | branch = extra.get(b"branch") |
|
552 | 552 | if branch in (b"default", b""): |
|
553 | 553 | del extra[b"branch"] |
|
554 | 554 | elif branch in (b".", b"null", b"tip"): |
|
555 | 555 | raise error.StorageError( |
|
556 | 556 | _(b'the name \'%s\' is reserved') % branch |
|
557 | 557 | ) |
|
558 | 558 | sortedfiles = sorted(files.touched) |
|
559 | 559 | sidedata = None |
|
560 | 560 | if self._copiesstorage == b'changeset-sidedata': |
|
561 | sidedata = {} | |
|
562 | p1copies = files.copied_from_p1 | |
|
563 | if p1copies: | |
|
564 | p1copies = metadata.encodecopies(sortedfiles, p1copies) | |
|
565 | sidedata[sidedatamod.SD_P1COPIES] = p1copies | |
|
566 | p2copies = files.copied_from_p2 | |
|
567 | if p2copies: | |
|
568 | p2copies = metadata.encodecopies(sortedfiles, p2copies) | |
|
569 | sidedata[sidedatamod.SD_P2COPIES] = p2copies | |
|
570 | filesadded = files.added | |
|
571 | if filesadded: | |
|
572 | filesadded = metadata.encodefileindices(sortedfiles, filesadded) | |
|
573 | sidedata[sidedatamod.SD_FILESADDED] = filesadded | |
|
574 | filesremoved = files.removed | |
|
575 | if filesremoved: | |
|
576 | filesremoved = metadata.encodefileindices( | |
|
577 | sortedfiles, filesremoved | |
|
578 | ) | |
|
579 | sidedata[sidedatamod.SD_FILESREMOVED] = filesremoved | |
|
580 | if not sidedata: | |
|
581 | sidedata = None | |
|
561 | sidedata = metadata.encode_copies_sidedata(files) | |
|
582 | 562 | |
|
583 | 563 | if extra: |
|
584 | 564 | extra = encodeextra(extra) |
|
585 | 565 | parseddate = b"%s %s" % (parseddate, extra) |
|
586 | 566 | l = [hex(manifest), user, parseddate] + sortedfiles + [b"", desc] |
|
587 | 567 | text = b"\n".join(l) |
|
588 | 568 | return self.addrevision( |
|
589 | 569 | text, transaction, len(self), p1, p2, sidedata=sidedata |
|
590 | 570 | ) |
|
591 | 571 | |
|
592 | 572 | def branchinfo(self, rev): |
|
593 | 573 | """return the branch name and open/close state of a revision |
|
594 | 574 | |
|
595 | 575 | This function exists because creating a changectx object |
|
596 | 576 | just to access this is costly.""" |
|
597 | 577 | extra = self.read(rev)[5] |
|
598 | 578 | return encoding.tolocal(extra.get(b"branch")), b'close' in extra |
|
599 | 579 | |
|
600 | 580 | def _nodeduplicatecallback(self, transaction, node): |
|
601 | 581 | # keep track of revisions that got "re-added", eg: unbunde of know rev. |
|
602 | 582 | # |
|
603 | 583 | # We track them in a list to preserve their order from the source bundle |
|
604 | 584 | duplicates = transaction.changes.setdefault(b'revduplicates', []) |
|
605 | 585 | duplicates.append(self.rev(node)) |
@@ -1,400 +1,424 b'' | |||
|
1 | 1 | # metadata.py -- code related to various metadata computation and access. |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2019 Google, Inc <martinvonz@google.com> |
|
4 | 4 | # Copyright 2020 Pierre-Yves David <pierre-yves.david@octobus.net> |
|
5 | 5 | # |
|
6 | 6 | # This software may be used and distributed according to the terms of the |
|
7 | 7 | # GNU General Public License version 2 or any later version. |
|
8 | 8 | from __future__ import absolute_import, print_function |
|
9 | 9 | |
|
10 | 10 | import multiprocessing |
|
11 | 11 | |
|
12 | 12 | from . import ( |
|
13 | 13 | error, |
|
14 | 14 | node, |
|
15 | 15 | pycompat, |
|
16 | 16 | util, |
|
17 | 17 | ) |
|
18 | 18 | |
|
19 | 19 | from .revlogutils import ( |
|
20 | 20 | flagutil as sidedataflag, |
|
21 | 21 | sidedata as sidedatamod, |
|
22 | 22 | ) |
|
23 | 23 | |
|
24 | 24 | |
|
25 | 25 | class ChangingFiles(object): |
|
26 | 26 | """A class recording the changes made to a file by a revision |
|
27 | 27 | """ |
|
28 | 28 | |
|
29 | 29 | def __init__( |
|
30 | 30 | self, touched=(), added=(), removed=(), p1_copies=(), p2_copies=(), |
|
31 | 31 | ): |
|
32 | 32 | self._added = set(added) |
|
33 | 33 | self._removed = set(removed) |
|
34 | 34 | self._touched = set(touched) |
|
35 | 35 | self._touched.update(self._added) |
|
36 | 36 | self._touched.update(self._removed) |
|
37 | 37 | self._p1_copies = dict(p1_copies) |
|
38 | 38 | self._p2_copies = dict(p2_copies) |
|
39 | 39 | |
|
40 | 40 | @property |
|
41 | 41 | def added(self): |
|
42 | 42 | return frozenset(self._added) |
|
43 | 43 | |
|
44 | 44 | def mark_added(self, filename): |
|
45 | 45 | self._added.add(filename) |
|
46 | 46 | self._touched.add(filename) |
|
47 | 47 | |
|
48 | 48 | def update_added(self, filenames): |
|
49 | 49 | for f in filenames: |
|
50 | 50 | self.mark_added(f) |
|
51 | 51 | |
|
52 | 52 | @property |
|
53 | 53 | def removed(self): |
|
54 | 54 | return frozenset(self._removed) |
|
55 | 55 | |
|
56 | 56 | def mark_removed(self, filename): |
|
57 | 57 | self._removed.add(filename) |
|
58 | 58 | self._touched.add(filename) |
|
59 | 59 | |
|
60 | 60 | def update_removed(self, filenames): |
|
61 | 61 | for f in filenames: |
|
62 | 62 | self.mark_removed(f) |
|
63 | 63 | |
|
64 | 64 | @property |
|
65 | 65 | def touched(self): |
|
66 | 66 | return frozenset(self._touched) |
|
67 | 67 | |
|
68 | 68 | def mark_touched(self, filename): |
|
69 | 69 | self._touched.add(filename) |
|
70 | 70 | |
|
71 | 71 | def update_touched(self, filenames): |
|
72 | 72 | for f in filenames: |
|
73 | 73 | self.mark_touched(f) |
|
74 | 74 | |
|
75 | 75 | @property |
|
76 | 76 | def copied_from_p1(self): |
|
77 | 77 | return self._p1_copies.copy() |
|
78 | 78 | |
|
79 | 79 | def mark_copied_from_p1(self, source, dest): |
|
80 | 80 | self._p1_copies[dest] = source |
|
81 | 81 | |
|
82 | 82 | def update_copies_from_p1(self, copies): |
|
83 | 83 | for dest, source in copies.items(): |
|
84 | 84 | self.mark_copied_from_p1(source, dest) |
|
85 | 85 | |
|
86 | 86 | @property |
|
87 | 87 | def copied_from_p2(self): |
|
88 | 88 | return self._p2_copies.copy() |
|
89 | 89 | |
|
90 | 90 | def mark_copied_from_p2(self, source, dest): |
|
91 | 91 | self._p2_copies[dest] = source |
|
92 | 92 | |
|
93 | 93 | def update_copies_from_p2(self, copies): |
|
94 | 94 | for dest, source in copies.items(): |
|
95 | 95 | self.mark_copied_from_p2(source, dest) |
|
96 | 96 | |
|
97 | 97 | |
|
98 | 98 | def computechangesetfilesadded(ctx): |
|
99 | 99 | """return the list of files added in a changeset |
|
100 | 100 | """ |
|
101 | 101 | added = [] |
|
102 | 102 | for f in ctx.files(): |
|
103 | 103 | if not any(f in p for p in ctx.parents()): |
|
104 | 104 | added.append(f) |
|
105 | 105 | return added |
|
106 | 106 | |
|
107 | 107 | |
|
108 | 108 | def get_removal_filter(ctx, x=None): |
|
109 | 109 | """return a function to detect files "wrongly" detected as `removed` |
|
110 | 110 | |
|
111 | 111 | When a file is removed relative to p1 in a merge, this |
|
112 | 112 | function determines whether the absence is due to a |
|
113 | 113 | deletion from a parent, or whether the merge commit |
|
114 | 114 | itself deletes the file. We decide this by doing a |
|
115 | 115 | simplified three way merge of the manifest entry for |
|
116 | 116 | the file. There are two ways we decide the merge |
|
117 | 117 | itself didn't delete a file: |
|
118 | 118 | - neither parent (nor the merge) contain the file |
|
119 | 119 | - exactly one parent contains the file, and that |
|
120 | 120 | parent has the same filelog entry as the merge |
|
121 | 121 | ancestor (or all of them if there two). In other |
|
122 | 122 | words, that parent left the file unchanged while the |
|
123 | 123 | other one deleted it. |
|
124 | 124 | One way to think about this is that deleting a file is |
|
125 | 125 | similar to emptying it, so the list of changed files |
|
126 | 126 | should be similar either way. The computation |
|
127 | 127 | described above is not done directly in _filecommit |
|
128 | 128 | when creating the list of changed files, however |
|
129 | 129 | it does something very similar by comparing filelog |
|
130 | 130 | nodes. |
|
131 | 131 | """ |
|
132 | 132 | |
|
133 | 133 | if x is not None: |
|
134 | 134 | p1, p2, m1, m2 = x |
|
135 | 135 | else: |
|
136 | 136 | p1 = ctx.p1() |
|
137 | 137 | p2 = ctx.p2() |
|
138 | 138 | m1 = p1.manifest() |
|
139 | 139 | m2 = p2.manifest() |
|
140 | 140 | |
|
141 | 141 | @util.cachefunc |
|
142 | 142 | def mas(): |
|
143 | 143 | p1n = p1.node() |
|
144 | 144 | p2n = p2.node() |
|
145 | 145 | cahs = ctx.repo().changelog.commonancestorsheads(p1n, p2n) |
|
146 | 146 | if not cahs: |
|
147 | 147 | cahs = [node.nullrev] |
|
148 | 148 | return [ctx.repo()[r].manifest() for r in cahs] |
|
149 | 149 | |
|
150 | 150 | def deletionfromparent(f): |
|
151 | 151 | if f in m1: |
|
152 | 152 | return f not in m2 and all( |
|
153 | 153 | f in ma and ma.find(f) == m1.find(f) for ma in mas() |
|
154 | 154 | ) |
|
155 | 155 | elif f in m2: |
|
156 | 156 | return all(f in ma and ma.find(f) == m2.find(f) for ma in mas()) |
|
157 | 157 | else: |
|
158 | 158 | return True |
|
159 | 159 | |
|
160 | 160 | return deletionfromparent |
|
161 | 161 | |
|
162 | 162 | |
|
163 | 163 | def computechangesetfilesremoved(ctx): |
|
164 | 164 | """return the list of files removed in a changeset |
|
165 | 165 | """ |
|
166 | 166 | removed = [] |
|
167 | 167 | for f in ctx.files(): |
|
168 | 168 | if f not in ctx: |
|
169 | 169 | removed.append(f) |
|
170 | 170 | if removed: |
|
171 | 171 | rf = get_removal_filter(ctx) |
|
172 | 172 | removed = [r for r in removed if not rf(r)] |
|
173 | 173 | return removed |
|
174 | 174 | |
|
175 | 175 | |
|
176 | 176 | def computechangesetcopies(ctx): |
|
177 | 177 | """return the copies data for a changeset |
|
178 | 178 | |
|
179 | 179 | The copies data are returned as a pair of dictionnary (p1copies, p2copies). |
|
180 | 180 | |
|
181 | 181 | Each dictionnary are in the form: `{newname: oldname}` |
|
182 | 182 | """ |
|
183 | 183 | p1copies = {} |
|
184 | 184 | p2copies = {} |
|
185 | 185 | p1 = ctx.p1() |
|
186 | 186 | p2 = ctx.p2() |
|
187 | 187 | narrowmatch = ctx._repo.narrowmatch() |
|
188 | 188 | for dst in ctx.files(): |
|
189 | 189 | if not narrowmatch(dst) or dst not in ctx: |
|
190 | 190 | continue |
|
191 | 191 | copied = ctx[dst].renamed() |
|
192 | 192 | if not copied: |
|
193 | 193 | continue |
|
194 | 194 | src, srcnode = copied |
|
195 | 195 | if src in p1 and p1[src].filenode() == srcnode: |
|
196 | 196 | p1copies[dst] = src |
|
197 | 197 | elif src in p2 and p2[src].filenode() == srcnode: |
|
198 | 198 | p2copies[dst] = src |
|
199 | 199 | return p1copies, p2copies |
|
200 | 200 | |
|
201 | 201 | |
|
202 | 202 | def encodecopies(files, copies): |
|
203 | 203 | items = [] |
|
204 | 204 | for i, dst in enumerate(files): |
|
205 | 205 | if dst in copies: |
|
206 | 206 | items.append(b'%d\0%s' % (i, copies[dst])) |
|
207 | 207 | if len(items) != len(copies): |
|
208 | 208 | raise error.ProgrammingError( |
|
209 | 209 | b'some copy targets missing from file list' |
|
210 | 210 | ) |
|
211 | 211 | return b"\n".join(items) |
|
212 | 212 | |
|
213 | 213 | |
|
214 | 214 | def decodecopies(files, data): |
|
215 | 215 | try: |
|
216 | 216 | copies = {} |
|
217 | 217 | if not data: |
|
218 | 218 | return copies |
|
219 | 219 | for l in data.split(b'\n'): |
|
220 | 220 | strindex, src = l.split(b'\0') |
|
221 | 221 | i = int(strindex) |
|
222 | 222 | dst = files[i] |
|
223 | 223 | copies[dst] = src |
|
224 | 224 | return copies |
|
225 | 225 | except (ValueError, IndexError): |
|
226 | 226 | # Perhaps someone had chosen the same key name (e.g. "p1copies") and |
|
227 | 227 | # used different syntax for the value. |
|
228 | 228 | return None |
|
229 | 229 | |
|
230 | 230 | |
|
231 | 231 | def encodefileindices(files, subset): |
|
232 | 232 | subset = set(subset) |
|
233 | 233 | indices = [] |
|
234 | 234 | for i, f in enumerate(files): |
|
235 | 235 | if f in subset: |
|
236 | 236 | indices.append(b'%d' % i) |
|
237 | 237 | return b'\n'.join(indices) |
|
238 | 238 | |
|
239 | 239 | |
|
240 | 240 | def decodefileindices(files, data): |
|
241 | 241 | try: |
|
242 | 242 | subset = [] |
|
243 | 243 | if not data: |
|
244 | 244 | return subset |
|
245 | 245 | for strindex in data.split(b'\n'): |
|
246 | 246 | i = int(strindex) |
|
247 | 247 | if i < 0 or i >= len(files): |
|
248 | 248 | return None |
|
249 | 249 | subset.append(files[i]) |
|
250 | 250 | return subset |
|
251 | 251 | except (ValueError, IndexError): |
|
252 | 252 | # Perhaps someone had chosen the same key name (e.g. "added") and |
|
253 | 253 | # used different syntax for the value. |
|
254 | 254 | return None |
|
255 | 255 | |
|
256 | 256 | |
|
257 | def encode_copies_sidedata(files): | |
|
258 | sortedfiles = sorted(files.touched) | |
|
259 | sidedata = {} | |
|
260 | p1copies = files.copied_from_p1 | |
|
261 | if p1copies: | |
|
262 | p1copies = encodecopies(sortedfiles, p1copies) | |
|
263 | sidedata[sidedatamod.SD_P1COPIES] = p1copies | |
|
264 | p2copies = files.copied_from_p2 | |
|
265 | if p2copies: | |
|
266 | p2copies = encodecopies(sortedfiles, p2copies) | |
|
267 | sidedata[sidedatamod.SD_P2COPIES] = p2copies | |
|
268 | filesadded = files.added | |
|
269 | if filesadded: | |
|
270 | filesadded = encodefileindices(sortedfiles, filesadded) | |
|
271 | sidedata[sidedatamod.SD_FILESADDED] = filesadded | |
|
272 | filesremoved = files.removed | |
|
273 | if filesremoved: | |
|
274 | filesremoved = encodefileindices(sortedfiles, filesremoved) | |
|
275 | sidedata[sidedatamod.SD_FILESREMOVED] = filesremoved | |
|
276 | if not sidedata: | |
|
277 | sidedata = None | |
|
278 | return sidedata | |
|
279 | ||
|
280 | ||
|
257 | 281 | def _getsidedata(srcrepo, rev): |
|
258 | 282 | ctx = srcrepo[rev] |
|
259 | 283 | filescopies = computechangesetcopies(ctx) |
|
260 | 284 | filesadded = computechangesetfilesadded(ctx) |
|
261 | 285 | filesremoved = computechangesetfilesremoved(ctx) |
|
262 | 286 | sidedata = {} |
|
263 | 287 | if any([filescopies, filesadded, filesremoved]): |
|
264 | 288 | sortedfiles = sorted(ctx.files()) |
|
265 | 289 | p1copies, p2copies = filescopies |
|
266 | 290 | p1copies = encodecopies(sortedfiles, p1copies) |
|
267 | 291 | p2copies = encodecopies(sortedfiles, p2copies) |
|
268 | 292 | filesadded = encodefileindices(sortedfiles, filesadded) |
|
269 | 293 | filesremoved = encodefileindices(sortedfiles, filesremoved) |
|
270 | 294 | if p1copies: |
|
271 | 295 | sidedata[sidedatamod.SD_P1COPIES] = p1copies |
|
272 | 296 | if p2copies: |
|
273 | 297 | sidedata[sidedatamod.SD_P2COPIES] = p2copies |
|
274 | 298 | if filesadded: |
|
275 | 299 | sidedata[sidedatamod.SD_FILESADDED] = filesadded |
|
276 | 300 | if filesremoved: |
|
277 | 301 | sidedata[sidedatamod.SD_FILESREMOVED] = filesremoved |
|
278 | 302 | return sidedata |
|
279 | 303 | |
|
280 | 304 | |
|
281 | 305 | def getsidedataadder(srcrepo, destrepo): |
|
282 | 306 | use_w = srcrepo.ui.configbool(b'experimental', b'worker.repository-upgrade') |
|
283 | 307 | if pycompat.iswindows or not use_w: |
|
284 | 308 | return _get_simple_sidedata_adder(srcrepo, destrepo) |
|
285 | 309 | else: |
|
286 | 310 | return _get_worker_sidedata_adder(srcrepo, destrepo) |
|
287 | 311 | |
|
288 | 312 | |
|
289 | 313 | def _sidedata_worker(srcrepo, revs_queue, sidedata_queue, tokens): |
|
290 | 314 | """The function used by worker precomputing sidedata |
|
291 | 315 | |
|
292 | 316 | It read an input queue containing revision numbers |
|
293 | 317 | It write in an output queue containing (rev, <sidedata-map>) |
|
294 | 318 | |
|
295 | 319 | The `None` input value is used as a stop signal. |
|
296 | 320 | |
|
297 | 321 | The `tokens` semaphore is user to avoid having too many unprocessed |
|
298 | 322 | entries. The workers needs to acquire one token before fetching a task. |
|
299 | 323 | They will be released by the consumer of the produced data. |
|
300 | 324 | """ |
|
301 | 325 | tokens.acquire() |
|
302 | 326 | rev = revs_queue.get() |
|
303 | 327 | while rev is not None: |
|
304 | 328 | data = _getsidedata(srcrepo, rev) |
|
305 | 329 | sidedata_queue.put((rev, data)) |
|
306 | 330 | tokens.acquire() |
|
307 | 331 | rev = revs_queue.get() |
|
308 | 332 | # processing of `None` is completed, release the token. |
|
309 | 333 | tokens.release() |
|
310 | 334 | |
|
311 | 335 | |
|
312 | 336 | BUFF_PER_WORKER = 50 |
|
313 | 337 | |
|
314 | 338 | |
|
315 | 339 | def _get_worker_sidedata_adder(srcrepo, destrepo): |
|
316 | 340 | """The parallel version of the sidedata computation |
|
317 | 341 | |
|
318 | 342 | This code spawn a pool of worker that precompute a buffer of sidedata |
|
319 | 343 | before we actually need them""" |
|
320 | 344 | # avoid circular import copies -> scmutil -> worker -> copies |
|
321 | 345 | from . import worker |
|
322 | 346 | |
|
323 | 347 | nbworkers = worker._numworkers(srcrepo.ui) |
|
324 | 348 | |
|
325 | 349 | tokens = multiprocessing.BoundedSemaphore(nbworkers * BUFF_PER_WORKER) |
|
326 | 350 | revsq = multiprocessing.Queue() |
|
327 | 351 | sidedataq = multiprocessing.Queue() |
|
328 | 352 | |
|
329 | 353 | assert srcrepo.filtername is None |
|
330 | 354 | # queue all tasks beforehand, revision numbers are small and it make |
|
331 | 355 | # synchronisation simpler |
|
332 | 356 | # |
|
333 | 357 | # Since the computation for each node can be quite expensive, the overhead |
|
334 | 358 | # of using a single queue is not revelant. In practice, most computation |
|
335 | 359 | # are fast but some are very expensive and dominate all the other smaller |
|
336 | 360 | # cost. |
|
337 | 361 | for r in srcrepo.changelog.revs(): |
|
338 | 362 | revsq.put(r) |
|
339 | 363 | # queue the "no more tasks" markers |
|
340 | 364 | for i in range(nbworkers): |
|
341 | 365 | revsq.put(None) |
|
342 | 366 | |
|
343 | 367 | allworkers = [] |
|
344 | 368 | for i in range(nbworkers): |
|
345 | 369 | args = (srcrepo, revsq, sidedataq, tokens) |
|
346 | 370 | w = multiprocessing.Process(target=_sidedata_worker, args=args) |
|
347 | 371 | allworkers.append(w) |
|
348 | 372 | w.start() |
|
349 | 373 | |
|
350 | 374 | # dictionnary to store results for revision higher than we one we are |
|
351 | 375 | # looking for. For example, if we need the sidedatamap for 42, and 43 is |
|
352 | 376 | # received, when shelve 43 for later use. |
|
353 | 377 | staging = {} |
|
354 | 378 | |
|
355 | 379 | def sidedata_companion(revlog, rev): |
|
356 | 380 | sidedata = {} |
|
357 | 381 | if util.safehasattr(revlog, b'filteredrevs'): # this is a changelog |
|
358 | 382 | # Is the data previously shelved ? |
|
359 | 383 | sidedata = staging.pop(rev, None) |
|
360 | 384 | if sidedata is None: |
|
361 | 385 | # look at the queued result until we find the one we are lookig |
|
362 | 386 | # for (shelve the other ones) |
|
363 | 387 | r, sidedata = sidedataq.get() |
|
364 | 388 | while r != rev: |
|
365 | 389 | staging[r] = sidedata |
|
366 | 390 | r, sidedata = sidedataq.get() |
|
367 | 391 | tokens.release() |
|
368 | 392 | return False, (), sidedata |
|
369 | 393 | |
|
370 | 394 | return sidedata_companion |
|
371 | 395 | |
|
372 | 396 | |
|
373 | 397 | def _get_simple_sidedata_adder(srcrepo, destrepo): |
|
374 | 398 | """The simple version of the sidedata computation |
|
375 | 399 | |
|
376 | 400 | It just compute it in the same thread on request""" |
|
377 | 401 | |
|
378 | 402 | def sidedatacompanion(revlog, rev): |
|
379 | 403 | sidedata = {} |
|
380 | 404 | if util.safehasattr(revlog, 'filteredrevs'): # this is a changelog |
|
381 | 405 | sidedata = _getsidedata(srcrepo, rev) |
|
382 | 406 | return False, (), sidedata |
|
383 | 407 | |
|
384 | 408 | return sidedatacompanion |
|
385 | 409 | |
|
386 | 410 | |
|
387 | 411 | def getsidedataremover(srcrepo, destrepo): |
|
388 | 412 | def sidedatacompanion(revlog, rev): |
|
389 | 413 | f = () |
|
390 | 414 | if util.safehasattr(revlog, 'filteredrevs'): # this is a changelog |
|
391 | 415 | if revlog.flags(rev) & sidedataflag.REVIDX_SIDEDATA: |
|
392 | 416 | f = ( |
|
393 | 417 | sidedatamod.SD_P1COPIES, |
|
394 | 418 | sidedatamod.SD_P2COPIES, |
|
395 | 419 | sidedatamod.SD_FILESADDED, |
|
396 | 420 | sidedatamod.SD_FILESREMOVED, |
|
397 | 421 | ) |
|
398 | 422 | return False, f, {} |
|
399 | 423 | |
|
400 | 424 | return sidedatacompanion |
General Comments 0
You need to be logged in to leave comments.
Login now