Show More
@@ -1,1299 +1,1301 b'' | |||
|
1 | 1 | # sqlitestore.py - Storage backend that uses SQLite |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2018 Gregory Szorc <gregory.szorc@gmail.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 | """store repository data in SQLite (EXPERIMENTAL) |
|
9 | 9 | |
|
10 | 10 | The sqlitestore extension enables the storage of repository data in SQLite. |
|
11 | 11 | |
|
12 | 12 | This extension is HIGHLY EXPERIMENTAL. There are NO BACKWARDS COMPATIBILITY |
|
13 | 13 | GUARANTEES. This means that repositories created with this extension may |
|
14 | 14 | only be usable with the exact version of this extension/Mercurial that was |
|
15 | 15 | used. The extension attempts to enforce this in order to prevent repository |
|
16 | 16 | corruption. |
|
17 | 17 | |
|
18 | 18 | In addition, several features are not yet supported or have known bugs: |
|
19 | 19 | |
|
20 | 20 | * Only some data is stored in SQLite. Changeset, manifest, and other repository |
|
21 | 21 | data is not yet stored in SQLite. |
|
22 | 22 | * Transactions are not robust. If the process is aborted at the right time |
|
23 | 23 | during transaction close/rollback, the repository could be in an inconsistent |
|
24 | 24 | state. This problem will diminish once all repository data is tracked by |
|
25 | 25 | SQLite. |
|
26 | 26 | * Bundle repositories do not work (the ability to use e.g. |
|
27 | 27 | `hg -R <bundle-file> log` to automatically overlay a bundle on top of the |
|
28 | 28 | existing repository). |
|
29 | 29 | * Various other features don't work. |
|
30 | 30 | |
|
31 | 31 | This extension should work for basic clone/pull, update, and commit workflows. |
|
32 | 32 | Some history rewriting operations may fail due to lack of support for bundle |
|
33 | 33 | repositories. |
|
34 | 34 | |
|
35 | 35 | To use, activate the extension and set the ``storage.new-repo-backend`` config |
|
36 | 36 | option to ``sqlite`` to enable new repositories to use SQLite for storage. |
|
37 | 37 | """ |
|
38 | 38 | |
|
39 | 39 | # To run the test suite with repos using SQLite by default, execute the |
|
40 | 40 | # following: |
|
41 | 41 | # |
|
42 | 42 | # HGREPOFEATURES="sqlitestore" run-tests.py \ |
|
43 | 43 | # --extra-config-opt extensions.sqlitestore= \ |
|
44 | 44 | # --extra-config-opt storage.new-repo-backend=sqlite |
|
45 | 45 | |
|
46 | 46 | from __future__ import absolute_import |
|
47 | 47 | |
|
48 | 48 | import sqlite3 |
|
49 | 49 | import struct |
|
50 | 50 | import threading |
|
51 | 51 | import zlib |
|
52 | 52 | |
|
53 | 53 | from mercurial.i18n import _ |
|
54 | 54 | from mercurial.node import ( |
|
55 | 55 | nullid, |
|
56 | 56 | nullrev, |
|
57 | 57 | short, |
|
58 | 58 | ) |
|
59 | 59 | from mercurial.thirdparty import attr |
|
60 | 60 | from mercurial import ( |
|
61 | 61 | ancestor, |
|
62 | 62 | dagop, |
|
63 | 63 | encoding, |
|
64 | 64 | error, |
|
65 | 65 | extensions, |
|
66 | 66 | localrepo, |
|
67 | 67 | mdiff, |
|
68 | 68 | pycompat, |
|
69 | 69 | registrar, |
|
70 | 70 | requirements, |
|
71 | 71 | util, |
|
72 | 72 | verify, |
|
73 | 73 | ) |
|
74 | 74 | from mercurial.interfaces import ( |
|
75 | 75 | repository, |
|
76 | 76 | util as interfaceutil, |
|
77 | 77 | ) |
|
78 | 78 | from mercurial.utils import ( |
|
79 | 79 | hashutil, |
|
80 | 80 | storageutil, |
|
81 | 81 | ) |
|
82 | 82 | |
|
83 | 83 | try: |
|
84 | 84 | from mercurial import zstd |
|
85 | 85 | |
|
86 | 86 | zstd.__version__ |
|
87 | 87 | except ImportError: |
|
88 | 88 | zstd = None |
|
89 | 89 | |
|
90 | 90 | configtable = {} |
|
91 | 91 | configitem = registrar.configitem(configtable) |
|
92 | 92 | |
|
93 | 93 | # experimental config: storage.sqlite.compression |
|
94 | 94 | configitem( |
|
95 | 95 | b'storage', |
|
96 | 96 | b'sqlite.compression', |
|
97 | 97 | default=b'zstd' if zstd else b'zlib', |
|
98 | 98 | experimental=True, |
|
99 | 99 | ) |
|
100 | 100 | |
|
101 | 101 | # Note for extension authors: ONLY specify testedwith = 'ships-with-hg-core' for |
|
102 | 102 | # extensions which SHIP WITH MERCURIAL. Non-mainline extensions should |
|
103 | 103 | # be specifying the version(s) of Mercurial they are tested with, or |
|
104 | 104 | # leave the attribute unspecified. |
|
105 | 105 | testedwith = b'ships-with-hg-core' |
|
106 | 106 | |
|
107 | 107 | REQUIREMENT = b'exp-sqlite-001' |
|
108 | 108 | REQUIREMENT_ZSTD = b'exp-sqlite-comp-001=zstd' |
|
109 | 109 | REQUIREMENT_ZLIB = b'exp-sqlite-comp-001=zlib' |
|
110 | 110 | REQUIREMENT_NONE = b'exp-sqlite-comp-001=none' |
|
111 | 111 | REQUIREMENT_SHALLOW_FILES = b'exp-sqlite-shallow-files' |
|
112 | 112 | |
|
113 | 113 | CURRENT_SCHEMA_VERSION = 1 |
|
114 | 114 | |
|
115 | 115 | COMPRESSION_NONE = 1 |
|
116 | 116 | COMPRESSION_ZSTD = 2 |
|
117 | 117 | COMPRESSION_ZLIB = 3 |
|
118 | 118 | |
|
119 | 119 | FLAG_CENSORED = 1 |
|
120 | 120 | FLAG_MISSING_P1 = 2 |
|
121 | 121 | FLAG_MISSING_P2 = 4 |
|
122 | 122 | |
|
123 | 123 | CREATE_SCHEMA = [ |
|
124 | 124 | # Deltas are stored as content-indexed blobs. |
|
125 | 125 | # compression column holds COMPRESSION_* constant for how the |
|
126 | 126 | # delta is encoded. |
|
127 | 127 | 'CREATE TABLE delta (' |
|
128 | 128 | ' id INTEGER PRIMARY KEY, ' |
|
129 | 129 | ' compression INTEGER NOT NULL, ' |
|
130 | 130 | ' hash BLOB UNIQUE ON CONFLICT ABORT, ' |
|
131 | 131 | ' delta BLOB NOT NULL ' |
|
132 | 132 | ')', |
|
133 | 133 | # Tracked paths are denormalized to integers to avoid redundant |
|
134 | 134 | # storage of the path name. |
|
135 | 135 | 'CREATE TABLE filepath (' |
|
136 | 136 | ' id INTEGER PRIMARY KEY, ' |
|
137 | 137 | ' path BLOB NOT NULL ' |
|
138 | 138 | ')', |
|
139 | 139 | 'CREATE UNIQUE INDEX filepath_path ON filepath (path)', |
|
140 | 140 | # We have a single table for all file revision data. |
|
141 | 141 | # Each file revision is uniquely described by a (path, rev) and |
|
142 | 142 | # (path, node). |
|
143 | 143 | # |
|
144 | 144 | # Revision data is stored as a pointer to the delta producing this |
|
145 | 145 | # revision and the file revision whose delta should be applied before |
|
146 | 146 | # that one. One can reconstruct the delta chain by recursively following |
|
147 | 147 | # the delta base revision pointers until one encounters NULL. |
|
148 | 148 | # |
|
149 | 149 | # flags column holds bitwise integer flags controlling storage options. |
|
150 | 150 | # These flags are defined by the FLAG_* constants. |
|
151 | 151 | 'CREATE TABLE fileindex (' |
|
152 | 152 | ' id INTEGER PRIMARY KEY, ' |
|
153 | 153 | ' pathid INTEGER REFERENCES filepath(id), ' |
|
154 | 154 | ' revnum INTEGER NOT NULL, ' |
|
155 | 155 | ' p1rev INTEGER NOT NULL, ' |
|
156 | 156 | ' p2rev INTEGER NOT NULL, ' |
|
157 | 157 | ' linkrev INTEGER NOT NULL, ' |
|
158 | 158 | ' flags INTEGER NOT NULL, ' |
|
159 | 159 | ' deltaid INTEGER REFERENCES delta(id), ' |
|
160 | 160 | ' deltabaseid INTEGER REFERENCES fileindex(id), ' |
|
161 | 161 | ' node BLOB NOT NULL ' |
|
162 | 162 | ')', |
|
163 | 163 | 'CREATE UNIQUE INDEX fileindex_pathrevnum ' |
|
164 | 164 | ' ON fileindex (pathid, revnum)', |
|
165 | 165 | 'CREATE UNIQUE INDEX fileindex_pathnode ON fileindex (pathid, node)', |
|
166 | 166 | # Provide a view over all file data for convenience. |
|
167 | 167 | 'CREATE VIEW filedata AS ' |
|
168 | 168 | 'SELECT ' |
|
169 | 169 | ' fileindex.id AS id, ' |
|
170 | 170 | ' filepath.id AS pathid, ' |
|
171 | 171 | ' filepath.path AS path, ' |
|
172 | 172 | ' fileindex.revnum AS revnum, ' |
|
173 | 173 | ' fileindex.node AS node, ' |
|
174 | 174 | ' fileindex.p1rev AS p1rev, ' |
|
175 | 175 | ' fileindex.p2rev AS p2rev, ' |
|
176 | 176 | ' fileindex.linkrev AS linkrev, ' |
|
177 | 177 | ' fileindex.flags AS flags, ' |
|
178 | 178 | ' fileindex.deltaid AS deltaid, ' |
|
179 | 179 | ' fileindex.deltabaseid AS deltabaseid ' |
|
180 | 180 | 'FROM filepath, fileindex ' |
|
181 | 181 | 'WHERE fileindex.pathid=filepath.id', |
|
182 | 182 | 'PRAGMA user_version=%d' % CURRENT_SCHEMA_VERSION, |
|
183 | 183 | ] |
|
184 | 184 | |
|
185 | 185 | |
|
186 | 186 | def resolvedeltachain(db, pathid, node, revisioncache, stoprids, zstddctx=None): |
|
187 | 187 | """Resolve a delta chain for a file node.""" |
|
188 | 188 | |
|
189 | 189 | # TODO the "not in ({stops})" here is possibly slowing down the query |
|
190 | 190 | # because it needs to perform the lookup on every recursive invocation. |
|
191 | 191 | # This could possibly be faster if we created a temporary query with |
|
192 | 192 | # baseid "poisoned" to null and limited the recursive filter to |
|
193 | 193 | # "is not null". |
|
194 | 194 | res = db.execute( |
|
195 | 195 | 'WITH RECURSIVE ' |
|
196 | 196 | ' deltachain(deltaid, baseid) AS (' |
|
197 | 197 | ' SELECT deltaid, deltabaseid FROM fileindex ' |
|
198 | 198 | ' WHERE pathid=? AND node=? ' |
|
199 | 199 | ' UNION ALL ' |
|
200 | 200 | ' SELECT fileindex.deltaid, deltabaseid ' |
|
201 | 201 | ' FROM fileindex, deltachain ' |
|
202 | 202 | ' WHERE ' |
|
203 | 203 | ' fileindex.id=deltachain.baseid ' |
|
204 | 204 | ' AND deltachain.baseid IS NOT NULL ' |
|
205 | 205 | ' AND fileindex.id NOT IN ({stops}) ' |
|
206 | 206 | ' ) ' |
|
207 | 207 | 'SELECT deltachain.baseid, compression, delta ' |
|
208 | 208 | 'FROM deltachain, delta ' |
|
209 | 209 | 'WHERE delta.id=deltachain.deltaid'.format( |
|
210 | 210 | stops=','.join(['?'] * len(stoprids)) |
|
211 | 211 | ), |
|
212 | 212 | tuple([pathid, node] + list(stoprids.keys())), |
|
213 | 213 | ) |
|
214 | 214 | |
|
215 | 215 | deltas = [] |
|
216 | 216 | lastdeltabaseid = None |
|
217 | 217 | |
|
218 | 218 | for deltabaseid, compression, delta in res: |
|
219 | 219 | lastdeltabaseid = deltabaseid |
|
220 | 220 | |
|
221 | 221 | if compression == COMPRESSION_ZSTD: |
|
222 | 222 | delta = zstddctx.decompress(delta) |
|
223 | 223 | elif compression == COMPRESSION_NONE: |
|
224 | 224 | delta = delta |
|
225 | 225 | elif compression == COMPRESSION_ZLIB: |
|
226 | 226 | delta = zlib.decompress(delta) |
|
227 | 227 | else: |
|
228 | 228 | raise SQLiteStoreError( |
|
229 | 229 | b'unhandled compression type: %d' % compression |
|
230 | 230 | ) |
|
231 | 231 | |
|
232 | 232 | deltas.append(delta) |
|
233 | 233 | |
|
234 | 234 | if lastdeltabaseid in stoprids: |
|
235 | 235 | basetext = revisioncache[stoprids[lastdeltabaseid]] |
|
236 | 236 | else: |
|
237 | 237 | basetext = deltas.pop() |
|
238 | 238 | |
|
239 | 239 | deltas.reverse() |
|
240 | 240 | fulltext = mdiff.patches(basetext, deltas) |
|
241 | 241 | |
|
242 | 242 | # SQLite returns buffer instances for blob columns on Python 2. This |
|
243 | 243 | # type can propagate through the delta application layer. Because |
|
244 | 244 | # downstream callers assume revisions are bytes, cast as needed. |
|
245 | 245 | if not isinstance(fulltext, bytes): |
|
246 | 246 | fulltext = bytes(delta) |
|
247 | 247 | |
|
248 | 248 | return fulltext |
|
249 | 249 | |
|
250 | 250 | |
|
251 | 251 | def insertdelta(db, compression, hash, delta): |
|
252 | 252 | try: |
|
253 | 253 | return db.execute( |
|
254 | 254 | 'INSERT INTO delta (compression, hash, delta) VALUES (?, ?, ?)', |
|
255 | 255 | (compression, hash, delta), |
|
256 | 256 | ).lastrowid |
|
257 | 257 | except sqlite3.IntegrityError: |
|
258 | 258 | return db.execute( |
|
259 | 259 | 'SELECT id FROM delta WHERE hash=?', (hash,) |
|
260 | 260 | ).fetchone()[0] |
|
261 | 261 | |
|
262 | 262 | |
|
263 | 263 | class SQLiteStoreError(error.StorageError): |
|
264 | 264 | pass |
|
265 | 265 | |
|
266 | 266 | |
|
267 | 267 | @attr.s |
|
268 | 268 | class revisionentry(object): |
|
269 | 269 | rid = attr.ib() |
|
270 | 270 | rev = attr.ib() |
|
271 | 271 | node = attr.ib() |
|
272 | 272 | p1rev = attr.ib() |
|
273 | 273 | p2rev = attr.ib() |
|
274 | 274 | p1node = attr.ib() |
|
275 | 275 | p2node = attr.ib() |
|
276 | 276 | linkrev = attr.ib() |
|
277 | 277 | flags = attr.ib() |
|
278 | 278 | |
|
279 | 279 | |
|
280 | 280 | @interfaceutil.implementer(repository.irevisiondelta) |
|
281 | 281 | @attr.s(slots=True) |
|
282 | 282 | class sqliterevisiondelta(object): |
|
283 | 283 | node = attr.ib() |
|
284 | 284 | p1node = attr.ib() |
|
285 | 285 | p2node = attr.ib() |
|
286 | 286 | basenode = attr.ib() |
|
287 | 287 | flags = attr.ib() |
|
288 | 288 | baserevisionsize = attr.ib() |
|
289 | 289 | revision = attr.ib() |
|
290 | 290 | delta = attr.ib() |
|
291 | 291 | linknode = attr.ib(default=None) |
|
292 | 292 | |
|
293 | 293 | |
|
294 | 294 | @interfaceutil.implementer(repository.iverifyproblem) |
|
295 | 295 | @attr.s(frozen=True) |
|
296 | 296 | class sqliteproblem(object): |
|
297 | 297 | warning = attr.ib(default=None) |
|
298 | 298 | error = attr.ib(default=None) |
|
299 | 299 | node = attr.ib(default=None) |
|
300 | 300 | |
|
301 | 301 | |
|
302 | 302 | @interfaceutil.implementer(repository.ifilestorage) |
|
303 | 303 | class sqlitefilestore(object): |
|
304 | 304 | """Implements storage for an individual tracked path.""" |
|
305 | 305 | |
|
306 | 306 | def __init__(self, db, path, compression): |
|
307 | 307 | self._db = db |
|
308 | 308 | self._path = path |
|
309 | 309 | |
|
310 | 310 | self._pathid = None |
|
311 | 311 | |
|
312 | 312 | # revnum -> node |
|
313 | 313 | self._revtonode = {} |
|
314 | 314 | # node -> revnum |
|
315 | 315 | self._nodetorev = {} |
|
316 | 316 | # node -> data structure |
|
317 | 317 | self._revisions = {} |
|
318 | 318 | |
|
319 | 319 | self._revisioncache = util.lrucachedict(10) |
|
320 | 320 | |
|
321 | 321 | self._compengine = compression |
|
322 | 322 | |
|
323 | 323 | if compression == b'zstd': |
|
324 | 324 | self._cctx = zstd.ZstdCompressor(level=3) |
|
325 | 325 | self._dctx = zstd.ZstdDecompressor() |
|
326 | 326 | else: |
|
327 | 327 | self._cctx = None |
|
328 | 328 | self._dctx = None |
|
329 | 329 | |
|
330 | 330 | self._refreshindex() |
|
331 | 331 | |
|
332 | 332 | def _refreshindex(self): |
|
333 | 333 | self._revtonode = {} |
|
334 | 334 | self._nodetorev = {} |
|
335 | 335 | self._revisions = {} |
|
336 | 336 | |
|
337 | 337 | res = list( |
|
338 | 338 | self._db.execute( |
|
339 | 339 | 'SELECT id FROM filepath WHERE path=?', (self._path,) |
|
340 | 340 | ) |
|
341 | 341 | ) |
|
342 | 342 | |
|
343 | 343 | if not res: |
|
344 | 344 | self._pathid = None |
|
345 | 345 | return |
|
346 | 346 | |
|
347 | 347 | self._pathid = res[0][0] |
|
348 | 348 | |
|
349 | 349 | res = self._db.execute( |
|
350 | 350 | 'SELECT id, revnum, node, p1rev, p2rev, linkrev, flags ' |
|
351 | 351 | 'FROM fileindex ' |
|
352 | 352 | 'WHERE pathid=? ' |
|
353 | 353 | 'ORDER BY revnum ASC', |
|
354 | 354 | (self._pathid,), |
|
355 | 355 | ) |
|
356 | 356 | |
|
357 | 357 | for i, row in enumerate(res): |
|
358 | 358 | rid, rev, node, p1rev, p2rev, linkrev, flags = row |
|
359 | 359 | |
|
360 | 360 | if i != rev: |
|
361 | 361 | raise SQLiteStoreError( |
|
362 | 362 | _(b'sqlite database has inconsistent revision numbers') |
|
363 | 363 | ) |
|
364 | 364 | |
|
365 | 365 | if p1rev == nullrev: |
|
366 | 366 | p1node = nullid |
|
367 | 367 | else: |
|
368 | 368 | p1node = self._revtonode[p1rev] |
|
369 | 369 | |
|
370 | 370 | if p2rev == nullrev: |
|
371 | 371 | p2node = nullid |
|
372 | 372 | else: |
|
373 | 373 | p2node = self._revtonode[p2rev] |
|
374 | 374 | |
|
375 | 375 | entry = revisionentry( |
|
376 | 376 | rid=rid, |
|
377 | 377 | rev=rev, |
|
378 | 378 | node=node, |
|
379 | 379 | p1rev=p1rev, |
|
380 | 380 | p2rev=p2rev, |
|
381 | 381 | p1node=p1node, |
|
382 | 382 | p2node=p2node, |
|
383 | 383 | linkrev=linkrev, |
|
384 | 384 | flags=flags, |
|
385 | 385 | ) |
|
386 | 386 | |
|
387 | 387 | self._revtonode[rev] = node |
|
388 | 388 | self._nodetorev[node] = rev |
|
389 | 389 | self._revisions[node] = entry |
|
390 | 390 | |
|
391 | 391 | # Start of ifileindex interface. |
|
392 | 392 | |
|
393 | 393 | def __len__(self): |
|
394 | 394 | return len(self._revisions) |
|
395 | 395 | |
|
396 | 396 | def __iter__(self): |
|
397 | 397 | return iter(pycompat.xrange(len(self._revisions))) |
|
398 | 398 | |
|
399 | 399 | def hasnode(self, node): |
|
400 | 400 | if node == nullid: |
|
401 | 401 | return False |
|
402 | 402 | |
|
403 | 403 | return node in self._nodetorev |
|
404 | 404 | |
|
405 | 405 | def revs(self, start=0, stop=None): |
|
406 | 406 | return storageutil.iterrevs( |
|
407 | 407 | len(self._revisions), start=start, stop=stop |
|
408 | 408 | ) |
|
409 | 409 | |
|
410 | 410 | def parents(self, node): |
|
411 | 411 | if node == nullid: |
|
412 | 412 | return nullid, nullid |
|
413 | 413 | |
|
414 | 414 | if node not in self._revisions: |
|
415 | 415 | raise error.LookupError(node, self._path, _(b'no node')) |
|
416 | 416 | |
|
417 | 417 | entry = self._revisions[node] |
|
418 | 418 | return entry.p1node, entry.p2node |
|
419 | 419 | |
|
420 | 420 | def parentrevs(self, rev): |
|
421 | 421 | if rev == nullrev: |
|
422 | 422 | return nullrev, nullrev |
|
423 | 423 | |
|
424 | 424 | if rev not in self._revtonode: |
|
425 | 425 | raise IndexError(rev) |
|
426 | 426 | |
|
427 | 427 | entry = self._revisions[self._revtonode[rev]] |
|
428 | 428 | return entry.p1rev, entry.p2rev |
|
429 | 429 | |
|
430 | 430 | def rev(self, node): |
|
431 | 431 | if node == nullid: |
|
432 | 432 | return nullrev |
|
433 | 433 | |
|
434 | 434 | if node not in self._nodetorev: |
|
435 | 435 | raise error.LookupError(node, self._path, _(b'no node')) |
|
436 | 436 | |
|
437 | 437 | return self._nodetorev[node] |
|
438 | 438 | |
|
439 | 439 | def node(self, rev): |
|
440 | 440 | if rev == nullrev: |
|
441 | 441 | return nullid |
|
442 | 442 | |
|
443 | 443 | if rev not in self._revtonode: |
|
444 | 444 | raise IndexError(rev) |
|
445 | 445 | |
|
446 | 446 | return self._revtonode[rev] |
|
447 | 447 | |
|
448 | 448 | def lookup(self, node): |
|
449 | 449 | return storageutil.fileidlookup(self, node, self._path) |
|
450 | 450 | |
|
451 | 451 | def linkrev(self, rev): |
|
452 | 452 | if rev == nullrev: |
|
453 | 453 | return nullrev |
|
454 | 454 | |
|
455 | 455 | if rev not in self._revtonode: |
|
456 | 456 | raise IndexError(rev) |
|
457 | 457 | |
|
458 | 458 | entry = self._revisions[self._revtonode[rev]] |
|
459 | 459 | return entry.linkrev |
|
460 | 460 | |
|
461 | 461 | def iscensored(self, rev): |
|
462 | 462 | if rev == nullrev: |
|
463 | 463 | return False |
|
464 | 464 | |
|
465 | 465 | if rev not in self._revtonode: |
|
466 | 466 | raise IndexError(rev) |
|
467 | 467 | |
|
468 | 468 | return self._revisions[self._revtonode[rev]].flags & FLAG_CENSORED |
|
469 | 469 | |
|
470 | 470 | def commonancestorsheads(self, node1, node2): |
|
471 | 471 | rev1 = self.rev(node1) |
|
472 | 472 | rev2 = self.rev(node2) |
|
473 | 473 | |
|
474 | 474 | ancestors = ancestor.commonancestorsheads(self.parentrevs, rev1, rev2) |
|
475 | 475 | return pycompat.maplist(self.node, ancestors) |
|
476 | 476 | |
|
477 | 477 | def descendants(self, revs): |
|
478 | 478 | # TODO we could implement this using a recursive SQL query, which |
|
479 | 479 | # might be faster. |
|
480 | 480 | return dagop.descendantrevs(revs, self.revs, self.parentrevs) |
|
481 | 481 | |
|
482 | 482 | def heads(self, start=None, stop=None): |
|
483 | 483 | if start is None and stop is None: |
|
484 | 484 | if not len(self): |
|
485 | 485 | return [nullid] |
|
486 | 486 | |
|
487 | 487 | startrev = self.rev(start) if start is not None else nullrev |
|
488 | 488 | stoprevs = {self.rev(n) for n in stop or []} |
|
489 | 489 | |
|
490 | 490 | revs = dagop.headrevssubset( |
|
491 | 491 | self.revs, self.parentrevs, startrev=startrev, stoprevs=stoprevs |
|
492 | 492 | ) |
|
493 | 493 | |
|
494 | 494 | return [self.node(rev) for rev in revs] |
|
495 | 495 | |
|
496 | 496 | def children(self, node): |
|
497 | 497 | rev = self.rev(node) |
|
498 | 498 | |
|
499 | 499 | res = self._db.execute( |
|
500 | 500 | 'SELECT' |
|
501 | 501 | ' node ' |
|
502 | 502 | ' FROM filedata ' |
|
503 | 503 | ' WHERE path=? AND (p1rev=? OR p2rev=?) ' |
|
504 | 504 | ' ORDER BY revnum ASC', |
|
505 | 505 | (self._path, rev, rev), |
|
506 | 506 | ) |
|
507 | 507 | |
|
508 | 508 | return [row[0] for row in res] |
|
509 | 509 | |
|
510 | 510 | # End of ifileindex interface. |
|
511 | 511 | |
|
512 | 512 | # Start of ifiledata interface. |
|
513 | 513 | |
|
514 | 514 | def size(self, rev): |
|
515 | 515 | if rev == nullrev: |
|
516 | 516 | return 0 |
|
517 | 517 | |
|
518 | 518 | if rev not in self._revtonode: |
|
519 | 519 | raise IndexError(rev) |
|
520 | 520 | |
|
521 | 521 | node = self._revtonode[rev] |
|
522 | 522 | |
|
523 | 523 | if self.renamed(node): |
|
524 | 524 | return len(self.read(node)) |
|
525 | 525 | |
|
526 | 526 | return len(self.revision(node)) |
|
527 | 527 | |
|
528 | 528 | def revision(self, node, raw=False, _verifyhash=True): |
|
529 | 529 | if node in (nullid, nullrev): |
|
530 | 530 | return b'' |
|
531 | 531 | |
|
532 | 532 | if isinstance(node, int): |
|
533 | 533 | node = self.node(node) |
|
534 | 534 | |
|
535 | 535 | if node not in self._nodetorev: |
|
536 | 536 | raise error.LookupError(node, self._path, _(b'no node')) |
|
537 | 537 | |
|
538 | 538 | if node in self._revisioncache: |
|
539 | 539 | return self._revisioncache[node] |
|
540 | 540 | |
|
541 | 541 | # Because we have a fulltext revision cache, we are able to |
|
542 | 542 | # short-circuit delta chain traversal and decompression as soon as |
|
543 | 543 | # we encounter a revision in the cache. |
|
544 | 544 | |
|
545 | 545 | stoprids = {self._revisions[n].rid: n for n in self._revisioncache} |
|
546 | 546 | |
|
547 | 547 | if not stoprids: |
|
548 | 548 | stoprids[-1] = None |
|
549 | 549 | |
|
550 | 550 | fulltext = resolvedeltachain( |
|
551 | 551 | self._db, |
|
552 | 552 | self._pathid, |
|
553 | 553 | node, |
|
554 | 554 | self._revisioncache, |
|
555 | 555 | stoprids, |
|
556 | 556 | zstddctx=self._dctx, |
|
557 | 557 | ) |
|
558 | 558 | |
|
559 | 559 | # Don't verify hashes if parent nodes were rewritten, as the hash |
|
560 | 560 | # wouldn't verify. |
|
561 | 561 | if self._revisions[node].flags & (FLAG_MISSING_P1 | FLAG_MISSING_P2): |
|
562 | 562 | _verifyhash = False |
|
563 | 563 | |
|
564 | 564 | if _verifyhash: |
|
565 | 565 | self._checkhash(fulltext, node) |
|
566 | 566 | self._revisioncache[node] = fulltext |
|
567 | 567 | |
|
568 | 568 | return fulltext |
|
569 | 569 | |
|
570 | 570 | def rawdata(self, *args, **kwargs): |
|
571 | 571 | return self.revision(*args, **kwargs) |
|
572 | 572 | |
|
573 | 573 | def read(self, node): |
|
574 | 574 | return storageutil.filtermetadata(self.revision(node)) |
|
575 | 575 | |
|
576 | 576 | def renamed(self, node): |
|
577 | 577 | return storageutil.filerevisioncopied(self, node) |
|
578 | 578 | |
|
579 | 579 | def cmp(self, node, fulltext): |
|
580 | 580 | return not storageutil.filedataequivalent(self, node, fulltext) |
|
581 | 581 | |
|
582 | 582 | def emitrevisions( |
|
583 | 583 | self, |
|
584 | 584 | nodes, |
|
585 | 585 | nodesorder=None, |
|
586 | 586 | revisiondata=False, |
|
587 | 587 | assumehaveparentrevisions=False, |
|
588 | 588 | deltamode=repository.CG_DELTAMODE_STD, |
|
589 | 589 | ): |
|
590 | 590 | if nodesorder not in (b'nodes', b'storage', b'linear', None): |
|
591 | 591 | raise error.ProgrammingError( |
|
592 | 592 | b'unhandled value for nodesorder: %s' % nodesorder |
|
593 | 593 | ) |
|
594 | 594 | |
|
595 | 595 | nodes = [n for n in nodes if n != nullid] |
|
596 | 596 | |
|
597 | 597 | if not nodes: |
|
598 | 598 | return |
|
599 | 599 | |
|
600 | 600 | # TODO perform in a single query. |
|
601 | 601 | res = self._db.execute( |
|
602 | 602 | 'SELECT revnum, deltaid FROM fileindex ' |
|
603 | 603 | 'WHERE pathid=? ' |
|
604 | 604 | ' AND node in (%s)' % (','.join(['?'] * len(nodes))), |
|
605 | 605 | tuple([self._pathid] + nodes), |
|
606 | 606 | ) |
|
607 | 607 | |
|
608 | 608 | deltabases = {} |
|
609 | 609 | |
|
610 | 610 | for rev, deltaid in res: |
|
611 | 611 | res = self._db.execute( |
|
612 | 612 | 'SELECT revnum from fileindex WHERE pathid=? AND deltaid=?', |
|
613 | 613 | (self._pathid, deltaid), |
|
614 | 614 | ) |
|
615 | 615 | deltabases[rev] = res.fetchone()[0] |
|
616 | 616 | |
|
617 | 617 | # TODO define revdifffn so we can use delta from storage. |
|
618 | 618 | for delta in storageutil.emitrevisions( |
|
619 | 619 | self, |
|
620 | 620 | nodes, |
|
621 | 621 | nodesorder, |
|
622 | 622 | sqliterevisiondelta, |
|
623 | 623 | deltaparentfn=deltabases.__getitem__, |
|
624 | 624 | revisiondata=revisiondata, |
|
625 | 625 | assumehaveparentrevisions=assumehaveparentrevisions, |
|
626 | 626 | deltamode=deltamode, |
|
627 | 627 | ): |
|
628 | 628 | |
|
629 | 629 | yield delta |
|
630 | 630 | |
|
631 | 631 | # End of ifiledata interface. |
|
632 | 632 | |
|
633 | 633 | # Start of ifilemutation interface. |
|
634 | 634 | |
|
635 | 635 | def add(self, filedata, meta, transaction, linkrev, p1, p2): |
|
636 | 636 | if meta or filedata.startswith(b'\x01\n'): |
|
637 | 637 | filedata = storageutil.packmeta(meta, filedata) |
|
638 | 638 | |
|
639 |
re |
|
|
639 | rev = self.addrevision(filedata, transaction, linkrev, p1, p2) | |
|
640 | return self.node(rev) | |
|
640 | 641 | |
|
641 | 642 | def addrevision( |
|
642 | 643 | self, |
|
643 | 644 | revisiondata, |
|
644 | 645 | transaction, |
|
645 | 646 | linkrev, |
|
646 | 647 | p1, |
|
647 | 648 | p2, |
|
648 | 649 | node=None, |
|
649 | 650 | flags=0, |
|
650 | 651 | cachedelta=None, |
|
651 | 652 | ): |
|
652 | 653 | if flags: |
|
653 | 654 | raise SQLiteStoreError(_(b'flags not supported on revisions')) |
|
654 | 655 | |
|
655 | 656 | validatehash = node is not None |
|
656 | 657 | node = node or storageutil.hashrevisionsha1(revisiondata, p1, p2) |
|
657 | 658 | |
|
658 | 659 | if validatehash: |
|
659 | 660 | self._checkhash(revisiondata, node, p1, p2) |
|
660 | 661 | |
|
661 |
|
|
|
662 |
|
|
|
662 | rev = self._nodetorev.get(node) | |
|
663 | if rev is not None: | |
|
664 | return rev | |
|
663 | 665 | |
|
664 |
|
|
|
666 | rev = self._addrawrevision( | |
|
665 | 667 | node, revisiondata, transaction, linkrev, p1, p2 |
|
666 | 668 | ) |
|
667 | 669 | |
|
668 | 670 | self._revisioncache[node] = revisiondata |
|
669 |
return |
|
|
671 | return rev | |
|
670 | 672 | |
|
671 | 673 | def addgroup( |
|
672 | 674 | self, |
|
673 | 675 | deltas, |
|
674 | 676 | linkmapper, |
|
675 | 677 | transaction, |
|
676 | 678 | addrevisioncb=None, |
|
677 | 679 | duplicaterevisioncb=None, |
|
678 | 680 | maybemissingparents=False, |
|
679 | 681 | ): |
|
680 | 682 | empty = True |
|
681 | 683 | |
|
682 | 684 | for node, p1, p2, linknode, deltabase, delta, wireflags in deltas: |
|
683 | 685 | storeflags = 0 |
|
684 | 686 | |
|
685 | 687 | if wireflags & repository.REVISION_FLAG_CENSORED: |
|
686 | 688 | storeflags |= FLAG_CENSORED |
|
687 | 689 | |
|
688 | 690 | if wireflags & ~repository.REVISION_FLAG_CENSORED: |
|
689 | 691 | raise SQLiteStoreError(b'unhandled revision flag') |
|
690 | 692 | |
|
691 | 693 | if maybemissingparents: |
|
692 | 694 | if p1 != nullid and not self.hasnode(p1): |
|
693 | 695 | p1 = nullid |
|
694 | 696 | storeflags |= FLAG_MISSING_P1 |
|
695 | 697 | |
|
696 | 698 | if p2 != nullid and not self.hasnode(p2): |
|
697 | 699 | p2 = nullid |
|
698 | 700 | storeflags |= FLAG_MISSING_P2 |
|
699 | 701 | |
|
700 | 702 | baserev = self.rev(deltabase) |
|
701 | 703 | |
|
702 | 704 | # If base is censored, delta must be full replacement in a single |
|
703 | 705 | # patch operation. |
|
704 | 706 | if baserev != nullrev and self.iscensored(baserev): |
|
705 | 707 | hlen = struct.calcsize(b'>lll') |
|
706 | 708 | oldlen = len(self.rawdata(deltabase, _verifyhash=False)) |
|
707 | 709 | newlen = len(delta) - hlen |
|
708 | 710 | |
|
709 | 711 | if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen): |
|
710 | 712 | raise error.CensoredBaseError(self._path, deltabase) |
|
711 | 713 | |
|
712 | 714 | if not (storeflags & FLAG_CENSORED) and storageutil.deltaiscensored( |
|
713 | 715 | delta, baserev, lambda x: len(self.rawdata(x)) |
|
714 | 716 | ): |
|
715 | 717 | storeflags |= FLAG_CENSORED |
|
716 | 718 | |
|
717 | 719 | linkrev = linkmapper(linknode) |
|
718 | 720 | |
|
719 | 721 | if node in self._revisions: |
|
720 | 722 | # Possibly reset parents to make them proper. |
|
721 | 723 | entry = self._revisions[node] |
|
722 | 724 | |
|
723 | 725 | if entry.flags & FLAG_MISSING_P1 and p1 != nullid: |
|
724 | 726 | entry.p1node = p1 |
|
725 | 727 | entry.p1rev = self._nodetorev[p1] |
|
726 | 728 | entry.flags &= ~FLAG_MISSING_P1 |
|
727 | 729 | |
|
728 | 730 | self._db.execute( |
|
729 | 731 | 'UPDATE fileindex SET p1rev=?, flags=? WHERE id=?', |
|
730 | 732 | (self._nodetorev[p1], entry.flags, entry.rid), |
|
731 | 733 | ) |
|
732 | 734 | |
|
733 | 735 | if entry.flags & FLAG_MISSING_P2 and p2 != nullid: |
|
734 | 736 | entry.p2node = p2 |
|
735 | 737 | entry.p2rev = self._nodetorev[p2] |
|
736 | 738 | entry.flags &= ~FLAG_MISSING_P2 |
|
737 | 739 | |
|
738 | 740 | self._db.execute( |
|
739 | 741 | 'UPDATE fileindex SET p2rev=?, flags=? WHERE id=?', |
|
740 | 742 | (self._nodetorev[p1], entry.flags, entry.rid), |
|
741 | 743 | ) |
|
742 | 744 | |
|
743 | 745 | if duplicaterevisioncb: |
|
744 | 746 | duplicaterevisioncb(self, node) |
|
745 | 747 | empty = False |
|
746 | 748 | continue |
|
747 | 749 | |
|
748 | 750 | if deltabase == nullid: |
|
749 | 751 | text = mdiff.patch(b'', delta) |
|
750 | 752 | storedelta = None |
|
751 | 753 | else: |
|
752 | 754 | text = None |
|
753 | 755 | storedelta = (deltabase, delta) |
|
754 | 756 | |
|
755 | 757 | self._addrawrevision( |
|
756 | 758 | node, |
|
757 | 759 | text, |
|
758 | 760 | transaction, |
|
759 | 761 | linkrev, |
|
760 | 762 | p1, |
|
761 | 763 | p2, |
|
762 | 764 | storedelta=storedelta, |
|
763 | 765 | flags=storeflags, |
|
764 | 766 | ) |
|
765 | 767 | |
|
766 | 768 | if addrevisioncb: |
|
767 | 769 | addrevisioncb(self, node) |
|
768 | 770 | empty = False |
|
769 | 771 | |
|
770 | 772 | return not empty |
|
771 | 773 | |
|
772 | 774 | def censorrevision(self, tr, censornode, tombstone=b''): |
|
773 | 775 | tombstone = storageutil.packmeta({b'censored': tombstone}, b'') |
|
774 | 776 | |
|
775 | 777 | # This restriction is cargo culted from revlogs and makes no sense for |
|
776 | 778 | # SQLite, since columns can be resized at will. |
|
777 | 779 | if len(tombstone) > len(self.rawdata(censornode)): |
|
778 | 780 | raise error.Abort( |
|
779 | 781 | _(b'censor tombstone must be no longer than censored data') |
|
780 | 782 | ) |
|
781 | 783 | |
|
782 | 784 | # We need to replace the censored revision's data with the tombstone. |
|
783 | 785 | # But replacing that data will have implications for delta chains that |
|
784 | 786 | # reference it. |
|
785 | 787 | # |
|
786 | 788 | # While "better," more complex strategies are possible, we do something |
|
787 | 789 | # simple: we find delta chain children of the censored revision and we |
|
788 | 790 | # replace those incremental deltas with fulltexts of their corresponding |
|
789 | 791 | # revision. Then we delete the now-unreferenced delta and original |
|
790 | 792 | # revision and insert a replacement. |
|
791 | 793 | |
|
792 | 794 | # Find the delta to be censored. |
|
793 | 795 | censoreddeltaid = self._db.execute( |
|
794 | 796 | 'SELECT deltaid FROM fileindex WHERE id=?', |
|
795 | 797 | (self._revisions[censornode].rid,), |
|
796 | 798 | ).fetchone()[0] |
|
797 | 799 | |
|
798 | 800 | # Find all its delta chain children. |
|
799 | 801 | # TODO once we support storing deltas for !files, we'll need to look |
|
800 | 802 | # for those delta chains too. |
|
801 | 803 | rows = list( |
|
802 | 804 | self._db.execute( |
|
803 | 805 | 'SELECT id, pathid, node FROM fileindex ' |
|
804 | 806 | 'WHERE deltabaseid=? OR deltaid=?', |
|
805 | 807 | (censoreddeltaid, censoreddeltaid), |
|
806 | 808 | ) |
|
807 | 809 | ) |
|
808 | 810 | |
|
809 | 811 | for row in rows: |
|
810 | 812 | rid, pathid, node = row |
|
811 | 813 | |
|
812 | 814 | fulltext = resolvedeltachain( |
|
813 | 815 | self._db, pathid, node, {}, {-1: None}, zstddctx=self._dctx |
|
814 | 816 | ) |
|
815 | 817 | |
|
816 | 818 | deltahash = hashutil.sha1(fulltext).digest() |
|
817 | 819 | |
|
818 | 820 | if self._compengine == b'zstd': |
|
819 | 821 | deltablob = self._cctx.compress(fulltext) |
|
820 | 822 | compression = COMPRESSION_ZSTD |
|
821 | 823 | elif self._compengine == b'zlib': |
|
822 | 824 | deltablob = zlib.compress(fulltext) |
|
823 | 825 | compression = COMPRESSION_ZLIB |
|
824 | 826 | elif self._compengine == b'none': |
|
825 | 827 | deltablob = fulltext |
|
826 | 828 | compression = COMPRESSION_NONE |
|
827 | 829 | else: |
|
828 | 830 | raise error.ProgrammingError( |
|
829 | 831 | b'unhandled compression engine: %s' % self._compengine |
|
830 | 832 | ) |
|
831 | 833 | |
|
832 | 834 | if len(deltablob) >= len(fulltext): |
|
833 | 835 | deltablob = fulltext |
|
834 | 836 | compression = COMPRESSION_NONE |
|
835 | 837 | |
|
836 | 838 | deltaid = insertdelta(self._db, compression, deltahash, deltablob) |
|
837 | 839 | |
|
838 | 840 | self._db.execute( |
|
839 | 841 | 'UPDATE fileindex SET deltaid=?, deltabaseid=NULL ' |
|
840 | 842 | 'WHERE id=?', |
|
841 | 843 | (deltaid, rid), |
|
842 | 844 | ) |
|
843 | 845 | |
|
844 | 846 | # Now create the tombstone delta and replace the delta on the censored |
|
845 | 847 | # node. |
|
846 | 848 | deltahash = hashutil.sha1(tombstone).digest() |
|
847 | 849 | tombstonedeltaid = insertdelta( |
|
848 | 850 | self._db, COMPRESSION_NONE, deltahash, tombstone |
|
849 | 851 | ) |
|
850 | 852 | |
|
851 | 853 | flags = self._revisions[censornode].flags |
|
852 | 854 | flags |= FLAG_CENSORED |
|
853 | 855 | |
|
854 | 856 | self._db.execute( |
|
855 | 857 | 'UPDATE fileindex SET flags=?, deltaid=?, deltabaseid=NULL ' |
|
856 | 858 | 'WHERE pathid=? AND node=?', |
|
857 | 859 | (flags, tombstonedeltaid, self._pathid, censornode), |
|
858 | 860 | ) |
|
859 | 861 | |
|
860 | 862 | self._db.execute('DELETE FROM delta WHERE id=?', (censoreddeltaid,)) |
|
861 | 863 | |
|
862 | 864 | self._refreshindex() |
|
863 | 865 | self._revisioncache.clear() |
|
864 | 866 | |
|
865 | 867 | def getstrippoint(self, minlink): |
|
866 | 868 | return storageutil.resolvestripinfo( |
|
867 | 869 | minlink, |
|
868 | 870 | len(self) - 1, |
|
869 | 871 | [self.rev(n) for n in self.heads()], |
|
870 | 872 | self.linkrev, |
|
871 | 873 | self.parentrevs, |
|
872 | 874 | ) |
|
873 | 875 | |
|
874 | 876 | def strip(self, minlink, transaction): |
|
875 | 877 | if not len(self): |
|
876 | 878 | return |
|
877 | 879 | |
|
878 | 880 | rev, _ignored = self.getstrippoint(minlink) |
|
879 | 881 | |
|
880 | 882 | if rev == len(self): |
|
881 | 883 | return |
|
882 | 884 | |
|
883 | 885 | for rev in self.revs(rev): |
|
884 | 886 | self._db.execute( |
|
885 | 887 | 'DELETE FROM fileindex WHERE pathid=? AND node=?', |
|
886 | 888 | (self._pathid, self.node(rev)), |
|
887 | 889 | ) |
|
888 | 890 | |
|
889 | 891 | # TODO how should we garbage collect data in delta table? |
|
890 | 892 | |
|
891 | 893 | self._refreshindex() |
|
892 | 894 | |
|
893 | 895 | # End of ifilemutation interface. |
|
894 | 896 | |
|
895 | 897 | # Start of ifilestorage interface. |
|
896 | 898 | |
|
897 | 899 | def files(self): |
|
898 | 900 | return [] |
|
899 | 901 | |
|
900 | 902 | def storageinfo( |
|
901 | 903 | self, |
|
902 | 904 | exclusivefiles=False, |
|
903 | 905 | sharedfiles=False, |
|
904 | 906 | revisionscount=False, |
|
905 | 907 | trackedsize=False, |
|
906 | 908 | storedsize=False, |
|
907 | 909 | ): |
|
908 | 910 | d = {} |
|
909 | 911 | |
|
910 | 912 | if exclusivefiles: |
|
911 | 913 | d[b'exclusivefiles'] = [] |
|
912 | 914 | |
|
913 | 915 | if sharedfiles: |
|
914 | 916 | # TODO list sqlite file(s) here. |
|
915 | 917 | d[b'sharedfiles'] = [] |
|
916 | 918 | |
|
917 | 919 | if revisionscount: |
|
918 | 920 | d[b'revisionscount'] = len(self) |
|
919 | 921 | |
|
920 | 922 | if trackedsize: |
|
921 | 923 | d[b'trackedsize'] = sum( |
|
922 | 924 | len(self.revision(node)) for node in self._nodetorev |
|
923 | 925 | ) |
|
924 | 926 | |
|
925 | 927 | if storedsize: |
|
926 | 928 | # TODO implement this? |
|
927 | 929 | d[b'storedsize'] = None |
|
928 | 930 | |
|
929 | 931 | return d |
|
930 | 932 | |
|
931 | 933 | def verifyintegrity(self, state): |
|
932 | 934 | state[b'skipread'] = set() |
|
933 | 935 | |
|
934 | 936 | for rev in self: |
|
935 | 937 | node = self.node(rev) |
|
936 | 938 | |
|
937 | 939 | try: |
|
938 | 940 | self.revision(node) |
|
939 | 941 | except Exception as e: |
|
940 | 942 | yield sqliteproblem( |
|
941 | 943 | error=_(b'unpacking %s: %s') % (short(node), e), node=node |
|
942 | 944 | ) |
|
943 | 945 | |
|
944 | 946 | state[b'skipread'].add(node) |
|
945 | 947 | |
|
946 | 948 | # End of ifilestorage interface. |
|
947 | 949 | |
|
948 | 950 | def _checkhash(self, fulltext, node, p1=None, p2=None): |
|
949 | 951 | if p1 is None and p2 is None: |
|
950 | 952 | p1, p2 = self.parents(node) |
|
951 | 953 | |
|
952 | 954 | if node == storageutil.hashrevisionsha1(fulltext, p1, p2): |
|
953 | 955 | return |
|
954 | 956 | |
|
955 | 957 | try: |
|
956 | 958 | del self._revisioncache[node] |
|
957 | 959 | except KeyError: |
|
958 | 960 | pass |
|
959 | 961 | |
|
960 | 962 | if storageutil.iscensoredtext(fulltext): |
|
961 | 963 | raise error.CensoredNodeError(self._path, node, fulltext) |
|
962 | 964 | |
|
963 | 965 | raise SQLiteStoreError(_(b'integrity check failed on %s') % self._path) |
|
964 | 966 | |
|
965 | 967 | def _addrawrevision( |
|
966 | 968 | self, |
|
967 | 969 | node, |
|
968 | 970 | revisiondata, |
|
969 | 971 | transaction, |
|
970 | 972 | linkrev, |
|
971 | 973 | p1, |
|
972 | 974 | p2, |
|
973 | 975 | storedelta=None, |
|
974 | 976 | flags=0, |
|
975 | 977 | ): |
|
976 | 978 | if self._pathid is None: |
|
977 | 979 | res = self._db.execute( |
|
978 | 980 | 'INSERT INTO filepath (path) VALUES (?)', (self._path,) |
|
979 | 981 | ) |
|
980 | 982 | self._pathid = res.lastrowid |
|
981 | 983 | |
|
982 | 984 | # For simplicity, always store a delta against p1. |
|
983 | 985 | # TODO we need a lot more logic here to make behavior reasonable. |
|
984 | 986 | |
|
985 | 987 | if storedelta: |
|
986 | 988 | deltabase, delta = storedelta |
|
987 | 989 | |
|
988 | 990 | if isinstance(deltabase, int): |
|
989 | 991 | deltabase = self.node(deltabase) |
|
990 | 992 | |
|
991 | 993 | else: |
|
992 | 994 | assert revisiondata is not None |
|
993 | 995 | deltabase = p1 |
|
994 | 996 | |
|
995 | 997 | if deltabase == nullid: |
|
996 | 998 | delta = revisiondata |
|
997 | 999 | else: |
|
998 | 1000 | delta = mdiff.textdiff( |
|
999 | 1001 | self.revision(self.rev(deltabase)), revisiondata |
|
1000 | 1002 | ) |
|
1001 | 1003 | |
|
1002 | 1004 | # File index stores a pointer to its delta and the parent delta. |
|
1003 | 1005 | # The parent delta is stored via a pointer to the fileindex PK. |
|
1004 | 1006 | if deltabase == nullid: |
|
1005 | 1007 | baseid = None |
|
1006 | 1008 | else: |
|
1007 | 1009 | baseid = self._revisions[deltabase].rid |
|
1008 | 1010 | |
|
1009 | 1011 | # Deltas are stored with a hash of their content. This allows |
|
1010 | 1012 | # us to de-duplicate. The table is configured to ignore conflicts |
|
1011 | 1013 | # and it is faster to just insert and silently noop than to look |
|
1012 | 1014 | # first. |
|
1013 | 1015 | deltahash = hashutil.sha1(delta).digest() |
|
1014 | 1016 | |
|
1015 | 1017 | if self._compengine == b'zstd': |
|
1016 | 1018 | deltablob = self._cctx.compress(delta) |
|
1017 | 1019 | compression = COMPRESSION_ZSTD |
|
1018 | 1020 | elif self._compengine == b'zlib': |
|
1019 | 1021 | deltablob = zlib.compress(delta) |
|
1020 | 1022 | compression = COMPRESSION_ZLIB |
|
1021 | 1023 | elif self._compengine == b'none': |
|
1022 | 1024 | deltablob = delta |
|
1023 | 1025 | compression = COMPRESSION_NONE |
|
1024 | 1026 | else: |
|
1025 | 1027 | raise error.ProgrammingError( |
|
1026 | 1028 | b'unhandled compression engine: %s' % self._compengine |
|
1027 | 1029 | ) |
|
1028 | 1030 | |
|
1029 | 1031 | # Don't store compressed data if it isn't practical. |
|
1030 | 1032 | if len(deltablob) >= len(delta): |
|
1031 | 1033 | deltablob = delta |
|
1032 | 1034 | compression = COMPRESSION_NONE |
|
1033 | 1035 | |
|
1034 | 1036 | deltaid = insertdelta(self._db, compression, deltahash, deltablob) |
|
1035 | 1037 | |
|
1036 | 1038 | rev = len(self) |
|
1037 | 1039 | |
|
1038 | 1040 | if p1 == nullid: |
|
1039 | 1041 | p1rev = nullrev |
|
1040 | 1042 | else: |
|
1041 | 1043 | p1rev = self._nodetorev[p1] |
|
1042 | 1044 | |
|
1043 | 1045 | if p2 == nullid: |
|
1044 | 1046 | p2rev = nullrev |
|
1045 | 1047 | else: |
|
1046 | 1048 | p2rev = self._nodetorev[p2] |
|
1047 | 1049 | |
|
1048 | 1050 | rid = self._db.execute( |
|
1049 | 1051 | 'INSERT INTO fileindex (' |
|
1050 | 1052 | ' pathid, revnum, node, p1rev, p2rev, linkrev, flags, ' |
|
1051 | 1053 | ' deltaid, deltabaseid) ' |
|
1052 | 1054 | ' VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)', |
|
1053 | 1055 | ( |
|
1054 | 1056 | self._pathid, |
|
1055 | 1057 | rev, |
|
1056 | 1058 | node, |
|
1057 | 1059 | p1rev, |
|
1058 | 1060 | p2rev, |
|
1059 | 1061 | linkrev, |
|
1060 | 1062 | flags, |
|
1061 | 1063 | deltaid, |
|
1062 | 1064 | baseid, |
|
1063 | 1065 | ), |
|
1064 | 1066 | ).lastrowid |
|
1065 | 1067 | |
|
1066 | 1068 | entry = revisionentry( |
|
1067 | 1069 | rid=rid, |
|
1068 | 1070 | rev=rev, |
|
1069 | 1071 | node=node, |
|
1070 | 1072 | p1rev=p1rev, |
|
1071 | 1073 | p2rev=p2rev, |
|
1072 | 1074 | p1node=p1, |
|
1073 | 1075 | p2node=p2, |
|
1074 | 1076 | linkrev=linkrev, |
|
1075 | 1077 | flags=flags, |
|
1076 | 1078 | ) |
|
1077 | 1079 | |
|
1078 | 1080 | self._nodetorev[node] = rev |
|
1079 | 1081 | self._revtonode[rev] = node |
|
1080 | 1082 | self._revisions[node] = entry |
|
1081 | 1083 | |
|
1082 |
return |
|
|
1084 | return rev | |
|
1083 | 1085 | |
|
1084 | 1086 | |
|
1085 | 1087 | class sqliterepository(localrepo.localrepository): |
|
1086 | 1088 | def cancopy(self): |
|
1087 | 1089 | return False |
|
1088 | 1090 | |
|
1089 | 1091 | def transaction(self, *args, **kwargs): |
|
1090 | 1092 | current = self.currenttransaction() |
|
1091 | 1093 | |
|
1092 | 1094 | tr = super(sqliterepository, self).transaction(*args, **kwargs) |
|
1093 | 1095 | |
|
1094 | 1096 | if current: |
|
1095 | 1097 | return tr |
|
1096 | 1098 | |
|
1097 | 1099 | self._dbconn.execute('BEGIN TRANSACTION') |
|
1098 | 1100 | |
|
1099 | 1101 | def committransaction(_): |
|
1100 | 1102 | self._dbconn.commit() |
|
1101 | 1103 | |
|
1102 | 1104 | tr.addfinalize(b'sqlitestore', committransaction) |
|
1103 | 1105 | |
|
1104 | 1106 | return tr |
|
1105 | 1107 | |
|
1106 | 1108 | @property |
|
1107 | 1109 | def _dbconn(self): |
|
1108 | 1110 | # SQLite connections can only be used on the thread that created |
|
1109 | 1111 | # them. In most cases, this "just works." However, hgweb uses |
|
1110 | 1112 | # multiple threads. |
|
1111 | 1113 | tid = threading.current_thread().ident |
|
1112 | 1114 | |
|
1113 | 1115 | if self._db: |
|
1114 | 1116 | if self._db[0] == tid: |
|
1115 | 1117 | return self._db[1] |
|
1116 | 1118 | |
|
1117 | 1119 | db = makedb(self.svfs.join(b'db.sqlite')) |
|
1118 | 1120 | self._db = (tid, db) |
|
1119 | 1121 | |
|
1120 | 1122 | return db |
|
1121 | 1123 | |
|
1122 | 1124 | |
|
1123 | 1125 | def makedb(path): |
|
1124 | 1126 | """Construct a database handle for a database at path.""" |
|
1125 | 1127 | |
|
1126 | 1128 | db = sqlite3.connect(encoding.strfromlocal(path)) |
|
1127 | 1129 | db.text_factory = bytes |
|
1128 | 1130 | |
|
1129 | 1131 | res = db.execute('PRAGMA user_version').fetchone()[0] |
|
1130 | 1132 | |
|
1131 | 1133 | # New database. |
|
1132 | 1134 | if res == 0: |
|
1133 | 1135 | for statement in CREATE_SCHEMA: |
|
1134 | 1136 | db.execute(statement) |
|
1135 | 1137 | |
|
1136 | 1138 | db.commit() |
|
1137 | 1139 | |
|
1138 | 1140 | elif res == CURRENT_SCHEMA_VERSION: |
|
1139 | 1141 | pass |
|
1140 | 1142 | |
|
1141 | 1143 | else: |
|
1142 | 1144 | raise error.Abort(_(b'sqlite database has unrecognized version')) |
|
1143 | 1145 | |
|
1144 | 1146 | db.execute('PRAGMA journal_mode=WAL') |
|
1145 | 1147 | |
|
1146 | 1148 | return db |
|
1147 | 1149 | |
|
1148 | 1150 | |
|
1149 | 1151 | def featuresetup(ui, supported): |
|
1150 | 1152 | supported.add(REQUIREMENT) |
|
1151 | 1153 | |
|
1152 | 1154 | if zstd: |
|
1153 | 1155 | supported.add(REQUIREMENT_ZSTD) |
|
1154 | 1156 | |
|
1155 | 1157 | supported.add(REQUIREMENT_ZLIB) |
|
1156 | 1158 | supported.add(REQUIREMENT_NONE) |
|
1157 | 1159 | supported.add(REQUIREMENT_SHALLOW_FILES) |
|
1158 | 1160 | supported.add(requirements.NARROW_REQUIREMENT) |
|
1159 | 1161 | |
|
1160 | 1162 | |
|
1161 | 1163 | def newreporequirements(orig, ui, createopts): |
|
1162 | 1164 | if createopts[b'backend'] != b'sqlite': |
|
1163 | 1165 | return orig(ui, createopts) |
|
1164 | 1166 | |
|
1165 | 1167 | # This restriction can be lifted once we have more confidence. |
|
1166 | 1168 | if b'sharedrepo' in createopts: |
|
1167 | 1169 | raise error.Abort( |
|
1168 | 1170 | _(b'shared repositories not supported with SQLite store') |
|
1169 | 1171 | ) |
|
1170 | 1172 | |
|
1171 | 1173 | # This filtering is out of an abundance of caution: we want to ensure |
|
1172 | 1174 | # we honor creation options and we do that by annotating exactly the |
|
1173 | 1175 | # creation options we recognize. |
|
1174 | 1176 | known = { |
|
1175 | 1177 | b'narrowfiles', |
|
1176 | 1178 | b'backend', |
|
1177 | 1179 | b'shallowfilestore', |
|
1178 | 1180 | } |
|
1179 | 1181 | |
|
1180 | 1182 | unsupported = set(createopts) - known |
|
1181 | 1183 | if unsupported: |
|
1182 | 1184 | raise error.Abort( |
|
1183 | 1185 | _(b'SQLite store does not support repo creation option: %s') |
|
1184 | 1186 | % b', '.join(sorted(unsupported)) |
|
1185 | 1187 | ) |
|
1186 | 1188 | |
|
1187 | 1189 | # Since we're a hybrid store that still relies on revlogs, we fall back |
|
1188 | 1190 | # to using the revlogv1 backend's storage requirements then adding our |
|
1189 | 1191 | # own requirement. |
|
1190 | 1192 | createopts[b'backend'] = b'revlogv1' |
|
1191 | 1193 | requirements = orig(ui, createopts) |
|
1192 | 1194 | requirements.add(REQUIREMENT) |
|
1193 | 1195 | |
|
1194 | 1196 | compression = ui.config(b'storage', b'sqlite.compression') |
|
1195 | 1197 | |
|
1196 | 1198 | if compression == b'zstd' and not zstd: |
|
1197 | 1199 | raise error.Abort( |
|
1198 | 1200 | _( |
|
1199 | 1201 | b'storage.sqlite.compression set to "zstd" but ' |
|
1200 | 1202 | b'zstandard compression not available to this ' |
|
1201 | 1203 | b'Mercurial install' |
|
1202 | 1204 | ) |
|
1203 | 1205 | ) |
|
1204 | 1206 | |
|
1205 | 1207 | if compression == b'zstd': |
|
1206 | 1208 | requirements.add(REQUIREMENT_ZSTD) |
|
1207 | 1209 | elif compression == b'zlib': |
|
1208 | 1210 | requirements.add(REQUIREMENT_ZLIB) |
|
1209 | 1211 | elif compression == b'none': |
|
1210 | 1212 | requirements.add(REQUIREMENT_NONE) |
|
1211 | 1213 | else: |
|
1212 | 1214 | raise error.Abort( |
|
1213 | 1215 | _( |
|
1214 | 1216 | b'unknown compression engine defined in ' |
|
1215 | 1217 | b'storage.sqlite.compression: %s' |
|
1216 | 1218 | ) |
|
1217 | 1219 | % compression |
|
1218 | 1220 | ) |
|
1219 | 1221 | |
|
1220 | 1222 | if createopts.get(b'shallowfilestore'): |
|
1221 | 1223 | requirements.add(REQUIREMENT_SHALLOW_FILES) |
|
1222 | 1224 | |
|
1223 | 1225 | return requirements |
|
1224 | 1226 | |
|
1225 | 1227 | |
|
1226 | 1228 | @interfaceutil.implementer(repository.ilocalrepositoryfilestorage) |
|
1227 | 1229 | class sqlitefilestorage(object): |
|
1228 | 1230 | """Repository file storage backed by SQLite.""" |
|
1229 | 1231 | |
|
1230 | 1232 | def file(self, path): |
|
1231 | 1233 | if path[0] == b'/': |
|
1232 | 1234 | path = path[1:] |
|
1233 | 1235 | |
|
1234 | 1236 | if REQUIREMENT_ZSTD in self.requirements: |
|
1235 | 1237 | compression = b'zstd' |
|
1236 | 1238 | elif REQUIREMENT_ZLIB in self.requirements: |
|
1237 | 1239 | compression = b'zlib' |
|
1238 | 1240 | elif REQUIREMENT_NONE in self.requirements: |
|
1239 | 1241 | compression = b'none' |
|
1240 | 1242 | else: |
|
1241 | 1243 | raise error.Abort( |
|
1242 | 1244 | _( |
|
1243 | 1245 | b'unable to determine what compression engine ' |
|
1244 | 1246 | b'to use for SQLite storage' |
|
1245 | 1247 | ) |
|
1246 | 1248 | ) |
|
1247 | 1249 | |
|
1248 | 1250 | return sqlitefilestore(self._dbconn, path, compression) |
|
1249 | 1251 | |
|
1250 | 1252 | |
|
1251 | 1253 | def makefilestorage(orig, requirements, features, **kwargs): |
|
1252 | 1254 | """Produce a type conforming to ``ilocalrepositoryfilestorage``.""" |
|
1253 | 1255 | if REQUIREMENT in requirements: |
|
1254 | 1256 | if REQUIREMENT_SHALLOW_FILES in requirements: |
|
1255 | 1257 | features.add(repository.REPO_FEATURE_SHALLOW_FILE_STORAGE) |
|
1256 | 1258 | |
|
1257 | 1259 | return sqlitefilestorage |
|
1258 | 1260 | else: |
|
1259 | 1261 | return orig(requirements=requirements, features=features, **kwargs) |
|
1260 | 1262 | |
|
1261 | 1263 | |
|
1262 | 1264 | def makemain(orig, ui, requirements, **kwargs): |
|
1263 | 1265 | if REQUIREMENT in requirements: |
|
1264 | 1266 | if REQUIREMENT_ZSTD in requirements and not zstd: |
|
1265 | 1267 | raise error.Abort( |
|
1266 | 1268 | _( |
|
1267 | 1269 | b'repository uses zstandard compression, which ' |
|
1268 | 1270 | b'is not available to this Mercurial install' |
|
1269 | 1271 | ) |
|
1270 | 1272 | ) |
|
1271 | 1273 | |
|
1272 | 1274 | return sqliterepository |
|
1273 | 1275 | |
|
1274 | 1276 | return orig(requirements=requirements, **kwargs) |
|
1275 | 1277 | |
|
1276 | 1278 | |
|
1277 | 1279 | def verifierinit(orig, self, *args, **kwargs): |
|
1278 | 1280 | orig(self, *args, **kwargs) |
|
1279 | 1281 | |
|
1280 | 1282 | # We don't care that files in the store don't align with what is |
|
1281 | 1283 | # advertised. So suppress these warnings. |
|
1282 | 1284 | self.warnorphanstorefiles = False |
|
1283 | 1285 | |
|
1284 | 1286 | |
|
1285 | 1287 | def extsetup(ui): |
|
1286 | 1288 | localrepo.featuresetupfuncs.add(featuresetup) |
|
1287 | 1289 | extensions.wrapfunction( |
|
1288 | 1290 | localrepo, b'newreporequirements', newreporequirements |
|
1289 | 1291 | ) |
|
1290 | 1292 | extensions.wrapfunction(localrepo, b'makefilestorage', makefilestorage) |
|
1291 | 1293 | extensions.wrapfunction(localrepo, b'makemain', makemain) |
|
1292 | 1294 | extensions.wrapfunction(verify.verifier, b'__init__', verifierinit) |
|
1293 | 1295 | |
|
1294 | 1296 | |
|
1295 | 1297 | def reposetup(ui, repo): |
|
1296 | 1298 | if isinstance(repo, sqliterepository): |
|
1297 | 1299 | repo._db = None |
|
1298 | 1300 | |
|
1299 | 1301 | # TODO check for bundlerepository? |
@@ -1,617 +1,618 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 | from .revlogutils import flagutil |
|
30 | 30 | |
|
31 | 31 | _defaultextra = {b'branch': b'default'} |
|
32 | 32 | |
|
33 | 33 | |
|
34 | 34 | def _string_escape(text): |
|
35 | 35 | """ |
|
36 | 36 | >>> from .pycompat import bytechr as chr |
|
37 | 37 | >>> d = {b'nl': chr(10), b'bs': chr(92), b'cr': chr(13), b'nul': chr(0)} |
|
38 | 38 | >>> s = b"ab%(nl)scd%(bs)s%(bs)sn%(nul)s12ab%(cr)scd%(bs)s%(nl)s" % d |
|
39 | 39 | >>> s |
|
40 | 40 | 'ab\\ncd\\\\\\\\n\\x0012ab\\rcd\\\\\\n' |
|
41 | 41 | >>> res = _string_escape(s) |
|
42 | 42 | >>> s == _string_unescape(res) |
|
43 | 43 | True |
|
44 | 44 | """ |
|
45 | 45 | # subset of the string_escape codec |
|
46 | 46 | text = ( |
|
47 | 47 | text.replace(b'\\', b'\\\\') |
|
48 | 48 | .replace(b'\n', b'\\n') |
|
49 | 49 | .replace(b'\r', b'\\r') |
|
50 | 50 | ) |
|
51 | 51 | return text.replace(b'\0', b'\\0') |
|
52 | 52 | |
|
53 | 53 | |
|
54 | 54 | def _string_unescape(text): |
|
55 | 55 | if b'\\0' in text: |
|
56 | 56 | # fix up \0 without getting into trouble with \\0 |
|
57 | 57 | text = text.replace(b'\\\\', b'\\\\\n') |
|
58 | 58 | text = text.replace(b'\\0', b'\0') |
|
59 | 59 | text = text.replace(b'\n', b'') |
|
60 | 60 | return stringutil.unescapestr(text) |
|
61 | 61 | |
|
62 | 62 | |
|
63 | 63 | def decodeextra(text): |
|
64 | 64 | """ |
|
65 | 65 | >>> from .pycompat import bytechr as chr |
|
66 | 66 | >>> sorted(decodeextra(encodeextra({b'foo': b'bar', b'baz': chr(0) + b'2'}) |
|
67 | 67 | ... ).items()) |
|
68 | 68 | [('baz', '\\x002'), ('branch', 'default'), ('foo', 'bar')] |
|
69 | 69 | >>> sorted(decodeextra(encodeextra({b'foo': b'bar', |
|
70 | 70 | ... b'baz': chr(92) + chr(0) + b'2'}) |
|
71 | 71 | ... ).items()) |
|
72 | 72 | [('baz', '\\\\\\x002'), ('branch', 'default'), ('foo', 'bar')] |
|
73 | 73 | """ |
|
74 | 74 | extra = _defaultextra.copy() |
|
75 | 75 | for l in text.split(b'\0'): |
|
76 | 76 | if l: |
|
77 | 77 | k, v = _string_unescape(l).split(b':', 1) |
|
78 | 78 | extra[k] = v |
|
79 | 79 | return extra |
|
80 | 80 | |
|
81 | 81 | |
|
82 | 82 | def encodeextra(d): |
|
83 | 83 | # keys must be sorted to produce a deterministic changelog entry |
|
84 | 84 | items = [_string_escape(b'%s:%s' % (k, d[k])) for k in sorted(d)] |
|
85 | 85 | return b"\0".join(items) |
|
86 | 86 | |
|
87 | 87 | |
|
88 | 88 | def stripdesc(desc): |
|
89 | 89 | """strip trailing whitespace and leading and trailing empty lines""" |
|
90 | 90 | return b'\n'.join([l.rstrip() for l in desc.splitlines()]).strip(b'\n') |
|
91 | 91 | |
|
92 | 92 | |
|
93 | 93 | class appender(object): |
|
94 | 94 | """the changelog index must be updated last on disk, so we use this class |
|
95 | 95 | to delay writes to it""" |
|
96 | 96 | |
|
97 | 97 | def __init__(self, vfs, name, mode, buf): |
|
98 | 98 | self.data = buf |
|
99 | 99 | fp = vfs(name, mode) |
|
100 | 100 | self.fp = fp |
|
101 | 101 | self.offset = fp.tell() |
|
102 | 102 | self.size = vfs.fstat(fp).st_size |
|
103 | 103 | self._end = self.size |
|
104 | 104 | |
|
105 | 105 | def end(self): |
|
106 | 106 | return self._end |
|
107 | 107 | |
|
108 | 108 | def tell(self): |
|
109 | 109 | return self.offset |
|
110 | 110 | |
|
111 | 111 | def flush(self): |
|
112 | 112 | pass |
|
113 | 113 | |
|
114 | 114 | @property |
|
115 | 115 | def closed(self): |
|
116 | 116 | return self.fp.closed |
|
117 | 117 | |
|
118 | 118 | def close(self): |
|
119 | 119 | self.fp.close() |
|
120 | 120 | |
|
121 | 121 | def seek(self, offset, whence=0): |
|
122 | 122 | '''virtual file offset spans real file and data''' |
|
123 | 123 | if whence == 0: |
|
124 | 124 | self.offset = offset |
|
125 | 125 | elif whence == 1: |
|
126 | 126 | self.offset += offset |
|
127 | 127 | elif whence == 2: |
|
128 | 128 | self.offset = self.end() + offset |
|
129 | 129 | if self.offset < self.size: |
|
130 | 130 | self.fp.seek(self.offset) |
|
131 | 131 | |
|
132 | 132 | def read(self, count=-1): |
|
133 | 133 | '''only trick here is reads that span real file and data''' |
|
134 | 134 | ret = b"" |
|
135 | 135 | if self.offset < self.size: |
|
136 | 136 | s = self.fp.read(count) |
|
137 | 137 | ret = s |
|
138 | 138 | self.offset += len(s) |
|
139 | 139 | if count > 0: |
|
140 | 140 | count -= len(s) |
|
141 | 141 | if count != 0: |
|
142 | 142 | doff = self.offset - self.size |
|
143 | 143 | self.data.insert(0, b"".join(self.data)) |
|
144 | 144 | del self.data[1:] |
|
145 | 145 | s = self.data[0][doff : doff + count] |
|
146 | 146 | self.offset += len(s) |
|
147 | 147 | ret += s |
|
148 | 148 | return ret |
|
149 | 149 | |
|
150 | 150 | def write(self, s): |
|
151 | 151 | self.data.append(bytes(s)) |
|
152 | 152 | self.offset += len(s) |
|
153 | 153 | self._end += len(s) |
|
154 | 154 | |
|
155 | 155 | def __enter__(self): |
|
156 | 156 | self.fp.__enter__() |
|
157 | 157 | return self |
|
158 | 158 | |
|
159 | 159 | def __exit__(self, *args): |
|
160 | 160 | return self.fp.__exit__(*args) |
|
161 | 161 | |
|
162 | 162 | |
|
163 | 163 | class _divertopener(object): |
|
164 | 164 | def __init__(self, opener, target): |
|
165 | 165 | self._opener = opener |
|
166 | 166 | self._target = target |
|
167 | 167 | |
|
168 | 168 | def __call__(self, name, mode=b'r', checkambig=False, **kwargs): |
|
169 | 169 | if name != self._target: |
|
170 | 170 | return self._opener(name, mode, **kwargs) |
|
171 | 171 | return self._opener(name + b".a", mode, **kwargs) |
|
172 | 172 | |
|
173 | 173 | def __getattr__(self, attr): |
|
174 | 174 | return getattr(self._opener, attr) |
|
175 | 175 | |
|
176 | 176 | |
|
177 | 177 | def _delayopener(opener, target, buf): |
|
178 | 178 | """build an opener that stores chunks in 'buf' instead of 'target'""" |
|
179 | 179 | |
|
180 | 180 | def _delay(name, mode=b'r', checkambig=False, **kwargs): |
|
181 | 181 | if name != target: |
|
182 | 182 | return opener(name, mode, **kwargs) |
|
183 | 183 | assert not kwargs |
|
184 | 184 | return appender(opener, name, mode, buf) |
|
185 | 185 | |
|
186 | 186 | return _delay |
|
187 | 187 | |
|
188 | 188 | |
|
189 | 189 | @attr.s |
|
190 | 190 | class _changelogrevision(object): |
|
191 | 191 | # Extensions might modify _defaultextra, so let the constructor below pass |
|
192 | 192 | # it in |
|
193 | 193 | extra = attr.ib() |
|
194 | 194 | manifest = attr.ib(default=nullid) |
|
195 | 195 | user = attr.ib(default=b'') |
|
196 | 196 | date = attr.ib(default=(0, 0)) |
|
197 | 197 | files = attr.ib(default=attr.Factory(list)) |
|
198 | 198 | filesadded = attr.ib(default=None) |
|
199 | 199 | filesremoved = attr.ib(default=None) |
|
200 | 200 | p1copies = attr.ib(default=None) |
|
201 | 201 | p2copies = attr.ib(default=None) |
|
202 | 202 | description = attr.ib(default=b'') |
|
203 | 203 | branchinfo = attr.ib(default=(_defaultextra[b'branch'], False)) |
|
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 | '_changes', |
|
220 | 220 | ) |
|
221 | 221 | |
|
222 | 222 | def __new__(cls, text, sidedata, cpsd): |
|
223 | 223 | if not text: |
|
224 | 224 | return _changelogrevision(extra=_defaultextra) |
|
225 | 225 | |
|
226 | 226 | self = super(changelogrevision, cls).__new__(cls) |
|
227 | 227 | # We could return here and implement the following as an __init__. |
|
228 | 228 | # But doing it here is equivalent and saves an extra function call. |
|
229 | 229 | |
|
230 | 230 | # format used: |
|
231 | 231 | # nodeid\n : manifest node in ascii |
|
232 | 232 | # user\n : user, no \n or \r allowed |
|
233 | 233 | # time tz extra\n : date (time is int or float, timezone is int) |
|
234 | 234 | # : extra is metadata, encoded and separated by '\0' |
|
235 | 235 | # : older versions ignore it |
|
236 | 236 | # files\n\n : files modified by the cset, no \n or \r allowed |
|
237 | 237 | # (.*) : comment (free text, ideally utf-8) |
|
238 | 238 | # |
|
239 | 239 | # changelog v0 doesn't use extra |
|
240 | 240 | |
|
241 | 241 | nl1 = text.index(b'\n') |
|
242 | 242 | nl2 = text.index(b'\n', nl1 + 1) |
|
243 | 243 | nl3 = text.index(b'\n', nl2 + 1) |
|
244 | 244 | |
|
245 | 245 | # The list of files may be empty. Which means nl3 is the first of the |
|
246 | 246 | # double newline that precedes the description. |
|
247 | 247 | if text[nl3 + 1 : nl3 + 2] == b'\n': |
|
248 | 248 | doublenl = nl3 |
|
249 | 249 | else: |
|
250 | 250 | doublenl = text.index(b'\n\n', nl3 + 1) |
|
251 | 251 | |
|
252 | 252 | self._offsets = (nl1, nl2, nl3, doublenl) |
|
253 | 253 | self._text = text |
|
254 | 254 | self._sidedata = sidedata |
|
255 | 255 | self._cpsd = cpsd |
|
256 | 256 | self._changes = None |
|
257 | 257 | |
|
258 | 258 | return self |
|
259 | 259 | |
|
260 | 260 | @property |
|
261 | 261 | def manifest(self): |
|
262 | 262 | return bin(self._text[0 : self._offsets[0]]) |
|
263 | 263 | |
|
264 | 264 | @property |
|
265 | 265 | def user(self): |
|
266 | 266 | off = self._offsets |
|
267 | 267 | return encoding.tolocal(self._text[off[0] + 1 : off[1]]) |
|
268 | 268 | |
|
269 | 269 | @property |
|
270 | 270 | def _rawdate(self): |
|
271 | 271 | off = self._offsets |
|
272 | 272 | dateextra = self._text[off[1] + 1 : off[2]] |
|
273 | 273 | return dateextra.split(b' ', 2)[0:2] |
|
274 | 274 | |
|
275 | 275 | @property |
|
276 | 276 | def _rawextra(self): |
|
277 | 277 | off = self._offsets |
|
278 | 278 | dateextra = self._text[off[1] + 1 : off[2]] |
|
279 | 279 | fields = dateextra.split(b' ', 2) |
|
280 | 280 | if len(fields) != 3: |
|
281 | 281 | return None |
|
282 | 282 | |
|
283 | 283 | return fields[2] |
|
284 | 284 | |
|
285 | 285 | @property |
|
286 | 286 | def date(self): |
|
287 | 287 | raw = self._rawdate |
|
288 | 288 | time = float(raw[0]) |
|
289 | 289 | # Various tools did silly things with the timezone. |
|
290 | 290 | try: |
|
291 | 291 | timezone = int(raw[1]) |
|
292 | 292 | except ValueError: |
|
293 | 293 | timezone = 0 |
|
294 | 294 | |
|
295 | 295 | return time, timezone |
|
296 | 296 | |
|
297 | 297 | @property |
|
298 | 298 | def extra(self): |
|
299 | 299 | raw = self._rawextra |
|
300 | 300 | if raw is None: |
|
301 | 301 | return _defaultextra |
|
302 | 302 | |
|
303 | 303 | return decodeextra(raw) |
|
304 | 304 | |
|
305 | 305 | @property |
|
306 | 306 | def changes(self): |
|
307 | 307 | if self._changes is not None: |
|
308 | 308 | return self._changes |
|
309 | 309 | if self._cpsd: |
|
310 | 310 | changes = metadata.decode_files_sidedata(self._sidedata) |
|
311 | 311 | else: |
|
312 | 312 | changes = metadata.ChangingFiles( |
|
313 | 313 | touched=self.files or (), |
|
314 | 314 | added=self.filesadded or (), |
|
315 | 315 | removed=self.filesremoved or (), |
|
316 | 316 | p1_copies=self.p1copies or {}, |
|
317 | 317 | p2_copies=self.p2copies or {}, |
|
318 | 318 | ) |
|
319 | 319 | self._changes = changes |
|
320 | 320 | return changes |
|
321 | 321 | |
|
322 | 322 | @property |
|
323 | 323 | def files(self): |
|
324 | 324 | if self._cpsd: |
|
325 | 325 | return sorted(self.changes.touched) |
|
326 | 326 | off = self._offsets |
|
327 | 327 | if off[2] == off[3]: |
|
328 | 328 | return [] |
|
329 | 329 | |
|
330 | 330 | return self._text[off[2] + 1 : off[3]].split(b'\n') |
|
331 | 331 | |
|
332 | 332 | @property |
|
333 | 333 | def filesadded(self): |
|
334 | 334 | if self._cpsd: |
|
335 | 335 | return self.changes.added |
|
336 | 336 | else: |
|
337 | 337 | rawindices = self.extra.get(b'filesadded') |
|
338 | 338 | if rawindices is None: |
|
339 | 339 | return None |
|
340 | 340 | return metadata.decodefileindices(self.files, rawindices) |
|
341 | 341 | |
|
342 | 342 | @property |
|
343 | 343 | def filesremoved(self): |
|
344 | 344 | if self._cpsd: |
|
345 | 345 | return self.changes.removed |
|
346 | 346 | else: |
|
347 | 347 | rawindices = self.extra.get(b'filesremoved') |
|
348 | 348 | if rawindices is None: |
|
349 | 349 | return None |
|
350 | 350 | return metadata.decodefileindices(self.files, rawindices) |
|
351 | 351 | |
|
352 | 352 | @property |
|
353 | 353 | def p1copies(self): |
|
354 | 354 | if self._cpsd: |
|
355 | 355 | return self.changes.copied_from_p1 |
|
356 | 356 | else: |
|
357 | 357 | rawcopies = self.extra.get(b'p1copies') |
|
358 | 358 | if rawcopies is None: |
|
359 | 359 | return None |
|
360 | 360 | return metadata.decodecopies(self.files, rawcopies) |
|
361 | 361 | |
|
362 | 362 | @property |
|
363 | 363 | def p2copies(self): |
|
364 | 364 | if self._cpsd: |
|
365 | 365 | return self.changes.copied_from_p2 |
|
366 | 366 | else: |
|
367 | 367 | rawcopies = self.extra.get(b'p2copies') |
|
368 | 368 | if rawcopies is None: |
|
369 | 369 | return None |
|
370 | 370 | return metadata.decodecopies(self.files, rawcopies) |
|
371 | 371 | |
|
372 | 372 | @property |
|
373 | 373 | def description(self): |
|
374 | 374 | return encoding.tolocal(self._text[self._offsets[3] + 2 :]) |
|
375 | 375 | |
|
376 | 376 | @property |
|
377 | 377 | def branchinfo(self): |
|
378 | 378 | extra = self.extra |
|
379 | 379 | return encoding.tolocal(extra.get(b"branch")), b'close' in extra |
|
380 | 380 | |
|
381 | 381 | |
|
382 | 382 | class changelog(revlog.revlog): |
|
383 | 383 | def __init__(self, opener, trypending=False): |
|
384 | 384 | """Load a changelog revlog using an opener. |
|
385 | 385 | |
|
386 | 386 | If ``trypending`` is true, we attempt to load the index from a |
|
387 | 387 | ``00changelog.i.a`` file instead of the default ``00changelog.i``. |
|
388 | 388 | The ``00changelog.i.a`` file contains index (and possibly inline |
|
389 | 389 | revision) data for a transaction that hasn't been finalized yet. |
|
390 | 390 | It exists in a separate file to facilitate readers (such as |
|
391 | 391 | hooks processes) accessing data before a transaction is finalized. |
|
392 | 392 | """ |
|
393 | 393 | if trypending and opener.exists(b'00changelog.i.a'): |
|
394 | 394 | indexfile = b'00changelog.i.a' |
|
395 | 395 | else: |
|
396 | 396 | indexfile = b'00changelog.i' |
|
397 | 397 | |
|
398 | 398 | datafile = b'00changelog.d' |
|
399 | 399 | revlog.revlog.__init__( |
|
400 | 400 | self, |
|
401 | 401 | opener, |
|
402 | 402 | indexfile, |
|
403 | 403 | datafile=datafile, |
|
404 | 404 | checkambig=True, |
|
405 | 405 | mmaplargeindex=True, |
|
406 | 406 | persistentnodemap=opener.options.get(b'persistent-nodemap', False), |
|
407 | 407 | ) |
|
408 | 408 | |
|
409 | 409 | if self._initempty and (self.version & 0xFFFF == revlog.REVLOGV1): |
|
410 | 410 | # changelogs don't benefit from generaldelta. |
|
411 | 411 | |
|
412 | 412 | self.version &= ~revlog.FLAG_GENERALDELTA |
|
413 | 413 | self._generaldelta = False |
|
414 | 414 | |
|
415 | 415 | # Delta chains for changelogs tend to be very small because entries |
|
416 | 416 | # tend to be small and don't delta well with each. So disable delta |
|
417 | 417 | # chains. |
|
418 | 418 | self._storedeltachains = False |
|
419 | 419 | |
|
420 | 420 | self._realopener = opener |
|
421 | 421 | self._delayed = False |
|
422 | 422 | self._delaybuf = None |
|
423 | 423 | self._divert = False |
|
424 | 424 | self._filteredrevs = frozenset() |
|
425 | 425 | self._filteredrevs_hashcache = {} |
|
426 | 426 | self._copiesstorage = opener.options.get(b'copies-storage') |
|
427 | 427 | |
|
428 | 428 | @property |
|
429 | 429 | def filteredrevs(self): |
|
430 | 430 | return self._filteredrevs |
|
431 | 431 | |
|
432 | 432 | @filteredrevs.setter |
|
433 | 433 | def filteredrevs(self, val): |
|
434 | 434 | # Ensure all updates go through this function |
|
435 | 435 | assert isinstance(val, frozenset) |
|
436 | 436 | self._filteredrevs = val |
|
437 | 437 | self._filteredrevs_hashcache = {} |
|
438 | 438 | |
|
439 | 439 | def delayupdate(self, tr): |
|
440 | 440 | """delay visibility of index updates to other readers""" |
|
441 | 441 | |
|
442 | 442 | if not self._delayed: |
|
443 | 443 | if len(self) == 0: |
|
444 | 444 | self._divert = True |
|
445 | 445 | if self._realopener.exists(self.indexfile + b'.a'): |
|
446 | 446 | self._realopener.unlink(self.indexfile + b'.a') |
|
447 | 447 | self.opener = _divertopener(self._realopener, self.indexfile) |
|
448 | 448 | else: |
|
449 | 449 | self._delaybuf = [] |
|
450 | 450 | self.opener = _delayopener( |
|
451 | 451 | self._realopener, self.indexfile, self._delaybuf |
|
452 | 452 | ) |
|
453 | 453 | self._delayed = True |
|
454 | 454 | tr.addpending(b'cl-%i' % id(self), self._writepending) |
|
455 | 455 | tr.addfinalize(b'cl-%i' % id(self), self._finalize) |
|
456 | 456 | |
|
457 | 457 | def _finalize(self, tr): |
|
458 | 458 | """finalize index updates""" |
|
459 | 459 | self._delayed = False |
|
460 | 460 | self.opener = self._realopener |
|
461 | 461 | # move redirected index data back into place |
|
462 | 462 | if self._divert: |
|
463 | 463 | assert not self._delaybuf |
|
464 | 464 | tmpname = self.indexfile + b".a" |
|
465 | 465 | nfile = self.opener.open(tmpname) |
|
466 | 466 | nfile.close() |
|
467 | 467 | self.opener.rename(tmpname, self.indexfile, checkambig=True) |
|
468 | 468 | elif self._delaybuf: |
|
469 | 469 | fp = self.opener(self.indexfile, b'a', checkambig=True) |
|
470 | 470 | fp.write(b"".join(self._delaybuf)) |
|
471 | 471 | fp.close() |
|
472 | 472 | self._delaybuf = None |
|
473 | 473 | self._divert = False |
|
474 | 474 | # split when we're done |
|
475 | 475 | self._enforceinlinesize(tr) |
|
476 | 476 | |
|
477 | 477 | def _writepending(self, tr): |
|
478 | 478 | """create a file containing the unfinalized state for |
|
479 | 479 | pretxnchangegroup""" |
|
480 | 480 | if self._delaybuf: |
|
481 | 481 | # make a temporary copy of the index |
|
482 | 482 | fp1 = self._realopener(self.indexfile) |
|
483 | 483 | pendingfilename = self.indexfile + b".a" |
|
484 | 484 | # register as a temp file to ensure cleanup on failure |
|
485 | 485 | tr.registertmp(pendingfilename) |
|
486 | 486 | # write existing data |
|
487 | 487 | fp2 = self._realopener(pendingfilename, b"w") |
|
488 | 488 | fp2.write(fp1.read()) |
|
489 | 489 | # add pending data |
|
490 | 490 | fp2.write(b"".join(self._delaybuf)) |
|
491 | 491 | fp2.close() |
|
492 | 492 | # switch modes so finalize can simply rename |
|
493 | 493 | self._delaybuf = None |
|
494 | 494 | self._divert = True |
|
495 | 495 | self.opener = _divertopener(self._realopener, self.indexfile) |
|
496 | 496 | |
|
497 | 497 | if self._divert: |
|
498 | 498 | return True |
|
499 | 499 | |
|
500 | 500 | return False |
|
501 | 501 | |
|
502 | 502 | def _enforceinlinesize(self, tr, fp=None): |
|
503 | 503 | if not self._delayed: |
|
504 | 504 | revlog.revlog._enforceinlinesize(self, tr, fp) |
|
505 | 505 | |
|
506 | 506 | def read(self, node): |
|
507 | 507 | """Obtain data from a parsed changelog revision. |
|
508 | 508 | |
|
509 | 509 | Returns a 6-tuple of: |
|
510 | 510 | |
|
511 | 511 | - manifest node in binary |
|
512 | 512 | - author/user as a localstr |
|
513 | 513 | - date as a 2-tuple of (time, timezone) |
|
514 | 514 | - list of files |
|
515 | 515 | - commit message as a localstr |
|
516 | 516 | - dict of extra metadata |
|
517 | 517 | |
|
518 | 518 | Unless you need to access all fields, consider calling |
|
519 | 519 | ``changelogrevision`` instead, as it is faster for partial object |
|
520 | 520 | access. |
|
521 | 521 | """ |
|
522 | 522 | d, s = self._revisiondata(node) |
|
523 | 523 | c = changelogrevision( |
|
524 | 524 | d, s, self._copiesstorage == b'changeset-sidedata' |
|
525 | 525 | ) |
|
526 | 526 | return (c.manifest, c.user, c.date, c.files, c.description, c.extra) |
|
527 | 527 | |
|
528 | 528 | def changelogrevision(self, nodeorrev): |
|
529 | 529 | """Obtain a ``changelogrevision`` for a node or revision.""" |
|
530 | 530 | text, sidedata = self._revisiondata(nodeorrev) |
|
531 | 531 | return changelogrevision( |
|
532 | 532 | text, sidedata, self._copiesstorage == b'changeset-sidedata' |
|
533 | 533 | ) |
|
534 | 534 | |
|
535 | 535 | def readfiles(self, node): |
|
536 | 536 | """ |
|
537 | 537 | short version of read that only returns the files modified by the cset |
|
538 | 538 | """ |
|
539 | 539 | text = self.revision(node) |
|
540 | 540 | if not text: |
|
541 | 541 | return [] |
|
542 | 542 | last = text.index(b"\n\n") |
|
543 | 543 | l = text[:last].split(b'\n') |
|
544 | 544 | return l[3:] |
|
545 | 545 | |
|
546 | 546 | def add( |
|
547 | 547 | self, |
|
548 | 548 | manifest, |
|
549 | 549 | files, |
|
550 | 550 | desc, |
|
551 | 551 | transaction, |
|
552 | 552 | p1, |
|
553 | 553 | p2, |
|
554 | 554 | user, |
|
555 | 555 | date=None, |
|
556 | 556 | extra=None, |
|
557 | 557 | ): |
|
558 | 558 | # Convert to UTF-8 encoded bytestrings as the very first |
|
559 | 559 | # thing: calling any method on a localstr object will turn it |
|
560 | 560 | # into a str object and the cached UTF-8 string is thus lost. |
|
561 | 561 | user, desc = encoding.fromlocal(user), encoding.fromlocal(desc) |
|
562 | 562 | |
|
563 | 563 | user = user.strip() |
|
564 | 564 | # An empty username or a username with a "\n" will make the |
|
565 | 565 | # revision text contain two "\n\n" sequences -> corrupt |
|
566 | 566 | # repository since read cannot unpack the revision. |
|
567 | 567 | if not user: |
|
568 | 568 | raise error.StorageError(_(b"empty username")) |
|
569 | 569 | if b"\n" in user: |
|
570 | 570 | raise error.StorageError( |
|
571 | 571 | _(b"username %r contains a newline") % pycompat.bytestr(user) |
|
572 | 572 | ) |
|
573 | 573 | |
|
574 | 574 | desc = stripdesc(desc) |
|
575 | 575 | |
|
576 | 576 | if date: |
|
577 | 577 | parseddate = b"%d %d" % dateutil.parsedate(date) |
|
578 | 578 | else: |
|
579 | 579 | parseddate = b"%d %d" % dateutil.makedate() |
|
580 | 580 | if extra: |
|
581 | 581 | branch = extra.get(b"branch") |
|
582 | 582 | if branch in (b"default", b""): |
|
583 | 583 | del extra[b"branch"] |
|
584 | 584 | elif branch in (b".", b"null", b"tip"): |
|
585 | 585 | raise error.StorageError( |
|
586 | 586 | _(b'the name \'%s\' is reserved') % branch |
|
587 | 587 | ) |
|
588 | 588 | sortedfiles = sorted(files.touched) |
|
589 | 589 | flags = 0 |
|
590 | 590 | sidedata = None |
|
591 | 591 | if self._copiesstorage == b'changeset-sidedata': |
|
592 | 592 | if files.has_copies_info: |
|
593 | 593 | flags |= flagutil.REVIDX_HASCOPIESINFO |
|
594 | 594 | sidedata = metadata.encode_files_sidedata(files) |
|
595 | 595 | |
|
596 | 596 | if extra: |
|
597 | 597 | extra = encodeextra(extra) |
|
598 | 598 | parseddate = b"%s %s" % (parseddate, extra) |
|
599 | 599 | l = [hex(manifest), user, parseddate] + sortedfiles + [b"", desc] |
|
600 | 600 | text = b"\n".join(l) |
|
601 |
re |
|
|
601 | rev = self.addrevision( | |
|
602 | 602 | text, transaction, len(self), p1, p2, sidedata=sidedata, flags=flags |
|
603 | 603 | ) |
|
604 | return self.node(rev) | |
|
604 | 605 | |
|
605 | 606 | def branchinfo(self, rev): |
|
606 | 607 | """return the branch name and open/close state of a revision |
|
607 | 608 | |
|
608 | 609 | This function exists because creating a changectx object |
|
609 | 610 | just to access this is costly.""" |
|
610 | 611 | return self.changelogrevision(rev).branchinfo |
|
611 | 612 | |
|
612 | 613 | def _nodeduplicatecallback(self, transaction, node): |
|
613 | 614 | # keep track of revisions that got "re-added", eg: unbunde of know rev. |
|
614 | 615 | # |
|
615 | 616 | # We track them in a list to preserve their order from the source bundle |
|
616 | 617 | duplicates = transaction.changes.setdefault(b'revduplicates', []) |
|
617 | 618 | duplicates.append(self.rev(node)) |
@@ -1,292 +1,293 b'' | |||
|
1 | 1 | # filelog.py - file history 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 | nullid, |
|
13 | 13 | nullrev, |
|
14 | 14 | ) |
|
15 | 15 | from . import ( |
|
16 | 16 | error, |
|
17 | 17 | revlog, |
|
18 | 18 | ) |
|
19 | 19 | from .interfaces import ( |
|
20 | 20 | repository, |
|
21 | 21 | util as interfaceutil, |
|
22 | 22 | ) |
|
23 | 23 | from .utils import storageutil |
|
24 | 24 | |
|
25 | 25 | |
|
26 | 26 | @interfaceutil.implementer(repository.ifilestorage) |
|
27 | 27 | class filelog(object): |
|
28 | 28 | def __init__(self, opener, path): |
|
29 | 29 | self._revlog = revlog.revlog( |
|
30 | 30 | opener, b'/'.join((b'data', path + b'.i')), censorable=True |
|
31 | 31 | ) |
|
32 | 32 | # Full name of the user visible file, relative to the repository root. |
|
33 | 33 | # Used by LFS. |
|
34 | 34 | self._revlog.filename = path |
|
35 | 35 | |
|
36 | 36 | def __len__(self): |
|
37 | 37 | return len(self._revlog) |
|
38 | 38 | |
|
39 | 39 | def __iter__(self): |
|
40 | 40 | return self._revlog.__iter__() |
|
41 | 41 | |
|
42 | 42 | def hasnode(self, node): |
|
43 | 43 | if node in (nullid, nullrev): |
|
44 | 44 | return False |
|
45 | 45 | |
|
46 | 46 | try: |
|
47 | 47 | self._revlog.rev(node) |
|
48 | 48 | return True |
|
49 | 49 | except (TypeError, ValueError, IndexError, error.LookupError): |
|
50 | 50 | return False |
|
51 | 51 | |
|
52 | 52 | def revs(self, start=0, stop=None): |
|
53 | 53 | return self._revlog.revs(start=start, stop=stop) |
|
54 | 54 | |
|
55 | 55 | def parents(self, node): |
|
56 | 56 | return self._revlog.parents(node) |
|
57 | 57 | |
|
58 | 58 | def parentrevs(self, rev): |
|
59 | 59 | return self._revlog.parentrevs(rev) |
|
60 | 60 | |
|
61 | 61 | def rev(self, node): |
|
62 | 62 | return self._revlog.rev(node) |
|
63 | 63 | |
|
64 | 64 | def node(self, rev): |
|
65 | 65 | return self._revlog.node(rev) |
|
66 | 66 | |
|
67 | 67 | def lookup(self, node): |
|
68 | 68 | return storageutil.fileidlookup( |
|
69 | 69 | self._revlog, node, self._revlog.indexfile |
|
70 | 70 | ) |
|
71 | 71 | |
|
72 | 72 | def linkrev(self, rev): |
|
73 | 73 | return self._revlog.linkrev(rev) |
|
74 | 74 | |
|
75 | 75 | def commonancestorsheads(self, node1, node2): |
|
76 | 76 | return self._revlog.commonancestorsheads(node1, node2) |
|
77 | 77 | |
|
78 | 78 | # Used by dagop.blockdescendants(). |
|
79 | 79 | def descendants(self, revs): |
|
80 | 80 | return self._revlog.descendants(revs) |
|
81 | 81 | |
|
82 | 82 | def heads(self, start=None, stop=None): |
|
83 | 83 | return self._revlog.heads(start, stop) |
|
84 | 84 | |
|
85 | 85 | # Used by hgweb, children extension. |
|
86 | 86 | def children(self, node): |
|
87 | 87 | return self._revlog.children(node) |
|
88 | 88 | |
|
89 | 89 | def iscensored(self, rev): |
|
90 | 90 | return self._revlog.iscensored(rev) |
|
91 | 91 | |
|
92 | 92 | def revision(self, node, _df=None, raw=False): |
|
93 | 93 | return self._revlog.revision(node, _df=_df, raw=raw) |
|
94 | 94 | |
|
95 | 95 | def rawdata(self, node, _df=None): |
|
96 | 96 | return self._revlog.rawdata(node, _df=_df) |
|
97 | 97 | |
|
98 | 98 | def emitrevisions( |
|
99 | 99 | self, |
|
100 | 100 | nodes, |
|
101 | 101 | nodesorder=None, |
|
102 | 102 | revisiondata=False, |
|
103 | 103 | assumehaveparentrevisions=False, |
|
104 | 104 | deltamode=repository.CG_DELTAMODE_STD, |
|
105 | 105 | ): |
|
106 | 106 | return self._revlog.emitrevisions( |
|
107 | 107 | nodes, |
|
108 | 108 | nodesorder=nodesorder, |
|
109 | 109 | revisiondata=revisiondata, |
|
110 | 110 | assumehaveparentrevisions=assumehaveparentrevisions, |
|
111 | 111 | deltamode=deltamode, |
|
112 | 112 | ) |
|
113 | 113 | |
|
114 | 114 | def addrevision( |
|
115 | 115 | self, |
|
116 | 116 | revisiondata, |
|
117 | 117 | transaction, |
|
118 | 118 | linkrev, |
|
119 | 119 | p1, |
|
120 | 120 | p2, |
|
121 | 121 | node=None, |
|
122 | 122 | flags=revlog.REVIDX_DEFAULT_FLAGS, |
|
123 | 123 | cachedelta=None, |
|
124 | 124 | ): |
|
125 | 125 | return self._revlog.addrevision( |
|
126 | 126 | revisiondata, |
|
127 | 127 | transaction, |
|
128 | 128 | linkrev, |
|
129 | 129 | p1, |
|
130 | 130 | p2, |
|
131 | 131 | node=node, |
|
132 | 132 | flags=flags, |
|
133 | 133 | cachedelta=cachedelta, |
|
134 | 134 | ) |
|
135 | 135 | |
|
136 | 136 | def addgroup( |
|
137 | 137 | self, |
|
138 | 138 | deltas, |
|
139 | 139 | linkmapper, |
|
140 | 140 | transaction, |
|
141 | 141 | addrevisioncb=None, |
|
142 | 142 | duplicaterevisioncb=None, |
|
143 | 143 | maybemissingparents=False, |
|
144 | 144 | ): |
|
145 | 145 | if maybemissingparents: |
|
146 | 146 | raise error.Abort( |
|
147 | 147 | _( |
|
148 | 148 | b'revlog storage does not support missing ' |
|
149 | 149 | b'parents write mode' |
|
150 | 150 | ) |
|
151 | 151 | ) |
|
152 | 152 | |
|
153 | 153 | return self._revlog.addgroup( |
|
154 | 154 | deltas, |
|
155 | 155 | linkmapper, |
|
156 | 156 | transaction, |
|
157 | 157 | addrevisioncb=addrevisioncb, |
|
158 | 158 | duplicaterevisioncb=duplicaterevisioncb, |
|
159 | 159 | ) |
|
160 | 160 | |
|
161 | 161 | def getstrippoint(self, minlink): |
|
162 | 162 | return self._revlog.getstrippoint(minlink) |
|
163 | 163 | |
|
164 | 164 | def strip(self, minlink, transaction): |
|
165 | 165 | return self._revlog.strip(minlink, transaction) |
|
166 | 166 | |
|
167 | 167 | def censorrevision(self, tr, node, tombstone=b''): |
|
168 | 168 | return self._revlog.censorrevision(tr, node, tombstone=tombstone) |
|
169 | 169 | |
|
170 | 170 | def files(self): |
|
171 | 171 | return self._revlog.files() |
|
172 | 172 | |
|
173 | 173 | def read(self, node): |
|
174 | 174 | return storageutil.filtermetadata(self.revision(node)) |
|
175 | 175 | |
|
176 | 176 | def add(self, text, meta, transaction, link, p1=None, p2=None): |
|
177 | 177 | if meta or text.startswith(b'\1\n'): |
|
178 | 178 | text = storageutil.packmeta(meta, text) |
|
179 |
re |
|
|
179 | rev = self.addrevision(text, transaction, link, p1, p2) | |
|
180 | return self.node(rev) | |
|
180 | 181 | |
|
181 | 182 | def renamed(self, node): |
|
182 | 183 | return storageutil.filerevisioncopied(self, node) |
|
183 | 184 | |
|
184 | 185 | def size(self, rev): |
|
185 | 186 | """return the size of a given revision""" |
|
186 | 187 | |
|
187 | 188 | # for revisions with renames, we have to go the slow way |
|
188 | 189 | node = self.node(rev) |
|
189 | 190 | if self.renamed(node): |
|
190 | 191 | return len(self.read(node)) |
|
191 | 192 | if self.iscensored(rev): |
|
192 | 193 | return 0 |
|
193 | 194 | |
|
194 | 195 | # XXX if self.read(node).startswith("\1\n"), this returns (size+4) |
|
195 | 196 | return self._revlog.size(rev) |
|
196 | 197 | |
|
197 | 198 | def cmp(self, node, text): |
|
198 | 199 | """compare text with a given file revision |
|
199 | 200 | |
|
200 | 201 | returns True if text is different than what is stored. |
|
201 | 202 | """ |
|
202 | 203 | return not storageutil.filedataequivalent(self, node, text) |
|
203 | 204 | |
|
204 | 205 | def verifyintegrity(self, state): |
|
205 | 206 | return self._revlog.verifyintegrity(state) |
|
206 | 207 | |
|
207 | 208 | def storageinfo( |
|
208 | 209 | self, |
|
209 | 210 | exclusivefiles=False, |
|
210 | 211 | sharedfiles=False, |
|
211 | 212 | revisionscount=False, |
|
212 | 213 | trackedsize=False, |
|
213 | 214 | storedsize=False, |
|
214 | 215 | ): |
|
215 | 216 | return self._revlog.storageinfo( |
|
216 | 217 | exclusivefiles=exclusivefiles, |
|
217 | 218 | sharedfiles=sharedfiles, |
|
218 | 219 | revisionscount=revisionscount, |
|
219 | 220 | trackedsize=trackedsize, |
|
220 | 221 | storedsize=storedsize, |
|
221 | 222 | ) |
|
222 | 223 | |
|
223 | 224 | # TODO these aren't part of the interface and aren't internal methods. |
|
224 | 225 | # Callers should be fixed to not use them. |
|
225 | 226 | |
|
226 | 227 | # Used by bundlefilelog, unionfilelog. |
|
227 | 228 | @property |
|
228 | 229 | def indexfile(self): |
|
229 | 230 | return self._revlog.indexfile |
|
230 | 231 | |
|
231 | 232 | @indexfile.setter |
|
232 | 233 | def indexfile(self, value): |
|
233 | 234 | self._revlog.indexfile = value |
|
234 | 235 | |
|
235 | 236 | # Used by repo upgrade. |
|
236 | 237 | def clone(self, tr, destrevlog, **kwargs): |
|
237 | 238 | if not isinstance(destrevlog, filelog): |
|
238 | 239 | raise error.ProgrammingError(b'expected filelog to clone()') |
|
239 | 240 | |
|
240 | 241 | return self._revlog.clone(tr, destrevlog._revlog, **kwargs) |
|
241 | 242 | |
|
242 | 243 | |
|
243 | 244 | class narrowfilelog(filelog): |
|
244 | 245 | """Filelog variation to be used with narrow stores.""" |
|
245 | 246 | |
|
246 | 247 | def __init__(self, opener, path, narrowmatch): |
|
247 | 248 | super(narrowfilelog, self).__init__(opener, path) |
|
248 | 249 | self._narrowmatch = narrowmatch |
|
249 | 250 | |
|
250 | 251 | def renamed(self, node): |
|
251 | 252 | res = super(narrowfilelog, self).renamed(node) |
|
252 | 253 | |
|
253 | 254 | # Renames that come from outside the narrowspec are problematic |
|
254 | 255 | # because we may lack the base text for the rename. This can result |
|
255 | 256 | # in code attempting to walk the ancestry or compute a diff |
|
256 | 257 | # encountering a missing revision. We address this by silently |
|
257 | 258 | # removing rename metadata if the source file is outside the |
|
258 | 259 | # narrow spec. |
|
259 | 260 | # |
|
260 | 261 | # A better solution would be to see if the base revision is available, |
|
261 | 262 | # rather than assuming it isn't. |
|
262 | 263 | # |
|
263 | 264 | # An even better solution would be to teach all consumers of rename |
|
264 | 265 | # metadata that the base revision may not be available. |
|
265 | 266 | # |
|
266 | 267 | # TODO consider better ways of doing this. |
|
267 | 268 | if res and not self._narrowmatch(res[0]): |
|
268 | 269 | return None |
|
269 | 270 | |
|
270 | 271 | return res |
|
271 | 272 | |
|
272 | 273 | def size(self, rev): |
|
273 | 274 | # Because we have a custom renamed() that may lie, we need to call |
|
274 | 275 | # the base renamed() to report accurate results. |
|
275 | 276 | node = self.node(rev) |
|
276 | 277 | if super(narrowfilelog, self).renamed(node): |
|
277 | 278 | return len(self.read(node)) |
|
278 | 279 | else: |
|
279 | 280 | return super(narrowfilelog, self).size(rev) |
|
280 | 281 | |
|
281 | 282 | def cmp(self, node, text): |
|
282 | 283 | different = super(narrowfilelog, self).cmp(node, text) |
|
283 | 284 | |
|
284 | 285 | # Because renamed() may lie, we may get false positives for |
|
285 | 286 | # different content. Check for this by comparing against the original |
|
286 | 287 | # renamed() implementation. |
|
287 | 288 | if different: |
|
288 | 289 | if super(narrowfilelog, self).renamed(node): |
|
289 | 290 | t2 = self.read(node) |
|
290 | 291 | return t2 != text |
|
291 | 292 | |
|
292 | 293 | return different |
@@ -1,1993 +1,1993 b'' | |||
|
1 | 1 | # repository.py - Interfaces and base classes for repositories and peers. |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2017 Gregory Szorc <gregory.szorc@gmail.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 .. import error |
|
12 | 12 | from . import util as interfaceutil |
|
13 | 13 | |
|
14 | 14 | # Local repository feature string. |
|
15 | 15 | |
|
16 | 16 | # Revlogs are being used for file storage. |
|
17 | 17 | REPO_FEATURE_REVLOG_FILE_STORAGE = b'revlogfilestorage' |
|
18 | 18 | # The storage part of the repository is shared from an external source. |
|
19 | 19 | REPO_FEATURE_SHARED_STORAGE = b'sharedstore' |
|
20 | 20 | # LFS supported for backing file storage. |
|
21 | 21 | REPO_FEATURE_LFS = b'lfs' |
|
22 | 22 | # Repository supports being stream cloned. |
|
23 | 23 | REPO_FEATURE_STREAM_CLONE = b'streamclone' |
|
24 | 24 | # Files storage may lack data for all ancestors. |
|
25 | 25 | REPO_FEATURE_SHALLOW_FILE_STORAGE = b'shallowfilestorage' |
|
26 | 26 | |
|
27 | 27 | REVISION_FLAG_CENSORED = 1 << 15 |
|
28 | 28 | REVISION_FLAG_ELLIPSIS = 1 << 14 |
|
29 | 29 | REVISION_FLAG_EXTSTORED = 1 << 13 |
|
30 | 30 | REVISION_FLAG_SIDEDATA = 1 << 12 |
|
31 | 31 | REVISION_FLAG_HASCOPIESINFO = 1 << 11 |
|
32 | 32 | |
|
33 | 33 | REVISION_FLAGS_KNOWN = ( |
|
34 | 34 | REVISION_FLAG_CENSORED |
|
35 | 35 | | REVISION_FLAG_ELLIPSIS |
|
36 | 36 | | REVISION_FLAG_EXTSTORED |
|
37 | 37 | | REVISION_FLAG_SIDEDATA |
|
38 | 38 | | REVISION_FLAG_HASCOPIESINFO |
|
39 | 39 | ) |
|
40 | 40 | |
|
41 | 41 | CG_DELTAMODE_STD = b'default' |
|
42 | 42 | CG_DELTAMODE_PREV = b'previous' |
|
43 | 43 | CG_DELTAMODE_FULL = b'fulltext' |
|
44 | 44 | CG_DELTAMODE_P1 = b'p1' |
|
45 | 45 | |
|
46 | 46 | |
|
47 | 47 | class ipeerconnection(interfaceutil.Interface): |
|
48 | 48 | """Represents a "connection" to a repository. |
|
49 | 49 | |
|
50 | 50 | This is the base interface for representing a connection to a repository. |
|
51 | 51 | It holds basic properties and methods applicable to all peer types. |
|
52 | 52 | |
|
53 | 53 | This is not a complete interface definition and should not be used |
|
54 | 54 | outside of this module. |
|
55 | 55 | """ |
|
56 | 56 | |
|
57 | 57 | ui = interfaceutil.Attribute("""ui.ui instance""") |
|
58 | 58 | |
|
59 | 59 | def url(): |
|
60 | 60 | """Returns a URL string representing this peer. |
|
61 | 61 | |
|
62 | 62 | Currently, implementations expose the raw URL used to construct the |
|
63 | 63 | instance. It may contain credentials as part of the URL. The |
|
64 | 64 | expectations of the value aren't well-defined and this could lead to |
|
65 | 65 | data leakage. |
|
66 | 66 | |
|
67 | 67 | TODO audit/clean consumers and more clearly define the contents of this |
|
68 | 68 | value. |
|
69 | 69 | """ |
|
70 | 70 | |
|
71 | 71 | def local(): |
|
72 | 72 | """Returns a local repository instance. |
|
73 | 73 | |
|
74 | 74 | If the peer represents a local repository, returns an object that |
|
75 | 75 | can be used to interface with it. Otherwise returns ``None``. |
|
76 | 76 | """ |
|
77 | 77 | |
|
78 | 78 | def peer(): |
|
79 | 79 | """Returns an object conforming to this interface. |
|
80 | 80 | |
|
81 | 81 | Most implementations will ``return self``. |
|
82 | 82 | """ |
|
83 | 83 | |
|
84 | 84 | def canpush(): |
|
85 | 85 | """Returns a boolean indicating if this peer can be pushed to.""" |
|
86 | 86 | |
|
87 | 87 | def close(): |
|
88 | 88 | """Close the connection to this peer. |
|
89 | 89 | |
|
90 | 90 | This is called when the peer will no longer be used. Resources |
|
91 | 91 | associated with the peer should be cleaned up. |
|
92 | 92 | """ |
|
93 | 93 | |
|
94 | 94 | |
|
95 | 95 | class ipeercapabilities(interfaceutil.Interface): |
|
96 | 96 | """Peer sub-interface related to capabilities.""" |
|
97 | 97 | |
|
98 | 98 | def capable(name): |
|
99 | 99 | """Determine support for a named capability. |
|
100 | 100 | |
|
101 | 101 | Returns ``False`` if capability not supported. |
|
102 | 102 | |
|
103 | 103 | Returns ``True`` if boolean capability is supported. Returns a string |
|
104 | 104 | if capability support is non-boolean. |
|
105 | 105 | |
|
106 | 106 | Capability strings may or may not map to wire protocol capabilities. |
|
107 | 107 | """ |
|
108 | 108 | |
|
109 | 109 | def requirecap(name, purpose): |
|
110 | 110 | """Require a capability to be present. |
|
111 | 111 | |
|
112 | 112 | Raises a ``CapabilityError`` if the capability isn't present. |
|
113 | 113 | """ |
|
114 | 114 | |
|
115 | 115 | |
|
116 | 116 | class ipeercommands(interfaceutil.Interface): |
|
117 | 117 | """Client-side interface for communicating over the wire protocol. |
|
118 | 118 | |
|
119 | 119 | This interface is used as a gateway to the Mercurial wire protocol. |
|
120 | 120 | methods commonly call wire protocol commands of the same name. |
|
121 | 121 | """ |
|
122 | 122 | |
|
123 | 123 | def branchmap(): |
|
124 | 124 | """Obtain heads in named branches. |
|
125 | 125 | |
|
126 | 126 | Returns a dict mapping branch name to an iterable of nodes that are |
|
127 | 127 | heads on that branch. |
|
128 | 128 | """ |
|
129 | 129 | |
|
130 | 130 | def capabilities(): |
|
131 | 131 | """Obtain capabilities of the peer. |
|
132 | 132 | |
|
133 | 133 | Returns a set of string capabilities. |
|
134 | 134 | """ |
|
135 | 135 | |
|
136 | 136 | def clonebundles(): |
|
137 | 137 | """Obtains the clone bundles manifest for the repo. |
|
138 | 138 | |
|
139 | 139 | Returns the manifest as unparsed bytes. |
|
140 | 140 | """ |
|
141 | 141 | |
|
142 | 142 | def debugwireargs(one, two, three=None, four=None, five=None): |
|
143 | 143 | """Used to facilitate debugging of arguments passed over the wire.""" |
|
144 | 144 | |
|
145 | 145 | def getbundle(source, **kwargs): |
|
146 | 146 | """Obtain remote repository data as a bundle. |
|
147 | 147 | |
|
148 | 148 | This command is how the bulk of repository data is transferred from |
|
149 | 149 | the peer to the local repository |
|
150 | 150 | |
|
151 | 151 | Returns a generator of bundle data. |
|
152 | 152 | """ |
|
153 | 153 | |
|
154 | 154 | def heads(): |
|
155 | 155 | """Determine all known head revisions in the peer. |
|
156 | 156 | |
|
157 | 157 | Returns an iterable of binary nodes. |
|
158 | 158 | """ |
|
159 | 159 | |
|
160 | 160 | def known(nodes): |
|
161 | 161 | """Determine whether multiple nodes are known. |
|
162 | 162 | |
|
163 | 163 | Accepts an iterable of nodes whose presence to check for. |
|
164 | 164 | |
|
165 | 165 | Returns an iterable of booleans indicating of the corresponding node |
|
166 | 166 | at that index is known to the peer. |
|
167 | 167 | """ |
|
168 | 168 | |
|
169 | 169 | def listkeys(namespace): |
|
170 | 170 | """Obtain all keys in a pushkey namespace. |
|
171 | 171 | |
|
172 | 172 | Returns an iterable of key names. |
|
173 | 173 | """ |
|
174 | 174 | |
|
175 | 175 | def lookup(key): |
|
176 | 176 | """Resolve a value to a known revision. |
|
177 | 177 | |
|
178 | 178 | Returns a binary node of the resolved revision on success. |
|
179 | 179 | """ |
|
180 | 180 | |
|
181 | 181 | def pushkey(namespace, key, old, new): |
|
182 | 182 | """Set a value using the ``pushkey`` protocol. |
|
183 | 183 | |
|
184 | 184 | Arguments correspond to the pushkey namespace and key to operate on and |
|
185 | 185 | the old and new values for that key. |
|
186 | 186 | |
|
187 | 187 | Returns a string with the peer result. The value inside varies by the |
|
188 | 188 | namespace. |
|
189 | 189 | """ |
|
190 | 190 | |
|
191 | 191 | def stream_out(): |
|
192 | 192 | """Obtain streaming clone data. |
|
193 | 193 | |
|
194 | 194 | Successful result should be a generator of data chunks. |
|
195 | 195 | """ |
|
196 | 196 | |
|
197 | 197 | def unbundle(bundle, heads, url): |
|
198 | 198 | """Transfer repository data to the peer. |
|
199 | 199 | |
|
200 | 200 | This is how the bulk of data during a push is transferred. |
|
201 | 201 | |
|
202 | 202 | Returns the integer number of heads added to the peer. |
|
203 | 203 | """ |
|
204 | 204 | |
|
205 | 205 | |
|
206 | 206 | class ipeerlegacycommands(interfaceutil.Interface): |
|
207 | 207 | """Interface for implementing support for legacy wire protocol commands. |
|
208 | 208 | |
|
209 | 209 | Wire protocol commands transition to legacy status when they are no longer |
|
210 | 210 | used by modern clients. To facilitate identifying which commands are |
|
211 | 211 | legacy, the interfaces are split. |
|
212 | 212 | """ |
|
213 | 213 | |
|
214 | 214 | def between(pairs): |
|
215 | 215 | """Obtain nodes between pairs of nodes. |
|
216 | 216 | |
|
217 | 217 | ``pairs`` is an iterable of node pairs. |
|
218 | 218 | |
|
219 | 219 | Returns an iterable of iterables of nodes corresponding to each |
|
220 | 220 | requested pair. |
|
221 | 221 | """ |
|
222 | 222 | |
|
223 | 223 | def branches(nodes): |
|
224 | 224 | """Obtain ancestor changesets of specific nodes back to a branch point. |
|
225 | 225 | |
|
226 | 226 | For each requested node, the peer finds the first ancestor node that is |
|
227 | 227 | a DAG root or is a merge. |
|
228 | 228 | |
|
229 | 229 | Returns an iterable of iterables with the resolved values for each node. |
|
230 | 230 | """ |
|
231 | 231 | |
|
232 | 232 | def changegroup(nodes, source): |
|
233 | 233 | """Obtain a changegroup with data for descendants of specified nodes.""" |
|
234 | 234 | |
|
235 | 235 | def changegroupsubset(bases, heads, source): |
|
236 | 236 | pass |
|
237 | 237 | |
|
238 | 238 | |
|
239 | 239 | class ipeercommandexecutor(interfaceutil.Interface): |
|
240 | 240 | """Represents a mechanism to execute remote commands. |
|
241 | 241 | |
|
242 | 242 | This is the primary interface for requesting that wire protocol commands |
|
243 | 243 | be executed. Instances of this interface are active in a context manager |
|
244 | 244 | and have a well-defined lifetime. When the context manager exits, all |
|
245 | 245 | outstanding requests are waited on. |
|
246 | 246 | """ |
|
247 | 247 | |
|
248 | 248 | def callcommand(name, args): |
|
249 | 249 | """Request that a named command be executed. |
|
250 | 250 | |
|
251 | 251 | Receives the command name and a dictionary of command arguments. |
|
252 | 252 | |
|
253 | 253 | Returns a ``concurrent.futures.Future`` that will resolve to the |
|
254 | 254 | result of that command request. That exact value is left up to |
|
255 | 255 | the implementation and possibly varies by command. |
|
256 | 256 | |
|
257 | 257 | Not all commands can coexist with other commands in an executor |
|
258 | 258 | instance: it depends on the underlying wire protocol transport being |
|
259 | 259 | used and the command itself. |
|
260 | 260 | |
|
261 | 261 | Implementations MAY call ``sendcommands()`` automatically if the |
|
262 | 262 | requested command can not coexist with other commands in this executor. |
|
263 | 263 | |
|
264 | 264 | Implementations MAY call ``sendcommands()`` automatically when the |
|
265 | 265 | future's ``result()`` is called. So, consumers using multiple |
|
266 | 266 | commands with an executor MUST ensure that ``result()`` is not called |
|
267 | 267 | until all command requests have been issued. |
|
268 | 268 | """ |
|
269 | 269 | |
|
270 | 270 | def sendcommands(): |
|
271 | 271 | """Trigger submission of queued command requests. |
|
272 | 272 | |
|
273 | 273 | Not all transports submit commands as soon as they are requested to |
|
274 | 274 | run. When called, this method forces queued command requests to be |
|
275 | 275 | issued. It will no-op if all commands have already been sent. |
|
276 | 276 | |
|
277 | 277 | When called, no more new commands may be issued with this executor. |
|
278 | 278 | """ |
|
279 | 279 | |
|
280 | 280 | def close(): |
|
281 | 281 | """Signal that this command request is finished. |
|
282 | 282 | |
|
283 | 283 | When called, no more new commands may be issued. All outstanding |
|
284 | 284 | commands that have previously been issued are waited on before |
|
285 | 285 | returning. This not only includes waiting for the futures to resolve, |
|
286 | 286 | but also waiting for all response data to arrive. In other words, |
|
287 | 287 | calling this waits for all on-wire state for issued command requests |
|
288 | 288 | to finish. |
|
289 | 289 | |
|
290 | 290 | When used as a context manager, this method is called when exiting the |
|
291 | 291 | context manager. |
|
292 | 292 | |
|
293 | 293 | This method may call ``sendcommands()`` if there are buffered commands. |
|
294 | 294 | """ |
|
295 | 295 | |
|
296 | 296 | |
|
297 | 297 | class ipeerrequests(interfaceutil.Interface): |
|
298 | 298 | """Interface for executing commands on a peer.""" |
|
299 | 299 | |
|
300 | 300 | limitedarguments = interfaceutil.Attribute( |
|
301 | 301 | """True if the peer cannot receive large argument value for commands.""" |
|
302 | 302 | ) |
|
303 | 303 | |
|
304 | 304 | def commandexecutor(): |
|
305 | 305 | """A context manager that resolves to an ipeercommandexecutor. |
|
306 | 306 | |
|
307 | 307 | The object this resolves to can be used to issue command requests |
|
308 | 308 | to the peer. |
|
309 | 309 | |
|
310 | 310 | Callers should call its ``callcommand`` method to issue command |
|
311 | 311 | requests. |
|
312 | 312 | |
|
313 | 313 | A new executor should be obtained for each distinct set of commands |
|
314 | 314 | (possibly just a single command) that the consumer wants to execute |
|
315 | 315 | as part of a single operation or round trip. This is because some |
|
316 | 316 | peers are half-duplex and/or don't support persistent connections. |
|
317 | 317 | e.g. in the case of HTTP peers, commands sent to an executor represent |
|
318 | 318 | a single HTTP request. While some peers may support multiple command |
|
319 | 319 | sends over the wire per executor, consumers need to code to the least |
|
320 | 320 | capable peer. So it should be assumed that command executors buffer |
|
321 | 321 | called commands until they are told to send them and that each |
|
322 | 322 | command executor could result in a new connection or wire-level request |
|
323 | 323 | being issued. |
|
324 | 324 | """ |
|
325 | 325 | |
|
326 | 326 | |
|
327 | 327 | class ipeerbase(ipeerconnection, ipeercapabilities, ipeerrequests): |
|
328 | 328 | """Unified interface for peer repositories. |
|
329 | 329 | |
|
330 | 330 | All peer instances must conform to this interface. |
|
331 | 331 | """ |
|
332 | 332 | |
|
333 | 333 | |
|
334 | 334 | class ipeerv2(ipeerconnection, ipeercapabilities, ipeerrequests): |
|
335 | 335 | """Unified peer interface for wire protocol version 2 peers.""" |
|
336 | 336 | |
|
337 | 337 | apidescriptor = interfaceutil.Attribute( |
|
338 | 338 | """Data structure holding description of server API.""" |
|
339 | 339 | ) |
|
340 | 340 | |
|
341 | 341 | |
|
342 | 342 | @interfaceutil.implementer(ipeerbase) |
|
343 | 343 | class peer(object): |
|
344 | 344 | """Base class for peer repositories.""" |
|
345 | 345 | |
|
346 | 346 | limitedarguments = False |
|
347 | 347 | |
|
348 | 348 | def capable(self, name): |
|
349 | 349 | caps = self.capabilities() |
|
350 | 350 | if name in caps: |
|
351 | 351 | return True |
|
352 | 352 | |
|
353 | 353 | name = b'%s=' % name |
|
354 | 354 | for cap in caps: |
|
355 | 355 | if cap.startswith(name): |
|
356 | 356 | return cap[len(name) :] |
|
357 | 357 | |
|
358 | 358 | return False |
|
359 | 359 | |
|
360 | 360 | def requirecap(self, name, purpose): |
|
361 | 361 | if self.capable(name): |
|
362 | 362 | return |
|
363 | 363 | |
|
364 | 364 | raise error.CapabilityError( |
|
365 | 365 | _( |
|
366 | 366 | b'cannot %s; remote repository does not support the ' |
|
367 | 367 | b'\'%s\' capability' |
|
368 | 368 | ) |
|
369 | 369 | % (purpose, name) |
|
370 | 370 | ) |
|
371 | 371 | |
|
372 | 372 | |
|
373 | 373 | class iverifyproblem(interfaceutil.Interface): |
|
374 | 374 | """Represents a problem with the integrity of the repository. |
|
375 | 375 | |
|
376 | 376 | Instances of this interface are emitted to describe an integrity issue |
|
377 | 377 | with a repository (e.g. corrupt storage, missing data, etc). |
|
378 | 378 | |
|
379 | 379 | Instances are essentially messages associated with severity. |
|
380 | 380 | """ |
|
381 | 381 | |
|
382 | 382 | warning = interfaceutil.Attribute( |
|
383 | 383 | """Message indicating a non-fatal problem.""" |
|
384 | 384 | ) |
|
385 | 385 | |
|
386 | 386 | error = interfaceutil.Attribute("""Message indicating a fatal problem.""") |
|
387 | 387 | |
|
388 | 388 | node = interfaceutil.Attribute( |
|
389 | 389 | """Revision encountering the problem. |
|
390 | 390 | |
|
391 | 391 | ``None`` means the problem doesn't apply to a single revision. |
|
392 | 392 | """ |
|
393 | 393 | ) |
|
394 | 394 | |
|
395 | 395 | |
|
396 | 396 | class irevisiondelta(interfaceutil.Interface): |
|
397 | 397 | """Represents a delta between one revision and another. |
|
398 | 398 | |
|
399 | 399 | Instances convey enough information to allow a revision to be exchanged |
|
400 | 400 | with another repository. |
|
401 | 401 | |
|
402 | 402 | Instances represent the fulltext revision data or a delta against |
|
403 | 403 | another revision. Therefore the ``revision`` and ``delta`` attributes |
|
404 | 404 | are mutually exclusive. |
|
405 | 405 | |
|
406 | 406 | Typically used for changegroup generation. |
|
407 | 407 | """ |
|
408 | 408 | |
|
409 | 409 | node = interfaceutil.Attribute("""20 byte node of this revision.""") |
|
410 | 410 | |
|
411 | 411 | p1node = interfaceutil.Attribute( |
|
412 | 412 | """20 byte node of 1st parent of this revision.""" |
|
413 | 413 | ) |
|
414 | 414 | |
|
415 | 415 | p2node = interfaceutil.Attribute( |
|
416 | 416 | """20 byte node of 2nd parent of this revision.""" |
|
417 | 417 | ) |
|
418 | 418 | |
|
419 | 419 | linknode = interfaceutil.Attribute( |
|
420 | 420 | """20 byte node of the changelog revision this node is linked to.""" |
|
421 | 421 | ) |
|
422 | 422 | |
|
423 | 423 | flags = interfaceutil.Attribute( |
|
424 | 424 | """2 bytes of integer flags that apply to this revision. |
|
425 | 425 | |
|
426 | 426 | This is a bitwise composition of the ``REVISION_FLAG_*`` constants. |
|
427 | 427 | """ |
|
428 | 428 | ) |
|
429 | 429 | |
|
430 | 430 | basenode = interfaceutil.Attribute( |
|
431 | 431 | """20 byte node of the revision this data is a delta against. |
|
432 | 432 | |
|
433 | 433 | ``nullid`` indicates that the revision is a full revision and not |
|
434 | 434 | a delta. |
|
435 | 435 | """ |
|
436 | 436 | ) |
|
437 | 437 | |
|
438 | 438 | baserevisionsize = interfaceutil.Attribute( |
|
439 | 439 | """Size of base revision this delta is against. |
|
440 | 440 | |
|
441 | 441 | May be ``None`` if ``basenode`` is ``nullid``. |
|
442 | 442 | """ |
|
443 | 443 | ) |
|
444 | 444 | |
|
445 | 445 | revision = interfaceutil.Attribute( |
|
446 | 446 | """Raw fulltext of revision data for this node.""" |
|
447 | 447 | ) |
|
448 | 448 | |
|
449 | 449 | delta = interfaceutil.Attribute( |
|
450 | 450 | """Delta between ``basenode`` and ``node``. |
|
451 | 451 | |
|
452 | 452 | Stored in the bdiff delta format. |
|
453 | 453 | """ |
|
454 | 454 | ) |
|
455 | 455 | |
|
456 | 456 | |
|
457 | 457 | class ifilerevisionssequence(interfaceutil.Interface): |
|
458 | 458 | """Contains index data for all revisions of a file. |
|
459 | 459 | |
|
460 | 460 | Types implementing this behave like lists of tuples. The index |
|
461 | 461 | in the list corresponds to the revision number. The values contain |
|
462 | 462 | index metadata. |
|
463 | 463 | |
|
464 | 464 | The *null* revision (revision number -1) is always the last item |
|
465 | 465 | in the index. |
|
466 | 466 | """ |
|
467 | 467 | |
|
468 | 468 | def __len__(): |
|
469 | 469 | """The total number of revisions.""" |
|
470 | 470 | |
|
471 | 471 | def __getitem__(rev): |
|
472 | 472 | """Returns the object having a specific revision number. |
|
473 | 473 | |
|
474 | 474 | Returns an 8-tuple with the following fields: |
|
475 | 475 | |
|
476 | 476 | offset+flags |
|
477 | 477 | Contains the offset and flags for the revision. 64-bit unsigned |
|
478 | 478 | integer where first 6 bytes are the offset and the next 2 bytes |
|
479 | 479 | are flags. The offset can be 0 if it is not used by the store. |
|
480 | 480 | compressed size |
|
481 | 481 | Size of the revision data in the store. It can be 0 if it isn't |
|
482 | 482 | needed by the store. |
|
483 | 483 | uncompressed size |
|
484 | 484 | Fulltext size. It can be 0 if it isn't needed by the store. |
|
485 | 485 | base revision |
|
486 | 486 | Revision number of revision the delta for storage is encoded |
|
487 | 487 | against. -1 indicates not encoded against a base revision. |
|
488 | 488 | link revision |
|
489 | 489 | Revision number of changelog revision this entry is related to. |
|
490 | 490 | p1 revision |
|
491 | 491 | Revision number of 1st parent. -1 if no 1st parent. |
|
492 | 492 | p2 revision |
|
493 | 493 | Revision number of 2nd parent. -1 if no 1st parent. |
|
494 | 494 | node |
|
495 | 495 | Binary node value for this revision number. |
|
496 | 496 | |
|
497 | 497 | Negative values should index off the end of the sequence. ``-1`` |
|
498 | 498 | should return the null revision. ``-2`` should return the most |
|
499 | 499 | recent revision. |
|
500 | 500 | """ |
|
501 | 501 | |
|
502 | 502 | def __contains__(rev): |
|
503 | 503 | """Whether a revision number exists.""" |
|
504 | 504 | |
|
505 | 505 | def insert(self, i, entry): |
|
506 | 506 | """Add an item to the index at specific revision.""" |
|
507 | 507 | |
|
508 | 508 | |
|
509 | 509 | class ifileindex(interfaceutil.Interface): |
|
510 | 510 | """Storage interface for index data of a single file. |
|
511 | 511 | |
|
512 | 512 | File storage data is divided into index metadata and data storage. |
|
513 | 513 | This interface defines the index portion of the interface. |
|
514 | 514 | |
|
515 | 515 | The index logically consists of: |
|
516 | 516 | |
|
517 | 517 | * A mapping between revision numbers and nodes. |
|
518 | 518 | * DAG data (storing and querying the relationship between nodes). |
|
519 | 519 | * Metadata to facilitate storage. |
|
520 | 520 | """ |
|
521 | 521 | |
|
522 | 522 | def __len__(): |
|
523 | 523 | """Obtain the number of revisions stored for this file.""" |
|
524 | 524 | |
|
525 | 525 | def __iter__(): |
|
526 | 526 | """Iterate over revision numbers for this file.""" |
|
527 | 527 | |
|
528 | 528 | def hasnode(node): |
|
529 | 529 | """Returns a bool indicating if a node is known to this store. |
|
530 | 530 | |
|
531 | 531 | Implementations must only return True for full, binary node values: |
|
532 | 532 | hex nodes, revision numbers, and partial node matches must be |
|
533 | 533 | rejected. |
|
534 | 534 | |
|
535 | 535 | The null node is never present. |
|
536 | 536 | """ |
|
537 | 537 | |
|
538 | 538 | def revs(start=0, stop=None): |
|
539 | 539 | """Iterate over revision numbers for this file, with control.""" |
|
540 | 540 | |
|
541 | 541 | def parents(node): |
|
542 | 542 | """Returns a 2-tuple of parent nodes for a revision. |
|
543 | 543 | |
|
544 | 544 | Values will be ``nullid`` if the parent is empty. |
|
545 | 545 | """ |
|
546 | 546 | |
|
547 | 547 | def parentrevs(rev): |
|
548 | 548 | """Like parents() but operates on revision numbers.""" |
|
549 | 549 | |
|
550 | 550 | def rev(node): |
|
551 | 551 | """Obtain the revision number given a node. |
|
552 | 552 | |
|
553 | 553 | Raises ``error.LookupError`` if the node is not known. |
|
554 | 554 | """ |
|
555 | 555 | |
|
556 | 556 | def node(rev): |
|
557 | 557 | """Obtain the node value given a revision number. |
|
558 | 558 | |
|
559 | 559 | Raises ``IndexError`` if the node is not known. |
|
560 | 560 | """ |
|
561 | 561 | |
|
562 | 562 | def lookup(node): |
|
563 | 563 | """Attempt to resolve a value to a node. |
|
564 | 564 | |
|
565 | 565 | Value can be a binary node, hex node, revision number, or a string |
|
566 | 566 | that can be converted to an integer. |
|
567 | 567 | |
|
568 | 568 | Raises ``error.LookupError`` if a node could not be resolved. |
|
569 | 569 | """ |
|
570 | 570 | |
|
571 | 571 | def linkrev(rev): |
|
572 | 572 | """Obtain the changeset revision number a revision is linked to.""" |
|
573 | 573 | |
|
574 | 574 | def iscensored(rev): |
|
575 | 575 | """Return whether a revision's content has been censored.""" |
|
576 | 576 | |
|
577 | 577 | def commonancestorsheads(node1, node2): |
|
578 | 578 | """Obtain an iterable of nodes containing heads of common ancestors. |
|
579 | 579 | |
|
580 | 580 | See ``ancestor.commonancestorsheads()``. |
|
581 | 581 | """ |
|
582 | 582 | |
|
583 | 583 | def descendants(revs): |
|
584 | 584 | """Obtain descendant revision numbers for a set of revision numbers. |
|
585 | 585 | |
|
586 | 586 | If ``nullrev`` is in the set, this is equivalent to ``revs()``. |
|
587 | 587 | """ |
|
588 | 588 | |
|
589 | 589 | def heads(start=None, stop=None): |
|
590 | 590 | """Obtain a list of nodes that are DAG heads, with control. |
|
591 | 591 | |
|
592 | 592 | The set of revisions examined can be limited by specifying |
|
593 | 593 | ``start`` and ``stop``. ``start`` is a node. ``stop`` is an |
|
594 | 594 | iterable of nodes. DAG traversal starts at earlier revision |
|
595 | 595 | ``start`` and iterates forward until any node in ``stop`` is |
|
596 | 596 | encountered. |
|
597 | 597 | """ |
|
598 | 598 | |
|
599 | 599 | def children(node): |
|
600 | 600 | """Obtain nodes that are children of a node. |
|
601 | 601 | |
|
602 | 602 | Returns a list of nodes. |
|
603 | 603 | """ |
|
604 | 604 | |
|
605 | 605 | |
|
606 | 606 | class ifiledata(interfaceutil.Interface): |
|
607 | 607 | """Storage interface for data storage of a specific file. |
|
608 | 608 | |
|
609 | 609 | This complements ``ifileindex`` and provides an interface for accessing |
|
610 | 610 | data for a tracked file. |
|
611 | 611 | """ |
|
612 | 612 | |
|
613 | 613 | def size(rev): |
|
614 | 614 | """Obtain the fulltext size of file data. |
|
615 | 615 | |
|
616 | 616 | Any metadata is excluded from size measurements. |
|
617 | 617 | """ |
|
618 | 618 | |
|
619 | 619 | def revision(node, raw=False): |
|
620 | 620 | """Obtain fulltext data for a node. |
|
621 | 621 | |
|
622 | 622 | By default, any storage transformations are applied before the data |
|
623 | 623 | is returned. If ``raw`` is True, non-raw storage transformations |
|
624 | 624 | are not applied. |
|
625 | 625 | |
|
626 | 626 | The fulltext data may contain a header containing metadata. Most |
|
627 | 627 | consumers should use ``read()`` to obtain the actual file data. |
|
628 | 628 | """ |
|
629 | 629 | |
|
630 | 630 | def rawdata(node): |
|
631 | 631 | """Obtain raw data for a node.""" |
|
632 | 632 | |
|
633 | 633 | def read(node): |
|
634 | 634 | """Resolve file fulltext data. |
|
635 | 635 | |
|
636 | 636 | This is similar to ``revision()`` except any metadata in the data |
|
637 | 637 | headers is stripped. |
|
638 | 638 | """ |
|
639 | 639 | |
|
640 | 640 | def renamed(node): |
|
641 | 641 | """Obtain copy metadata for a node. |
|
642 | 642 | |
|
643 | 643 | Returns ``False`` if no copy metadata is stored or a 2-tuple of |
|
644 | 644 | (path, node) from which this revision was copied. |
|
645 | 645 | """ |
|
646 | 646 | |
|
647 | 647 | def cmp(node, fulltext): |
|
648 | 648 | """Compare fulltext to another revision. |
|
649 | 649 | |
|
650 | 650 | Returns True if the fulltext is different from what is stored. |
|
651 | 651 | |
|
652 | 652 | This takes copy metadata into account. |
|
653 | 653 | |
|
654 | 654 | TODO better document the copy metadata and censoring logic. |
|
655 | 655 | """ |
|
656 | 656 | |
|
657 | 657 | def emitrevisions( |
|
658 | 658 | nodes, |
|
659 | 659 | nodesorder=None, |
|
660 | 660 | revisiondata=False, |
|
661 | 661 | assumehaveparentrevisions=False, |
|
662 | 662 | deltamode=CG_DELTAMODE_STD, |
|
663 | 663 | ): |
|
664 | 664 | """Produce ``irevisiondelta`` for revisions. |
|
665 | 665 | |
|
666 | 666 | Given an iterable of nodes, emits objects conforming to the |
|
667 | 667 | ``irevisiondelta`` interface that describe revisions in storage. |
|
668 | 668 | |
|
669 | 669 | This method is a generator. |
|
670 | 670 | |
|
671 | 671 | The input nodes may be unordered. Implementations must ensure that a |
|
672 | 672 | node's parents are emitted before the node itself. Transitively, this |
|
673 | 673 | means that a node may only be emitted once all its ancestors in |
|
674 | 674 | ``nodes`` have also been emitted. |
|
675 | 675 | |
|
676 | 676 | By default, emits "index" data (the ``node``, ``p1node``, and |
|
677 | 677 | ``p2node`` attributes). If ``revisiondata`` is set, revision data |
|
678 | 678 | will also be present on the emitted objects. |
|
679 | 679 | |
|
680 | 680 | With default argument values, implementations can choose to emit |
|
681 | 681 | either fulltext revision data or a delta. When emitting deltas, |
|
682 | 682 | implementations must consider whether the delta's base revision |
|
683 | 683 | fulltext is available to the receiver. |
|
684 | 684 | |
|
685 | 685 | The base revision fulltext is guaranteed to be available if any of |
|
686 | 686 | the following are met: |
|
687 | 687 | |
|
688 | 688 | * Its fulltext revision was emitted by this method call. |
|
689 | 689 | * A delta for that revision was emitted by this method call. |
|
690 | 690 | * ``assumehaveparentrevisions`` is True and the base revision is a |
|
691 | 691 | parent of the node. |
|
692 | 692 | |
|
693 | 693 | ``nodesorder`` can be used to control the order that revisions are |
|
694 | 694 | emitted. By default, revisions can be reordered as long as they are |
|
695 | 695 | in DAG topological order (see above). If the value is ``nodes``, |
|
696 | 696 | the iteration order from ``nodes`` should be used. If the value is |
|
697 | 697 | ``storage``, then the native order from the backing storage layer |
|
698 | 698 | is used. (Not all storage layers will have strong ordering and behavior |
|
699 | 699 | of this mode is storage-dependent.) ``nodes`` ordering can force |
|
700 | 700 | revisions to be emitted before their ancestors, so consumers should |
|
701 | 701 | use it with care. |
|
702 | 702 | |
|
703 | 703 | The ``linknode`` attribute on the returned ``irevisiondelta`` may not |
|
704 | 704 | be set and it is the caller's responsibility to resolve it, if needed. |
|
705 | 705 | |
|
706 | 706 | If ``deltamode`` is CG_DELTAMODE_PREV and revision data is requested, |
|
707 | 707 | all revision data should be emitted as deltas against the revision |
|
708 | 708 | emitted just prior. The initial revision should be a delta against its |
|
709 | 709 | 1st parent. |
|
710 | 710 | """ |
|
711 | 711 | |
|
712 | 712 | |
|
713 | 713 | class ifilemutation(interfaceutil.Interface): |
|
714 | 714 | """Storage interface for mutation events of a tracked file.""" |
|
715 | 715 | |
|
716 | 716 | def add(filedata, meta, transaction, linkrev, p1, p2): |
|
717 | 717 | """Add a new revision to the store. |
|
718 | 718 | |
|
719 | 719 | Takes file data, dictionary of metadata, a transaction, linkrev, |
|
720 | 720 | and parent nodes. |
|
721 | 721 | |
|
722 | 722 | Returns the node that was added. |
|
723 | 723 | |
|
724 | 724 | May no-op if a revision matching the supplied data is already stored. |
|
725 | 725 | """ |
|
726 | 726 | |
|
727 | 727 | def addrevision( |
|
728 | 728 | revisiondata, |
|
729 | 729 | transaction, |
|
730 | 730 | linkrev, |
|
731 | 731 | p1, |
|
732 | 732 | p2, |
|
733 | 733 | node=None, |
|
734 | 734 | flags=0, |
|
735 | 735 | cachedelta=None, |
|
736 | 736 | ): |
|
737 | """Add a new revision to the store. | |
|
737 | """Add a new revision to the store and return its number. | |
|
738 | 738 | |
|
739 | 739 | This is similar to ``add()`` except it operates at a lower level. |
|
740 | 740 | |
|
741 | 741 | The data passed in already contains a metadata header, if any. |
|
742 | 742 | |
|
743 | 743 | ``node`` and ``flags`` can be used to define the expected node and |
|
744 | 744 | the flags to use with storage. ``flags`` is a bitwise value composed |
|
745 | 745 | of the various ``REVISION_FLAG_*`` constants. |
|
746 | 746 | |
|
747 | 747 | ``add()`` is usually called when adding files from e.g. the working |
|
748 | 748 | directory. ``addrevision()`` is often called by ``add()`` and for |
|
749 | 749 | scenarios where revision data has already been computed, such as when |
|
750 | 750 | applying raw data from a peer repo. |
|
751 | 751 | """ |
|
752 | 752 | |
|
753 | 753 | def addgroup( |
|
754 | 754 | deltas, |
|
755 | 755 | linkmapper, |
|
756 | 756 | transaction, |
|
757 | 757 | addrevisioncb=None, |
|
758 | 758 | duplicaterevisioncb=None, |
|
759 | 759 | maybemissingparents=False, |
|
760 | 760 | ): |
|
761 | 761 | """Process a series of deltas for storage. |
|
762 | 762 | |
|
763 | 763 | ``deltas`` is an iterable of 7-tuples of |
|
764 | 764 | (node, p1, p2, linknode, deltabase, delta, flags) defining revisions |
|
765 | 765 | to add. |
|
766 | 766 | |
|
767 | 767 | The ``delta`` field contains ``mpatch`` data to apply to a base |
|
768 | 768 | revision, identified by ``deltabase``. The base node can be |
|
769 | 769 | ``nullid``, in which case the header from the delta can be ignored |
|
770 | 770 | and the delta used as the fulltext. |
|
771 | 771 | |
|
772 | 772 | ``alwayscache`` instructs the lower layers to cache the content of the |
|
773 | 773 | newly added revision, even if it needs to be explicitly computed. |
|
774 | 774 | This used to be the default when ``addrevisioncb`` was provided up to |
|
775 | 775 | Mercurial 5.8. |
|
776 | 776 | |
|
777 | 777 | ``addrevisioncb`` should be called for each node as it is committed. |
|
778 | 778 | ``duplicaterevisioncb`` should be called for each pre-existing node. |
|
779 | 779 | |
|
780 | 780 | ``maybemissingparents`` is a bool indicating whether the incoming |
|
781 | 781 | data may reference parents/ancestor revisions that aren't present. |
|
782 | 782 | This flag is set when receiving data into a "shallow" store that |
|
783 | 783 | doesn't hold all history. |
|
784 | 784 | |
|
785 | 785 | Returns a list of nodes that were processed. A node will be in the list |
|
786 | 786 | even if it existed in the store previously. |
|
787 | 787 | """ |
|
788 | 788 | |
|
789 | 789 | def censorrevision(tr, node, tombstone=b''): |
|
790 | 790 | """Remove the content of a single revision. |
|
791 | 791 | |
|
792 | 792 | The specified ``node`` will have its content purged from storage. |
|
793 | 793 | Future attempts to access the revision data for this node will |
|
794 | 794 | result in failure. |
|
795 | 795 | |
|
796 | 796 | A ``tombstone`` message can optionally be stored. This message may be |
|
797 | 797 | displayed to users when they attempt to access the missing revision |
|
798 | 798 | data. |
|
799 | 799 | |
|
800 | 800 | Storage backends may have stored deltas against the previous content |
|
801 | 801 | in this revision. As part of censoring a revision, these storage |
|
802 | 802 | backends are expected to rewrite any internally stored deltas such |
|
803 | 803 | that they no longer reference the deleted content. |
|
804 | 804 | """ |
|
805 | 805 | |
|
806 | 806 | def getstrippoint(minlink): |
|
807 | 807 | """Find the minimum revision that must be stripped to strip a linkrev. |
|
808 | 808 | |
|
809 | 809 | Returns a 2-tuple containing the minimum revision number and a set |
|
810 | 810 | of all revisions numbers that would be broken by this strip. |
|
811 | 811 | |
|
812 | 812 | TODO this is highly revlog centric and should be abstracted into |
|
813 | 813 | a higher-level deletion API. ``repair.strip()`` relies on this. |
|
814 | 814 | """ |
|
815 | 815 | |
|
816 | 816 | def strip(minlink, transaction): |
|
817 | 817 | """Remove storage of items starting at a linkrev. |
|
818 | 818 | |
|
819 | 819 | This uses ``getstrippoint()`` to determine the first node to remove. |
|
820 | 820 | Then it effectively truncates storage for all revisions after that. |
|
821 | 821 | |
|
822 | 822 | TODO this is highly revlog centric and should be abstracted into a |
|
823 | 823 | higher-level deletion API. |
|
824 | 824 | """ |
|
825 | 825 | |
|
826 | 826 | |
|
827 | 827 | class ifilestorage(ifileindex, ifiledata, ifilemutation): |
|
828 | 828 | """Complete storage interface for a single tracked file.""" |
|
829 | 829 | |
|
830 | 830 | def files(): |
|
831 | 831 | """Obtain paths that are backing storage for this file. |
|
832 | 832 | |
|
833 | 833 | TODO this is used heavily by verify code and there should probably |
|
834 | 834 | be a better API for that. |
|
835 | 835 | """ |
|
836 | 836 | |
|
837 | 837 | def storageinfo( |
|
838 | 838 | exclusivefiles=False, |
|
839 | 839 | sharedfiles=False, |
|
840 | 840 | revisionscount=False, |
|
841 | 841 | trackedsize=False, |
|
842 | 842 | storedsize=False, |
|
843 | 843 | ): |
|
844 | 844 | """Obtain information about storage for this file's data. |
|
845 | 845 | |
|
846 | 846 | Returns a dict describing storage for this tracked path. The keys |
|
847 | 847 | in the dict map to arguments of the same. The arguments are bools |
|
848 | 848 | indicating whether to calculate and obtain that data. |
|
849 | 849 | |
|
850 | 850 | exclusivefiles |
|
851 | 851 | Iterable of (vfs, path) describing files that are exclusively |
|
852 | 852 | used to back storage for this tracked path. |
|
853 | 853 | |
|
854 | 854 | sharedfiles |
|
855 | 855 | Iterable of (vfs, path) describing files that are used to back |
|
856 | 856 | storage for this tracked path. Those files may also provide storage |
|
857 | 857 | for other stored entities. |
|
858 | 858 | |
|
859 | 859 | revisionscount |
|
860 | 860 | Number of revisions available for retrieval. |
|
861 | 861 | |
|
862 | 862 | trackedsize |
|
863 | 863 | Total size in bytes of all tracked revisions. This is a sum of the |
|
864 | 864 | length of the fulltext of all revisions. |
|
865 | 865 | |
|
866 | 866 | storedsize |
|
867 | 867 | Total size in bytes used to store data for all tracked revisions. |
|
868 | 868 | This is commonly less than ``trackedsize`` due to internal usage |
|
869 | 869 | of deltas rather than fulltext revisions. |
|
870 | 870 | |
|
871 | 871 | Not all storage backends may support all queries are have a reasonable |
|
872 | 872 | value to use. In that case, the value should be set to ``None`` and |
|
873 | 873 | callers are expected to handle this special value. |
|
874 | 874 | """ |
|
875 | 875 | |
|
876 | 876 | def verifyintegrity(state): |
|
877 | 877 | """Verifies the integrity of file storage. |
|
878 | 878 | |
|
879 | 879 | ``state`` is a dict holding state of the verifier process. It can be |
|
880 | 880 | used to communicate data between invocations of multiple storage |
|
881 | 881 | primitives. |
|
882 | 882 | |
|
883 | 883 | If individual revisions cannot have their revision content resolved, |
|
884 | 884 | the method is expected to set the ``skipread`` key to a set of nodes |
|
885 | 885 | that encountered problems. If set, the method can also add the node(s) |
|
886 | 886 | to ``safe_renamed`` in order to indicate nodes that may perform the |
|
887 | 887 | rename checks with currently accessible data. |
|
888 | 888 | |
|
889 | 889 | The method yields objects conforming to the ``iverifyproblem`` |
|
890 | 890 | interface. |
|
891 | 891 | """ |
|
892 | 892 | |
|
893 | 893 | |
|
894 | 894 | class idirs(interfaceutil.Interface): |
|
895 | 895 | """Interface representing a collection of directories from paths. |
|
896 | 896 | |
|
897 | 897 | This interface is essentially a derived data structure representing |
|
898 | 898 | directories from a collection of paths. |
|
899 | 899 | """ |
|
900 | 900 | |
|
901 | 901 | def addpath(path): |
|
902 | 902 | """Add a path to the collection. |
|
903 | 903 | |
|
904 | 904 | All directories in the path will be added to the collection. |
|
905 | 905 | """ |
|
906 | 906 | |
|
907 | 907 | def delpath(path): |
|
908 | 908 | """Remove a path from the collection. |
|
909 | 909 | |
|
910 | 910 | If the removal was the last path in a particular directory, the |
|
911 | 911 | directory is removed from the collection. |
|
912 | 912 | """ |
|
913 | 913 | |
|
914 | 914 | def __iter__(): |
|
915 | 915 | """Iterate over the directories in this collection of paths.""" |
|
916 | 916 | |
|
917 | 917 | def __contains__(path): |
|
918 | 918 | """Whether a specific directory is in this collection.""" |
|
919 | 919 | |
|
920 | 920 | |
|
921 | 921 | class imanifestdict(interfaceutil.Interface): |
|
922 | 922 | """Interface representing a manifest data structure. |
|
923 | 923 | |
|
924 | 924 | A manifest is effectively a dict mapping paths to entries. Each entry |
|
925 | 925 | consists of a binary node and extra flags affecting that entry. |
|
926 | 926 | """ |
|
927 | 927 | |
|
928 | 928 | def __getitem__(path): |
|
929 | 929 | """Returns the binary node value for a path in the manifest. |
|
930 | 930 | |
|
931 | 931 | Raises ``KeyError`` if the path does not exist in the manifest. |
|
932 | 932 | |
|
933 | 933 | Equivalent to ``self.find(path)[0]``. |
|
934 | 934 | """ |
|
935 | 935 | |
|
936 | 936 | def find(path): |
|
937 | 937 | """Returns the entry for a path in the manifest. |
|
938 | 938 | |
|
939 | 939 | Returns a 2-tuple of (node, flags). |
|
940 | 940 | |
|
941 | 941 | Raises ``KeyError`` if the path does not exist in the manifest. |
|
942 | 942 | """ |
|
943 | 943 | |
|
944 | 944 | def __len__(): |
|
945 | 945 | """Return the number of entries in the manifest.""" |
|
946 | 946 | |
|
947 | 947 | def __nonzero__(): |
|
948 | 948 | """Returns True if the manifest has entries, False otherwise.""" |
|
949 | 949 | |
|
950 | 950 | __bool__ = __nonzero__ |
|
951 | 951 | |
|
952 | 952 | def __setitem__(path, node): |
|
953 | 953 | """Define the node value for a path in the manifest. |
|
954 | 954 | |
|
955 | 955 | If the path is already in the manifest, its flags will be copied to |
|
956 | 956 | the new entry. |
|
957 | 957 | """ |
|
958 | 958 | |
|
959 | 959 | def __contains__(path): |
|
960 | 960 | """Whether a path exists in the manifest.""" |
|
961 | 961 | |
|
962 | 962 | def __delitem__(path): |
|
963 | 963 | """Remove a path from the manifest. |
|
964 | 964 | |
|
965 | 965 | Raises ``KeyError`` if the path is not in the manifest. |
|
966 | 966 | """ |
|
967 | 967 | |
|
968 | 968 | def __iter__(): |
|
969 | 969 | """Iterate over paths in the manifest.""" |
|
970 | 970 | |
|
971 | 971 | def iterkeys(): |
|
972 | 972 | """Iterate over paths in the manifest.""" |
|
973 | 973 | |
|
974 | 974 | def keys(): |
|
975 | 975 | """Obtain a list of paths in the manifest.""" |
|
976 | 976 | |
|
977 | 977 | def filesnotin(other, match=None): |
|
978 | 978 | """Obtain the set of paths in this manifest but not in another. |
|
979 | 979 | |
|
980 | 980 | ``match`` is an optional matcher function to be applied to both |
|
981 | 981 | manifests. |
|
982 | 982 | |
|
983 | 983 | Returns a set of paths. |
|
984 | 984 | """ |
|
985 | 985 | |
|
986 | 986 | def dirs(): |
|
987 | 987 | """Returns an object implementing the ``idirs`` interface.""" |
|
988 | 988 | |
|
989 | 989 | def hasdir(dir): |
|
990 | 990 | """Returns a bool indicating if a directory is in this manifest.""" |
|
991 | 991 | |
|
992 | 992 | def walk(match): |
|
993 | 993 | """Generator of paths in manifest satisfying a matcher. |
|
994 | 994 | |
|
995 | 995 | If the matcher has explicit files listed and they don't exist in |
|
996 | 996 | the manifest, ``match.bad()`` is called for each missing file. |
|
997 | 997 | """ |
|
998 | 998 | |
|
999 | 999 | def diff(other, match=None, clean=False): |
|
1000 | 1000 | """Find differences between this manifest and another. |
|
1001 | 1001 | |
|
1002 | 1002 | This manifest is compared to ``other``. |
|
1003 | 1003 | |
|
1004 | 1004 | If ``match`` is provided, the two manifests are filtered against this |
|
1005 | 1005 | matcher and only entries satisfying the matcher are compared. |
|
1006 | 1006 | |
|
1007 | 1007 | If ``clean`` is True, unchanged files are included in the returned |
|
1008 | 1008 | object. |
|
1009 | 1009 | |
|
1010 | 1010 | Returns a dict with paths as keys and values of 2-tuples of 2-tuples of |
|
1011 | 1011 | the form ``((node1, flag1), (node2, flag2))`` where ``(node1, flag1)`` |
|
1012 | 1012 | represents the node and flags for this manifest and ``(node2, flag2)`` |
|
1013 | 1013 | are the same for the other manifest. |
|
1014 | 1014 | """ |
|
1015 | 1015 | |
|
1016 | 1016 | def setflag(path, flag): |
|
1017 | 1017 | """Set the flag value for a given path. |
|
1018 | 1018 | |
|
1019 | 1019 | Raises ``KeyError`` if the path is not already in the manifest. |
|
1020 | 1020 | """ |
|
1021 | 1021 | |
|
1022 | 1022 | def get(path, default=None): |
|
1023 | 1023 | """Obtain the node value for a path or a default value if missing.""" |
|
1024 | 1024 | |
|
1025 | 1025 | def flags(path): |
|
1026 | 1026 | """Return the flags value for a path (default: empty bytestring).""" |
|
1027 | 1027 | |
|
1028 | 1028 | def copy(): |
|
1029 | 1029 | """Return a copy of this manifest.""" |
|
1030 | 1030 | |
|
1031 | 1031 | def items(): |
|
1032 | 1032 | """Returns an iterable of (path, node) for items in this manifest.""" |
|
1033 | 1033 | |
|
1034 | 1034 | def iteritems(): |
|
1035 | 1035 | """Identical to items().""" |
|
1036 | 1036 | |
|
1037 | 1037 | def iterentries(): |
|
1038 | 1038 | """Returns an iterable of (path, node, flags) for this manifest. |
|
1039 | 1039 | |
|
1040 | 1040 | Similar to ``iteritems()`` except items are a 3-tuple and include |
|
1041 | 1041 | flags. |
|
1042 | 1042 | """ |
|
1043 | 1043 | |
|
1044 | 1044 | def text(): |
|
1045 | 1045 | """Obtain the raw data representation for this manifest. |
|
1046 | 1046 | |
|
1047 | 1047 | Result is used to create a manifest revision. |
|
1048 | 1048 | """ |
|
1049 | 1049 | |
|
1050 | 1050 | def fastdelta(base, changes): |
|
1051 | 1051 | """Obtain a delta between this manifest and another given changes. |
|
1052 | 1052 | |
|
1053 | 1053 | ``base`` in the raw data representation for another manifest. |
|
1054 | 1054 | |
|
1055 | 1055 | ``changes`` is an iterable of ``(path, to_delete)``. |
|
1056 | 1056 | |
|
1057 | 1057 | Returns a 2-tuple containing ``bytearray(self.text())`` and the |
|
1058 | 1058 | delta between ``base`` and this manifest. |
|
1059 | 1059 | |
|
1060 | 1060 | If this manifest implementation can't support ``fastdelta()``, |
|
1061 | 1061 | raise ``mercurial.manifest.FastdeltaUnavailable``. |
|
1062 | 1062 | """ |
|
1063 | 1063 | |
|
1064 | 1064 | |
|
1065 | 1065 | class imanifestrevisionbase(interfaceutil.Interface): |
|
1066 | 1066 | """Base interface representing a single revision of a manifest. |
|
1067 | 1067 | |
|
1068 | 1068 | Should not be used as a primary interface: should always be inherited |
|
1069 | 1069 | as part of a larger interface. |
|
1070 | 1070 | """ |
|
1071 | 1071 | |
|
1072 | 1072 | def copy(): |
|
1073 | 1073 | """Obtain a copy of this manifest instance. |
|
1074 | 1074 | |
|
1075 | 1075 | Returns an object conforming to the ``imanifestrevisionwritable`` |
|
1076 | 1076 | interface. The instance will be associated with the same |
|
1077 | 1077 | ``imanifestlog`` collection as this instance. |
|
1078 | 1078 | """ |
|
1079 | 1079 | |
|
1080 | 1080 | def read(): |
|
1081 | 1081 | """Obtain the parsed manifest data structure. |
|
1082 | 1082 | |
|
1083 | 1083 | The returned object conforms to the ``imanifestdict`` interface. |
|
1084 | 1084 | """ |
|
1085 | 1085 | |
|
1086 | 1086 | |
|
1087 | 1087 | class imanifestrevisionstored(imanifestrevisionbase): |
|
1088 | 1088 | """Interface representing a manifest revision committed to storage.""" |
|
1089 | 1089 | |
|
1090 | 1090 | def node(): |
|
1091 | 1091 | """The binary node for this manifest.""" |
|
1092 | 1092 | |
|
1093 | 1093 | parents = interfaceutil.Attribute( |
|
1094 | 1094 | """List of binary nodes that are parents for this manifest revision.""" |
|
1095 | 1095 | ) |
|
1096 | 1096 | |
|
1097 | 1097 | def readdelta(shallow=False): |
|
1098 | 1098 | """Obtain the manifest data structure representing changes from parent. |
|
1099 | 1099 | |
|
1100 | 1100 | This manifest is compared to its 1st parent. A new manifest representing |
|
1101 | 1101 | those differences is constructed. |
|
1102 | 1102 | |
|
1103 | 1103 | The returned object conforms to the ``imanifestdict`` interface. |
|
1104 | 1104 | """ |
|
1105 | 1105 | |
|
1106 | 1106 | def readfast(shallow=False): |
|
1107 | 1107 | """Calls either ``read()`` or ``readdelta()``. |
|
1108 | 1108 | |
|
1109 | 1109 | The faster of the two options is called. |
|
1110 | 1110 | """ |
|
1111 | 1111 | |
|
1112 | 1112 | def find(key): |
|
1113 | 1113 | """Calls self.read().find(key)``. |
|
1114 | 1114 | |
|
1115 | 1115 | Returns a 2-tuple of ``(node, flags)`` or raises ``KeyError``. |
|
1116 | 1116 | """ |
|
1117 | 1117 | |
|
1118 | 1118 | |
|
1119 | 1119 | class imanifestrevisionwritable(imanifestrevisionbase): |
|
1120 | 1120 | """Interface representing a manifest revision that can be committed.""" |
|
1121 | 1121 | |
|
1122 | 1122 | def write(transaction, linkrev, p1node, p2node, added, removed, match=None): |
|
1123 | 1123 | """Add this revision to storage. |
|
1124 | 1124 | |
|
1125 | 1125 | Takes a transaction object, the changeset revision number it will |
|
1126 | 1126 | be associated with, its parent nodes, and lists of added and |
|
1127 | 1127 | removed paths. |
|
1128 | 1128 | |
|
1129 | 1129 | If match is provided, storage can choose not to inspect or write out |
|
1130 | 1130 | items that do not match. Storage is still required to be able to provide |
|
1131 | 1131 | the full manifest in the future for any directories written (these |
|
1132 | 1132 | manifests should not be "narrowed on disk"). |
|
1133 | 1133 | |
|
1134 | 1134 | Returns the binary node of the created revision. |
|
1135 | 1135 | """ |
|
1136 | 1136 | |
|
1137 | 1137 | |
|
1138 | 1138 | class imanifeststorage(interfaceutil.Interface): |
|
1139 | 1139 | """Storage interface for manifest data.""" |
|
1140 | 1140 | |
|
1141 | 1141 | tree = interfaceutil.Attribute( |
|
1142 | 1142 | """The path to the directory this manifest tracks. |
|
1143 | 1143 | |
|
1144 | 1144 | The empty bytestring represents the root manifest. |
|
1145 | 1145 | """ |
|
1146 | 1146 | ) |
|
1147 | 1147 | |
|
1148 | 1148 | index = interfaceutil.Attribute( |
|
1149 | 1149 | """An ``ifilerevisionssequence`` instance.""" |
|
1150 | 1150 | ) |
|
1151 | 1151 | |
|
1152 | 1152 | indexfile = interfaceutil.Attribute( |
|
1153 | 1153 | """Path of revlog index file. |
|
1154 | 1154 | |
|
1155 | 1155 | TODO this is revlog specific and should not be exposed. |
|
1156 | 1156 | """ |
|
1157 | 1157 | ) |
|
1158 | 1158 | |
|
1159 | 1159 | opener = interfaceutil.Attribute( |
|
1160 | 1160 | """VFS opener to use to access underlying files used for storage. |
|
1161 | 1161 | |
|
1162 | 1162 | TODO this is revlog specific and should not be exposed. |
|
1163 | 1163 | """ |
|
1164 | 1164 | ) |
|
1165 | 1165 | |
|
1166 | 1166 | version = interfaceutil.Attribute( |
|
1167 | 1167 | """Revlog version number. |
|
1168 | 1168 | |
|
1169 | 1169 | TODO this is revlog specific and should not be exposed. |
|
1170 | 1170 | """ |
|
1171 | 1171 | ) |
|
1172 | 1172 | |
|
1173 | 1173 | _generaldelta = interfaceutil.Attribute( |
|
1174 | 1174 | """Whether generaldelta storage is being used. |
|
1175 | 1175 | |
|
1176 | 1176 | TODO this is revlog specific and should not be exposed. |
|
1177 | 1177 | """ |
|
1178 | 1178 | ) |
|
1179 | 1179 | |
|
1180 | 1180 | fulltextcache = interfaceutil.Attribute( |
|
1181 | 1181 | """Dict with cache of fulltexts. |
|
1182 | 1182 | |
|
1183 | 1183 | TODO this doesn't feel appropriate for the storage interface. |
|
1184 | 1184 | """ |
|
1185 | 1185 | ) |
|
1186 | 1186 | |
|
1187 | 1187 | def __len__(): |
|
1188 | 1188 | """Obtain the number of revisions stored for this manifest.""" |
|
1189 | 1189 | |
|
1190 | 1190 | def __iter__(): |
|
1191 | 1191 | """Iterate over revision numbers for this manifest.""" |
|
1192 | 1192 | |
|
1193 | 1193 | def rev(node): |
|
1194 | 1194 | """Obtain the revision number given a binary node. |
|
1195 | 1195 | |
|
1196 | 1196 | Raises ``error.LookupError`` if the node is not known. |
|
1197 | 1197 | """ |
|
1198 | 1198 | |
|
1199 | 1199 | def node(rev): |
|
1200 | 1200 | """Obtain the node value given a revision number. |
|
1201 | 1201 | |
|
1202 | 1202 | Raises ``error.LookupError`` if the revision is not known. |
|
1203 | 1203 | """ |
|
1204 | 1204 | |
|
1205 | 1205 | def lookup(value): |
|
1206 | 1206 | """Attempt to resolve a value to a node. |
|
1207 | 1207 | |
|
1208 | 1208 | Value can be a binary node, hex node, revision number, or a bytes |
|
1209 | 1209 | that can be converted to an integer. |
|
1210 | 1210 | |
|
1211 | 1211 | Raises ``error.LookupError`` if a ndoe could not be resolved. |
|
1212 | 1212 | """ |
|
1213 | 1213 | |
|
1214 | 1214 | def parents(node): |
|
1215 | 1215 | """Returns a 2-tuple of parent nodes for a node. |
|
1216 | 1216 | |
|
1217 | 1217 | Values will be ``nullid`` if the parent is empty. |
|
1218 | 1218 | """ |
|
1219 | 1219 | |
|
1220 | 1220 | def parentrevs(rev): |
|
1221 | 1221 | """Like parents() but operates on revision numbers.""" |
|
1222 | 1222 | |
|
1223 | 1223 | def linkrev(rev): |
|
1224 | 1224 | """Obtain the changeset revision number a revision is linked to.""" |
|
1225 | 1225 | |
|
1226 | 1226 | def revision(node, _df=None, raw=False): |
|
1227 | 1227 | """Obtain fulltext data for a node.""" |
|
1228 | 1228 | |
|
1229 | 1229 | def rawdata(node, _df=None): |
|
1230 | 1230 | """Obtain raw data for a node.""" |
|
1231 | 1231 | |
|
1232 | 1232 | def revdiff(rev1, rev2): |
|
1233 | 1233 | """Obtain a delta between two revision numbers. |
|
1234 | 1234 | |
|
1235 | 1235 | The returned data is the result of ``bdiff.bdiff()`` on the raw |
|
1236 | 1236 | revision data. |
|
1237 | 1237 | """ |
|
1238 | 1238 | |
|
1239 | 1239 | def cmp(node, fulltext): |
|
1240 | 1240 | """Compare fulltext to another revision. |
|
1241 | 1241 | |
|
1242 | 1242 | Returns True if the fulltext is different from what is stored. |
|
1243 | 1243 | """ |
|
1244 | 1244 | |
|
1245 | 1245 | def emitrevisions( |
|
1246 | 1246 | nodes, |
|
1247 | 1247 | nodesorder=None, |
|
1248 | 1248 | revisiondata=False, |
|
1249 | 1249 | assumehaveparentrevisions=False, |
|
1250 | 1250 | ): |
|
1251 | 1251 | """Produce ``irevisiondelta`` describing revisions. |
|
1252 | 1252 | |
|
1253 | 1253 | See the documentation for ``ifiledata`` for more. |
|
1254 | 1254 | """ |
|
1255 | 1255 | |
|
1256 | 1256 | def addgroup( |
|
1257 | 1257 | deltas, |
|
1258 | 1258 | linkmapper, |
|
1259 | 1259 | transaction, |
|
1260 | 1260 | addrevisioncb=None, |
|
1261 | 1261 | duplicaterevisioncb=None, |
|
1262 | 1262 | ): |
|
1263 | 1263 | """Process a series of deltas for storage. |
|
1264 | 1264 | |
|
1265 | 1265 | See the documentation in ``ifilemutation`` for more. |
|
1266 | 1266 | """ |
|
1267 | 1267 | |
|
1268 | 1268 | def rawsize(rev): |
|
1269 | 1269 | """Obtain the size of tracked data. |
|
1270 | 1270 | |
|
1271 | 1271 | Is equivalent to ``len(m.rawdata(node))``. |
|
1272 | 1272 | |
|
1273 | 1273 | TODO this method is only used by upgrade code and may be removed. |
|
1274 | 1274 | """ |
|
1275 | 1275 | |
|
1276 | 1276 | def getstrippoint(minlink): |
|
1277 | 1277 | """Find minimum revision that must be stripped to strip a linkrev. |
|
1278 | 1278 | |
|
1279 | 1279 | See the documentation in ``ifilemutation`` for more. |
|
1280 | 1280 | """ |
|
1281 | 1281 | |
|
1282 | 1282 | def strip(minlink, transaction): |
|
1283 | 1283 | """Remove storage of items starting at a linkrev. |
|
1284 | 1284 | |
|
1285 | 1285 | See the documentation in ``ifilemutation`` for more. |
|
1286 | 1286 | """ |
|
1287 | 1287 | |
|
1288 | 1288 | def checksize(): |
|
1289 | 1289 | """Obtain the expected sizes of backing files. |
|
1290 | 1290 | |
|
1291 | 1291 | TODO this is used by verify and it should not be part of the interface. |
|
1292 | 1292 | """ |
|
1293 | 1293 | |
|
1294 | 1294 | def files(): |
|
1295 | 1295 | """Obtain paths that are backing storage for this manifest. |
|
1296 | 1296 | |
|
1297 | 1297 | TODO this is used by verify and there should probably be a better API |
|
1298 | 1298 | for this functionality. |
|
1299 | 1299 | """ |
|
1300 | 1300 | |
|
1301 | 1301 | def deltaparent(rev): |
|
1302 | 1302 | """Obtain the revision that a revision is delta'd against. |
|
1303 | 1303 | |
|
1304 | 1304 | TODO delta encoding is an implementation detail of storage and should |
|
1305 | 1305 | not be exposed to the storage interface. |
|
1306 | 1306 | """ |
|
1307 | 1307 | |
|
1308 | 1308 | def clone(tr, dest, **kwargs): |
|
1309 | 1309 | """Clone this instance to another.""" |
|
1310 | 1310 | |
|
1311 | 1311 | def clearcaches(clear_persisted_data=False): |
|
1312 | 1312 | """Clear any caches associated with this instance.""" |
|
1313 | 1313 | |
|
1314 | 1314 | def dirlog(d): |
|
1315 | 1315 | """Obtain a manifest storage instance for a tree.""" |
|
1316 | 1316 | |
|
1317 | 1317 | def add( |
|
1318 | 1318 | m, transaction, link, p1, p2, added, removed, readtree=None, match=None |
|
1319 | 1319 | ): |
|
1320 | 1320 | """Add a revision to storage. |
|
1321 | 1321 | |
|
1322 | 1322 | ``m`` is an object conforming to ``imanifestdict``. |
|
1323 | 1323 | |
|
1324 | 1324 | ``link`` is the linkrev revision number. |
|
1325 | 1325 | |
|
1326 | 1326 | ``p1`` and ``p2`` are the parent revision numbers. |
|
1327 | 1327 | |
|
1328 | 1328 | ``added`` and ``removed`` are iterables of added and removed paths, |
|
1329 | 1329 | respectively. |
|
1330 | 1330 | |
|
1331 | 1331 | ``readtree`` is a function that can be used to read the child tree(s) |
|
1332 | 1332 | when recursively writing the full tree structure when using |
|
1333 | 1333 | treemanifets. |
|
1334 | 1334 | |
|
1335 | 1335 | ``match`` is a matcher that can be used to hint to storage that not all |
|
1336 | 1336 | paths must be inspected; this is an optimization and can be safely |
|
1337 | 1337 | ignored. Note that the storage must still be able to reproduce a full |
|
1338 | 1338 | manifest including files that did not match. |
|
1339 | 1339 | """ |
|
1340 | 1340 | |
|
1341 | 1341 | def storageinfo( |
|
1342 | 1342 | exclusivefiles=False, |
|
1343 | 1343 | sharedfiles=False, |
|
1344 | 1344 | revisionscount=False, |
|
1345 | 1345 | trackedsize=False, |
|
1346 | 1346 | storedsize=False, |
|
1347 | 1347 | ): |
|
1348 | 1348 | """Obtain information about storage for this manifest's data. |
|
1349 | 1349 | |
|
1350 | 1350 | See ``ifilestorage.storageinfo()`` for a description of this method. |
|
1351 | 1351 | This one behaves the same way, except for manifest data. |
|
1352 | 1352 | """ |
|
1353 | 1353 | |
|
1354 | 1354 | |
|
1355 | 1355 | class imanifestlog(interfaceutil.Interface): |
|
1356 | 1356 | """Interface representing a collection of manifest snapshots. |
|
1357 | 1357 | |
|
1358 | 1358 | Represents the root manifest in a repository. |
|
1359 | 1359 | |
|
1360 | 1360 | Also serves as a means to access nested tree manifests and to cache |
|
1361 | 1361 | tree manifests. |
|
1362 | 1362 | """ |
|
1363 | 1363 | |
|
1364 | 1364 | def __getitem__(node): |
|
1365 | 1365 | """Obtain a manifest instance for a given binary node. |
|
1366 | 1366 | |
|
1367 | 1367 | Equivalent to calling ``self.get('', node)``. |
|
1368 | 1368 | |
|
1369 | 1369 | The returned object conforms to the ``imanifestrevisionstored`` |
|
1370 | 1370 | interface. |
|
1371 | 1371 | """ |
|
1372 | 1372 | |
|
1373 | 1373 | def get(tree, node, verify=True): |
|
1374 | 1374 | """Retrieve the manifest instance for a given directory and binary node. |
|
1375 | 1375 | |
|
1376 | 1376 | ``node`` always refers to the node of the root manifest (which will be |
|
1377 | 1377 | the only manifest if flat manifests are being used). |
|
1378 | 1378 | |
|
1379 | 1379 | If ``tree`` is the empty string, the root manifest is returned. |
|
1380 | 1380 | Otherwise the manifest for the specified directory will be returned |
|
1381 | 1381 | (requires tree manifests). |
|
1382 | 1382 | |
|
1383 | 1383 | If ``verify`` is True, ``LookupError`` is raised if the node is not |
|
1384 | 1384 | known. |
|
1385 | 1385 | |
|
1386 | 1386 | The returned object conforms to the ``imanifestrevisionstored`` |
|
1387 | 1387 | interface. |
|
1388 | 1388 | """ |
|
1389 | 1389 | |
|
1390 | 1390 | def getstorage(tree): |
|
1391 | 1391 | """Retrieve an interface to storage for a particular tree. |
|
1392 | 1392 | |
|
1393 | 1393 | If ``tree`` is the empty bytestring, storage for the root manifest will |
|
1394 | 1394 | be returned. Otherwise storage for a tree manifest is returned. |
|
1395 | 1395 | |
|
1396 | 1396 | TODO formalize interface for returned object. |
|
1397 | 1397 | """ |
|
1398 | 1398 | |
|
1399 | 1399 | def clearcaches(): |
|
1400 | 1400 | """Clear caches associated with this collection.""" |
|
1401 | 1401 | |
|
1402 | 1402 | def rev(node): |
|
1403 | 1403 | """Obtain the revision number for a binary node. |
|
1404 | 1404 | |
|
1405 | 1405 | Raises ``error.LookupError`` if the node is not known. |
|
1406 | 1406 | """ |
|
1407 | 1407 | |
|
1408 | 1408 | def update_caches(transaction): |
|
1409 | 1409 | """update whatever cache are relevant for the used storage.""" |
|
1410 | 1410 | |
|
1411 | 1411 | |
|
1412 | 1412 | class ilocalrepositoryfilestorage(interfaceutil.Interface): |
|
1413 | 1413 | """Local repository sub-interface providing access to tracked file storage. |
|
1414 | 1414 | |
|
1415 | 1415 | This interface defines how a repository accesses storage for a single |
|
1416 | 1416 | tracked file path. |
|
1417 | 1417 | """ |
|
1418 | 1418 | |
|
1419 | 1419 | def file(f): |
|
1420 | 1420 | """Obtain a filelog for a tracked path. |
|
1421 | 1421 | |
|
1422 | 1422 | The returned type conforms to the ``ifilestorage`` interface. |
|
1423 | 1423 | """ |
|
1424 | 1424 | |
|
1425 | 1425 | |
|
1426 | 1426 | class ilocalrepositorymain(interfaceutil.Interface): |
|
1427 | 1427 | """Main interface for local repositories. |
|
1428 | 1428 | |
|
1429 | 1429 | This currently captures the reality of things - not how things should be. |
|
1430 | 1430 | """ |
|
1431 | 1431 | |
|
1432 | 1432 | supportedformats = interfaceutil.Attribute( |
|
1433 | 1433 | """Set of requirements that apply to stream clone. |
|
1434 | 1434 | |
|
1435 | 1435 | This is actually a class attribute and is shared among all instances. |
|
1436 | 1436 | """ |
|
1437 | 1437 | ) |
|
1438 | 1438 | |
|
1439 | 1439 | supported = interfaceutil.Attribute( |
|
1440 | 1440 | """Set of requirements that this repo is capable of opening.""" |
|
1441 | 1441 | ) |
|
1442 | 1442 | |
|
1443 | 1443 | requirements = interfaceutil.Attribute( |
|
1444 | 1444 | """Set of requirements this repo uses.""" |
|
1445 | 1445 | ) |
|
1446 | 1446 | |
|
1447 | 1447 | features = interfaceutil.Attribute( |
|
1448 | 1448 | """Set of "features" this repository supports. |
|
1449 | 1449 | |
|
1450 | 1450 | A "feature" is a loosely-defined term. It can refer to a feature |
|
1451 | 1451 | in the classical sense or can describe an implementation detail |
|
1452 | 1452 | of the repository. For example, a ``readonly`` feature may denote |
|
1453 | 1453 | the repository as read-only. Or a ``revlogfilestore`` feature may |
|
1454 | 1454 | denote that the repository is using revlogs for file storage. |
|
1455 | 1455 | |
|
1456 | 1456 | The intent of features is to provide a machine-queryable mechanism |
|
1457 | 1457 | for repo consumers to test for various repository characteristics. |
|
1458 | 1458 | |
|
1459 | 1459 | Features are similar to ``requirements``. The main difference is that |
|
1460 | 1460 | requirements are stored on-disk and represent requirements to open the |
|
1461 | 1461 | repository. Features are more run-time capabilities of the repository |
|
1462 | 1462 | and more granular capabilities (which may be derived from requirements). |
|
1463 | 1463 | """ |
|
1464 | 1464 | ) |
|
1465 | 1465 | |
|
1466 | 1466 | filtername = interfaceutil.Attribute( |
|
1467 | 1467 | """Name of the repoview that is active on this repo.""" |
|
1468 | 1468 | ) |
|
1469 | 1469 | |
|
1470 | 1470 | wvfs = interfaceutil.Attribute( |
|
1471 | 1471 | """VFS used to access the working directory.""" |
|
1472 | 1472 | ) |
|
1473 | 1473 | |
|
1474 | 1474 | vfs = interfaceutil.Attribute( |
|
1475 | 1475 | """VFS rooted at the .hg directory. |
|
1476 | 1476 | |
|
1477 | 1477 | Used to access repository data not in the store. |
|
1478 | 1478 | """ |
|
1479 | 1479 | ) |
|
1480 | 1480 | |
|
1481 | 1481 | svfs = interfaceutil.Attribute( |
|
1482 | 1482 | """VFS rooted at the store. |
|
1483 | 1483 | |
|
1484 | 1484 | Used to access repository data in the store. Typically .hg/store. |
|
1485 | 1485 | But can point elsewhere if the store is shared. |
|
1486 | 1486 | """ |
|
1487 | 1487 | ) |
|
1488 | 1488 | |
|
1489 | 1489 | root = interfaceutil.Attribute( |
|
1490 | 1490 | """Path to the root of the working directory.""" |
|
1491 | 1491 | ) |
|
1492 | 1492 | |
|
1493 | 1493 | path = interfaceutil.Attribute("""Path to the .hg directory.""") |
|
1494 | 1494 | |
|
1495 | 1495 | origroot = interfaceutil.Attribute( |
|
1496 | 1496 | """The filesystem path that was used to construct the repo.""" |
|
1497 | 1497 | ) |
|
1498 | 1498 | |
|
1499 | 1499 | auditor = interfaceutil.Attribute( |
|
1500 | 1500 | """A pathauditor for the working directory. |
|
1501 | 1501 | |
|
1502 | 1502 | This checks if a path refers to a nested repository. |
|
1503 | 1503 | |
|
1504 | 1504 | Operates on the filesystem. |
|
1505 | 1505 | """ |
|
1506 | 1506 | ) |
|
1507 | 1507 | |
|
1508 | 1508 | nofsauditor = interfaceutil.Attribute( |
|
1509 | 1509 | """A pathauditor for the working directory. |
|
1510 | 1510 | |
|
1511 | 1511 | This is like ``auditor`` except it doesn't do filesystem checks. |
|
1512 | 1512 | """ |
|
1513 | 1513 | ) |
|
1514 | 1514 | |
|
1515 | 1515 | baseui = interfaceutil.Attribute( |
|
1516 | 1516 | """Original ui instance passed into constructor.""" |
|
1517 | 1517 | ) |
|
1518 | 1518 | |
|
1519 | 1519 | ui = interfaceutil.Attribute("""Main ui instance for this instance.""") |
|
1520 | 1520 | |
|
1521 | 1521 | sharedpath = interfaceutil.Attribute( |
|
1522 | 1522 | """Path to the .hg directory of the repo this repo was shared from.""" |
|
1523 | 1523 | ) |
|
1524 | 1524 | |
|
1525 | 1525 | store = interfaceutil.Attribute("""A store instance.""") |
|
1526 | 1526 | |
|
1527 | 1527 | spath = interfaceutil.Attribute("""Path to the store.""") |
|
1528 | 1528 | |
|
1529 | 1529 | sjoin = interfaceutil.Attribute("""Alias to self.store.join.""") |
|
1530 | 1530 | |
|
1531 | 1531 | cachevfs = interfaceutil.Attribute( |
|
1532 | 1532 | """A VFS used to access the cache directory. |
|
1533 | 1533 | |
|
1534 | 1534 | Typically .hg/cache. |
|
1535 | 1535 | """ |
|
1536 | 1536 | ) |
|
1537 | 1537 | |
|
1538 | 1538 | wcachevfs = interfaceutil.Attribute( |
|
1539 | 1539 | """A VFS used to access the cache directory dedicated to working copy |
|
1540 | 1540 | |
|
1541 | 1541 | Typically .hg/wcache. |
|
1542 | 1542 | """ |
|
1543 | 1543 | ) |
|
1544 | 1544 | |
|
1545 | 1545 | filteredrevcache = interfaceutil.Attribute( |
|
1546 | 1546 | """Holds sets of revisions to be filtered.""" |
|
1547 | 1547 | ) |
|
1548 | 1548 | |
|
1549 | 1549 | names = interfaceutil.Attribute("""A ``namespaces`` instance.""") |
|
1550 | 1550 | |
|
1551 | 1551 | filecopiesmode = interfaceutil.Attribute( |
|
1552 | 1552 | """The way files copies should be dealt with in this repo.""" |
|
1553 | 1553 | ) |
|
1554 | 1554 | |
|
1555 | 1555 | def close(): |
|
1556 | 1556 | """Close the handle on this repository.""" |
|
1557 | 1557 | |
|
1558 | 1558 | def peer(): |
|
1559 | 1559 | """Obtain an object conforming to the ``peer`` interface.""" |
|
1560 | 1560 | |
|
1561 | 1561 | def unfiltered(): |
|
1562 | 1562 | """Obtain an unfiltered/raw view of this repo.""" |
|
1563 | 1563 | |
|
1564 | 1564 | def filtered(name, visibilityexceptions=None): |
|
1565 | 1565 | """Obtain a named view of this repository.""" |
|
1566 | 1566 | |
|
1567 | 1567 | obsstore = interfaceutil.Attribute("""A store of obsolescence data.""") |
|
1568 | 1568 | |
|
1569 | 1569 | changelog = interfaceutil.Attribute("""A handle on the changelog revlog.""") |
|
1570 | 1570 | |
|
1571 | 1571 | manifestlog = interfaceutil.Attribute( |
|
1572 | 1572 | """An instance conforming to the ``imanifestlog`` interface. |
|
1573 | 1573 | |
|
1574 | 1574 | Provides access to manifests for the repository. |
|
1575 | 1575 | """ |
|
1576 | 1576 | ) |
|
1577 | 1577 | |
|
1578 | 1578 | dirstate = interfaceutil.Attribute("""Working directory state.""") |
|
1579 | 1579 | |
|
1580 | 1580 | narrowpats = interfaceutil.Attribute( |
|
1581 | 1581 | """Matcher patterns for this repository's narrowspec.""" |
|
1582 | 1582 | ) |
|
1583 | 1583 | |
|
1584 | 1584 | def narrowmatch(match=None, includeexact=False): |
|
1585 | 1585 | """Obtain a matcher for the narrowspec.""" |
|
1586 | 1586 | |
|
1587 | 1587 | def setnarrowpats(newincludes, newexcludes): |
|
1588 | 1588 | """Define the narrowspec for this repository.""" |
|
1589 | 1589 | |
|
1590 | 1590 | def __getitem__(changeid): |
|
1591 | 1591 | """Try to resolve a changectx.""" |
|
1592 | 1592 | |
|
1593 | 1593 | def __contains__(changeid): |
|
1594 | 1594 | """Whether a changeset exists.""" |
|
1595 | 1595 | |
|
1596 | 1596 | def __nonzero__(): |
|
1597 | 1597 | """Always returns True.""" |
|
1598 | 1598 | return True |
|
1599 | 1599 | |
|
1600 | 1600 | __bool__ = __nonzero__ |
|
1601 | 1601 | |
|
1602 | 1602 | def __len__(): |
|
1603 | 1603 | """Returns the number of changesets in the repo.""" |
|
1604 | 1604 | |
|
1605 | 1605 | def __iter__(): |
|
1606 | 1606 | """Iterate over revisions in the changelog.""" |
|
1607 | 1607 | |
|
1608 | 1608 | def revs(expr, *args): |
|
1609 | 1609 | """Evaluate a revset. |
|
1610 | 1610 | |
|
1611 | 1611 | Emits revisions. |
|
1612 | 1612 | """ |
|
1613 | 1613 | |
|
1614 | 1614 | def set(expr, *args): |
|
1615 | 1615 | """Evaluate a revset. |
|
1616 | 1616 | |
|
1617 | 1617 | Emits changectx instances. |
|
1618 | 1618 | """ |
|
1619 | 1619 | |
|
1620 | 1620 | def anyrevs(specs, user=False, localalias=None): |
|
1621 | 1621 | """Find revisions matching one of the given revsets.""" |
|
1622 | 1622 | |
|
1623 | 1623 | def url(): |
|
1624 | 1624 | """Returns a string representing the location of this repo.""" |
|
1625 | 1625 | |
|
1626 | 1626 | def hook(name, throw=False, **args): |
|
1627 | 1627 | """Call a hook.""" |
|
1628 | 1628 | |
|
1629 | 1629 | def tags(): |
|
1630 | 1630 | """Return a mapping of tag to node.""" |
|
1631 | 1631 | |
|
1632 | 1632 | def tagtype(tagname): |
|
1633 | 1633 | """Return the type of a given tag.""" |
|
1634 | 1634 | |
|
1635 | 1635 | def tagslist(): |
|
1636 | 1636 | """Return a list of tags ordered by revision.""" |
|
1637 | 1637 | |
|
1638 | 1638 | def nodetags(node): |
|
1639 | 1639 | """Return the tags associated with a node.""" |
|
1640 | 1640 | |
|
1641 | 1641 | def nodebookmarks(node): |
|
1642 | 1642 | """Return the list of bookmarks pointing to the specified node.""" |
|
1643 | 1643 | |
|
1644 | 1644 | def branchmap(): |
|
1645 | 1645 | """Return a mapping of branch to heads in that branch.""" |
|
1646 | 1646 | |
|
1647 | 1647 | def revbranchcache(): |
|
1648 | 1648 | pass |
|
1649 | 1649 | |
|
1650 | 1650 | def register_changeset(rev, changelogrevision): |
|
1651 | 1651 | """Extension point for caches for new nodes. |
|
1652 | 1652 | |
|
1653 | 1653 | Multiple consumers are expected to need parts of the changelogrevision, |
|
1654 | 1654 | so it is provided as optimization to avoid duplicate lookups. A simple |
|
1655 | 1655 | cache would be fragile when other revisions are accessed, too.""" |
|
1656 | 1656 | pass |
|
1657 | 1657 | |
|
1658 | 1658 | def branchtip(branchtip, ignoremissing=False): |
|
1659 | 1659 | """Return the tip node for a given branch.""" |
|
1660 | 1660 | |
|
1661 | 1661 | def lookup(key): |
|
1662 | 1662 | """Resolve the node for a revision.""" |
|
1663 | 1663 | |
|
1664 | 1664 | def lookupbranch(key): |
|
1665 | 1665 | """Look up the branch name of the given revision or branch name.""" |
|
1666 | 1666 | |
|
1667 | 1667 | def known(nodes): |
|
1668 | 1668 | """Determine whether a series of nodes is known. |
|
1669 | 1669 | |
|
1670 | 1670 | Returns a list of bools. |
|
1671 | 1671 | """ |
|
1672 | 1672 | |
|
1673 | 1673 | def local(): |
|
1674 | 1674 | """Whether the repository is local.""" |
|
1675 | 1675 | return True |
|
1676 | 1676 | |
|
1677 | 1677 | def publishing(): |
|
1678 | 1678 | """Whether the repository is a publishing repository.""" |
|
1679 | 1679 | |
|
1680 | 1680 | def cancopy(): |
|
1681 | 1681 | pass |
|
1682 | 1682 | |
|
1683 | 1683 | def shared(): |
|
1684 | 1684 | """The type of shared repository or None.""" |
|
1685 | 1685 | |
|
1686 | 1686 | def wjoin(f, *insidef): |
|
1687 | 1687 | """Calls self.vfs.reljoin(self.root, f, *insidef)""" |
|
1688 | 1688 | |
|
1689 | 1689 | def setparents(p1, p2): |
|
1690 | 1690 | """Set the parent nodes of the working directory.""" |
|
1691 | 1691 | |
|
1692 | 1692 | def filectx(path, changeid=None, fileid=None): |
|
1693 | 1693 | """Obtain a filectx for the given file revision.""" |
|
1694 | 1694 | |
|
1695 | 1695 | def getcwd(): |
|
1696 | 1696 | """Obtain the current working directory from the dirstate.""" |
|
1697 | 1697 | |
|
1698 | 1698 | def pathto(f, cwd=None): |
|
1699 | 1699 | """Obtain the relative path to a file.""" |
|
1700 | 1700 | |
|
1701 | 1701 | def adddatafilter(name, fltr): |
|
1702 | 1702 | pass |
|
1703 | 1703 | |
|
1704 | 1704 | def wread(filename): |
|
1705 | 1705 | """Read a file from wvfs, using data filters.""" |
|
1706 | 1706 | |
|
1707 | 1707 | def wwrite(filename, data, flags, backgroundclose=False, **kwargs): |
|
1708 | 1708 | """Write data to a file in the wvfs, using data filters.""" |
|
1709 | 1709 | |
|
1710 | 1710 | def wwritedata(filename, data): |
|
1711 | 1711 | """Resolve data for writing to the wvfs, using data filters.""" |
|
1712 | 1712 | |
|
1713 | 1713 | def currenttransaction(): |
|
1714 | 1714 | """Obtain the current transaction instance or None.""" |
|
1715 | 1715 | |
|
1716 | 1716 | def transaction(desc, report=None): |
|
1717 | 1717 | """Open a new transaction to write to the repository.""" |
|
1718 | 1718 | |
|
1719 | 1719 | def undofiles(): |
|
1720 | 1720 | """Returns a list of (vfs, path) for files to undo transactions.""" |
|
1721 | 1721 | |
|
1722 | 1722 | def recover(): |
|
1723 | 1723 | """Roll back an interrupted transaction.""" |
|
1724 | 1724 | |
|
1725 | 1725 | def rollback(dryrun=False, force=False): |
|
1726 | 1726 | """Undo the last transaction. |
|
1727 | 1727 | |
|
1728 | 1728 | DANGEROUS. |
|
1729 | 1729 | """ |
|
1730 | 1730 | |
|
1731 | 1731 | def updatecaches(tr=None, full=False): |
|
1732 | 1732 | """Warm repo caches.""" |
|
1733 | 1733 | |
|
1734 | 1734 | def invalidatecaches(): |
|
1735 | 1735 | """Invalidate cached data due to the repository mutating.""" |
|
1736 | 1736 | |
|
1737 | 1737 | def invalidatevolatilesets(): |
|
1738 | 1738 | pass |
|
1739 | 1739 | |
|
1740 | 1740 | def invalidatedirstate(): |
|
1741 | 1741 | """Invalidate the dirstate.""" |
|
1742 | 1742 | |
|
1743 | 1743 | def invalidate(clearfilecache=False): |
|
1744 | 1744 | pass |
|
1745 | 1745 | |
|
1746 | 1746 | def invalidateall(): |
|
1747 | 1747 | pass |
|
1748 | 1748 | |
|
1749 | 1749 | def lock(wait=True): |
|
1750 | 1750 | """Lock the repository store and return a lock instance.""" |
|
1751 | 1751 | |
|
1752 | 1752 | def wlock(wait=True): |
|
1753 | 1753 | """Lock the non-store parts of the repository.""" |
|
1754 | 1754 | |
|
1755 | 1755 | def currentwlock(): |
|
1756 | 1756 | """Return the wlock if it's held or None.""" |
|
1757 | 1757 | |
|
1758 | 1758 | def checkcommitpatterns(wctx, match, status, fail): |
|
1759 | 1759 | pass |
|
1760 | 1760 | |
|
1761 | 1761 | def commit( |
|
1762 | 1762 | text=b'', |
|
1763 | 1763 | user=None, |
|
1764 | 1764 | date=None, |
|
1765 | 1765 | match=None, |
|
1766 | 1766 | force=False, |
|
1767 | 1767 | editor=False, |
|
1768 | 1768 | extra=None, |
|
1769 | 1769 | ): |
|
1770 | 1770 | """Add a new revision to the repository.""" |
|
1771 | 1771 | |
|
1772 | 1772 | def commitctx(ctx, error=False, origctx=None): |
|
1773 | 1773 | """Commit a commitctx instance to the repository.""" |
|
1774 | 1774 | |
|
1775 | 1775 | def destroying(): |
|
1776 | 1776 | """Inform the repository that nodes are about to be destroyed.""" |
|
1777 | 1777 | |
|
1778 | 1778 | def destroyed(): |
|
1779 | 1779 | """Inform the repository that nodes have been destroyed.""" |
|
1780 | 1780 | |
|
1781 | 1781 | def status( |
|
1782 | 1782 | node1=b'.', |
|
1783 | 1783 | node2=None, |
|
1784 | 1784 | match=None, |
|
1785 | 1785 | ignored=False, |
|
1786 | 1786 | clean=False, |
|
1787 | 1787 | unknown=False, |
|
1788 | 1788 | listsubrepos=False, |
|
1789 | 1789 | ): |
|
1790 | 1790 | """Convenience method to call repo[x].status().""" |
|
1791 | 1791 | |
|
1792 | 1792 | def addpostdsstatus(ps): |
|
1793 | 1793 | pass |
|
1794 | 1794 | |
|
1795 | 1795 | def postdsstatus(): |
|
1796 | 1796 | pass |
|
1797 | 1797 | |
|
1798 | 1798 | def clearpostdsstatus(): |
|
1799 | 1799 | pass |
|
1800 | 1800 | |
|
1801 | 1801 | def heads(start=None): |
|
1802 | 1802 | """Obtain list of nodes that are DAG heads.""" |
|
1803 | 1803 | |
|
1804 | 1804 | def branchheads(branch=None, start=None, closed=False): |
|
1805 | 1805 | pass |
|
1806 | 1806 | |
|
1807 | 1807 | def branches(nodes): |
|
1808 | 1808 | pass |
|
1809 | 1809 | |
|
1810 | 1810 | def between(pairs): |
|
1811 | 1811 | pass |
|
1812 | 1812 | |
|
1813 | 1813 | def checkpush(pushop): |
|
1814 | 1814 | pass |
|
1815 | 1815 | |
|
1816 | 1816 | prepushoutgoinghooks = interfaceutil.Attribute("""util.hooks instance.""") |
|
1817 | 1817 | |
|
1818 | 1818 | def pushkey(namespace, key, old, new): |
|
1819 | 1819 | pass |
|
1820 | 1820 | |
|
1821 | 1821 | def listkeys(namespace): |
|
1822 | 1822 | pass |
|
1823 | 1823 | |
|
1824 | 1824 | def debugwireargs(one, two, three=None, four=None, five=None): |
|
1825 | 1825 | pass |
|
1826 | 1826 | |
|
1827 | 1827 | def savecommitmessage(text): |
|
1828 | 1828 | pass |
|
1829 | 1829 | |
|
1830 | 1830 | |
|
1831 | 1831 | class completelocalrepository( |
|
1832 | 1832 | ilocalrepositorymain, ilocalrepositoryfilestorage |
|
1833 | 1833 | ): |
|
1834 | 1834 | """Complete interface for a local repository.""" |
|
1835 | 1835 | |
|
1836 | 1836 | |
|
1837 | 1837 | class iwireprotocolcommandcacher(interfaceutil.Interface): |
|
1838 | 1838 | """Represents a caching backend for wire protocol commands. |
|
1839 | 1839 | |
|
1840 | 1840 | Wire protocol version 2 supports transparent caching of many commands. |
|
1841 | 1841 | To leverage this caching, servers can activate objects that cache |
|
1842 | 1842 | command responses. Objects handle both cache writing and reading. |
|
1843 | 1843 | This interface defines how that response caching mechanism works. |
|
1844 | 1844 | |
|
1845 | 1845 | Wire protocol version 2 commands emit a series of objects that are |
|
1846 | 1846 | serialized and sent to the client. The caching layer exists between |
|
1847 | 1847 | the invocation of the command function and the sending of its output |
|
1848 | 1848 | objects to an output layer. |
|
1849 | 1849 | |
|
1850 | 1850 | Instances of this interface represent a binding to a cache that |
|
1851 | 1851 | can serve a response (in place of calling a command function) and/or |
|
1852 | 1852 | write responses to a cache for subsequent use. |
|
1853 | 1853 | |
|
1854 | 1854 | When a command request arrives, the following happens with regards |
|
1855 | 1855 | to this interface: |
|
1856 | 1856 | |
|
1857 | 1857 | 1. The server determines whether the command request is cacheable. |
|
1858 | 1858 | 2. If it is, an instance of this interface is spawned. |
|
1859 | 1859 | 3. The cacher is activated in a context manager (``__enter__`` is called). |
|
1860 | 1860 | 4. A cache *key* for that request is derived. This will call the |
|
1861 | 1861 | instance's ``adjustcachekeystate()`` method so the derivation |
|
1862 | 1862 | can be influenced. |
|
1863 | 1863 | 5. The cacher is informed of the derived cache key via a call to |
|
1864 | 1864 | ``setcachekey()``. |
|
1865 | 1865 | 6. The cacher's ``lookup()`` method is called to test for presence of |
|
1866 | 1866 | the derived key in the cache. |
|
1867 | 1867 | 7. If ``lookup()`` returns a hit, that cached result is used in place |
|
1868 | 1868 | of invoking the command function. ``__exit__`` is called and the instance |
|
1869 | 1869 | is discarded. |
|
1870 | 1870 | 8. The command function is invoked. |
|
1871 | 1871 | 9. ``onobject()`` is called for each object emitted by the command |
|
1872 | 1872 | function. |
|
1873 | 1873 | 10. After the final object is seen, ``onfinished()`` is called. |
|
1874 | 1874 | 11. ``__exit__`` is called to signal the end of use of the instance. |
|
1875 | 1875 | |
|
1876 | 1876 | Cache *key* derivation can be influenced by the instance. |
|
1877 | 1877 | |
|
1878 | 1878 | Cache keys are initially derived by a deterministic representation of |
|
1879 | 1879 | the command request. This includes the command name, arguments, protocol |
|
1880 | 1880 | version, etc. This initial key derivation is performed by CBOR-encoding a |
|
1881 | 1881 | data structure and feeding that output into a hasher. |
|
1882 | 1882 | |
|
1883 | 1883 | Instances of this interface can influence this initial key derivation |
|
1884 | 1884 | via ``adjustcachekeystate()``. |
|
1885 | 1885 | |
|
1886 | 1886 | The instance is informed of the derived cache key via a call to |
|
1887 | 1887 | ``setcachekey()``. The instance must store the key locally so it can |
|
1888 | 1888 | be consulted on subsequent operations that may require it. |
|
1889 | 1889 | |
|
1890 | 1890 | When constructed, the instance has access to a callable that can be used |
|
1891 | 1891 | for encoding response objects. This callable receives as its single |
|
1892 | 1892 | argument an object emitted by a command function. It returns an iterable |
|
1893 | 1893 | of bytes chunks representing the encoded object. Unless the cacher is |
|
1894 | 1894 | caching native Python objects in memory or has a way of reconstructing |
|
1895 | 1895 | the original Python objects, implementations typically call this function |
|
1896 | 1896 | to produce bytes from the output objects and then store those bytes in |
|
1897 | 1897 | the cache. When it comes time to re-emit those bytes, they are wrapped |
|
1898 | 1898 | in a ``wireprototypes.encodedresponse`` instance to tell the output |
|
1899 | 1899 | layer that they are pre-encoded. |
|
1900 | 1900 | |
|
1901 | 1901 | When receiving the objects emitted by the command function, instances |
|
1902 | 1902 | can choose what to do with those objects. The simplest thing to do is |
|
1903 | 1903 | re-emit the original objects. They will be forwarded to the output |
|
1904 | 1904 | layer and will be processed as if the cacher did not exist. |
|
1905 | 1905 | |
|
1906 | 1906 | Implementations could also choose to not emit objects - instead locally |
|
1907 | 1907 | buffering objects or their encoded representation. They could then emit |
|
1908 | 1908 | a single "coalesced" object when ``onfinished()`` is called. In |
|
1909 | 1909 | this way, the implementation would function as a filtering layer of |
|
1910 | 1910 | sorts. |
|
1911 | 1911 | |
|
1912 | 1912 | When caching objects, typically the encoded form of the object will |
|
1913 | 1913 | be stored. Keep in mind that if the original object is forwarded to |
|
1914 | 1914 | the output layer, it will need to be encoded there as well. For large |
|
1915 | 1915 | output, this redundant encoding could add overhead. Implementations |
|
1916 | 1916 | could wrap the encoded object data in ``wireprototypes.encodedresponse`` |
|
1917 | 1917 | instances to avoid this overhead. |
|
1918 | 1918 | """ |
|
1919 | 1919 | |
|
1920 | 1920 | def __enter__(): |
|
1921 | 1921 | """Marks the instance as active. |
|
1922 | 1922 | |
|
1923 | 1923 | Should return self. |
|
1924 | 1924 | """ |
|
1925 | 1925 | |
|
1926 | 1926 | def __exit__(exctype, excvalue, exctb): |
|
1927 | 1927 | """Called when cacher is no longer used. |
|
1928 | 1928 | |
|
1929 | 1929 | This can be used by implementations to perform cleanup actions (e.g. |
|
1930 | 1930 | disconnecting network sockets, aborting a partially cached response. |
|
1931 | 1931 | """ |
|
1932 | 1932 | |
|
1933 | 1933 | def adjustcachekeystate(state): |
|
1934 | 1934 | """Influences cache key derivation by adjusting state to derive key. |
|
1935 | 1935 | |
|
1936 | 1936 | A dict defining the state used to derive the cache key is passed. |
|
1937 | 1937 | |
|
1938 | 1938 | Implementations can modify this dict to record additional state that |
|
1939 | 1939 | is wanted to influence key derivation. |
|
1940 | 1940 | |
|
1941 | 1941 | Implementations are *highly* encouraged to not modify or delete |
|
1942 | 1942 | existing keys. |
|
1943 | 1943 | """ |
|
1944 | 1944 | |
|
1945 | 1945 | def setcachekey(key): |
|
1946 | 1946 | """Record the derived cache key for this request. |
|
1947 | 1947 | |
|
1948 | 1948 | Instances may mutate the key for internal usage, as desired. e.g. |
|
1949 | 1949 | instances may wish to prepend the repo name, introduce path |
|
1950 | 1950 | components for filesystem or URL addressing, etc. Behavior is up to |
|
1951 | 1951 | the cache. |
|
1952 | 1952 | |
|
1953 | 1953 | Returns a bool indicating if the request is cacheable by this |
|
1954 | 1954 | instance. |
|
1955 | 1955 | """ |
|
1956 | 1956 | |
|
1957 | 1957 | def lookup(): |
|
1958 | 1958 | """Attempt to resolve an entry in the cache. |
|
1959 | 1959 | |
|
1960 | 1960 | The instance is instructed to look for the cache key that it was |
|
1961 | 1961 | informed about via the call to ``setcachekey()``. |
|
1962 | 1962 | |
|
1963 | 1963 | If there's no cache hit or the cacher doesn't wish to use the cached |
|
1964 | 1964 | entry, ``None`` should be returned. |
|
1965 | 1965 | |
|
1966 | 1966 | Else, a dict defining the cached result should be returned. The |
|
1967 | 1967 | dict may have the following keys: |
|
1968 | 1968 | |
|
1969 | 1969 | objs |
|
1970 | 1970 | An iterable of objects that should be sent to the client. That |
|
1971 | 1971 | iterable of objects is expected to be what the command function |
|
1972 | 1972 | would return if invoked or an equivalent representation thereof. |
|
1973 | 1973 | """ |
|
1974 | 1974 | |
|
1975 | 1975 | def onobject(obj): |
|
1976 | 1976 | """Called when a new object is emitted from the command function. |
|
1977 | 1977 | |
|
1978 | 1978 | Receives as its argument the object that was emitted from the |
|
1979 | 1979 | command function. |
|
1980 | 1980 | |
|
1981 | 1981 | This method returns an iterator of objects to forward to the output |
|
1982 | 1982 | layer. The easiest implementation is a generator that just |
|
1983 | 1983 | ``yield obj``. |
|
1984 | 1984 | """ |
|
1985 | 1985 | |
|
1986 | 1986 | def onfinished(): |
|
1987 | 1987 | """Called after all objects have been emitted from the command function. |
|
1988 | 1988 | |
|
1989 | 1989 | Implementations should return an iterator of objects to forward to |
|
1990 | 1990 | the output layer. |
|
1991 | 1991 | |
|
1992 | 1992 | This method can be a generator. |
|
1993 | 1993 | """ |
@@ -1,2351 +1,2354 b'' | |||
|
1 | 1 | # manifest.py - manifest revision 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 | import heapq |
|
11 | 11 | import itertools |
|
12 | 12 | import struct |
|
13 | 13 | import weakref |
|
14 | 14 | |
|
15 | 15 | from .i18n import _ |
|
16 | 16 | from .node import ( |
|
17 | 17 | bin, |
|
18 | 18 | hex, |
|
19 | 19 | nullid, |
|
20 | 20 | nullrev, |
|
21 | 21 | ) |
|
22 | 22 | from .pycompat import getattr |
|
23 | 23 | from . import ( |
|
24 | 24 | encoding, |
|
25 | 25 | error, |
|
26 | 26 | match as matchmod, |
|
27 | 27 | mdiff, |
|
28 | 28 | pathutil, |
|
29 | 29 | policy, |
|
30 | 30 | pycompat, |
|
31 | 31 | revlog, |
|
32 | 32 | util, |
|
33 | 33 | ) |
|
34 | 34 | from .interfaces import ( |
|
35 | 35 | repository, |
|
36 | 36 | util as interfaceutil, |
|
37 | 37 | ) |
|
38 | 38 | |
|
39 | 39 | parsers = policy.importmod('parsers') |
|
40 | 40 | propertycache = util.propertycache |
|
41 | 41 | |
|
42 | 42 | # Allow tests to more easily test the alternate path in manifestdict.fastdelta() |
|
43 | 43 | FASTDELTA_TEXTDIFF_THRESHOLD = 1000 |
|
44 | 44 | |
|
45 | 45 | |
|
46 | 46 | def _parse(data): |
|
47 | 47 | # This method does a little bit of excessive-looking |
|
48 | 48 | # precondition checking. This is so that the behavior of this |
|
49 | 49 | # class exactly matches its C counterpart to try and help |
|
50 | 50 | # prevent surprise breakage for anyone that develops against |
|
51 | 51 | # the pure version. |
|
52 | 52 | if data and data[-1:] != b'\n': |
|
53 | 53 | raise ValueError(b'Manifest did not end in a newline.') |
|
54 | 54 | prev = None |
|
55 | 55 | for l in data.splitlines(): |
|
56 | 56 | if prev is not None and prev > l: |
|
57 | 57 | raise ValueError(b'Manifest lines not in sorted order.') |
|
58 | 58 | prev = l |
|
59 | 59 | f, n = l.split(b'\0') |
|
60 | 60 | nl = len(n) |
|
61 | 61 | flags = n[-1:] |
|
62 | 62 | if flags in _manifestflags: |
|
63 | 63 | n = n[:-1] |
|
64 | 64 | nl -= 1 |
|
65 | 65 | else: |
|
66 | 66 | flags = b'' |
|
67 | 67 | if nl not in (40, 64): |
|
68 | 68 | raise ValueError(b'Invalid manifest line') |
|
69 | 69 | |
|
70 | 70 | yield f, bin(n), flags |
|
71 | 71 | |
|
72 | 72 | |
|
73 | 73 | def _text(it): |
|
74 | 74 | files = [] |
|
75 | 75 | lines = [] |
|
76 | 76 | for f, n, fl in it: |
|
77 | 77 | files.append(f) |
|
78 | 78 | # if this is changed to support newlines in filenames, |
|
79 | 79 | # be sure to check the templates/ dir again (especially *-raw.tmpl) |
|
80 | 80 | lines.append(b"%s\0%s%s\n" % (f, hex(n), fl)) |
|
81 | 81 | |
|
82 | 82 | _checkforbidden(files) |
|
83 | 83 | return b''.join(lines) |
|
84 | 84 | |
|
85 | 85 | |
|
86 | 86 | class lazymanifestiter(object): |
|
87 | 87 | def __init__(self, lm): |
|
88 | 88 | self.pos = 0 |
|
89 | 89 | self.lm = lm |
|
90 | 90 | |
|
91 | 91 | def __iter__(self): |
|
92 | 92 | return self |
|
93 | 93 | |
|
94 | 94 | def next(self): |
|
95 | 95 | try: |
|
96 | 96 | data, pos = self.lm._get(self.pos) |
|
97 | 97 | except IndexError: |
|
98 | 98 | raise StopIteration |
|
99 | 99 | if pos == -1: |
|
100 | 100 | self.pos += 1 |
|
101 | 101 | return data[0] |
|
102 | 102 | self.pos += 1 |
|
103 | 103 | zeropos = data.find(b'\x00', pos) |
|
104 | 104 | return data[pos:zeropos] |
|
105 | 105 | |
|
106 | 106 | __next__ = next |
|
107 | 107 | |
|
108 | 108 | |
|
109 | 109 | class lazymanifestiterentries(object): |
|
110 | 110 | def __init__(self, lm): |
|
111 | 111 | self.lm = lm |
|
112 | 112 | self.pos = 0 |
|
113 | 113 | |
|
114 | 114 | def __iter__(self): |
|
115 | 115 | return self |
|
116 | 116 | |
|
117 | 117 | def next(self): |
|
118 | 118 | try: |
|
119 | 119 | data, pos = self.lm._get(self.pos) |
|
120 | 120 | except IndexError: |
|
121 | 121 | raise StopIteration |
|
122 | 122 | if pos == -1: |
|
123 | 123 | self.pos += 1 |
|
124 | 124 | return data |
|
125 | 125 | zeropos = data.find(b'\x00', pos) |
|
126 | 126 | nlpos = data.find(b'\n', pos) |
|
127 | 127 | if zeropos == -1 or nlpos == -1 or nlpos < zeropos: |
|
128 | 128 | raise error.StorageError(b'Invalid manifest line') |
|
129 | 129 | flags = data[nlpos - 1 : nlpos] |
|
130 | 130 | if flags in _manifestflags: |
|
131 | 131 | hlen = nlpos - zeropos - 2 |
|
132 | 132 | else: |
|
133 | 133 | hlen = nlpos - zeropos - 1 |
|
134 | 134 | flags = b'' |
|
135 | 135 | if hlen not in (40, 64): |
|
136 | 136 | raise error.StorageError(b'Invalid manifest line') |
|
137 | 137 | hashval = unhexlify( |
|
138 | 138 | data, self.lm.extrainfo[self.pos], zeropos + 1, hlen |
|
139 | 139 | ) |
|
140 | 140 | self.pos += 1 |
|
141 | 141 | return (data[pos:zeropos], hashval, flags) |
|
142 | 142 | |
|
143 | 143 | __next__ = next |
|
144 | 144 | |
|
145 | 145 | |
|
146 | 146 | def unhexlify(data, extra, pos, length): |
|
147 | 147 | s = bin(data[pos : pos + length]) |
|
148 | 148 | if extra: |
|
149 | 149 | s += chr(extra & 0xFF) |
|
150 | 150 | return s |
|
151 | 151 | |
|
152 | 152 | |
|
153 | 153 | def _cmp(a, b): |
|
154 | 154 | return (a > b) - (a < b) |
|
155 | 155 | |
|
156 | 156 | |
|
157 | 157 | _manifestflags = {b'', b'l', b't', b'x'} |
|
158 | 158 | |
|
159 | 159 | |
|
160 | 160 | class _lazymanifest(object): |
|
161 | 161 | """A pure python manifest backed by a byte string. It is supplimented with |
|
162 | 162 | internal lists as it is modified, until it is compacted back to a pure byte |
|
163 | 163 | string. |
|
164 | 164 | |
|
165 | 165 | ``data`` is the initial manifest data. |
|
166 | 166 | |
|
167 | 167 | ``positions`` is a list of offsets, one per manifest entry. Positive |
|
168 | 168 | values are offsets into ``data``, negative values are offsets into the |
|
169 | 169 | ``extradata`` list. When an entry is removed, its entry is dropped from |
|
170 | 170 | ``positions``. The values are encoded such that when walking the list and |
|
171 | 171 | indexing into ``data`` or ``extradata`` as appropriate, the entries are |
|
172 | 172 | sorted by filename. |
|
173 | 173 | |
|
174 | 174 | ``extradata`` is a list of (key, hash, flags) for entries that were added or |
|
175 | 175 | modified since the manifest was created or compacted. |
|
176 | 176 | """ |
|
177 | 177 | |
|
178 | 178 | def __init__( |
|
179 | 179 | self, |
|
180 | 180 | data, |
|
181 | 181 | positions=None, |
|
182 | 182 | extrainfo=None, |
|
183 | 183 | extradata=None, |
|
184 | 184 | hasremovals=False, |
|
185 | 185 | ): |
|
186 | 186 | if positions is None: |
|
187 | 187 | self.positions = self.findlines(data) |
|
188 | 188 | self.extrainfo = [0] * len(self.positions) |
|
189 | 189 | self.data = data |
|
190 | 190 | self.extradata = [] |
|
191 | 191 | self.hasremovals = False |
|
192 | 192 | else: |
|
193 | 193 | self.positions = positions[:] |
|
194 | 194 | self.extrainfo = extrainfo[:] |
|
195 | 195 | self.extradata = extradata[:] |
|
196 | 196 | self.data = data |
|
197 | 197 | self.hasremovals = hasremovals |
|
198 | 198 | |
|
199 | 199 | def findlines(self, data): |
|
200 | 200 | if not data: |
|
201 | 201 | return [] |
|
202 | 202 | pos = data.find(b"\n") |
|
203 | 203 | if pos == -1 or data[-1:] != b'\n': |
|
204 | 204 | raise ValueError(b"Manifest did not end in a newline.") |
|
205 | 205 | positions = [0] |
|
206 | 206 | prev = data[: data.find(b'\x00')] |
|
207 | 207 | while pos < len(data) - 1 and pos != -1: |
|
208 | 208 | positions.append(pos + 1) |
|
209 | 209 | nexts = data[pos + 1 : data.find(b'\x00', pos + 1)] |
|
210 | 210 | if nexts < prev: |
|
211 | 211 | raise ValueError(b"Manifest lines not in sorted order.") |
|
212 | 212 | prev = nexts |
|
213 | 213 | pos = data.find(b"\n", pos + 1) |
|
214 | 214 | return positions |
|
215 | 215 | |
|
216 | 216 | def _get(self, index): |
|
217 | 217 | # get the position encoded in pos: |
|
218 | 218 | # positive number is an index in 'data' |
|
219 | 219 | # negative number is in extrapieces |
|
220 | 220 | pos = self.positions[index] |
|
221 | 221 | if pos >= 0: |
|
222 | 222 | return self.data, pos |
|
223 | 223 | return self.extradata[-pos - 1], -1 |
|
224 | 224 | |
|
225 | 225 | def _getkey(self, pos): |
|
226 | 226 | if pos >= 0: |
|
227 | 227 | return self.data[pos : self.data.find(b'\x00', pos + 1)] |
|
228 | 228 | return self.extradata[-pos - 1][0] |
|
229 | 229 | |
|
230 | 230 | def bsearch(self, key): |
|
231 | 231 | first = 0 |
|
232 | 232 | last = len(self.positions) - 1 |
|
233 | 233 | |
|
234 | 234 | while first <= last: |
|
235 | 235 | midpoint = (first + last) // 2 |
|
236 | 236 | nextpos = self.positions[midpoint] |
|
237 | 237 | candidate = self._getkey(nextpos) |
|
238 | 238 | r = _cmp(key, candidate) |
|
239 | 239 | if r == 0: |
|
240 | 240 | return midpoint |
|
241 | 241 | else: |
|
242 | 242 | if r < 0: |
|
243 | 243 | last = midpoint - 1 |
|
244 | 244 | else: |
|
245 | 245 | first = midpoint + 1 |
|
246 | 246 | return -1 |
|
247 | 247 | |
|
248 | 248 | def bsearch2(self, key): |
|
249 | 249 | # same as the above, but will always return the position |
|
250 | 250 | # done for performance reasons |
|
251 | 251 | first = 0 |
|
252 | 252 | last = len(self.positions) - 1 |
|
253 | 253 | |
|
254 | 254 | while first <= last: |
|
255 | 255 | midpoint = (first + last) // 2 |
|
256 | 256 | nextpos = self.positions[midpoint] |
|
257 | 257 | candidate = self._getkey(nextpos) |
|
258 | 258 | r = _cmp(key, candidate) |
|
259 | 259 | if r == 0: |
|
260 | 260 | return (midpoint, True) |
|
261 | 261 | else: |
|
262 | 262 | if r < 0: |
|
263 | 263 | last = midpoint - 1 |
|
264 | 264 | else: |
|
265 | 265 | first = midpoint + 1 |
|
266 | 266 | return (first, False) |
|
267 | 267 | |
|
268 | 268 | def __contains__(self, key): |
|
269 | 269 | return self.bsearch(key) != -1 |
|
270 | 270 | |
|
271 | 271 | def __getitem__(self, key): |
|
272 | 272 | if not isinstance(key, bytes): |
|
273 | 273 | raise TypeError(b"getitem: manifest keys must be a bytes.") |
|
274 | 274 | needle = self.bsearch(key) |
|
275 | 275 | if needle == -1: |
|
276 | 276 | raise KeyError |
|
277 | 277 | data, pos = self._get(needle) |
|
278 | 278 | if pos == -1: |
|
279 | 279 | return (data[1], data[2]) |
|
280 | 280 | zeropos = data.find(b'\x00', pos) |
|
281 | 281 | nlpos = data.find(b'\n', zeropos) |
|
282 | 282 | assert 0 <= needle <= len(self.positions) |
|
283 | 283 | assert len(self.extrainfo) == len(self.positions) |
|
284 | 284 | if zeropos == -1 or nlpos == -1 or nlpos < zeropos: |
|
285 | 285 | raise error.StorageError(b'Invalid manifest line') |
|
286 | 286 | hlen = nlpos - zeropos - 1 |
|
287 | 287 | flags = data[nlpos - 1 : nlpos] |
|
288 | 288 | if flags in _manifestflags: |
|
289 | 289 | hlen -= 1 |
|
290 | 290 | else: |
|
291 | 291 | flags = b'' |
|
292 | 292 | if hlen not in (40, 64): |
|
293 | 293 | raise error.StorageError(b'Invalid manifest line') |
|
294 | 294 | hashval = unhexlify(data, self.extrainfo[needle], zeropos + 1, hlen) |
|
295 | 295 | return (hashval, flags) |
|
296 | 296 | |
|
297 | 297 | def __delitem__(self, key): |
|
298 | 298 | needle, found = self.bsearch2(key) |
|
299 | 299 | if not found: |
|
300 | 300 | raise KeyError |
|
301 | 301 | cur = self.positions[needle] |
|
302 | 302 | self.positions = self.positions[:needle] + self.positions[needle + 1 :] |
|
303 | 303 | self.extrainfo = self.extrainfo[:needle] + self.extrainfo[needle + 1 :] |
|
304 | 304 | if cur >= 0: |
|
305 | 305 | # This does NOT unsort the list as far as the search functions are |
|
306 | 306 | # concerned, as they only examine lines mapped by self.positions. |
|
307 | 307 | self.data = self.data[:cur] + b'\x00' + self.data[cur + 1 :] |
|
308 | 308 | self.hasremovals = True |
|
309 | 309 | |
|
310 | 310 | def __setitem__(self, key, value): |
|
311 | 311 | if not isinstance(key, bytes): |
|
312 | 312 | raise TypeError(b"setitem: manifest keys must be a byte string.") |
|
313 | 313 | if not isinstance(value, tuple) or len(value) != 2: |
|
314 | 314 | raise TypeError( |
|
315 | 315 | b"Manifest values must be a tuple of (node, flags)." |
|
316 | 316 | ) |
|
317 | 317 | hashval = value[0] |
|
318 | 318 | if not isinstance(hashval, bytes) or len(hashval) not in (20, 32): |
|
319 | 319 | raise TypeError(b"node must be a 20-byte or 32-byte byte string") |
|
320 | 320 | flags = value[1] |
|
321 | 321 | if not isinstance(flags, bytes) or len(flags) > 1: |
|
322 | 322 | raise TypeError(b"flags must a 0 or 1 byte string, got %r", flags) |
|
323 | 323 | needle, found = self.bsearch2(key) |
|
324 | 324 | if found: |
|
325 | 325 | # put the item |
|
326 | 326 | pos = self.positions[needle] |
|
327 | 327 | if pos < 0: |
|
328 | 328 | self.extradata[-pos - 1] = (key, hashval, value[1]) |
|
329 | 329 | else: |
|
330 | 330 | # just don't bother |
|
331 | 331 | self.extradata.append((key, hashval, value[1])) |
|
332 | 332 | self.positions[needle] = -len(self.extradata) |
|
333 | 333 | else: |
|
334 | 334 | # not found, put it in with extra positions |
|
335 | 335 | self.extradata.append((key, hashval, value[1])) |
|
336 | 336 | self.positions = ( |
|
337 | 337 | self.positions[:needle] |
|
338 | 338 | + [-len(self.extradata)] |
|
339 | 339 | + self.positions[needle:] |
|
340 | 340 | ) |
|
341 | 341 | self.extrainfo = ( |
|
342 | 342 | self.extrainfo[:needle] + [0] + self.extrainfo[needle:] |
|
343 | 343 | ) |
|
344 | 344 | |
|
345 | 345 | def copy(self): |
|
346 | 346 | # XXX call _compact like in C? |
|
347 | 347 | return _lazymanifest( |
|
348 | 348 | self.data, |
|
349 | 349 | self.positions, |
|
350 | 350 | self.extrainfo, |
|
351 | 351 | self.extradata, |
|
352 | 352 | self.hasremovals, |
|
353 | 353 | ) |
|
354 | 354 | |
|
355 | 355 | def _compact(self): |
|
356 | 356 | # hopefully not called TOO often |
|
357 | 357 | if len(self.extradata) == 0 and not self.hasremovals: |
|
358 | 358 | return |
|
359 | 359 | l = [] |
|
360 | 360 | i = 0 |
|
361 | 361 | offset = 0 |
|
362 | 362 | self.extrainfo = [0] * len(self.positions) |
|
363 | 363 | while i < len(self.positions): |
|
364 | 364 | if self.positions[i] >= 0: |
|
365 | 365 | cur = self.positions[i] |
|
366 | 366 | last_cut = cur |
|
367 | 367 | |
|
368 | 368 | # Collect all contiguous entries in the buffer at the current |
|
369 | 369 | # offset, breaking out only for added/modified items held in |
|
370 | 370 | # extradata, or a deleted line prior to the next position. |
|
371 | 371 | while True: |
|
372 | 372 | self.positions[i] = offset |
|
373 | 373 | i += 1 |
|
374 | 374 | if i == len(self.positions) or self.positions[i] < 0: |
|
375 | 375 | break |
|
376 | 376 | |
|
377 | 377 | # A removed file has no positions[] entry, but does have an |
|
378 | 378 | # overwritten first byte. Break out and find the end of the |
|
379 | 379 | # current good entry/entries if there is a removed file |
|
380 | 380 | # before the next position. |
|
381 | 381 | if ( |
|
382 | 382 | self.hasremovals |
|
383 | 383 | and self.data.find(b'\n\x00', cur, self.positions[i]) |
|
384 | 384 | != -1 |
|
385 | 385 | ): |
|
386 | 386 | break |
|
387 | 387 | |
|
388 | 388 | offset += self.positions[i] - cur |
|
389 | 389 | cur = self.positions[i] |
|
390 | 390 | end_cut = self.data.find(b'\n', cur) |
|
391 | 391 | if end_cut != -1: |
|
392 | 392 | end_cut += 1 |
|
393 | 393 | offset += end_cut - cur |
|
394 | 394 | l.append(self.data[last_cut:end_cut]) |
|
395 | 395 | else: |
|
396 | 396 | while i < len(self.positions) and self.positions[i] < 0: |
|
397 | 397 | cur = self.positions[i] |
|
398 | 398 | t = self.extradata[-cur - 1] |
|
399 | 399 | l.append(self._pack(t)) |
|
400 | 400 | self.positions[i] = offset |
|
401 | 401 | # Hashes are either 20 bytes (old sha1s) or 32 |
|
402 | 402 | # bytes (new non-sha1). |
|
403 | 403 | hlen = 20 |
|
404 | 404 | if len(t[1]) > 25: |
|
405 | 405 | hlen = 32 |
|
406 | 406 | if len(t[1]) > hlen: |
|
407 | 407 | self.extrainfo[i] = ord(t[1][hlen + 1]) |
|
408 | 408 | offset += len(l[-1]) |
|
409 | 409 | i += 1 |
|
410 | 410 | self.data = b''.join(l) |
|
411 | 411 | self.hasremovals = False |
|
412 | 412 | self.extradata = [] |
|
413 | 413 | |
|
414 | 414 | def _pack(self, d): |
|
415 | 415 | n = d[1] |
|
416 | 416 | assert len(n) in (20, 32) |
|
417 | 417 | return d[0] + b'\x00' + hex(n) + d[2] + b'\n' |
|
418 | 418 | |
|
419 | 419 | def text(self): |
|
420 | 420 | self._compact() |
|
421 | 421 | return self.data |
|
422 | 422 | |
|
423 | 423 | def diff(self, m2, clean=False): |
|
424 | 424 | '''Finds changes between the current manifest and m2.''' |
|
425 | 425 | # XXX think whether efficiency matters here |
|
426 | 426 | diff = {} |
|
427 | 427 | |
|
428 | 428 | for fn, e1, flags in self.iterentries(): |
|
429 | 429 | if fn not in m2: |
|
430 | 430 | diff[fn] = (e1, flags), (None, b'') |
|
431 | 431 | else: |
|
432 | 432 | e2 = m2[fn] |
|
433 | 433 | if (e1, flags) != e2: |
|
434 | 434 | diff[fn] = (e1, flags), e2 |
|
435 | 435 | elif clean: |
|
436 | 436 | diff[fn] = None |
|
437 | 437 | |
|
438 | 438 | for fn, e2, flags in m2.iterentries(): |
|
439 | 439 | if fn not in self: |
|
440 | 440 | diff[fn] = (None, b''), (e2, flags) |
|
441 | 441 | |
|
442 | 442 | return diff |
|
443 | 443 | |
|
444 | 444 | def iterentries(self): |
|
445 | 445 | return lazymanifestiterentries(self) |
|
446 | 446 | |
|
447 | 447 | def iterkeys(self): |
|
448 | 448 | return lazymanifestiter(self) |
|
449 | 449 | |
|
450 | 450 | def __iter__(self): |
|
451 | 451 | return lazymanifestiter(self) |
|
452 | 452 | |
|
453 | 453 | def __len__(self): |
|
454 | 454 | return len(self.positions) |
|
455 | 455 | |
|
456 | 456 | def filtercopy(self, filterfn): |
|
457 | 457 | # XXX should be optimized |
|
458 | 458 | c = _lazymanifest(b'') |
|
459 | 459 | for f, n, fl in self.iterentries(): |
|
460 | 460 | if filterfn(f): |
|
461 | 461 | c[f] = n, fl |
|
462 | 462 | return c |
|
463 | 463 | |
|
464 | 464 | |
|
465 | 465 | try: |
|
466 | 466 | _lazymanifest = parsers.lazymanifest |
|
467 | 467 | except AttributeError: |
|
468 | 468 | pass |
|
469 | 469 | |
|
470 | 470 | |
|
471 | 471 | @interfaceutil.implementer(repository.imanifestdict) |
|
472 | 472 | class manifestdict(object): |
|
473 | 473 | def __init__(self, data=b''): |
|
474 | 474 | self._lm = _lazymanifest(data) |
|
475 | 475 | |
|
476 | 476 | def __getitem__(self, key): |
|
477 | 477 | return self._lm[key][0] |
|
478 | 478 | |
|
479 | 479 | def find(self, key): |
|
480 | 480 | return self._lm[key] |
|
481 | 481 | |
|
482 | 482 | def __len__(self): |
|
483 | 483 | return len(self._lm) |
|
484 | 484 | |
|
485 | 485 | def __nonzero__(self): |
|
486 | 486 | # nonzero is covered by the __len__ function, but implementing it here |
|
487 | 487 | # makes it easier for extensions to override. |
|
488 | 488 | return len(self._lm) != 0 |
|
489 | 489 | |
|
490 | 490 | __bool__ = __nonzero__ |
|
491 | 491 | |
|
492 | 492 | def __setitem__(self, key, node): |
|
493 | 493 | self._lm[key] = node, self.flags(key) |
|
494 | 494 | |
|
495 | 495 | def __contains__(self, key): |
|
496 | 496 | if key is None: |
|
497 | 497 | return False |
|
498 | 498 | return key in self._lm |
|
499 | 499 | |
|
500 | 500 | def __delitem__(self, key): |
|
501 | 501 | del self._lm[key] |
|
502 | 502 | |
|
503 | 503 | def __iter__(self): |
|
504 | 504 | return self._lm.__iter__() |
|
505 | 505 | |
|
506 | 506 | def iterkeys(self): |
|
507 | 507 | return self._lm.iterkeys() |
|
508 | 508 | |
|
509 | 509 | def keys(self): |
|
510 | 510 | return list(self.iterkeys()) |
|
511 | 511 | |
|
512 | 512 | def filesnotin(self, m2, match=None): |
|
513 | 513 | '''Set of files in this manifest that are not in the other''' |
|
514 | 514 | if match is not None: |
|
515 | 515 | match = matchmod.badmatch(match, lambda path, msg: None) |
|
516 | 516 | sm2 = set(m2.walk(match)) |
|
517 | 517 | return {f for f in self.walk(match) if f not in sm2} |
|
518 | 518 | return {f for f in self if f not in m2} |
|
519 | 519 | |
|
520 | 520 | @propertycache |
|
521 | 521 | def _dirs(self): |
|
522 | 522 | return pathutil.dirs(self) |
|
523 | 523 | |
|
524 | 524 | def dirs(self): |
|
525 | 525 | return self._dirs |
|
526 | 526 | |
|
527 | 527 | def hasdir(self, dir): |
|
528 | 528 | return dir in self._dirs |
|
529 | 529 | |
|
530 | 530 | def _filesfastpath(self, match): |
|
531 | 531 | """Checks whether we can correctly and quickly iterate over matcher |
|
532 | 532 | files instead of over manifest files.""" |
|
533 | 533 | files = match.files() |
|
534 | 534 | return len(files) < 100 and ( |
|
535 | 535 | match.isexact() |
|
536 | 536 | or (match.prefix() and all(fn in self for fn in files)) |
|
537 | 537 | ) |
|
538 | 538 | |
|
539 | 539 | def walk(self, match): |
|
540 | 540 | """Generates matching file names. |
|
541 | 541 | |
|
542 | 542 | Equivalent to manifest.matches(match).iterkeys(), but without creating |
|
543 | 543 | an entirely new manifest. |
|
544 | 544 | |
|
545 | 545 | It also reports nonexistent files by marking them bad with match.bad(). |
|
546 | 546 | """ |
|
547 | 547 | if match.always(): |
|
548 | 548 | for f in iter(self): |
|
549 | 549 | yield f |
|
550 | 550 | return |
|
551 | 551 | |
|
552 | 552 | fset = set(match.files()) |
|
553 | 553 | |
|
554 | 554 | # avoid the entire walk if we're only looking for specific files |
|
555 | 555 | if self._filesfastpath(match): |
|
556 | 556 | for fn in sorted(fset): |
|
557 | 557 | if fn in self: |
|
558 | 558 | yield fn |
|
559 | 559 | return |
|
560 | 560 | |
|
561 | 561 | for fn in self: |
|
562 | 562 | if fn in fset: |
|
563 | 563 | # specified pattern is the exact name |
|
564 | 564 | fset.remove(fn) |
|
565 | 565 | if match(fn): |
|
566 | 566 | yield fn |
|
567 | 567 | |
|
568 | 568 | # for dirstate.walk, files=[''] means "walk the whole tree". |
|
569 | 569 | # follow that here, too |
|
570 | 570 | fset.discard(b'') |
|
571 | 571 | |
|
572 | 572 | for fn in sorted(fset): |
|
573 | 573 | if not self.hasdir(fn): |
|
574 | 574 | match.bad(fn, None) |
|
575 | 575 | |
|
576 | 576 | def _matches(self, match): |
|
577 | 577 | '''generate a new manifest filtered by the match argument''' |
|
578 | 578 | if match.always(): |
|
579 | 579 | return self.copy() |
|
580 | 580 | |
|
581 | 581 | if self._filesfastpath(match): |
|
582 | 582 | m = manifestdict() |
|
583 | 583 | lm = self._lm |
|
584 | 584 | for fn in match.files(): |
|
585 | 585 | if fn in lm: |
|
586 | 586 | m._lm[fn] = lm[fn] |
|
587 | 587 | return m |
|
588 | 588 | |
|
589 | 589 | m = manifestdict() |
|
590 | 590 | m._lm = self._lm.filtercopy(match) |
|
591 | 591 | return m |
|
592 | 592 | |
|
593 | 593 | def diff(self, m2, match=None, clean=False): |
|
594 | 594 | """Finds changes between the current manifest and m2. |
|
595 | 595 | |
|
596 | 596 | Args: |
|
597 | 597 | m2: the manifest to which this manifest should be compared. |
|
598 | 598 | clean: if true, include files unchanged between these manifests |
|
599 | 599 | with a None value in the returned dictionary. |
|
600 | 600 | |
|
601 | 601 | The result is returned as a dict with filename as key and |
|
602 | 602 | values of the form ((n1,fl1),(n2,fl2)), where n1/n2 is the |
|
603 | 603 | nodeid in the current/other manifest and fl1/fl2 is the flag |
|
604 | 604 | in the current/other manifest. Where the file does not exist, |
|
605 | 605 | the nodeid will be None and the flags will be the empty |
|
606 | 606 | string. |
|
607 | 607 | """ |
|
608 | 608 | if match: |
|
609 | 609 | m1 = self._matches(match) |
|
610 | 610 | m2 = m2._matches(match) |
|
611 | 611 | return m1.diff(m2, clean=clean) |
|
612 | 612 | return self._lm.diff(m2._lm, clean) |
|
613 | 613 | |
|
614 | 614 | def setflag(self, key, flag): |
|
615 | 615 | if flag not in _manifestflags: |
|
616 | 616 | raise TypeError(b"Invalid manifest flag set.") |
|
617 | 617 | self._lm[key] = self[key], flag |
|
618 | 618 | |
|
619 | 619 | def get(self, key, default=None): |
|
620 | 620 | try: |
|
621 | 621 | return self._lm[key][0] |
|
622 | 622 | except KeyError: |
|
623 | 623 | return default |
|
624 | 624 | |
|
625 | 625 | def flags(self, key): |
|
626 | 626 | try: |
|
627 | 627 | return self._lm[key][1] |
|
628 | 628 | except KeyError: |
|
629 | 629 | return b'' |
|
630 | 630 | |
|
631 | 631 | def copy(self): |
|
632 | 632 | c = manifestdict() |
|
633 | 633 | c._lm = self._lm.copy() |
|
634 | 634 | return c |
|
635 | 635 | |
|
636 | 636 | def items(self): |
|
637 | 637 | return (x[:2] for x in self._lm.iterentries()) |
|
638 | 638 | |
|
639 | 639 | def iteritems(self): |
|
640 | 640 | return (x[:2] for x in self._lm.iterentries()) |
|
641 | 641 | |
|
642 | 642 | def iterentries(self): |
|
643 | 643 | return self._lm.iterentries() |
|
644 | 644 | |
|
645 | 645 | def text(self): |
|
646 | 646 | # most likely uses native version |
|
647 | 647 | return self._lm.text() |
|
648 | 648 | |
|
649 | 649 | def fastdelta(self, base, changes): |
|
650 | 650 | """Given a base manifest text as a bytearray and a list of changes |
|
651 | 651 | relative to that text, compute a delta that can be used by revlog. |
|
652 | 652 | """ |
|
653 | 653 | delta = [] |
|
654 | 654 | dstart = None |
|
655 | 655 | dend = None |
|
656 | 656 | dline = [b""] |
|
657 | 657 | start = 0 |
|
658 | 658 | # zero copy representation of base as a buffer |
|
659 | 659 | addbuf = util.buffer(base) |
|
660 | 660 | |
|
661 | 661 | changes = list(changes) |
|
662 | 662 | if len(changes) < FASTDELTA_TEXTDIFF_THRESHOLD: |
|
663 | 663 | # start with a readonly loop that finds the offset of |
|
664 | 664 | # each line and creates the deltas |
|
665 | 665 | for f, todelete in changes: |
|
666 | 666 | # bs will either be the index of the item or the insert point |
|
667 | 667 | start, end = _msearch(addbuf, f, start) |
|
668 | 668 | if not todelete: |
|
669 | 669 | h, fl = self._lm[f] |
|
670 | 670 | l = b"%s\0%s%s\n" % (f, hex(h), fl) |
|
671 | 671 | else: |
|
672 | 672 | if start == end: |
|
673 | 673 | # item we want to delete was not found, error out |
|
674 | 674 | raise AssertionError( |
|
675 | 675 | _(b"failed to remove %s from manifest") % f |
|
676 | 676 | ) |
|
677 | 677 | l = b"" |
|
678 | 678 | if dstart is not None and dstart <= start and dend >= start: |
|
679 | 679 | if dend < end: |
|
680 | 680 | dend = end |
|
681 | 681 | if l: |
|
682 | 682 | dline.append(l) |
|
683 | 683 | else: |
|
684 | 684 | if dstart is not None: |
|
685 | 685 | delta.append([dstart, dend, b"".join(dline)]) |
|
686 | 686 | dstart = start |
|
687 | 687 | dend = end |
|
688 | 688 | dline = [l] |
|
689 | 689 | |
|
690 | 690 | if dstart is not None: |
|
691 | 691 | delta.append([dstart, dend, b"".join(dline)]) |
|
692 | 692 | # apply the delta to the base, and get a delta for addrevision |
|
693 | 693 | deltatext, arraytext = _addlistdelta(base, delta) |
|
694 | 694 | else: |
|
695 | 695 | # For large changes, it's much cheaper to just build the text and |
|
696 | 696 | # diff it. |
|
697 | 697 | arraytext = bytearray(self.text()) |
|
698 | 698 | deltatext = mdiff.textdiff( |
|
699 | 699 | util.buffer(base), util.buffer(arraytext) |
|
700 | 700 | ) |
|
701 | 701 | |
|
702 | 702 | return arraytext, deltatext |
|
703 | 703 | |
|
704 | 704 | |
|
705 | 705 | def _msearch(m, s, lo=0, hi=None): |
|
706 | 706 | """return a tuple (start, end) that says where to find s within m. |
|
707 | 707 | |
|
708 | 708 | If the string is found m[start:end] are the line containing |
|
709 | 709 | that string. If start == end the string was not found and |
|
710 | 710 | they indicate the proper sorted insertion point. |
|
711 | 711 | |
|
712 | 712 | m should be a buffer, a memoryview or a byte string. |
|
713 | 713 | s is a byte string""" |
|
714 | 714 | |
|
715 | 715 | def advance(i, c): |
|
716 | 716 | while i < lenm and m[i : i + 1] != c: |
|
717 | 717 | i += 1 |
|
718 | 718 | return i |
|
719 | 719 | |
|
720 | 720 | if not s: |
|
721 | 721 | return (lo, lo) |
|
722 | 722 | lenm = len(m) |
|
723 | 723 | if not hi: |
|
724 | 724 | hi = lenm |
|
725 | 725 | while lo < hi: |
|
726 | 726 | mid = (lo + hi) // 2 |
|
727 | 727 | start = mid |
|
728 | 728 | while start > 0 and m[start - 1 : start] != b'\n': |
|
729 | 729 | start -= 1 |
|
730 | 730 | end = advance(start, b'\0') |
|
731 | 731 | if bytes(m[start:end]) < s: |
|
732 | 732 | # we know that after the null there are 40 bytes of sha1 |
|
733 | 733 | # this translates to the bisect lo = mid + 1 |
|
734 | 734 | lo = advance(end + 40, b'\n') + 1 |
|
735 | 735 | else: |
|
736 | 736 | # this translates to the bisect hi = mid |
|
737 | 737 | hi = start |
|
738 | 738 | end = advance(lo, b'\0') |
|
739 | 739 | found = m[lo:end] |
|
740 | 740 | if s == found: |
|
741 | 741 | # we know that after the null there are 40 bytes of sha1 |
|
742 | 742 | end = advance(end + 40, b'\n') |
|
743 | 743 | return (lo, end + 1) |
|
744 | 744 | else: |
|
745 | 745 | return (lo, lo) |
|
746 | 746 | |
|
747 | 747 | |
|
748 | 748 | def _checkforbidden(l): |
|
749 | 749 | """Check filenames for illegal characters.""" |
|
750 | 750 | for f in l: |
|
751 | 751 | if b'\n' in f or b'\r' in f: |
|
752 | 752 | raise error.StorageError( |
|
753 | 753 | _(b"'\\n' and '\\r' disallowed in filenames: %r") |
|
754 | 754 | % pycompat.bytestr(f) |
|
755 | 755 | ) |
|
756 | 756 | |
|
757 | 757 | |
|
758 | 758 | # apply the changes collected during the bisect loop to our addlist |
|
759 | 759 | # return a delta suitable for addrevision |
|
760 | 760 | def _addlistdelta(addlist, x): |
|
761 | 761 | # for large addlist arrays, building a new array is cheaper |
|
762 | 762 | # than repeatedly modifying the existing one |
|
763 | 763 | currentposition = 0 |
|
764 | 764 | newaddlist = bytearray() |
|
765 | 765 | |
|
766 | 766 | for start, end, content in x: |
|
767 | 767 | newaddlist += addlist[currentposition:start] |
|
768 | 768 | if content: |
|
769 | 769 | newaddlist += bytearray(content) |
|
770 | 770 | |
|
771 | 771 | currentposition = end |
|
772 | 772 | |
|
773 | 773 | newaddlist += addlist[currentposition:] |
|
774 | 774 | |
|
775 | 775 | deltatext = b"".join( |
|
776 | 776 | struct.pack(b">lll", start, end, len(content)) + content |
|
777 | 777 | for start, end, content in x |
|
778 | 778 | ) |
|
779 | 779 | return deltatext, newaddlist |
|
780 | 780 | |
|
781 | 781 | |
|
782 | 782 | def _splittopdir(f): |
|
783 | 783 | if b'/' in f: |
|
784 | 784 | dir, subpath = f.split(b'/', 1) |
|
785 | 785 | return dir + b'/', subpath |
|
786 | 786 | else: |
|
787 | 787 | return b'', f |
|
788 | 788 | |
|
789 | 789 | |
|
790 | 790 | _noop = lambda s: None |
|
791 | 791 | |
|
792 | 792 | |
|
793 | 793 | @interfaceutil.implementer(repository.imanifestdict) |
|
794 | 794 | class treemanifest(object): |
|
795 | 795 | def __init__(self, dir=b'', text=b''): |
|
796 | 796 | self._dir = dir |
|
797 | 797 | self._node = nullid |
|
798 | 798 | self._loadfunc = _noop |
|
799 | 799 | self._copyfunc = _noop |
|
800 | 800 | self._dirty = False |
|
801 | 801 | self._dirs = {} |
|
802 | 802 | self._lazydirs = {} |
|
803 | 803 | # Using _lazymanifest here is a little slower than plain old dicts |
|
804 | 804 | self._files = {} |
|
805 | 805 | self._flags = {} |
|
806 | 806 | if text: |
|
807 | 807 | |
|
808 | 808 | def readsubtree(subdir, subm): |
|
809 | 809 | raise AssertionError( |
|
810 | 810 | b'treemanifest constructor only accepts flat manifests' |
|
811 | 811 | ) |
|
812 | 812 | |
|
813 | 813 | self.parse(text, readsubtree) |
|
814 | 814 | self._dirty = True # Mark flat manifest dirty after parsing |
|
815 | 815 | |
|
816 | 816 | def _subpath(self, path): |
|
817 | 817 | return self._dir + path |
|
818 | 818 | |
|
819 | 819 | def _loadalllazy(self): |
|
820 | 820 | selfdirs = self._dirs |
|
821 | 821 | subpath = self._subpath |
|
822 | 822 | for d, (node, readsubtree, docopy) in pycompat.iteritems( |
|
823 | 823 | self._lazydirs |
|
824 | 824 | ): |
|
825 | 825 | if docopy: |
|
826 | 826 | selfdirs[d] = readsubtree(subpath(d), node).copy() |
|
827 | 827 | else: |
|
828 | 828 | selfdirs[d] = readsubtree(subpath(d), node) |
|
829 | 829 | self._lazydirs = {} |
|
830 | 830 | |
|
831 | 831 | def _loadlazy(self, d): |
|
832 | 832 | v = self._lazydirs.get(d) |
|
833 | 833 | if v: |
|
834 | 834 | node, readsubtree, docopy = v |
|
835 | 835 | if docopy: |
|
836 | 836 | self._dirs[d] = readsubtree(self._subpath(d), node).copy() |
|
837 | 837 | else: |
|
838 | 838 | self._dirs[d] = readsubtree(self._subpath(d), node) |
|
839 | 839 | del self._lazydirs[d] |
|
840 | 840 | |
|
841 | 841 | def _loadchildrensetlazy(self, visit): |
|
842 | 842 | if not visit: |
|
843 | 843 | return None |
|
844 | 844 | if visit == b'all' or visit == b'this': |
|
845 | 845 | self._loadalllazy() |
|
846 | 846 | return None |
|
847 | 847 | |
|
848 | 848 | loadlazy = self._loadlazy |
|
849 | 849 | for k in visit: |
|
850 | 850 | loadlazy(k + b'/') |
|
851 | 851 | return visit |
|
852 | 852 | |
|
853 | 853 | def _loaddifflazy(self, t1, t2): |
|
854 | 854 | """load items in t1 and t2 if they're needed for diffing. |
|
855 | 855 | |
|
856 | 856 | The criteria currently is: |
|
857 | 857 | - if it's not present in _lazydirs in either t1 or t2, load it in the |
|
858 | 858 | other (it may already be loaded or it may not exist, doesn't matter) |
|
859 | 859 | - if it's present in _lazydirs in both, compare the nodeid; if it |
|
860 | 860 | differs, load it in both |
|
861 | 861 | """ |
|
862 | 862 | toloadlazy = [] |
|
863 | 863 | for d, v1 in pycompat.iteritems(t1._lazydirs): |
|
864 | 864 | v2 = t2._lazydirs.get(d) |
|
865 | 865 | if not v2 or v2[0] != v1[0]: |
|
866 | 866 | toloadlazy.append(d) |
|
867 | 867 | for d, v1 in pycompat.iteritems(t2._lazydirs): |
|
868 | 868 | if d not in t1._lazydirs: |
|
869 | 869 | toloadlazy.append(d) |
|
870 | 870 | |
|
871 | 871 | for d in toloadlazy: |
|
872 | 872 | t1._loadlazy(d) |
|
873 | 873 | t2._loadlazy(d) |
|
874 | 874 | |
|
875 | 875 | def __len__(self): |
|
876 | 876 | self._load() |
|
877 | 877 | size = len(self._files) |
|
878 | 878 | self._loadalllazy() |
|
879 | 879 | for m in self._dirs.values(): |
|
880 | 880 | size += m.__len__() |
|
881 | 881 | return size |
|
882 | 882 | |
|
883 | 883 | def __nonzero__(self): |
|
884 | 884 | # Faster than "__len() != 0" since it avoids loading sub-manifests |
|
885 | 885 | return not self._isempty() |
|
886 | 886 | |
|
887 | 887 | __bool__ = __nonzero__ |
|
888 | 888 | |
|
889 | 889 | def _isempty(self): |
|
890 | 890 | self._load() # for consistency; already loaded by all callers |
|
891 | 891 | # See if we can skip loading everything. |
|
892 | 892 | if self._files or ( |
|
893 | 893 | self._dirs and any(not m._isempty() for m in self._dirs.values()) |
|
894 | 894 | ): |
|
895 | 895 | return False |
|
896 | 896 | self._loadalllazy() |
|
897 | 897 | return not self._dirs or all(m._isempty() for m in self._dirs.values()) |
|
898 | 898 | |
|
899 | 899 | @encoding.strmethod |
|
900 | 900 | def __repr__(self): |
|
901 | 901 | return ( |
|
902 | 902 | b'<treemanifest dir=%s, node=%s, loaded=%r, dirty=%r at 0x%x>' |
|
903 | 903 | % ( |
|
904 | 904 | self._dir, |
|
905 | 905 | hex(self._node), |
|
906 | 906 | bool(self._loadfunc is _noop), |
|
907 | 907 | self._dirty, |
|
908 | 908 | id(self), |
|
909 | 909 | ) |
|
910 | 910 | ) |
|
911 | 911 | |
|
912 | 912 | def dir(self): |
|
913 | 913 | """The directory that this tree manifest represents, including a |
|
914 | 914 | trailing '/'. Empty string for the repo root directory.""" |
|
915 | 915 | return self._dir |
|
916 | 916 | |
|
917 | 917 | def node(self): |
|
918 | 918 | """This node of this instance. nullid for unsaved instances. Should |
|
919 | 919 | be updated when the instance is read or written from a revlog. |
|
920 | 920 | """ |
|
921 | 921 | assert not self._dirty |
|
922 | 922 | return self._node |
|
923 | 923 | |
|
924 | 924 | def setnode(self, node): |
|
925 | 925 | self._node = node |
|
926 | 926 | self._dirty = False |
|
927 | 927 | |
|
928 | 928 | def iterentries(self): |
|
929 | 929 | self._load() |
|
930 | 930 | self._loadalllazy() |
|
931 | 931 | for p, n in sorted( |
|
932 | 932 | itertools.chain(self._dirs.items(), self._files.items()) |
|
933 | 933 | ): |
|
934 | 934 | if p in self._files: |
|
935 | 935 | yield self._subpath(p), n, self._flags.get(p, b'') |
|
936 | 936 | else: |
|
937 | 937 | for x in n.iterentries(): |
|
938 | 938 | yield x |
|
939 | 939 | |
|
940 | 940 | def items(self): |
|
941 | 941 | self._load() |
|
942 | 942 | self._loadalllazy() |
|
943 | 943 | for p, n in sorted( |
|
944 | 944 | itertools.chain(self._dirs.items(), self._files.items()) |
|
945 | 945 | ): |
|
946 | 946 | if p in self._files: |
|
947 | 947 | yield self._subpath(p), n |
|
948 | 948 | else: |
|
949 | 949 | for f, sn in pycompat.iteritems(n): |
|
950 | 950 | yield f, sn |
|
951 | 951 | |
|
952 | 952 | iteritems = items |
|
953 | 953 | |
|
954 | 954 | def iterkeys(self): |
|
955 | 955 | self._load() |
|
956 | 956 | self._loadalllazy() |
|
957 | 957 | for p in sorted(itertools.chain(self._dirs, self._files)): |
|
958 | 958 | if p in self._files: |
|
959 | 959 | yield self._subpath(p) |
|
960 | 960 | else: |
|
961 | 961 | for f in self._dirs[p]: |
|
962 | 962 | yield f |
|
963 | 963 | |
|
964 | 964 | def keys(self): |
|
965 | 965 | return list(self.iterkeys()) |
|
966 | 966 | |
|
967 | 967 | def __iter__(self): |
|
968 | 968 | return self.iterkeys() |
|
969 | 969 | |
|
970 | 970 | def __contains__(self, f): |
|
971 | 971 | if f is None: |
|
972 | 972 | return False |
|
973 | 973 | self._load() |
|
974 | 974 | dir, subpath = _splittopdir(f) |
|
975 | 975 | if dir: |
|
976 | 976 | self._loadlazy(dir) |
|
977 | 977 | |
|
978 | 978 | if dir not in self._dirs: |
|
979 | 979 | return False |
|
980 | 980 | |
|
981 | 981 | return self._dirs[dir].__contains__(subpath) |
|
982 | 982 | else: |
|
983 | 983 | return f in self._files |
|
984 | 984 | |
|
985 | 985 | def get(self, f, default=None): |
|
986 | 986 | self._load() |
|
987 | 987 | dir, subpath = _splittopdir(f) |
|
988 | 988 | if dir: |
|
989 | 989 | self._loadlazy(dir) |
|
990 | 990 | |
|
991 | 991 | if dir not in self._dirs: |
|
992 | 992 | return default |
|
993 | 993 | return self._dirs[dir].get(subpath, default) |
|
994 | 994 | else: |
|
995 | 995 | return self._files.get(f, default) |
|
996 | 996 | |
|
997 | 997 | def __getitem__(self, f): |
|
998 | 998 | self._load() |
|
999 | 999 | dir, subpath = _splittopdir(f) |
|
1000 | 1000 | if dir: |
|
1001 | 1001 | self._loadlazy(dir) |
|
1002 | 1002 | |
|
1003 | 1003 | return self._dirs[dir].__getitem__(subpath) |
|
1004 | 1004 | else: |
|
1005 | 1005 | return self._files[f] |
|
1006 | 1006 | |
|
1007 | 1007 | def flags(self, f): |
|
1008 | 1008 | self._load() |
|
1009 | 1009 | dir, subpath = _splittopdir(f) |
|
1010 | 1010 | if dir: |
|
1011 | 1011 | self._loadlazy(dir) |
|
1012 | 1012 | |
|
1013 | 1013 | if dir not in self._dirs: |
|
1014 | 1014 | return b'' |
|
1015 | 1015 | return self._dirs[dir].flags(subpath) |
|
1016 | 1016 | else: |
|
1017 | 1017 | if f in self._lazydirs or f in self._dirs: |
|
1018 | 1018 | return b'' |
|
1019 | 1019 | return self._flags.get(f, b'') |
|
1020 | 1020 | |
|
1021 | 1021 | def find(self, f): |
|
1022 | 1022 | self._load() |
|
1023 | 1023 | dir, subpath = _splittopdir(f) |
|
1024 | 1024 | if dir: |
|
1025 | 1025 | self._loadlazy(dir) |
|
1026 | 1026 | |
|
1027 | 1027 | return self._dirs[dir].find(subpath) |
|
1028 | 1028 | else: |
|
1029 | 1029 | return self._files[f], self._flags.get(f, b'') |
|
1030 | 1030 | |
|
1031 | 1031 | def __delitem__(self, f): |
|
1032 | 1032 | self._load() |
|
1033 | 1033 | dir, subpath = _splittopdir(f) |
|
1034 | 1034 | if dir: |
|
1035 | 1035 | self._loadlazy(dir) |
|
1036 | 1036 | |
|
1037 | 1037 | self._dirs[dir].__delitem__(subpath) |
|
1038 | 1038 | # If the directory is now empty, remove it |
|
1039 | 1039 | if self._dirs[dir]._isempty(): |
|
1040 | 1040 | del self._dirs[dir] |
|
1041 | 1041 | else: |
|
1042 | 1042 | del self._files[f] |
|
1043 | 1043 | if f in self._flags: |
|
1044 | 1044 | del self._flags[f] |
|
1045 | 1045 | self._dirty = True |
|
1046 | 1046 | |
|
1047 | 1047 | def __setitem__(self, f, n): |
|
1048 | 1048 | assert n is not None |
|
1049 | 1049 | self._load() |
|
1050 | 1050 | dir, subpath = _splittopdir(f) |
|
1051 | 1051 | if dir: |
|
1052 | 1052 | self._loadlazy(dir) |
|
1053 | 1053 | if dir not in self._dirs: |
|
1054 | 1054 | self._dirs[dir] = treemanifest(self._subpath(dir)) |
|
1055 | 1055 | self._dirs[dir].__setitem__(subpath, n) |
|
1056 | 1056 | else: |
|
1057 | 1057 | # manifest nodes are either 20 bytes or 32 bytes, |
|
1058 | 1058 | # depending on the hash in use. Assert this as historically |
|
1059 | 1059 | # sometimes extra bytes were added. |
|
1060 | 1060 | assert len(n) in (20, 32) |
|
1061 | 1061 | self._files[f] = n |
|
1062 | 1062 | self._dirty = True |
|
1063 | 1063 | |
|
1064 | 1064 | def _load(self): |
|
1065 | 1065 | if self._loadfunc is not _noop: |
|
1066 | 1066 | lf, self._loadfunc = self._loadfunc, _noop |
|
1067 | 1067 | lf(self) |
|
1068 | 1068 | elif self._copyfunc is not _noop: |
|
1069 | 1069 | cf, self._copyfunc = self._copyfunc, _noop |
|
1070 | 1070 | cf(self) |
|
1071 | 1071 | |
|
1072 | 1072 | def setflag(self, f, flags): |
|
1073 | 1073 | """Set the flags (symlink, executable) for path f.""" |
|
1074 | 1074 | if flags not in _manifestflags: |
|
1075 | 1075 | raise TypeError(b"Invalid manifest flag set.") |
|
1076 | 1076 | self._load() |
|
1077 | 1077 | dir, subpath = _splittopdir(f) |
|
1078 | 1078 | if dir: |
|
1079 | 1079 | self._loadlazy(dir) |
|
1080 | 1080 | if dir not in self._dirs: |
|
1081 | 1081 | self._dirs[dir] = treemanifest(self._subpath(dir)) |
|
1082 | 1082 | self._dirs[dir].setflag(subpath, flags) |
|
1083 | 1083 | else: |
|
1084 | 1084 | self._flags[f] = flags |
|
1085 | 1085 | self._dirty = True |
|
1086 | 1086 | |
|
1087 | 1087 | def copy(self): |
|
1088 | 1088 | copy = treemanifest(self._dir) |
|
1089 | 1089 | copy._node = self._node |
|
1090 | 1090 | copy._dirty = self._dirty |
|
1091 | 1091 | if self._copyfunc is _noop: |
|
1092 | 1092 | |
|
1093 | 1093 | def _copyfunc(s): |
|
1094 | 1094 | self._load() |
|
1095 | 1095 | s._lazydirs = { |
|
1096 | 1096 | d: (n, r, True) |
|
1097 | 1097 | for d, (n, r, c) in pycompat.iteritems(self._lazydirs) |
|
1098 | 1098 | } |
|
1099 | 1099 | sdirs = s._dirs |
|
1100 | 1100 | for d, v in pycompat.iteritems(self._dirs): |
|
1101 | 1101 | sdirs[d] = v.copy() |
|
1102 | 1102 | s._files = dict.copy(self._files) |
|
1103 | 1103 | s._flags = dict.copy(self._flags) |
|
1104 | 1104 | |
|
1105 | 1105 | if self._loadfunc is _noop: |
|
1106 | 1106 | _copyfunc(copy) |
|
1107 | 1107 | else: |
|
1108 | 1108 | copy._copyfunc = _copyfunc |
|
1109 | 1109 | else: |
|
1110 | 1110 | copy._copyfunc = self._copyfunc |
|
1111 | 1111 | return copy |
|
1112 | 1112 | |
|
1113 | 1113 | def filesnotin(self, m2, match=None): |
|
1114 | 1114 | '''Set of files in this manifest that are not in the other''' |
|
1115 | 1115 | if match and not match.always(): |
|
1116 | 1116 | m1 = self._matches(match) |
|
1117 | 1117 | m2 = m2._matches(match) |
|
1118 | 1118 | return m1.filesnotin(m2) |
|
1119 | 1119 | |
|
1120 | 1120 | files = set() |
|
1121 | 1121 | |
|
1122 | 1122 | def _filesnotin(t1, t2): |
|
1123 | 1123 | if t1._node == t2._node and not t1._dirty and not t2._dirty: |
|
1124 | 1124 | return |
|
1125 | 1125 | t1._load() |
|
1126 | 1126 | t2._load() |
|
1127 | 1127 | self._loaddifflazy(t1, t2) |
|
1128 | 1128 | for d, m1 in pycompat.iteritems(t1._dirs): |
|
1129 | 1129 | if d in t2._dirs: |
|
1130 | 1130 | m2 = t2._dirs[d] |
|
1131 | 1131 | _filesnotin(m1, m2) |
|
1132 | 1132 | else: |
|
1133 | 1133 | files.update(m1.iterkeys()) |
|
1134 | 1134 | |
|
1135 | 1135 | for fn in t1._files: |
|
1136 | 1136 | if fn not in t2._files: |
|
1137 | 1137 | files.add(t1._subpath(fn)) |
|
1138 | 1138 | |
|
1139 | 1139 | _filesnotin(self, m2) |
|
1140 | 1140 | return files |
|
1141 | 1141 | |
|
1142 | 1142 | @propertycache |
|
1143 | 1143 | def _alldirs(self): |
|
1144 | 1144 | return pathutil.dirs(self) |
|
1145 | 1145 | |
|
1146 | 1146 | def dirs(self): |
|
1147 | 1147 | return self._alldirs |
|
1148 | 1148 | |
|
1149 | 1149 | def hasdir(self, dir): |
|
1150 | 1150 | self._load() |
|
1151 | 1151 | topdir, subdir = _splittopdir(dir) |
|
1152 | 1152 | if topdir: |
|
1153 | 1153 | self._loadlazy(topdir) |
|
1154 | 1154 | if topdir in self._dirs: |
|
1155 | 1155 | return self._dirs[topdir].hasdir(subdir) |
|
1156 | 1156 | return False |
|
1157 | 1157 | dirslash = dir + b'/' |
|
1158 | 1158 | return dirslash in self._dirs or dirslash in self._lazydirs |
|
1159 | 1159 | |
|
1160 | 1160 | def walk(self, match): |
|
1161 | 1161 | """Generates matching file names. |
|
1162 | 1162 | |
|
1163 | 1163 | It also reports nonexistent files by marking them bad with match.bad(). |
|
1164 | 1164 | """ |
|
1165 | 1165 | if match.always(): |
|
1166 | 1166 | for f in iter(self): |
|
1167 | 1167 | yield f |
|
1168 | 1168 | return |
|
1169 | 1169 | |
|
1170 | 1170 | fset = set(match.files()) |
|
1171 | 1171 | |
|
1172 | 1172 | for fn in self._walk(match): |
|
1173 | 1173 | if fn in fset: |
|
1174 | 1174 | # specified pattern is the exact name |
|
1175 | 1175 | fset.remove(fn) |
|
1176 | 1176 | yield fn |
|
1177 | 1177 | |
|
1178 | 1178 | # for dirstate.walk, files=[''] means "walk the whole tree". |
|
1179 | 1179 | # follow that here, too |
|
1180 | 1180 | fset.discard(b'') |
|
1181 | 1181 | |
|
1182 | 1182 | for fn in sorted(fset): |
|
1183 | 1183 | if not self.hasdir(fn): |
|
1184 | 1184 | match.bad(fn, None) |
|
1185 | 1185 | |
|
1186 | 1186 | def _walk(self, match): |
|
1187 | 1187 | '''Recursively generates matching file names for walk().''' |
|
1188 | 1188 | visit = match.visitchildrenset(self._dir[:-1]) |
|
1189 | 1189 | if not visit: |
|
1190 | 1190 | return |
|
1191 | 1191 | |
|
1192 | 1192 | # yield this dir's files and walk its submanifests |
|
1193 | 1193 | self._load() |
|
1194 | 1194 | visit = self._loadchildrensetlazy(visit) |
|
1195 | 1195 | for p in sorted(list(self._dirs) + list(self._files)): |
|
1196 | 1196 | if p in self._files: |
|
1197 | 1197 | fullp = self._subpath(p) |
|
1198 | 1198 | if match(fullp): |
|
1199 | 1199 | yield fullp |
|
1200 | 1200 | else: |
|
1201 | 1201 | if not visit or p[:-1] in visit: |
|
1202 | 1202 | for f in self._dirs[p]._walk(match): |
|
1203 | 1203 | yield f |
|
1204 | 1204 | |
|
1205 | 1205 | def _matches(self, match): |
|
1206 | 1206 | """recursively generate a new manifest filtered by the match argument.""" |
|
1207 | 1207 | if match.always(): |
|
1208 | 1208 | return self.copy() |
|
1209 | 1209 | return self._matches_inner(match) |
|
1210 | 1210 | |
|
1211 | 1211 | def _matches_inner(self, match): |
|
1212 | 1212 | if match.always(): |
|
1213 | 1213 | return self.copy() |
|
1214 | 1214 | |
|
1215 | 1215 | visit = match.visitchildrenset(self._dir[:-1]) |
|
1216 | 1216 | if visit == b'all': |
|
1217 | 1217 | return self.copy() |
|
1218 | 1218 | ret = treemanifest(self._dir) |
|
1219 | 1219 | if not visit: |
|
1220 | 1220 | return ret |
|
1221 | 1221 | |
|
1222 | 1222 | self._load() |
|
1223 | 1223 | for fn in self._files: |
|
1224 | 1224 | # While visitchildrenset *usually* lists only subdirs, this is |
|
1225 | 1225 | # actually up to the matcher and may have some files in the set(). |
|
1226 | 1226 | # If visit == 'this', we should obviously look at the files in this |
|
1227 | 1227 | # directory; if visit is a set, and fn is in it, we should inspect |
|
1228 | 1228 | # fn (but no need to inspect things not in the set). |
|
1229 | 1229 | if visit != b'this' and fn not in visit: |
|
1230 | 1230 | continue |
|
1231 | 1231 | fullp = self._subpath(fn) |
|
1232 | 1232 | # visitchildrenset isn't perfect, we still need to call the regular |
|
1233 | 1233 | # matcher code to further filter results. |
|
1234 | 1234 | if not match(fullp): |
|
1235 | 1235 | continue |
|
1236 | 1236 | ret._files[fn] = self._files[fn] |
|
1237 | 1237 | if fn in self._flags: |
|
1238 | 1238 | ret._flags[fn] = self._flags[fn] |
|
1239 | 1239 | |
|
1240 | 1240 | visit = self._loadchildrensetlazy(visit) |
|
1241 | 1241 | for dir, subm in pycompat.iteritems(self._dirs): |
|
1242 | 1242 | if visit and dir[:-1] not in visit: |
|
1243 | 1243 | continue |
|
1244 | 1244 | m = subm._matches_inner(match) |
|
1245 | 1245 | if not m._isempty(): |
|
1246 | 1246 | ret._dirs[dir] = m |
|
1247 | 1247 | |
|
1248 | 1248 | if not ret._isempty(): |
|
1249 | 1249 | ret._dirty = True |
|
1250 | 1250 | return ret |
|
1251 | 1251 | |
|
1252 | 1252 | def fastdelta(self, base, changes): |
|
1253 | 1253 | raise FastdeltaUnavailable() |
|
1254 | 1254 | |
|
1255 | 1255 | def diff(self, m2, match=None, clean=False): |
|
1256 | 1256 | """Finds changes between the current manifest and m2. |
|
1257 | 1257 | |
|
1258 | 1258 | Args: |
|
1259 | 1259 | m2: the manifest to which this manifest should be compared. |
|
1260 | 1260 | clean: if true, include files unchanged between these manifests |
|
1261 | 1261 | with a None value in the returned dictionary. |
|
1262 | 1262 | |
|
1263 | 1263 | The result is returned as a dict with filename as key and |
|
1264 | 1264 | values of the form ((n1,fl1),(n2,fl2)), where n1/n2 is the |
|
1265 | 1265 | nodeid in the current/other manifest and fl1/fl2 is the flag |
|
1266 | 1266 | in the current/other manifest. Where the file does not exist, |
|
1267 | 1267 | the nodeid will be None and the flags will be the empty |
|
1268 | 1268 | string. |
|
1269 | 1269 | """ |
|
1270 | 1270 | if match and not match.always(): |
|
1271 | 1271 | m1 = self._matches(match) |
|
1272 | 1272 | m2 = m2._matches(match) |
|
1273 | 1273 | return m1.diff(m2, clean=clean) |
|
1274 | 1274 | result = {} |
|
1275 | 1275 | emptytree = treemanifest() |
|
1276 | 1276 | |
|
1277 | 1277 | def _iterativediff(t1, t2, stack): |
|
1278 | 1278 | """compares two tree manifests and append new tree-manifests which |
|
1279 | 1279 | needs to be compared to stack""" |
|
1280 | 1280 | if t1._node == t2._node and not t1._dirty and not t2._dirty: |
|
1281 | 1281 | return |
|
1282 | 1282 | t1._load() |
|
1283 | 1283 | t2._load() |
|
1284 | 1284 | self._loaddifflazy(t1, t2) |
|
1285 | 1285 | |
|
1286 | 1286 | for d, m1 in pycompat.iteritems(t1._dirs): |
|
1287 | 1287 | m2 = t2._dirs.get(d, emptytree) |
|
1288 | 1288 | stack.append((m1, m2)) |
|
1289 | 1289 | |
|
1290 | 1290 | for d, m2 in pycompat.iteritems(t2._dirs): |
|
1291 | 1291 | if d not in t1._dirs: |
|
1292 | 1292 | stack.append((emptytree, m2)) |
|
1293 | 1293 | |
|
1294 | 1294 | for fn, n1 in pycompat.iteritems(t1._files): |
|
1295 | 1295 | fl1 = t1._flags.get(fn, b'') |
|
1296 | 1296 | n2 = t2._files.get(fn, None) |
|
1297 | 1297 | fl2 = t2._flags.get(fn, b'') |
|
1298 | 1298 | if n1 != n2 or fl1 != fl2: |
|
1299 | 1299 | result[t1._subpath(fn)] = ((n1, fl1), (n2, fl2)) |
|
1300 | 1300 | elif clean: |
|
1301 | 1301 | result[t1._subpath(fn)] = None |
|
1302 | 1302 | |
|
1303 | 1303 | for fn, n2 in pycompat.iteritems(t2._files): |
|
1304 | 1304 | if fn not in t1._files: |
|
1305 | 1305 | fl2 = t2._flags.get(fn, b'') |
|
1306 | 1306 | result[t2._subpath(fn)] = ((None, b''), (n2, fl2)) |
|
1307 | 1307 | |
|
1308 | 1308 | stackls = [] |
|
1309 | 1309 | _iterativediff(self, m2, stackls) |
|
1310 | 1310 | while stackls: |
|
1311 | 1311 | t1, t2 = stackls.pop() |
|
1312 | 1312 | # stackls is populated in the function call |
|
1313 | 1313 | _iterativediff(t1, t2, stackls) |
|
1314 | 1314 | return result |
|
1315 | 1315 | |
|
1316 | 1316 | def unmodifiedsince(self, m2): |
|
1317 | 1317 | return not self._dirty and not m2._dirty and self._node == m2._node |
|
1318 | 1318 | |
|
1319 | 1319 | def parse(self, text, readsubtree): |
|
1320 | 1320 | selflazy = self._lazydirs |
|
1321 | 1321 | for f, n, fl in _parse(text): |
|
1322 | 1322 | if fl == b't': |
|
1323 | 1323 | f = f + b'/' |
|
1324 | 1324 | # False below means "doesn't need to be copied" and can use the |
|
1325 | 1325 | # cached value from readsubtree directly. |
|
1326 | 1326 | selflazy[f] = (n, readsubtree, False) |
|
1327 | 1327 | elif b'/' in f: |
|
1328 | 1328 | # This is a flat manifest, so use __setitem__ and setflag rather |
|
1329 | 1329 | # than assigning directly to _files and _flags, so we can |
|
1330 | 1330 | # assign a path in a subdirectory, and to mark dirty (compared |
|
1331 | 1331 | # to nullid). |
|
1332 | 1332 | self[f] = n |
|
1333 | 1333 | if fl: |
|
1334 | 1334 | self.setflag(f, fl) |
|
1335 | 1335 | else: |
|
1336 | 1336 | # Assigning to _files and _flags avoids marking as dirty, |
|
1337 | 1337 | # and should be a little faster. |
|
1338 | 1338 | self._files[f] = n |
|
1339 | 1339 | if fl: |
|
1340 | 1340 | self._flags[f] = fl |
|
1341 | 1341 | |
|
1342 | 1342 | def text(self): |
|
1343 | 1343 | """Get the full data of this manifest as a bytestring.""" |
|
1344 | 1344 | self._load() |
|
1345 | 1345 | return _text(self.iterentries()) |
|
1346 | 1346 | |
|
1347 | 1347 | def dirtext(self): |
|
1348 | 1348 | """Get the full data of this directory as a bytestring. Make sure that |
|
1349 | 1349 | any submanifests have been written first, so their nodeids are correct. |
|
1350 | 1350 | """ |
|
1351 | 1351 | self._load() |
|
1352 | 1352 | flags = self.flags |
|
1353 | 1353 | lazydirs = [ |
|
1354 | 1354 | (d[:-1], v[0], b't') for d, v in pycompat.iteritems(self._lazydirs) |
|
1355 | 1355 | ] |
|
1356 | 1356 | dirs = [(d[:-1], self._dirs[d]._node, b't') for d in self._dirs] |
|
1357 | 1357 | files = [(f, self._files[f], flags(f)) for f in self._files] |
|
1358 | 1358 | return _text(sorted(dirs + files + lazydirs)) |
|
1359 | 1359 | |
|
1360 | 1360 | def read(self, gettext, readsubtree): |
|
1361 | 1361 | def _load_for_read(s): |
|
1362 | 1362 | s.parse(gettext(), readsubtree) |
|
1363 | 1363 | s._dirty = False |
|
1364 | 1364 | |
|
1365 | 1365 | self._loadfunc = _load_for_read |
|
1366 | 1366 | |
|
1367 | 1367 | def writesubtrees(self, m1, m2, writesubtree, match): |
|
1368 | 1368 | self._load() # for consistency; should never have any effect here |
|
1369 | 1369 | m1._load() |
|
1370 | 1370 | m2._load() |
|
1371 | 1371 | emptytree = treemanifest() |
|
1372 | 1372 | |
|
1373 | 1373 | def getnode(m, d): |
|
1374 | 1374 | ld = m._lazydirs.get(d) |
|
1375 | 1375 | if ld: |
|
1376 | 1376 | return ld[0] |
|
1377 | 1377 | return m._dirs.get(d, emptytree)._node |
|
1378 | 1378 | |
|
1379 | 1379 | # let's skip investigating things that `match` says we do not need. |
|
1380 | 1380 | visit = match.visitchildrenset(self._dir[:-1]) |
|
1381 | 1381 | visit = self._loadchildrensetlazy(visit) |
|
1382 | 1382 | if visit == b'this' or visit == b'all': |
|
1383 | 1383 | visit = None |
|
1384 | 1384 | for d, subm in pycompat.iteritems(self._dirs): |
|
1385 | 1385 | if visit and d[:-1] not in visit: |
|
1386 | 1386 | continue |
|
1387 | 1387 | subp1 = getnode(m1, d) |
|
1388 | 1388 | subp2 = getnode(m2, d) |
|
1389 | 1389 | if subp1 == nullid: |
|
1390 | 1390 | subp1, subp2 = subp2, subp1 |
|
1391 | 1391 | writesubtree(subm, subp1, subp2, match) |
|
1392 | 1392 | |
|
1393 | 1393 | def walksubtrees(self, matcher=None): |
|
1394 | 1394 | """Returns an iterator of the subtrees of this manifest, including this |
|
1395 | 1395 | manifest itself. |
|
1396 | 1396 | |
|
1397 | 1397 | If `matcher` is provided, it only returns subtrees that match. |
|
1398 | 1398 | """ |
|
1399 | 1399 | if matcher and not matcher.visitdir(self._dir[:-1]): |
|
1400 | 1400 | return |
|
1401 | 1401 | if not matcher or matcher(self._dir[:-1]): |
|
1402 | 1402 | yield self |
|
1403 | 1403 | |
|
1404 | 1404 | self._load() |
|
1405 | 1405 | # OPT: use visitchildrenset to avoid loading everything. |
|
1406 | 1406 | self._loadalllazy() |
|
1407 | 1407 | for d, subm in pycompat.iteritems(self._dirs): |
|
1408 | 1408 | for subtree in subm.walksubtrees(matcher=matcher): |
|
1409 | 1409 | yield subtree |
|
1410 | 1410 | |
|
1411 | 1411 | |
|
1412 | 1412 | class manifestfulltextcache(util.lrucachedict): |
|
1413 | 1413 | """File-backed LRU cache for the manifest cache |
|
1414 | 1414 | |
|
1415 | 1415 | File consists of entries, up to EOF: |
|
1416 | 1416 | |
|
1417 | 1417 | - 20 bytes node, 4 bytes length, <length> manifest data |
|
1418 | 1418 | |
|
1419 | 1419 | These are written in reverse cache order (oldest to newest). |
|
1420 | 1420 | |
|
1421 | 1421 | """ |
|
1422 | 1422 | |
|
1423 | 1423 | _file = b'manifestfulltextcache' |
|
1424 | 1424 | |
|
1425 | 1425 | def __init__(self, max): |
|
1426 | 1426 | super(manifestfulltextcache, self).__init__(max) |
|
1427 | 1427 | self._dirty = False |
|
1428 | 1428 | self._read = False |
|
1429 | 1429 | self._opener = None |
|
1430 | 1430 | |
|
1431 | 1431 | def read(self): |
|
1432 | 1432 | if self._read or self._opener is None: |
|
1433 | 1433 | return |
|
1434 | 1434 | |
|
1435 | 1435 | try: |
|
1436 | 1436 | with self._opener(self._file) as fp: |
|
1437 | 1437 | set = super(manifestfulltextcache, self).__setitem__ |
|
1438 | 1438 | # ignore trailing data, this is a cache, corruption is skipped |
|
1439 | 1439 | while True: |
|
1440 | 1440 | # TODO do we need to do work here for sha1 portability? |
|
1441 | 1441 | node = fp.read(20) |
|
1442 | 1442 | if len(node) < 20: |
|
1443 | 1443 | break |
|
1444 | 1444 | try: |
|
1445 | 1445 | size = struct.unpack(b'>L', fp.read(4))[0] |
|
1446 | 1446 | except struct.error: |
|
1447 | 1447 | break |
|
1448 | 1448 | value = bytearray(fp.read(size)) |
|
1449 | 1449 | if len(value) != size: |
|
1450 | 1450 | break |
|
1451 | 1451 | set(node, value) |
|
1452 | 1452 | except IOError: |
|
1453 | 1453 | # the file is allowed to be missing |
|
1454 | 1454 | pass |
|
1455 | 1455 | |
|
1456 | 1456 | self._read = True |
|
1457 | 1457 | self._dirty = False |
|
1458 | 1458 | |
|
1459 | 1459 | def write(self): |
|
1460 | 1460 | if not self._dirty or self._opener is None: |
|
1461 | 1461 | return |
|
1462 | 1462 | # rotate backwards to the first used node |
|
1463 | 1463 | try: |
|
1464 | 1464 | with self._opener( |
|
1465 | 1465 | self._file, b'w', atomictemp=True, checkambig=True |
|
1466 | 1466 | ) as fp: |
|
1467 | 1467 | node = self._head.prev |
|
1468 | 1468 | while True: |
|
1469 | 1469 | if node.key in self._cache: |
|
1470 | 1470 | fp.write(node.key) |
|
1471 | 1471 | fp.write(struct.pack(b'>L', len(node.value))) |
|
1472 | 1472 | fp.write(node.value) |
|
1473 | 1473 | if node is self._head: |
|
1474 | 1474 | break |
|
1475 | 1475 | node = node.prev |
|
1476 | 1476 | except IOError: |
|
1477 | 1477 | # We could not write the cache (eg: permission error) |
|
1478 | 1478 | # the content can be missing. |
|
1479 | 1479 | # |
|
1480 | 1480 | # We could try harder and see if we could recreate a wcache |
|
1481 | 1481 | # directory were we coudl write too. |
|
1482 | 1482 | # |
|
1483 | 1483 | # XXX the error pass silently, having some way to issue an error |
|
1484 | 1484 | # log `ui.log` would be nice. |
|
1485 | 1485 | pass |
|
1486 | 1486 | |
|
1487 | 1487 | def __len__(self): |
|
1488 | 1488 | if not self._read: |
|
1489 | 1489 | self.read() |
|
1490 | 1490 | return super(manifestfulltextcache, self).__len__() |
|
1491 | 1491 | |
|
1492 | 1492 | def __contains__(self, k): |
|
1493 | 1493 | if not self._read: |
|
1494 | 1494 | self.read() |
|
1495 | 1495 | return super(manifestfulltextcache, self).__contains__(k) |
|
1496 | 1496 | |
|
1497 | 1497 | def __iter__(self): |
|
1498 | 1498 | if not self._read: |
|
1499 | 1499 | self.read() |
|
1500 | 1500 | return super(manifestfulltextcache, self).__iter__() |
|
1501 | 1501 | |
|
1502 | 1502 | def __getitem__(self, k): |
|
1503 | 1503 | if not self._read: |
|
1504 | 1504 | self.read() |
|
1505 | 1505 | # the cache lru order can change on read |
|
1506 | 1506 | setdirty = self._cache.get(k) is not self._head |
|
1507 | 1507 | value = super(manifestfulltextcache, self).__getitem__(k) |
|
1508 | 1508 | if setdirty: |
|
1509 | 1509 | self._dirty = True |
|
1510 | 1510 | return value |
|
1511 | 1511 | |
|
1512 | 1512 | def __setitem__(self, k, v): |
|
1513 | 1513 | if not self._read: |
|
1514 | 1514 | self.read() |
|
1515 | 1515 | super(manifestfulltextcache, self).__setitem__(k, v) |
|
1516 | 1516 | self._dirty = True |
|
1517 | 1517 | |
|
1518 | 1518 | def __delitem__(self, k): |
|
1519 | 1519 | if not self._read: |
|
1520 | 1520 | self.read() |
|
1521 | 1521 | super(manifestfulltextcache, self).__delitem__(k) |
|
1522 | 1522 | self._dirty = True |
|
1523 | 1523 | |
|
1524 | 1524 | def get(self, k, default=None): |
|
1525 | 1525 | if not self._read: |
|
1526 | 1526 | self.read() |
|
1527 | 1527 | return super(manifestfulltextcache, self).get(k, default=default) |
|
1528 | 1528 | |
|
1529 | 1529 | def clear(self, clear_persisted_data=False): |
|
1530 | 1530 | super(manifestfulltextcache, self).clear() |
|
1531 | 1531 | if clear_persisted_data: |
|
1532 | 1532 | self._dirty = True |
|
1533 | 1533 | self.write() |
|
1534 | 1534 | self._read = False |
|
1535 | 1535 | |
|
1536 | 1536 | |
|
1537 | 1537 | # and upper bound of what we expect from compression |
|
1538 | 1538 | # (real live value seems to be "3") |
|
1539 | 1539 | MAXCOMPRESSION = 3 |
|
1540 | 1540 | |
|
1541 | 1541 | |
|
1542 | 1542 | class FastdeltaUnavailable(Exception): |
|
1543 | 1543 | """Exception raised when fastdelta isn't usable on a manifest.""" |
|
1544 | 1544 | |
|
1545 | 1545 | |
|
1546 | 1546 | @interfaceutil.implementer(repository.imanifeststorage) |
|
1547 | 1547 | class manifestrevlog(object): |
|
1548 | 1548 | """A revlog that stores manifest texts. This is responsible for caching the |
|
1549 | 1549 | full-text manifest contents. |
|
1550 | 1550 | """ |
|
1551 | 1551 | |
|
1552 | 1552 | def __init__( |
|
1553 | 1553 | self, |
|
1554 | 1554 | opener, |
|
1555 | 1555 | tree=b'', |
|
1556 | 1556 | dirlogcache=None, |
|
1557 | 1557 | indexfile=None, |
|
1558 | 1558 | treemanifest=False, |
|
1559 | 1559 | ): |
|
1560 | 1560 | """Constructs a new manifest revlog |
|
1561 | 1561 | |
|
1562 | 1562 | `indexfile` - used by extensions to have two manifests at once, like |
|
1563 | 1563 | when transitioning between flatmanifeset and treemanifests. |
|
1564 | 1564 | |
|
1565 | 1565 | `treemanifest` - used to indicate this is a tree manifest revlog. Opener |
|
1566 | 1566 | options can also be used to make this a tree manifest revlog. The opener |
|
1567 | 1567 | option takes precedence, so if it is set to True, we ignore whatever |
|
1568 | 1568 | value is passed in to the constructor. |
|
1569 | 1569 | """ |
|
1570 | 1570 | # During normal operations, we expect to deal with not more than four |
|
1571 | 1571 | # revs at a time (such as during commit --amend). When rebasing large |
|
1572 | 1572 | # stacks of commits, the number can go up, hence the config knob below. |
|
1573 | 1573 | cachesize = 4 |
|
1574 | 1574 | optiontreemanifest = False |
|
1575 | 1575 | opts = getattr(opener, 'options', None) |
|
1576 | 1576 | if opts is not None: |
|
1577 | 1577 | cachesize = opts.get(b'manifestcachesize', cachesize) |
|
1578 | 1578 | optiontreemanifest = opts.get(b'treemanifest', False) |
|
1579 | 1579 | |
|
1580 | 1580 | self._treeondisk = optiontreemanifest or treemanifest |
|
1581 | 1581 | |
|
1582 | 1582 | self._fulltextcache = manifestfulltextcache(cachesize) |
|
1583 | 1583 | |
|
1584 | 1584 | if tree: |
|
1585 | 1585 | assert self._treeondisk, b'opts is %r' % opts |
|
1586 | 1586 | |
|
1587 | 1587 | if indexfile is None: |
|
1588 | 1588 | indexfile = b'00manifest.i' |
|
1589 | 1589 | if tree: |
|
1590 | 1590 | indexfile = b"meta/" + tree + indexfile |
|
1591 | 1591 | |
|
1592 | 1592 | self.tree = tree |
|
1593 | 1593 | |
|
1594 | 1594 | # The dirlogcache is kept on the root manifest log |
|
1595 | 1595 | if tree: |
|
1596 | 1596 | self._dirlogcache = dirlogcache |
|
1597 | 1597 | else: |
|
1598 | 1598 | self._dirlogcache = {b'': self} |
|
1599 | 1599 | |
|
1600 | 1600 | self._revlog = revlog.revlog( |
|
1601 | 1601 | opener, |
|
1602 | 1602 | indexfile, |
|
1603 | 1603 | # only root indexfile is cached |
|
1604 | 1604 | checkambig=not bool(tree), |
|
1605 | 1605 | mmaplargeindex=True, |
|
1606 | 1606 | upperboundcomp=MAXCOMPRESSION, |
|
1607 | 1607 | persistentnodemap=opener.options.get(b'persistent-nodemap', False), |
|
1608 | 1608 | ) |
|
1609 | 1609 | |
|
1610 | 1610 | self.index = self._revlog.index |
|
1611 | 1611 | self.version = self._revlog.version |
|
1612 | 1612 | self._generaldelta = self._revlog._generaldelta |
|
1613 | 1613 | |
|
1614 | 1614 | def _setupmanifestcachehooks(self, repo): |
|
1615 | 1615 | """Persist the manifestfulltextcache on lock release""" |
|
1616 | 1616 | if not util.safehasattr(repo, b'_wlockref'): |
|
1617 | 1617 | return |
|
1618 | 1618 | |
|
1619 | 1619 | self._fulltextcache._opener = repo.wcachevfs |
|
1620 | 1620 | if repo._currentlock(repo._wlockref) is None: |
|
1621 | 1621 | return |
|
1622 | 1622 | |
|
1623 | 1623 | reporef = weakref.ref(repo) |
|
1624 | 1624 | manifestrevlogref = weakref.ref(self) |
|
1625 | 1625 | |
|
1626 | 1626 | def persistmanifestcache(success): |
|
1627 | 1627 | # Repo is in an unknown state, do not persist. |
|
1628 | 1628 | if not success: |
|
1629 | 1629 | return |
|
1630 | 1630 | |
|
1631 | 1631 | repo = reporef() |
|
1632 | 1632 | self = manifestrevlogref() |
|
1633 | 1633 | if repo is None or self is None: |
|
1634 | 1634 | return |
|
1635 | 1635 | if repo.manifestlog.getstorage(b'') is not self: |
|
1636 | 1636 | # there's a different manifest in play now, abort |
|
1637 | 1637 | return |
|
1638 | 1638 | self._fulltextcache.write() |
|
1639 | 1639 | |
|
1640 | 1640 | repo._afterlock(persistmanifestcache) |
|
1641 | 1641 | |
|
1642 | 1642 | @property |
|
1643 | 1643 | def fulltextcache(self): |
|
1644 | 1644 | return self._fulltextcache |
|
1645 | 1645 | |
|
1646 | 1646 | def clearcaches(self, clear_persisted_data=False): |
|
1647 | 1647 | self._revlog.clearcaches() |
|
1648 | 1648 | self._fulltextcache.clear(clear_persisted_data=clear_persisted_data) |
|
1649 | 1649 | self._dirlogcache = {self.tree: self} |
|
1650 | 1650 | |
|
1651 | 1651 | def dirlog(self, d): |
|
1652 | 1652 | if d: |
|
1653 | 1653 | assert self._treeondisk |
|
1654 | 1654 | if d not in self._dirlogcache: |
|
1655 | 1655 | mfrevlog = manifestrevlog( |
|
1656 | 1656 | self.opener, d, self._dirlogcache, treemanifest=self._treeondisk |
|
1657 | 1657 | ) |
|
1658 | 1658 | self._dirlogcache[d] = mfrevlog |
|
1659 | 1659 | return self._dirlogcache[d] |
|
1660 | 1660 | |
|
1661 | 1661 | def add( |
|
1662 | 1662 | self, |
|
1663 | 1663 | m, |
|
1664 | 1664 | transaction, |
|
1665 | 1665 | link, |
|
1666 | 1666 | p1, |
|
1667 | 1667 | p2, |
|
1668 | 1668 | added, |
|
1669 | 1669 | removed, |
|
1670 | 1670 | readtree=None, |
|
1671 | 1671 | match=None, |
|
1672 | 1672 | ): |
|
1673 | 1673 | """add some manifest entry in to the manifest log |
|
1674 | 1674 | |
|
1675 | 1675 | input: |
|
1676 | 1676 | |
|
1677 | 1677 | m: the manifest dict we want to store |
|
1678 | 1678 | transaction: the open transaction |
|
1679 | 1679 | p1: manifest-node of p1 |
|
1680 | 1680 | p2: manifest-node of p2 |
|
1681 | 1681 | added: file added/changed compared to parent |
|
1682 | 1682 | removed: file removed compared to parent |
|
1683 | 1683 | |
|
1684 | 1684 | tree manifest input: |
|
1685 | 1685 | |
|
1686 | 1686 | readtree: a function to read a subtree |
|
1687 | 1687 | match: a filematcher for the subpart of the tree manifest |
|
1688 | 1688 | """ |
|
1689 | 1689 | try: |
|
1690 | 1690 | if p1 not in self.fulltextcache: |
|
1691 | 1691 | raise FastdeltaUnavailable() |
|
1692 | 1692 | # If our first parent is in the manifest cache, we can |
|
1693 | 1693 | # compute a delta here using properties we know about the |
|
1694 | 1694 | # manifest up-front, which may save time later for the |
|
1695 | 1695 | # revlog layer. |
|
1696 | 1696 | |
|
1697 | 1697 | _checkforbidden(added) |
|
1698 | 1698 | # combine the changed lists into one sorted iterator |
|
1699 | 1699 | work = heapq.merge( |
|
1700 | 1700 | [(x, False) for x in sorted(added)], |
|
1701 | 1701 | [(x, True) for x in sorted(removed)], |
|
1702 | 1702 | ) |
|
1703 | 1703 | |
|
1704 | 1704 | arraytext, deltatext = m.fastdelta(self.fulltextcache[p1], work) |
|
1705 | 1705 | cachedelta = self._revlog.rev(p1), deltatext |
|
1706 | 1706 | text = util.buffer(arraytext) |
|
1707 |
|
|
|
1707 | rev = self._revlog.addrevision( | |
|
1708 | 1708 | text, transaction, link, p1, p2, cachedelta |
|
1709 | 1709 | ) |
|
1710 | n = self._revlog.node(rev) | |
|
1710 | 1711 | except FastdeltaUnavailable: |
|
1711 | 1712 | # The first parent manifest isn't already loaded or the |
|
1712 | 1713 | # manifest implementation doesn't support fastdelta, so |
|
1713 | 1714 | # we'll just encode a fulltext of the manifest and pass |
|
1714 | 1715 | # that through to the revlog layer, and let it handle the |
|
1715 | 1716 | # delta process. |
|
1716 | 1717 | if self._treeondisk: |
|
1717 | 1718 | assert readtree, b"readtree must be set for treemanifest writes" |
|
1718 | 1719 | assert match, b"match must be specified for treemanifest writes" |
|
1719 | 1720 | m1 = readtree(self.tree, p1) |
|
1720 | 1721 | m2 = readtree(self.tree, p2) |
|
1721 | 1722 | n = self._addtree( |
|
1722 | 1723 | m, transaction, link, m1, m2, readtree, match=match |
|
1723 | 1724 | ) |
|
1724 | 1725 | arraytext = None |
|
1725 | 1726 | else: |
|
1726 | 1727 | text = m.text() |
|
1727 |
|
|
|
1728 | rev = self._revlog.addrevision(text, transaction, link, p1, p2) | |
|
1729 | n = self._revlog.node(rev) | |
|
1728 | 1730 | arraytext = bytearray(text) |
|
1729 | 1731 | |
|
1730 | 1732 | if arraytext is not None: |
|
1731 | 1733 | self.fulltextcache[n] = arraytext |
|
1732 | 1734 | |
|
1733 | 1735 | return n |
|
1734 | 1736 | |
|
1735 | 1737 | def _addtree(self, m, transaction, link, m1, m2, readtree, match): |
|
1736 | 1738 | # If the manifest is unchanged compared to one parent, |
|
1737 | 1739 | # don't write a new revision |
|
1738 | 1740 | if self.tree != b'' and ( |
|
1739 | 1741 | m.unmodifiedsince(m1) or m.unmodifiedsince(m2) |
|
1740 | 1742 | ): |
|
1741 | 1743 | return m.node() |
|
1742 | 1744 | |
|
1743 | 1745 | def writesubtree(subm, subp1, subp2, match): |
|
1744 | 1746 | sublog = self.dirlog(subm.dir()) |
|
1745 | 1747 | sublog.add( |
|
1746 | 1748 | subm, |
|
1747 | 1749 | transaction, |
|
1748 | 1750 | link, |
|
1749 | 1751 | subp1, |
|
1750 | 1752 | subp2, |
|
1751 | 1753 | None, |
|
1752 | 1754 | None, |
|
1753 | 1755 | readtree=readtree, |
|
1754 | 1756 | match=match, |
|
1755 | 1757 | ) |
|
1756 | 1758 | |
|
1757 | 1759 | m.writesubtrees(m1, m2, writesubtree, match) |
|
1758 | 1760 | text = m.dirtext() |
|
1759 | 1761 | n = None |
|
1760 | 1762 | if self.tree != b'': |
|
1761 | 1763 | # Double-check whether contents are unchanged to one parent |
|
1762 | 1764 | if text == m1.dirtext(): |
|
1763 | 1765 | n = m1.node() |
|
1764 | 1766 | elif text == m2.dirtext(): |
|
1765 | 1767 | n = m2.node() |
|
1766 | 1768 | |
|
1767 | 1769 | if not n: |
|
1768 |
|
|
|
1770 | rev = self._revlog.addrevision( | |
|
1769 | 1771 | text, transaction, link, m1.node(), m2.node() |
|
1770 | 1772 | ) |
|
1773 | n = self._revlog.node(rev) | |
|
1771 | 1774 | |
|
1772 | 1775 | # Save nodeid so parent manifest can calculate its nodeid |
|
1773 | 1776 | m.setnode(n) |
|
1774 | 1777 | return n |
|
1775 | 1778 | |
|
1776 | 1779 | def __len__(self): |
|
1777 | 1780 | return len(self._revlog) |
|
1778 | 1781 | |
|
1779 | 1782 | def __iter__(self): |
|
1780 | 1783 | return self._revlog.__iter__() |
|
1781 | 1784 | |
|
1782 | 1785 | def rev(self, node): |
|
1783 | 1786 | return self._revlog.rev(node) |
|
1784 | 1787 | |
|
1785 | 1788 | def node(self, rev): |
|
1786 | 1789 | return self._revlog.node(rev) |
|
1787 | 1790 | |
|
1788 | 1791 | def lookup(self, value): |
|
1789 | 1792 | return self._revlog.lookup(value) |
|
1790 | 1793 | |
|
1791 | 1794 | def parentrevs(self, rev): |
|
1792 | 1795 | return self._revlog.parentrevs(rev) |
|
1793 | 1796 | |
|
1794 | 1797 | def parents(self, node): |
|
1795 | 1798 | return self._revlog.parents(node) |
|
1796 | 1799 | |
|
1797 | 1800 | def linkrev(self, rev): |
|
1798 | 1801 | return self._revlog.linkrev(rev) |
|
1799 | 1802 | |
|
1800 | 1803 | def checksize(self): |
|
1801 | 1804 | return self._revlog.checksize() |
|
1802 | 1805 | |
|
1803 | 1806 | def revision(self, node, _df=None, raw=False): |
|
1804 | 1807 | return self._revlog.revision(node, _df=_df, raw=raw) |
|
1805 | 1808 | |
|
1806 | 1809 | def rawdata(self, node, _df=None): |
|
1807 | 1810 | return self._revlog.rawdata(node, _df=_df) |
|
1808 | 1811 | |
|
1809 | 1812 | def revdiff(self, rev1, rev2): |
|
1810 | 1813 | return self._revlog.revdiff(rev1, rev2) |
|
1811 | 1814 | |
|
1812 | 1815 | def cmp(self, node, text): |
|
1813 | 1816 | return self._revlog.cmp(node, text) |
|
1814 | 1817 | |
|
1815 | 1818 | def deltaparent(self, rev): |
|
1816 | 1819 | return self._revlog.deltaparent(rev) |
|
1817 | 1820 | |
|
1818 | 1821 | def emitrevisions( |
|
1819 | 1822 | self, |
|
1820 | 1823 | nodes, |
|
1821 | 1824 | nodesorder=None, |
|
1822 | 1825 | revisiondata=False, |
|
1823 | 1826 | assumehaveparentrevisions=False, |
|
1824 | 1827 | deltamode=repository.CG_DELTAMODE_STD, |
|
1825 | 1828 | ): |
|
1826 | 1829 | return self._revlog.emitrevisions( |
|
1827 | 1830 | nodes, |
|
1828 | 1831 | nodesorder=nodesorder, |
|
1829 | 1832 | revisiondata=revisiondata, |
|
1830 | 1833 | assumehaveparentrevisions=assumehaveparentrevisions, |
|
1831 | 1834 | deltamode=deltamode, |
|
1832 | 1835 | ) |
|
1833 | 1836 | |
|
1834 | 1837 | def addgroup( |
|
1835 | 1838 | self, |
|
1836 | 1839 | deltas, |
|
1837 | 1840 | linkmapper, |
|
1838 | 1841 | transaction, |
|
1839 | 1842 | alwayscache=False, |
|
1840 | 1843 | addrevisioncb=None, |
|
1841 | 1844 | duplicaterevisioncb=None, |
|
1842 | 1845 | ): |
|
1843 | 1846 | return self._revlog.addgroup( |
|
1844 | 1847 | deltas, |
|
1845 | 1848 | linkmapper, |
|
1846 | 1849 | transaction, |
|
1847 | 1850 | alwayscache=alwayscache, |
|
1848 | 1851 | addrevisioncb=addrevisioncb, |
|
1849 | 1852 | duplicaterevisioncb=duplicaterevisioncb, |
|
1850 | 1853 | ) |
|
1851 | 1854 | |
|
1852 | 1855 | def rawsize(self, rev): |
|
1853 | 1856 | return self._revlog.rawsize(rev) |
|
1854 | 1857 | |
|
1855 | 1858 | def getstrippoint(self, minlink): |
|
1856 | 1859 | return self._revlog.getstrippoint(minlink) |
|
1857 | 1860 | |
|
1858 | 1861 | def strip(self, minlink, transaction): |
|
1859 | 1862 | return self._revlog.strip(minlink, transaction) |
|
1860 | 1863 | |
|
1861 | 1864 | def files(self): |
|
1862 | 1865 | return self._revlog.files() |
|
1863 | 1866 | |
|
1864 | 1867 | def clone(self, tr, destrevlog, **kwargs): |
|
1865 | 1868 | if not isinstance(destrevlog, manifestrevlog): |
|
1866 | 1869 | raise error.ProgrammingError(b'expected manifestrevlog to clone()') |
|
1867 | 1870 | |
|
1868 | 1871 | return self._revlog.clone(tr, destrevlog._revlog, **kwargs) |
|
1869 | 1872 | |
|
1870 | 1873 | def storageinfo( |
|
1871 | 1874 | self, |
|
1872 | 1875 | exclusivefiles=False, |
|
1873 | 1876 | sharedfiles=False, |
|
1874 | 1877 | revisionscount=False, |
|
1875 | 1878 | trackedsize=False, |
|
1876 | 1879 | storedsize=False, |
|
1877 | 1880 | ): |
|
1878 | 1881 | return self._revlog.storageinfo( |
|
1879 | 1882 | exclusivefiles=exclusivefiles, |
|
1880 | 1883 | sharedfiles=sharedfiles, |
|
1881 | 1884 | revisionscount=revisionscount, |
|
1882 | 1885 | trackedsize=trackedsize, |
|
1883 | 1886 | storedsize=storedsize, |
|
1884 | 1887 | ) |
|
1885 | 1888 | |
|
1886 | 1889 | @property |
|
1887 | 1890 | def indexfile(self): |
|
1888 | 1891 | return self._revlog.indexfile |
|
1889 | 1892 | |
|
1890 | 1893 | @indexfile.setter |
|
1891 | 1894 | def indexfile(self, value): |
|
1892 | 1895 | self._revlog.indexfile = value |
|
1893 | 1896 | |
|
1894 | 1897 | @property |
|
1895 | 1898 | def opener(self): |
|
1896 | 1899 | return self._revlog.opener |
|
1897 | 1900 | |
|
1898 | 1901 | @opener.setter |
|
1899 | 1902 | def opener(self, value): |
|
1900 | 1903 | self._revlog.opener = value |
|
1901 | 1904 | |
|
1902 | 1905 | |
|
1903 | 1906 | @interfaceutil.implementer(repository.imanifestlog) |
|
1904 | 1907 | class manifestlog(object): |
|
1905 | 1908 | """A collection class representing the collection of manifest snapshots |
|
1906 | 1909 | referenced by commits in the repository. |
|
1907 | 1910 | |
|
1908 | 1911 | In this situation, 'manifest' refers to the abstract concept of a snapshot |
|
1909 | 1912 | of the list of files in the given commit. Consumers of the output of this |
|
1910 | 1913 | class do not care about the implementation details of the actual manifests |
|
1911 | 1914 | they receive (i.e. tree or flat or lazily loaded, etc).""" |
|
1912 | 1915 | |
|
1913 | 1916 | def __init__(self, opener, repo, rootstore, narrowmatch): |
|
1914 | 1917 | usetreemanifest = False |
|
1915 | 1918 | cachesize = 4 |
|
1916 | 1919 | |
|
1917 | 1920 | opts = getattr(opener, 'options', None) |
|
1918 | 1921 | if opts is not None: |
|
1919 | 1922 | usetreemanifest = opts.get(b'treemanifest', usetreemanifest) |
|
1920 | 1923 | cachesize = opts.get(b'manifestcachesize', cachesize) |
|
1921 | 1924 | |
|
1922 | 1925 | self._treemanifests = usetreemanifest |
|
1923 | 1926 | |
|
1924 | 1927 | self._rootstore = rootstore |
|
1925 | 1928 | self._rootstore._setupmanifestcachehooks(repo) |
|
1926 | 1929 | self._narrowmatch = narrowmatch |
|
1927 | 1930 | |
|
1928 | 1931 | # A cache of the manifestctx or treemanifestctx for each directory |
|
1929 | 1932 | self._dirmancache = {} |
|
1930 | 1933 | self._dirmancache[b''] = util.lrucachedict(cachesize) |
|
1931 | 1934 | |
|
1932 | 1935 | self._cachesize = cachesize |
|
1933 | 1936 | |
|
1934 | 1937 | def __getitem__(self, node): |
|
1935 | 1938 | """Retrieves the manifest instance for the given node. Throws a |
|
1936 | 1939 | LookupError if not found. |
|
1937 | 1940 | """ |
|
1938 | 1941 | return self.get(b'', node) |
|
1939 | 1942 | |
|
1940 | 1943 | def get(self, tree, node, verify=True): |
|
1941 | 1944 | """Retrieves the manifest instance for the given node. Throws a |
|
1942 | 1945 | LookupError if not found. |
|
1943 | 1946 | |
|
1944 | 1947 | `verify` - if True an exception will be thrown if the node is not in |
|
1945 | 1948 | the revlog |
|
1946 | 1949 | """ |
|
1947 | 1950 | if node in self._dirmancache.get(tree, ()): |
|
1948 | 1951 | return self._dirmancache[tree][node] |
|
1949 | 1952 | |
|
1950 | 1953 | if not self._narrowmatch.always(): |
|
1951 | 1954 | if not self._narrowmatch.visitdir(tree[:-1]): |
|
1952 | 1955 | return excludeddirmanifestctx(tree, node) |
|
1953 | 1956 | if tree: |
|
1954 | 1957 | if self._rootstore._treeondisk: |
|
1955 | 1958 | if verify: |
|
1956 | 1959 | # Side-effect is LookupError is raised if node doesn't |
|
1957 | 1960 | # exist. |
|
1958 | 1961 | self.getstorage(tree).rev(node) |
|
1959 | 1962 | |
|
1960 | 1963 | m = treemanifestctx(self, tree, node) |
|
1961 | 1964 | else: |
|
1962 | 1965 | raise error.Abort( |
|
1963 | 1966 | _( |
|
1964 | 1967 | b"cannot ask for manifest directory '%s' in a flat " |
|
1965 | 1968 | b"manifest" |
|
1966 | 1969 | ) |
|
1967 | 1970 | % tree |
|
1968 | 1971 | ) |
|
1969 | 1972 | else: |
|
1970 | 1973 | if verify: |
|
1971 | 1974 | # Side-effect is LookupError is raised if node doesn't exist. |
|
1972 | 1975 | self._rootstore.rev(node) |
|
1973 | 1976 | |
|
1974 | 1977 | if self._treemanifests: |
|
1975 | 1978 | m = treemanifestctx(self, b'', node) |
|
1976 | 1979 | else: |
|
1977 | 1980 | m = manifestctx(self, node) |
|
1978 | 1981 | |
|
1979 | 1982 | if node != nullid: |
|
1980 | 1983 | mancache = self._dirmancache.get(tree) |
|
1981 | 1984 | if not mancache: |
|
1982 | 1985 | mancache = util.lrucachedict(self._cachesize) |
|
1983 | 1986 | self._dirmancache[tree] = mancache |
|
1984 | 1987 | mancache[node] = m |
|
1985 | 1988 | return m |
|
1986 | 1989 | |
|
1987 | 1990 | def getstorage(self, tree): |
|
1988 | 1991 | return self._rootstore.dirlog(tree) |
|
1989 | 1992 | |
|
1990 | 1993 | def clearcaches(self, clear_persisted_data=False): |
|
1991 | 1994 | self._dirmancache.clear() |
|
1992 | 1995 | self._rootstore.clearcaches(clear_persisted_data=clear_persisted_data) |
|
1993 | 1996 | |
|
1994 | 1997 | def rev(self, node): |
|
1995 | 1998 | return self._rootstore.rev(node) |
|
1996 | 1999 | |
|
1997 | 2000 | def update_caches(self, transaction): |
|
1998 | 2001 | return self._rootstore._revlog.update_caches(transaction=transaction) |
|
1999 | 2002 | |
|
2000 | 2003 | |
|
2001 | 2004 | @interfaceutil.implementer(repository.imanifestrevisionwritable) |
|
2002 | 2005 | class memmanifestctx(object): |
|
2003 | 2006 | def __init__(self, manifestlog): |
|
2004 | 2007 | self._manifestlog = manifestlog |
|
2005 | 2008 | self._manifestdict = manifestdict() |
|
2006 | 2009 | |
|
2007 | 2010 | def _storage(self): |
|
2008 | 2011 | return self._manifestlog.getstorage(b'') |
|
2009 | 2012 | |
|
2010 | 2013 | def copy(self): |
|
2011 | 2014 | memmf = memmanifestctx(self._manifestlog) |
|
2012 | 2015 | memmf._manifestdict = self.read().copy() |
|
2013 | 2016 | return memmf |
|
2014 | 2017 | |
|
2015 | 2018 | def read(self): |
|
2016 | 2019 | return self._manifestdict |
|
2017 | 2020 | |
|
2018 | 2021 | def write(self, transaction, link, p1, p2, added, removed, match=None): |
|
2019 | 2022 | return self._storage().add( |
|
2020 | 2023 | self._manifestdict, |
|
2021 | 2024 | transaction, |
|
2022 | 2025 | link, |
|
2023 | 2026 | p1, |
|
2024 | 2027 | p2, |
|
2025 | 2028 | added, |
|
2026 | 2029 | removed, |
|
2027 | 2030 | match=match, |
|
2028 | 2031 | ) |
|
2029 | 2032 | |
|
2030 | 2033 | |
|
2031 | 2034 | @interfaceutil.implementer(repository.imanifestrevisionstored) |
|
2032 | 2035 | class manifestctx(object): |
|
2033 | 2036 | """A class representing a single revision of a manifest, including its |
|
2034 | 2037 | contents, its parent revs, and its linkrev. |
|
2035 | 2038 | """ |
|
2036 | 2039 | |
|
2037 | 2040 | def __init__(self, manifestlog, node): |
|
2038 | 2041 | self._manifestlog = manifestlog |
|
2039 | 2042 | self._data = None |
|
2040 | 2043 | |
|
2041 | 2044 | self._node = node |
|
2042 | 2045 | |
|
2043 | 2046 | # TODO: We eventually want p1, p2, and linkrev exposed on this class, |
|
2044 | 2047 | # but let's add it later when something needs it and we can load it |
|
2045 | 2048 | # lazily. |
|
2046 | 2049 | # self.p1, self.p2 = store.parents(node) |
|
2047 | 2050 | # rev = store.rev(node) |
|
2048 | 2051 | # self.linkrev = store.linkrev(rev) |
|
2049 | 2052 | |
|
2050 | 2053 | def _storage(self): |
|
2051 | 2054 | return self._manifestlog.getstorage(b'') |
|
2052 | 2055 | |
|
2053 | 2056 | def node(self): |
|
2054 | 2057 | return self._node |
|
2055 | 2058 | |
|
2056 | 2059 | def copy(self): |
|
2057 | 2060 | memmf = memmanifestctx(self._manifestlog) |
|
2058 | 2061 | memmf._manifestdict = self.read().copy() |
|
2059 | 2062 | return memmf |
|
2060 | 2063 | |
|
2061 | 2064 | @propertycache |
|
2062 | 2065 | def parents(self): |
|
2063 | 2066 | return self._storage().parents(self._node) |
|
2064 | 2067 | |
|
2065 | 2068 | def read(self): |
|
2066 | 2069 | if self._data is None: |
|
2067 | 2070 | if self._node == nullid: |
|
2068 | 2071 | self._data = manifestdict() |
|
2069 | 2072 | else: |
|
2070 | 2073 | store = self._storage() |
|
2071 | 2074 | if self._node in store.fulltextcache: |
|
2072 | 2075 | text = pycompat.bytestr(store.fulltextcache[self._node]) |
|
2073 | 2076 | else: |
|
2074 | 2077 | text = store.revision(self._node) |
|
2075 | 2078 | arraytext = bytearray(text) |
|
2076 | 2079 | store.fulltextcache[self._node] = arraytext |
|
2077 | 2080 | self._data = manifestdict(text) |
|
2078 | 2081 | return self._data |
|
2079 | 2082 | |
|
2080 | 2083 | def readfast(self, shallow=False): |
|
2081 | 2084 | """Calls either readdelta or read, based on which would be less work. |
|
2082 | 2085 | readdelta is called if the delta is against the p1, and therefore can be |
|
2083 | 2086 | read quickly. |
|
2084 | 2087 | |
|
2085 | 2088 | If `shallow` is True, nothing changes since this is a flat manifest. |
|
2086 | 2089 | """ |
|
2087 | 2090 | store = self._storage() |
|
2088 | 2091 | r = store.rev(self._node) |
|
2089 | 2092 | deltaparent = store.deltaparent(r) |
|
2090 | 2093 | if deltaparent != nullrev and deltaparent in store.parentrevs(r): |
|
2091 | 2094 | return self.readdelta() |
|
2092 | 2095 | return self.read() |
|
2093 | 2096 | |
|
2094 | 2097 | def readdelta(self, shallow=False): |
|
2095 | 2098 | """Returns a manifest containing just the entries that are present |
|
2096 | 2099 | in this manifest, but not in its p1 manifest. This is efficient to read |
|
2097 | 2100 | if the revlog delta is already p1. |
|
2098 | 2101 | |
|
2099 | 2102 | Changing the value of `shallow` has no effect on flat manifests. |
|
2100 | 2103 | """ |
|
2101 | 2104 | store = self._storage() |
|
2102 | 2105 | r = store.rev(self._node) |
|
2103 | 2106 | d = mdiff.patchtext(store.revdiff(store.deltaparent(r), r)) |
|
2104 | 2107 | return manifestdict(d) |
|
2105 | 2108 | |
|
2106 | 2109 | def find(self, key): |
|
2107 | 2110 | return self.read().find(key) |
|
2108 | 2111 | |
|
2109 | 2112 | |
|
2110 | 2113 | @interfaceutil.implementer(repository.imanifestrevisionwritable) |
|
2111 | 2114 | class memtreemanifestctx(object): |
|
2112 | 2115 | def __init__(self, manifestlog, dir=b''): |
|
2113 | 2116 | self._manifestlog = manifestlog |
|
2114 | 2117 | self._dir = dir |
|
2115 | 2118 | self._treemanifest = treemanifest() |
|
2116 | 2119 | |
|
2117 | 2120 | def _storage(self): |
|
2118 | 2121 | return self._manifestlog.getstorage(b'') |
|
2119 | 2122 | |
|
2120 | 2123 | def copy(self): |
|
2121 | 2124 | memmf = memtreemanifestctx(self._manifestlog, dir=self._dir) |
|
2122 | 2125 | memmf._treemanifest = self._treemanifest.copy() |
|
2123 | 2126 | return memmf |
|
2124 | 2127 | |
|
2125 | 2128 | def read(self): |
|
2126 | 2129 | return self._treemanifest |
|
2127 | 2130 | |
|
2128 | 2131 | def write(self, transaction, link, p1, p2, added, removed, match=None): |
|
2129 | 2132 | def readtree(dir, node): |
|
2130 | 2133 | return self._manifestlog.get(dir, node).read() |
|
2131 | 2134 | |
|
2132 | 2135 | return self._storage().add( |
|
2133 | 2136 | self._treemanifest, |
|
2134 | 2137 | transaction, |
|
2135 | 2138 | link, |
|
2136 | 2139 | p1, |
|
2137 | 2140 | p2, |
|
2138 | 2141 | added, |
|
2139 | 2142 | removed, |
|
2140 | 2143 | readtree=readtree, |
|
2141 | 2144 | match=match, |
|
2142 | 2145 | ) |
|
2143 | 2146 | |
|
2144 | 2147 | |
|
2145 | 2148 | @interfaceutil.implementer(repository.imanifestrevisionstored) |
|
2146 | 2149 | class treemanifestctx(object): |
|
2147 | 2150 | def __init__(self, manifestlog, dir, node): |
|
2148 | 2151 | self._manifestlog = manifestlog |
|
2149 | 2152 | self._dir = dir |
|
2150 | 2153 | self._data = None |
|
2151 | 2154 | |
|
2152 | 2155 | self._node = node |
|
2153 | 2156 | |
|
2154 | 2157 | # TODO: Load p1/p2/linkrev lazily. They need to be lazily loaded so that |
|
2155 | 2158 | # we can instantiate treemanifestctx objects for directories we don't |
|
2156 | 2159 | # have on disk. |
|
2157 | 2160 | # self.p1, self.p2 = store.parents(node) |
|
2158 | 2161 | # rev = store.rev(node) |
|
2159 | 2162 | # self.linkrev = store.linkrev(rev) |
|
2160 | 2163 | |
|
2161 | 2164 | def _storage(self): |
|
2162 | 2165 | narrowmatch = self._manifestlog._narrowmatch |
|
2163 | 2166 | if not narrowmatch.always(): |
|
2164 | 2167 | if not narrowmatch.visitdir(self._dir[:-1]): |
|
2165 | 2168 | return excludedmanifestrevlog(self._dir) |
|
2166 | 2169 | return self._manifestlog.getstorage(self._dir) |
|
2167 | 2170 | |
|
2168 | 2171 | def read(self): |
|
2169 | 2172 | if self._data is None: |
|
2170 | 2173 | store = self._storage() |
|
2171 | 2174 | if self._node == nullid: |
|
2172 | 2175 | self._data = treemanifest() |
|
2173 | 2176 | # TODO accessing non-public API |
|
2174 | 2177 | elif store._treeondisk: |
|
2175 | 2178 | m = treemanifest(dir=self._dir) |
|
2176 | 2179 | |
|
2177 | 2180 | def gettext(): |
|
2178 | 2181 | return store.revision(self._node) |
|
2179 | 2182 | |
|
2180 | 2183 | def readsubtree(dir, subm): |
|
2181 | 2184 | # Set verify to False since we need to be able to create |
|
2182 | 2185 | # subtrees for trees that don't exist on disk. |
|
2183 | 2186 | return self._manifestlog.get(dir, subm, verify=False).read() |
|
2184 | 2187 | |
|
2185 | 2188 | m.read(gettext, readsubtree) |
|
2186 | 2189 | m.setnode(self._node) |
|
2187 | 2190 | self._data = m |
|
2188 | 2191 | else: |
|
2189 | 2192 | if self._node in store.fulltextcache: |
|
2190 | 2193 | text = pycompat.bytestr(store.fulltextcache[self._node]) |
|
2191 | 2194 | else: |
|
2192 | 2195 | text = store.revision(self._node) |
|
2193 | 2196 | arraytext = bytearray(text) |
|
2194 | 2197 | store.fulltextcache[self._node] = arraytext |
|
2195 | 2198 | self._data = treemanifest(dir=self._dir, text=text) |
|
2196 | 2199 | |
|
2197 | 2200 | return self._data |
|
2198 | 2201 | |
|
2199 | 2202 | def node(self): |
|
2200 | 2203 | return self._node |
|
2201 | 2204 | |
|
2202 | 2205 | def copy(self): |
|
2203 | 2206 | memmf = memtreemanifestctx(self._manifestlog, dir=self._dir) |
|
2204 | 2207 | memmf._treemanifest = self.read().copy() |
|
2205 | 2208 | return memmf |
|
2206 | 2209 | |
|
2207 | 2210 | @propertycache |
|
2208 | 2211 | def parents(self): |
|
2209 | 2212 | return self._storage().parents(self._node) |
|
2210 | 2213 | |
|
2211 | 2214 | def readdelta(self, shallow=False): |
|
2212 | 2215 | """Returns a manifest containing just the entries that are present |
|
2213 | 2216 | in this manifest, but not in its p1 manifest. This is efficient to read |
|
2214 | 2217 | if the revlog delta is already p1. |
|
2215 | 2218 | |
|
2216 | 2219 | If `shallow` is True, this will read the delta for this directory, |
|
2217 | 2220 | without recursively reading subdirectory manifests. Instead, any |
|
2218 | 2221 | subdirectory entry will be reported as it appears in the manifest, i.e. |
|
2219 | 2222 | the subdirectory will be reported among files and distinguished only by |
|
2220 | 2223 | its 't' flag. |
|
2221 | 2224 | """ |
|
2222 | 2225 | store = self._storage() |
|
2223 | 2226 | if shallow: |
|
2224 | 2227 | r = store.rev(self._node) |
|
2225 | 2228 | d = mdiff.patchtext(store.revdiff(store.deltaparent(r), r)) |
|
2226 | 2229 | return manifestdict(d) |
|
2227 | 2230 | else: |
|
2228 | 2231 | # Need to perform a slow delta |
|
2229 | 2232 | r0 = store.deltaparent(store.rev(self._node)) |
|
2230 | 2233 | m0 = self._manifestlog.get(self._dir, store.node(r0)).read() |
|
2231 | 2234 | m1 = self.read() |
|
2232 | 2235 | md = treemanifest(dir=self._dir) |
|
2233 | 2236 | for f, ((n0, fl0), (n1, fl1)) in pycompat.iteritems(m0.diff(m1)): |
|
2234 | 2237 | if n1: |
|
2235 | 2238 | md[f] = n1 |
|
2236 | 2239 | if fl1: |
|
2237 | 2240 | md.setflag(f, fl1) |
|
2238 | 2241 | return md |
|
2239 | 2242 | |
|
2240 | 2243 | def readfast(self, shallow=False): |
|
2241 | 2244 | """Calls either readdelta or read, based on which would be less work. |
|
2242 | 2245 | readdelta is called if the delta is against the p1, and therefore can be |
|
2243 | 2246 | read quickly. |
|
2244 | 2247 | |
|
2245 | 2248 | If `shallow` is True, it only returns the entries from this manifest, |
|
2246 | 2249 | and not any submanifests. |
|
2247 | 2250 | """ |
|
2248 | 2251 | store = self._storage() |
|
2249 | 2252 | r = store.rev(self._node) |
|
2250 | 2253 | deltaparent = store.deltaparent(r) |
|
2251 | 2254 | if deltaparent != nullrev and deltaparent in store.parentrevs(r): |
|
2252 | 2255 | return self.readdelta(shallow=shallow) |
|
2253 | 2256 | |
|
2254 | 2257 | if shallow: |
|
2255 | 2258 | return manifestdict(store.revision(self._node)) |
|
2256 | 2259 | else: |
|
2257 | 2260 | return self.read() |
|
2258 | 2261 | |
|
2259 | 2262 | def find(self, key): |
|
2260 | 2263 | return self.read().find(key) |
|
2261 | 2264 | |
|
2262 | 2265 | |
|
2263 | 2266 | class excludeddir(treemanifest): |
|
2264 | 2267 | """Stand-in for a directory that is excluded from the repository. |
|
2265 | 2268 | |
|
2266 | 2269 | With narrowing active on a repository that uses treemanifests, |
|
2267 | 2270 | some of the directory revlogs will be excluded from the resulting |
|
2268 | 2271 | clone. This is a huge storage win for clients, but means we need |
|
2269 | 2272 | some sort of pseudo-manifest to surface to internals so we can |
|
2270 | 2273 | detect a merge conflict outside the narrowspec. That's what this |
|
2271 | 2274 | class is: it stands in for a directory whose node is known, but |
|
2272 | 2275 | whose contents are unknown. |
|
2273 | 2276 | """ |
|
2274 | 2277 | |
|
2275 | 2278 | def __init__(self, dir, node): |
|
2276 | 2279 | super(excludeddir, self).__init__(dir) |
|
2277 | 2280 | self._node = node |
|
2278 | 2281 | # Add an empty file, which will be included by iterators and such, |
|
2279 | 2282 | # appearing as the directory itself (i.e. something like "dir/") |
|
2280 | 2283 | self._files[b''] = node |
|
2281 | 2284 | self._flags[b''] = b't' |
|
2282 | 2285 | |
|
2283 | 2286 | # Manifests outside the narrowspec should never be modified, so avoid |
|
2284 | 2287 | # copying. This makes a noticeable difference when there are very many |
|
2285 | 2288 | # directories outside the narrowspec. Also, it makes sense for the copy to |
|
2286 | 2289 | # be of the same type as the original, which would not happen with the |
|
2287 | 2290 | # super type's copy(). |
|
2288 | 2291 | def copy(self): |
|
2289 | 2292 | return self |
|
2290 | 2293 | |
|
2291 | 2294 | |
|
2292 | 2295 | class excludeddirmanifestctx(treemanifestctx): |
|
2293 | 2296 | """context wrapper for excludeddir - see that docstring for rationale""" |
|
2294 | 2297 | |
|
2295 | 2298 | def __init__(self, dir, node): |
|
2296 | 2299 | self._dir = dir |
|
2297 | 2300 | self._node = node |
|
2298 | 2301 | |
|
2299 | 2302 | def read(self): |
|
2300 | 2303 | return excludeddir(self._dir, self._node) |
|
2301 | 2304 | |
|
2302 | 2305 | def readfast(self, shallow=False): |
|
2303 | 2306 | # special version of readfast since we don't have underlying storage |
|
2304 | 2307 | return self.read() |
|
2305 | 2308 | |
|
2306 | 2309 | def write(self, *args): |
|
2307 | 2310 | raise error.ProgrammingError( |
|
2308 | 2311 | b'attempt to write manifest from excluded dir %s' % self._dir |
|
2309 | 2312 | ) |
|
2310 | 2313 | |
|
2311 | 2314 | |
|
2312 | 2315 | class excludedmanifestrevlog(manifestrevlog): |
|
2313 | 2316 | """Stand-in for excluded treemanifest revlogs. |
|
2314 | 2317 | |
|
2315 | 2318 | When narrowing is active on a treemanifest repository, we'll have |
|
2316 | 2319 | references to directories we can't see due to the revlog being |
|
2317 | 2320 | skipped. This class exists to conform to the manifestrevlog |
|
2318 | 2321 | interface for those directories and proactively prevent writes to |
|
2319 | 2322 | outside the narrowspec. |
|
2320 | 2323 | """ |
|
2321 | 2324 | |
|
2322 | 2325 | def __init__(self, dir): |
|
2323 | 2326 | self._dir = dir |
|
2324 | 2327 | |
|
2325 | 2328 | def __len__(self): |
|
2326 | 2329 | raise error.ProgrammingError( |
|
2327 | 2330 | b'attempt to get length of excluded dir %s' % self._dir |
|
2328 | 2331 | ) |
|
2329 | 2332 | |
|
2330 | 2333 | def rev(self, node): |
|
2331 | 2334 | raise error.ProgrammingError( |
|
2332 | 2335 | b'attempt to get rev from excluded dir %s' % self._dir |
|
2333 | 2336 | ) |
|
2334 | 2337 | |
|
2335 | 2338 | def linkrev(self, node): |
|
2336 | 2339 | raise error.ProgrammingError( |
|
2337 | 2340 | b'attempt to get linkrev from excluded dir %s' % self._dir |
|
2338 | 2341 | ) |
|
2339 | 2342 | |
|
2340 | 2343 | def node(self, rev): |
|
2341 | 2344 | raise error.ProgrammingError( |
|
2342 | 2345 | b'attempt to get node from excluded dir %s' % self._dir |
|
2343 | 2346 | ) |
|
2344 | 2347 | |
|
2345 | 2348 | def add(self, *args, **kwargs): |
|
2346 | 2349 | # We should never write entries in dirlogs outside the narrow clone. |
|
2347 | 2350 | # However, the method still gets called from writesubtree() in |
|
2348 | 2351 | # _addtree(), so we need to handle it. We should possibly make that |
|
2349 | 2352 | # avoid calling add() with a clean manifest (_dirty is always False |
|
2350 | 2353 | # in excludeddir instances). |
|
2351 | 2354 | pass |
@@ -1,3086 +1,3086 b'' | |||
|
1 | 1 | # revlog.py - storage back-end 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 | """Storage back-end for Mercurial. |
|
9 | 9 | |
|
10 | 10 | This provides efficient delta storage with O(1) retrieve and append |
|
11 | 11 | and O(changes) merge between branches. |
|
12 | 12 | """ |
|
13 | 13 | |
|
14 | 14 | from __future__ import absolute_import |
|
15 | 15 | |
|
16 | 16 | import collections |
|
17 | 17 | import contextlib |
|
18 | 18 | import errno |
|
19 | 19 | import io |
|
20 | 20 | import os |
|
21 | 21 | import struct |
|
22 | 22 | import zlib |
|
23 | 23 | |
|
24 | 24 | # import stuff from node for others to import from revlog |
|
25 | 25 | from .node import ( |
|
26 | 26 | bin, |
|
27 | 27 | hex, |
|
28 | 28 | nullhex, |
|
29 | 29 | nullid, |
|
30 | 30 | nullrev, |
|
31 | 31 | short, |
|
32 | 32 | wdirfilenodeids, |
|
33 | 33 | wdirhex, |
|
34 | 34 | wdirid, |
|
35 | 35 | wdirrev, |
|
36 | 36 | ) |
|
37 | 37 | from .i18n import _ |
|
38 | 38 | from .pycompat import getattr |
|
39 | 39 | from .revlogutils.constants import ( |
|
40 | 40 | FLAG_GENERALDELTA, |
|
41 | 41 | FLAG_INLINE_DATA, |
|
42 | 42 | REVLOGV0, |
|
43 | 43 | REVLOGV1, |
|
44 | 44 | REVLOGV1_FLAGS, |
|
45 | 45 | REVLOGV2, |
|
46 | 46 | REVLOGV2_FLAGS, |
|
47 | 47 | REVLOG_DEFAULT_FLAGS, |
|
48 | 48 | REVLOG_DEFAULT_FORMAT, |
|
49 | 49 | REVLOG_DEFAULT_VERSION, |
|
50 | 50 | ) |
|
51 | 51 | from .revlogutils.flagutil import ( |
|
52 | 52 | REVIDX_DEFAULT_FLAGS, |
|
53 | 53 | REVIDX_ELLIPSIS, |
|
54 | 54 | REVIDX_EXTSTORED, |
|
55 | 55 | REVIDX_FLAGS_ORDER, |
|
56 | 56 | REVIDX_HASCOPIESINFO, |
|
57 | 57 | REVIDX_ISCENSORED, |
|
58 | 58 | REVIDX_RAWTEXT_CHANGING_FLAGS, |
|
59 | 59 | REVIDX_SIDEDATA, |
|
60 | 60 | ) |
|
61 | 61 | from .thirdparty import attr |
|
62 | 62 | from . import ( |
|
63 | 63 | ancestor, |
|
64 | 64 | dagop, |
|
65 | 65 | error, |
|
66 | 66 | mdiff, |
|
67 | 67 | policy, |
|
68 | 68 | pycompat, |
|
69 | 69 | templatefilters, |
|
70 | 70 | util, |
|
71 | 71 | ) |
|
72 | 72 | from .interfaces import ( |
|
73 | 73 | repository, |
|
74 | 74 | util as interfaceutil, |
|
75 | 75 | ) |
|
76 | 76 | from .revlogutils import ( |
|
77 | 77 | deltas as deltautil, |
|
78 | 78 | flagutil, |
|
79 | 79 | nodemap as nodemaputil, |
|
80 | 80 | sidedata as sidedatautil, |
|
81 | 81 | ) |
|
82 | 82 | from .utils import ( |
|
83 | 83 | storageutil, |
|
84 | 84 | stringutil, |
|
85 | 85 | ) |
|
86 | 86 | |
|
87 | 87 | # blanked usage of all the name to prevent pyflakes constraints |
|
88 | 88 | # We need these name available in the module for extensions. |
|
89 | 89 | REVLOGV0 |
|
90 | 90 | REVLOGV1 |
|
91 | 91 | REVLOGV2 |
|
92 | 92 | FLAG_INLINE_DATA |
|
93 | 93 | FLAG_GENERALDELTA |
|
94 | 94 | REVLOG_DEFAULT_FLAGS |
|
95 | 95 | REVLOG_DEFAULT_FORMAT |
|
96 | 96 | REVLOG_DEFAULT_VERSION |
|
97 | 97 | REVLOGV1_FLAGS |
|
98 | 98 | REVLOGV2_FLAGS |
|
99 | 99 | REVIDX_ISCENSORED |
|
100 | 100 | REVIDX_ELLIPSIS |
|
101 | 101 | REVIDX_SIDEDATA |
|
102 | 102 | REVIDX_HASCOPIESINFO |
|
103 | 103 | REVIDX_EXTSTORED |
|
104 | 104 | REVIDX_DEFAULT_FLAGS |
|
105 | 105 | REVIDX_FLAGS_ORDER |
|
106 | 106 | REVIDX_RAWTEXT_CHANGING_FLAGS |
|
107 | 107 | |
|
108 | 108 | parsers = policy.importmod('parsers') |
|
109 | 109 | rustancestor = policy.importrust('ancestor') |
|
110 | 110 | rustdagop = policy.importrust('dagop') |
|
111 | 111 | rustrevlog = policy.importrust('revlog') |
|
112 | 112 | |
|
113 | 113 | # Aliased for performance. |
|
114 | 114 | _zlibdecompress = zlib.decompress |
|
115 | 115 | |
|
116 | 116 | # max size of revlog with inline data |
|
117 | 117 | _maxinline = 131072 |
|
118 | 118 | _chunksize = 1048576 |
|
119 | 119 | |
|
120 | 120 | # Flag processors for REVIDX_ELLIPSIS. |
|
121 | 121 | def ellipsisreadprocessor(rl, text): |
|
122 | 122 | return text, False, {} |
|
123 | 123 | |
|
124 | 124 | |
|
125 | 125 | def ellipsiswriteprocessor(rl, text, sidedata): |
|
126 | 126 | return text, False |
|
127 | 127 | |
|
128 | 128 | |
|
129 | 129 | def ellipsisrawprocessor(rl, text): |
|
130 | 130 | return False |
|
131 | 131 | |
|
132 | 132 | |
|
133 | 133 | ellipsisprocessor = ( |
|
134 | 134 | ellipsisreadprocessor, |
|
135 | 135 | ellipsiswriteprocessor, |
|
136 | 136 | ellipsisrawprocessor, |
|
137 | 137 | ) |
|
138 | 138 | |
|
139 | 139 | |
|
140 | 140 | def getoffset(q): |
|
141 | 141 | return int(q >> 16) |
|
142 | 142 | |
|
143 | 143 | |
|
144 | 144 | def gettype(q): |
|
145 | 145 | return int(q & 0xFFFF) |
|
146 | 146 | |
|
147 | 147 | |
|
148 | 148 | def offset_type(offset, type): |
|
149 | 149 | if (type & ~flagutil.REVIDX_KNOWN_FLAGS) != 0: |
|
150 | 150 | raise ValueError(b'unknown revlog index flags') |
|
151 | 151 | return int(int(offset) << 16 | type) |
|
152 | 152 | |
|
153 | 153 | |
|
154 | 154 | def _verify_revision(rl, skipflags, state, node): |
|
155 | 155 | """Verify the integrity of the given revlog ``node`` while providing a hook |
|
156 | 156 | point for extensions to influence the operation.""" |
|
157 | 157 | if skipflags: |
|
158 | 158 | state[b'skipread'].add(node) |
|
159 | 159 | else: |
|
160 | 160 | # Side-effect: read content and verify hash. |
|
161 | 161 | rl.revision(node) |
|
162 | 162 | |
|
163 | 163 | |
|
164 | 164 | # True if a fast implementation for persistent-nodemap is available |
|
165 | 165 | # |
|
166 | 166 | # We also consider we have a "fast" implementation in "pure" python because |
|
167 | 167 | # people using pure don't really have performance consideration (and a |
|
168 | 168 | # wheelbarrow of other slowness source) |
|
169 | 169 | HAS_FAST_PERSISTENT_NODEMAP = rustrevlog is not None or util.safehasattr( |
|
170 | 170 | parsers, 'BaseIndexObject' |
|
171 | 171 | ) |
|
172 | 172 | |
|
173 | 173 | |
|
174 | 174 | @attr.s(slots=True, frozen=True) |
|
175 | 175 | class _revisioninfo(object): |
|
176 | 176 | """Information about a revision that allows building its fulltext |
|
177 | 177 | node: expected hash of the revision |
|
178 | 178 | p1, p2: parent revs of the revision |
|
179 | 179 | btext: built text cache consisting of a one-element list |
|
180 | 180 | cachedelta: (baserev, uncompressed_delta) or None |
|
181 | 181 | flags: flags associated to the revision storage |
|
182 | 182 | |
|
183 | 183 | One of btext[0] or cachedelta must be set. |
|
184 | 184 | """ |
|
185 | 185 | |
|
186 | 186 | node = attr.ib() |
|
187 | 187 | p1 = attr.ib() |
|
188 | 188 | p2 = attr.ib() |
|
189 | 189 | btext = attr.ib() |
|
190 | 190 | textlen = attr.ib() |
|
191 | 191 | cachedelta = attr.ib() |
|
192 | 192 | flags = attr.ib() |
|
193 | 193 | |
|
194 | 194 | |
|
195 | 195 | @interfaceutil.implementer(repository.irevisiondelta) |
|
196 | 196 | @attr.s(slots=True) |
|
197 | 197 | class revlogrevisiondelta(object): |
|
198 | 198 | node = attr.ib() |
|
199 | 199 | p1node = attr.ib() |
|
200 | 200 | p2node = attr.ib() |
|
201 | 201 | basenode = attr.ib() |
|
202 | 202 | flags = attr.ib() |
|
203 | 203 | baserevisionsize = attr.ib() |
|
204 | 204 | revision = attr.ib() |
|
205 | 205 | delta = attr.ib() |
|
206 | 206 | linknode = attr.ib(default=None) |
|
207 | 207 | |
|
208 | 208 | |
|
209 | 209 | @interfaceutil.implementer(repository.iverifyproblem) |
|
210 | 210 | @attr.s(frozen=True) |
|
211 | 211 | class revlogproblem(object): |
|
212 | 212 | warning = attr.ib(default=None) |
|
213 | 213 | error = attr.ib(default=None) |
|
214 | 214 | node = attr.ib(default=None) |
|
215 | 215 | |
|
216 | 216 | |
|
217 | 217 | # index v0: |
|
218 | 218 | # 4 bytes: offset |
|
219 | 219 | # 4 bytes: compressed length |
|
220 | 220 | # 4 bytes: base rev |
|
221 | 221 | # 4 bytes: link rev |
|
222 | 222 | # 20 bytes: parent 1 nodeid |
|
223 | 223 | # 20 bytes: parent 2 nodeid |
|
224 | 224 | # 20 bytes: nodeid |
|
225 | 225 | indexformatv0 = struct.Struct(b">4l20s20s20s") |
|
226 | 226 | indexformatv0_pack = indexformatv0.pack |
|
227 | 227 | indexformatv0_unpack = indexformatv0.unpack |
|
228 | 228 | |
|
229 | 229 | |
|
230 | 230 | class revlogoldindex(list): |
|
231 | 231 | @property |
|
232 | 232 | def nodemap(self): |
|
233 | 233 | msg = b"index.nodemap is deprecated, use index.[has_node|rev|get_rev]" |
|
234 | 234 | util.nouideprecwarn(msg, b'5.3', stacklevel=2) |
|
235 | 235 | return self._nodemap |
|
236 | 236 | |
|
237 | 237 | @util.propertycache |
|
238 | 238 | def _nodemap(self): |
|
239 | 239 | nodemap = nodemaputil.NodeMap({nullid: nullrev}) |
|
240 | 240 | for r in range(0, len(self)): |
|
241 | 241 | n = self[r][7] |
|
242 | 242 | nodemap[n] = r |
|
243 | 243 | return nodemap |
|
244 | 244 | |
|
245 | 245 | def has_node(self, node): |
|
246 | 246 | """return True if the node exist in the index""" |
|
247 | 247 | return node in self._nodemap |
|
248 | 248 | |
|
249 | 249 | def rev(self, node): |
|
250 | 250 | """return a revision for a node |
|
251 | 251 | |
|
252 | 252 | If the node is unknown, raise a RevlogError""" |
|
253 | 253 | return self._nodemap[node] |
|
254 | 254 | |
|
255 | 255 | def get_rev(self, node): |
|
256 | 256 | """return a revision for a node |
|
257 | 257 | |
|
258 | 258 | If the node is unknown, return None""" |
|
259 | 259 | return self._nodemap.get(node) |
|
260 | 260 | |
|
261 | 261 | def append(self, tup): |
|
262 | 262 | self._nodemap[tup[7]] = len(self) |
|
263 | 263 | super(revlogoldindex, self).append(tup) |
|
264 | 264 | |
|
265 | 265 | def __delitem__(self, i): |
|
266 | 266 | if not isinstance(i, slice) or not i.stop == -1 or i.step is not None: |
|
267 | 267 | raise ValueError(b"deleting slices only supports a:-1 with step 1") |
|
268 | 268 | for r in pycompat.xrange(i.start, len(self)): |
|
269 | 269 | del self._nodemap[self[r][7]] |
|
270 | 270 | super(revlogoldindex, self).__delitem__(i) |
|
271 | 271 | |
|
272 | 272 | def clearcaches(self): |
|
273 | 273 | self.__dict__.pop('_nodemap', None) |
|
274 | 274 | |
|
275 | 275 | def __getitem__(self, i): |
|
276 | 276 | if i == -1: |
|
277 | 277 | return (0, 0, 0, -1, -1, -1, -1, nullid) |
|
278 | 278 | return list.__getitem__(self, i) |
|
279 | 279 | |
|
280 | 280 | |
|
281 | 281 | class revlogoldio(object): |
|
282 | 282 | def __init__(self): |
|
283 | 283 | self.size = indexformatv0.size |
|
284 | 284 | |
|
285 | 285 | def parseindex(self, data, inline): |
|
286 | 286 | s = self.size |
|
287 | 287 | index = [] |
|
288 | 288 | nodemap = nodemaputil.NodeMap({nullid: nullrev}) |
|
289 | 289 | n = off = 0 |
|
290 | 290 | l = len(data) |
|
291 | 291 | while off + s <= l: |
|
292 | 292 | cur = data[off : off + s] |
|
293 | 293 | off += s |
|
294 | 294 | e = indexformatv0_unpack(cur) |
|
295 | 295 | # transform to revlogv1 format |
|
296 | 296 | e2 = ( |
|
297 | 297 | offset_type(e[0], 0), |
|
298 | 298 | e[1], |
|
299 | 299 | -1, |
|
300 | 300 | e[2], |
|
301 | 301 | e[3], |
|
302 | 302 | nodemap.get(e[4], nullrev), |
|
303 | 303 | nodemap.get(e[5], nullrev), |
|
304 | 304 | e[6], |
|
305 | 305 | ) |
|
306 | 306 | index.append(e2) |
|
307 | 307 | nodemap[e[6]] = n |
|
308 | 308 | n += 1 |
|
309 | 309 | |
|
310 | 310 | index = revlogoldindex(index) |
|
311 | 311 | return index, None |
|
312 | 312 | |
|
313 | 313 | def packentry(self, entry, node, version, rev): |
|
314 | 314 | if gettype(entry[0]): |
|
315 | 315 | raise error.RevlogError( |
|
316 | 316 | _(b'index entry flags need revlog version 1') |
|
317 | 317 | ) |
|
318 | 318 | e2 = ( |
|
319 | 319 | getoffset(entry[0]), |
|
320 | 320 | entry[1], |
|
321 | 321 | entry[3], |
|
322 | 322 | entry[4], |
|
323 | 323 | node(entry[5]), |
|
324 | 324 | node(entry[6]), |
|
325 | 325 | entry[7], |
|
326 | 326 | ) |
|
327 | 327 | return indexformatv0_pack(*e2) |
|
328 | 328 | |
|
329 | 329 | |
|
330 | 330 | # index ng: |
|
331 | 331 | # 6 bytes: offset |
|
332 | 332 | # 2 bytes: flags |
|
333 | 333 | # 4 bytes: compressed length |
|
334 | 334 | # 4 bytes: uncompressed length |
|
335 | 335 | # 4 bytes: base rev |
|
336 | 336 | # 4 bytes: link rev |
|
337 | 337 | # 4 bytes: parent 1 rev |
|
338 | 338 | # 4 bytes: parent 2 rev |
|
339 | 339 | # 32 bytes: nodeid |
|
340 | 340 | indexformatng = struct.Struct(b">Qiiiiii20s12x") |
|
341 | 341 | indexformatng_pack = indexformatng.pack |
|
342 | 342 | versionformat = struct.Struct(b">I") |
|
343 | 343 | versionformat_pack = versionformat.pack |
|
344 | 344 | versionformat_unpack = versionformat.unpack |
|
345 | 345 | |
|
346 | 346 | # corresponds to uncompressed length of indexformatng (2 gigs, 4-byte |
|
347 | 347 | # signed integer) |
|
348 | 348 | _maxentrysize = 0x7FFFFFFF |
|
349 | 349 | |
|
350 | 350 | |
|
351 | 351 | class revlogio(object): |
|
352 | 352 | def __init__(self): |
|
353 | 353 | self.size = indexformatng.size |
|
354 | 354 | |
|
355 | 355 | def parseindex(self, data, inline): |
|
356 | 356 | # call the C implementation to parse the index data |
|
357 | 357 | index, cache = parsers.parse_index2(data, inline) |
|
358 | 358 | return index, cache |
|
359 | 359 | |
|
360 | 360 | def packentry(self, entry, node, version, rev): |
|
361 | 361 | p = indexformatng_pack(*entry) |
|
362 | 362 | if rev == 0: |
|
363 | 363 | p = versionformat_pack(version) + p[4:] |
|
364 | 364 | return p |
|
365 | 365 | |
|
366 | 366 | |
|
367 | 367 | NodemapRevlogIO = None |
|
368 | 368 | |
|
369 | 369 | if util.safehasattr(parsers, 'parse_index_devel_nodemap'): |
|
370 | 370 | |
|
371 | 371 | class NodemapRevlogIO(revlogio): |
|
372 | 372 | """A debug oriented IO class that return a PersistentNodeMapIndexObject |
|
373 | 373 | |
|
374 | 374 | The PersistentNodeMapIndexObject object is meant to test the persistent nodemap feature. |
|
375 | 375 | """ |
|
376 | 376 | |
|
377 | 377 | def parseindex(self, data, inline): |
|
378 | 378 | index, cache = parsers.parse_index_devel_nodemap(data, inline) |
|
379 | 379 | return index, cache |
|
380 | 380 | |
|
381 | 381 | |
|
382 | 382 | class rustrevlogio(revlogio): |
|
383 | 383 | def parseindex(self, data, inline): |
|
384 | 384 | index, cache = super(rustrevlogio, self).parseindex(data, inline) |
|
385 | 385 | return rustrevlog.MixedIndex(index), cache |
|
386 | 386 | |
|
387 | 387 | |
|
388 | 388 | class revlog(object): |
|
389 | 389 | """ |
|
390 | 390 | the underlying revision storage object |
|
391 | 391 | |
|
392 | 392 | A revlog consists of two parts, an index and the revision data. |
|
393 | 393 | |
|
394 | 394 | The index is a file with a fixed record size containing |
|
395 | 395 | information on each revision, including its nodeid (hash), the |
|
396 | 396 | nodeids of its parents, the position and offset of its data within |
|
397 | 397 | the data file, and the revision it's based on. Finally, each entry |
|
398 | 398 | contains a linkrev entry that can serve as a pointer to external |
|
399 | 399 | data. |
|
400 | 400 | |
|
401 | 401 | The revision data itself is a linear collection of data chunks. |
|
402 | 402 | Each chunk represents a revision and is usually represented as a |
|
403 | 403 | delta against the previous chunk. To bound lookup time, runs of |
|
404 | 404 | deltas are limited to about 2 times the length of the original |
|
405 | 405 | version data. This makes retrieval of a version proportional to |
|
406 | 406 | its size, or O(1) relative to the number of revisions. |
|
407 | 407 | |
|
408 | 408 | Both pieces of the revlog are written to in an append-only |
|
409 | 409 | fashion, which means we never need to rewrite a file to insert or |
|
410 | 410 | remove data, and can use some simple techniques to avoid the need |
|
411 | 411 | for locking while reading. |
|
412 | 412 | |
|
413 | 413 | If checkambig, indexfile is opened with checkambig=True at |
|
414 | 414 | writing, to avoid file stat ambiguity. |
|
415 | 415 | |
|
416 | 416 | If mmaplargeindex is True, and an mmapindexthreshold is set, the |
|
417 | 417 | index will be mmapped rather than read if it is larger than the |
|
418 | 418 | configured threshold. |
|
419 | 419 | |
|
420 | 420 | If censorable is True, the revlog can have censored revisions. |
|
421 | 421 | |
|
422 | 422 | If `upperboundcomp` is not None, this is the expected maximal gain from |
|
423 | 423 | compression for the data content. |
|
424 | 424 | """ |
|
425 | 425 | |
|
426 | 426 | _flagserrorclass = error.RevlogError |
|
427 | 427 | |
|
428 | 428 | def __init__( |
|
429 | 429 | self, |
|
430 | 430 | opener, |
|
431 | 431 | indexfile, |
|
432 | 432 | datafile=None, |
|
433 | 433 | checkambig=False, |
|
434 | 434 | mmaplargeindex=False, |
|
435 | 435 | censorable=False, |
|
436 | 436 | upperboundcomp=None, |
|
437 | 437 | persistentnodemap=False, |
|
438 | 438 | ): |
|
439 | 439 | """ |
|
440 | 440 | create a revlog object |
|
441 | 441 | |
|
442 | 442 | opener is a function that abstracts the file opening operation |
|
443 | 443 | and can be used to implement COW semantics or the like. |
|
444 | 444 | |
|
445 | 445 | """ |
|
446 | 446 | self.upperboundcomp = upperboundcomp |
|
447 | 447 | self.indexfile = indexfile |
|
448 | 448 | self.datafile = datafile or (indexfile[:-2] + b".d") |
|
449 | 449 | self.nodemap_file = None |
|
450 | 450 | if persistentnodemap: |
|
451 | 451 | self.nodemap_file = nodemaputil.get_nodemap_file( |
|
452 | 452 | opener, self.indexfile |
|
453 | 453 | ) |
|
454 | 454 | |
|
455 | 455 | self.opener = opener |
|
456 | 456 | # When True, indexfile is opened with checkambig=True at writing, to |
|
457 | 457 | # avoid file stat ambiguity. |
|
458 | 458 | self._checkambig = checkambig |
|
459 | 459 | self._mmaplargeindex = mmaplargeindex |
|
460 | 460 | self._censorable = censorable |
|
461 | 461 | # 3-tuple of (node, rev, text) for a raw revision. |
|
462 | 462 | self._revisioncache = None |
|
463 | 463 | # Maps rev to chain base rev. |
|
464 | 464 | self._chainbasecache = util.lrucachedict(100) |
|
465 | 465 | # 2-tuple of (offset, data) of raw data from the revlog at an offset. |
|
466 | 466 | self._chunkcache = (0, b'') |
|
467 | 467 | # How much data to read and cache into the raw revlog data cache. |
|
468 | 468 | self._chunkcachesize = 65536 |
|
469 | 469 | self._maxchainlen = None |
|
470 | 470 | self._deltabothparents = True |
|
471 | 471 | self.index = None |
|
472 | 472 | self._nodemap_docket = None |
|
473 | 473 | # Mapping of partial identifiers to full nodes. |
|
474 | 474 | self._pcache = {} |
|
475 | 475 | # Mapping of revision integer to full node. |
|
476 | 476 | self._compengine = b'zlib' |
|
477 | 477 | self._compengineopts = {} |
|
478 | 478 | self._maxdeltachainspan = -1 |
|
479 | 479 | self._withsparseread = False |
|
480 | 480 | self._sparserevlog = False |
|
481 | 481 | self._srdensitythreshold = 0.50 |
|
482 | 482 | self._srmingapsize = 262144 |
|
483 | 483 | |
|
484 | 484 | # Make copy of flag processors so each revlog instance can support |
|
485 | 485 | # custom flags. |
|
486 | 486 | self._flagprocessors = dict(flagutil.flagprocessors) |
|
487 | 487 | |
|
488 | 488 | # 2-tuple of file handles being used for active writing. |
|
489 | 489 | self._writinghandles = None |
|
490 | 490 | |
|
491 | 491 | self._loadindex() |
|
492 | 492 | |
|
493 | 493 | def _loadindex(self): |
|
494 | 494 | mmapindexthreshold = None |
|
495 | 495 | opts = self.opener.options |
|
496 | 496 | |
|
497 | 497 | if b'revlogv2' in opts: |
|
498 | 498 | newversionflags = REVLOGV2 | FLAG_INLINE_DATA |
|
499 | 499 | elif b'revlogv1' in opts: |
|
500 | 500 | newversionflags = REVLOGV1 | FLAG_INLINE_DATA |
|
501 | 501 | if b'generaldelta' in opts: |
|
502 | 502 | newversionflags |= FLAG_GENERALDELTA |
|
503 | 503 | elif b'revlogv0' in self.opener.options: |
|
504 | 504 | newversionflags = REVLOGV0 |
|
505 | 505 | else: |
|
506 | 506 | newversionflags = REVLOG_DEFAULT_VERSION |
|
507 | 507 | |
|
508 | 508 | if b'chunkcachesize' in opts: |
|
509 | 509 | self._chunkcachesize = opts[b'chunkcachesize'] |
|
510 | 510 | if b'maxchainlen' in opts: |
|
511 | 511 | self._maxchainlen = opts[b'maxchainlen'] |
|
512 | 512 | if b'deltabothparents' in opts: |
|
513 | 513 | self._deltabothparents = opts[b'deltabothparents'] |
|
514 | 514 | self._lazydelta = bool(opts.get(b'lazydelta', True)) |
|
515 | 515 | self._lazydeltabase = False |
|
516 | 516 | if self._lazydelta: |
|
517 | 517 | self._lazydeltabase = bool(opts.get(b'lazydeltabase', False)) |
|
518 | 518 | if b'compengine' in opts: |
|
519 | 519 | self._compengine = opts[b'compengine'] |
|
520 | 520 | if b'zlib.level' in opts: |
|
521 | 521 | self._compengineopts[b'zlib.level'] = opts[b'zlib.level'] |
|
522 | 522 | if b'zstd.level' in opts: |
|
523 | 523 | self._compengineopts[b'zstd.level'] = opts[b'zstd.level'] |
|
524 | 524 | if b'maxdeltachainspan' in opts: |
|
525 | 525 | self._maxdeltachainspan = opts[b'maxdeltachainspan'] |
|
526 | 526 | if self._mmaplargeindex and b'mmapindexthreshold' in opts: |
|
527 | 527 | mmapindexthreshold = opts[b'mmapindexthreshold'] |
|
528 | 528 | self.hassidedata = bool(opts.get(b'side-data', False)) |
|
529 | 529 | if self.hassidedata: |
|
530 | 530 | self._flagprocessors[REVIDX_SIDEDATA] = sidedatautil.processors |
|
531 | 531 | self._sparserevlog = bool(opts.get(b'sparse-revlog', False)) |
|
532 | 532 | withsparseread = bool(opts.get(b'with-sparse-read', False)) |
|
533 | 533 | # sparse-revlog forces sparse-read |
|
534 | 534 | self._withsparseread = self._sparserevlog or withsparseread |
|
535 | 535 | if b'sparse-read-density-threshold' in opts: |
|
536 | 536 | self._srdensitythreshold = opts[b'sparse-read-density-threshold'] |
|
537 | 537 | if b'sparse-read-min-gap-size' in opts: |
|
538 | 538 | self._srmingapsize = opts[b'sparse-read-min-gap-size'] |
|
539 | 539 | if opts.get(b'enableellipsis'): |
|
540 | 540 | self._flagprocessors[REVIDX_ELLIPSIS] = ellipsisprocessor |
|
541 | 541 | |
|
542 | 542 | # revlog v0 doesn't have flag processors |
|
543 | 543 | for flag, processor in pycompat.iteritems( |
|
544 | 544 | opts.get(b'flagprocessors', {}) |
|
545 | 545 | ): |
|
546 | 546 | flagutil.insertflagprocessor(flag, processor, self._flagprocessors) |
|
547 | 547 | |
|
548 | 548 | if self._chunkcachesize <= 0: |
|
549 | 549 | raise error.RevlogError( |
|
550 | 550 | _(b'revlog chunk cache size %r is not greater than 0') |
|
551 | 551 | % self._chunkcachesize |
|
552 | 552 | ) |
|
553 | 553 | elif self._chunkcachesize & (self._chunkcachesize - 1): |
|
554 | 554 | raise error.RevlogError( |
|
555 | 555 | _(b'revlog chunk cache size %r is not a power of 2') |
|
556 | 556 | % self._chunkcachesize |
|
557 | 557 | ) |
|
558 | 558 | |
|
559 | 559 | indexdata = b'' |
|
560 | 560 | self._initempty = True |
|
561 | 561 | try: |
|
562 | 562 | with self._indexfp() as f: |
|
563 | 563 | if ( |
|
564 | 564 | mmapindexthreshold is not None |
|
565 | 565 | and self.opener.fstat(f).st_size >= mmapindexthreshold |
|
566 | 566 | ): |
|
567 | 567 | # TODO: should .close() to release resources without |
|
568 | 568 | # relying on Python GC |
|
569 | 569 | indexdata = util.buffer(util.mmapread(f)) |
|
570 | 570 | else: |
|
571 | 571 | indexdata = f.read() |
|
572 | 572 | if len(indexdata) > 0: |
|
573 | 573 | versionflags = versionformat_unpack(indexdata[:4])[0] |
|
574 | 574 | self._initempty = False |
|
575 | 575 | else: |
|
576 | 576 | versionflags = newversionflags |
|
577 | 577 | except IOError as inst: |
|
578 | 578 | if inst.errno != errno.ENOENT: |
|
579 | 579 | raise |
|
580 | 580 | |
|
581 | 581 | versionflags = newversionflags |
|
582 | 582 | |
|
583 | 583 | self.version = versionflags |
|
584 | 584 | |
|
585 | 585 | flags = versionflags & ~0xFFFF |
|
586 | 586 | fmt = versionflags & 0xFFFF |
|
587 | 587 | |
|
588 | 588 | if fmt == REVLOGV0: |
|
589 | 589 | if flags: |
|
590 | 590 | raise error.RevlogError( |
|
591 | 591 | _(b'unknown flags (%#04x) in version %d revlog %s') |
|
592 | 592 | % (flags >> 16, fmt, self.indexfile) |
|
593 | 593 | ) |
|
594 | 594 | |
|
595 | 595 | self._inline = False |
|
596 | 596 | self._generaldelta = False |
|
597 | 597 | |
|
598 | 598 | elif fmt == REVLOGV1: |
|
599 | 599 | if flags & ~REVLOGV1_FLAGS: |
|
600 | 600 | raise error.RevlogError( |
|
601 | 601 | _(b'unknown flags (%#04x) in version %d revlog %s') |
|
602 | 602 | % (flags >> 16, fmt, self.indexfile) |
|
603 | 603 | ) |
|
604 | 604 | |
|
605 | 605 | self._inline = versionflags & FLAG_INLINE_DATA |
|
606 | 606 | self._generaldelta = versionflags & FLAG_GENERALDELTA |
|
607 | 607 | |
|
608 | 608 | elif fmt == REVLOGV2: |
|
609 | 609 | if flags & ~REVLOGV2_FLAGS: |
|
610 | 610 | raise error.RevlogError( |
|
611 | 611 | _(b'unknown flags (%#04x) in version %d revlog %s') |
|
612 | 612 | % (flags >> 16, fmt, self.indexfile) |
|
613 | 613 | ) |
|
614 | 614 | |
|
615 | 615 | self._inline = versionflags & FLAG_INLINE_DATA |
|
616 | 616 | # generaldelta implied by version 2 revlogs. |
|
617 | 617 | self._generaldelta = True |
|
618 | 618 | |
|
619 | 619 | else: |
|
620 | 620 | raise error.RevlogError( |
|
621 | 621 | _(b'unknown version (%d) in revlog %s') % (fmt, self.indexfile) |
|
622 | 622 | ) |
|
623 | 623 | # sparse-revlog can't be on without general-delta (issue6056) |
|
624 | 624 | if not self._generaldelta: |
|
625 | 625 | self._sparserevlog = False |
|
626 | 626 | |
|
627 | 627 | self._storedeltachains = True |
|
628 | 628 | |
|
629 | 629 | devel_nodemap = ( |
|
630 | 630 | self.nodemap_file |
|
631 | 631 | and opts.get(b'devel-force-nodemap', False) |
|
632 | 632 | and NodemapRevlogIO is not None |
|
633 | 633 | ) |
|
634 | 634 | |
|
635 | 635 | use_rust_index = False |
|
636 | 636 | if rustrevlog is not None: |
|
637 | 637 | if self.nodemap_file is not None: |
|
638 | 638 | use_rust_index = True |
|
639 | 639 | else: |
|
640 | 640 | use_rust_index = self.opener.options.get(b'rust.index') |
|
641 | 641 | |
|
642 | 642 | self._io = revlogio() |
|
643 | 643 | if self.version == REVLOGV0: |
|
644 | 644 | self._io = revlogoldio() |
|
645 | 645 | elif devel_nodemap: |
|
646 | 646 | self._io = NodemapRevlogIO() |
|
647 | 647 | elif use_rust_index: |
|
648 | 648 | self._io = rustrevlogio() |
|
649 | 649 | try: |
|
650 | 650 | d = self._io.parseindex(indexdata, self._inline) |
|
651 | 651 | index, _chunkcache = d |
|
652 | 652 | use_nodemap = ( |
|
653 | 653 | not self._inline |
|
654 | 654 | and self.nodemap_file is not None |
|
655 | 655 | and util.safehasattr(index, 'update_nodemap_data') |
|
656 | 656 | ) |
|
657 | 657 | if use_nodemap: |
|
658 | 658 | nodemap_data = nodemaputil.persisted_data(self) |
|
659 | 659 | if nodemap_data is not None: |
|
660 | 660 | docket = nodemap_data[0] |
|
661 | 661 | if ( |
|
662 | 662 | len(d[0]) > docket.tip_rev |
|
663 | 663 | and d[0][docket.tip_rev][7] == docket.tip_node |
|
664 | 664 | ): |
|
665 | 665 | # no changelog tampering |
|
666 | 666 | self._nodemap_docket = docket |
|
667 | 667 | index.update_nodemap_data(*nodemap_data) |
|
668 | 668 | except (ValueError, IndexError): |
|
669 | 669 | raise error.RevlogError( |
|
670 | 670 | _(b"index %s is corrupted") % self.indexfile |
|
671 | 671 | ) |
|
672 | 672 | self.index, self._chunkcache = d |
|
673 | 673 | if not self._chunkcache: |
|
674 | 674 | self._chunkclear() |
|
675 | 675 | # revnum -> (chain-length, sum-delta-length) |
|
676 | 676 | self._chaininfocache = util.lrucachedict(500) |
|
677 | 677 | # revlog header -> revlog compressor |
|
678 | 678 | self._decompressors = {} |
|
679 | 679 | |
|
680 | 680 | @util.propertycache |
|
681 | 681 | def _compressor(self): |
|
682 | 682 | engine = util.compengines[self._compengine] |
|
683 | 683 | return engine.revlogcompressor(self._compengineopts) |
|
684 | 684 | |
|
685 | 685 | def _indexfp(self, mode=b'r'): |
|
686 | 686 | """file object for the revlog's index file""" |
|
687 | 687 | args = {'mode': mode} |
|
688 | 688 | if mode != b'r': |
|
689 | 689 | args['checkambig'] = self._checkambig |
|
690 | 690 | if mode == b'w': |
|
691 | 691 | args['atomictemp'] = True |
|
692 | 692 | return self.opener(self.indexfile, **args) |
|
693 | 693 | |
|
694 | 694 | def _datafp(self, mode=b'r'): |
|
695 | 695 | """file object for the revlog's data file""" |
|
696 | 696 | return self.opener(self.datafile, mode=mode) |
|
697 | 697 | |
|
698 | 698 | @contextlib.contextmanager |
|
699 | 699 | def _datareadfp(self, existingfp=None): |
|
700 | 700 | """file object suitable to read data""" |
|
701 | 701 | # Use explicit file handle, if given. |
|
702 | 702 | if existingfp is not None: |
|
703 | 703 | yield existingfp |
|
704 | 704 | |
|
705 | 705 | # Use a file handle being actively used for writes, if available. |
|
706 | 706 | # There is some danger to doing this because reads will seek the |
|
707 | 707 | # file. However, _writeentry() performs a SEEK_END before all writes, |
|
708 | 708 | # so we should be safe. |
|
709 | 709 | elif self._writinghandles: |
|
710 | 710 | if self._inline: |
|
711 | 711 | yield self._writinghandles[0] |
|
712 | 712 | else: |
|
713 | 713 | yield self._writinghandles[1] |
|
714 | 714 | |
|
715 | 715 | # Otherwise open a new file handle. |
|
716 | 716 | else: |
|
717 | 717 | if self._inline: |
|
718 | 718 | func = self._indexfp |
|
719 | 719 | else: |
|
720 | 720 | func = self._datafp |
|
721 | 721 | with func() as fp: |
|
722 | 722 | yield fp |
|
723 | 723 | |
|
724 | 724 | def tiprev(self): |
|
725 | 725 | return len(self.index) - 1 |
|
726 | 726 | |
|
727 | 727 | def tip(self): |
|
728 | 728 | return self.node(self.tiprev()) |
|
729 | 729 | |
|
730 | 730 | def __contains__(self, rev): |
|
731 | 731 | return 0 <= rev < len(self) |
|
732 | 732 | |
|
733 | 733 | def __len__(self): |
|
734 | 734 | return len(self.index) |
|
735 | 735 | |
|
736 | 736 | def __iter__(self): |
|
737 | 737 | return iter(pycompat.xrange(len(self))) |
|
738 | 738 | |
|
739 | 739 | def revs(self, start=0, stop=None): |
|
740 | 740 | """iterate over all rev in this revlog (from start to stop)""" |
|
741 | 741 | return storageutil.iterrevs(len(self), start=start, stop=stop) |
|
742 | 742 | |
|
743 | 743 | @property |
|
744 | 744 | def nodemap(self): |
|
745 | 745 | msg = ( |
|
746 | 746 | b"revlog.nodemap is deprecated, " |
|
747 | 747 | b"use revlog.index.[has_node|rev|get_rev]" |
|
748 | 748 | ) |
|
749 | 749 | util.nouideprecwarn(msg, b'5.3', stacklevel=2) |
|
750 | 750 | return self.index.nodemap |
|
751 | 751 | |
|
752 | 752 | @property |
|
753 | 753 | def _nodecache(self): |
|
754 | 754 | msg = b"revlog._nodecache is deprecated, use revlog.index.nodemap" |
|
755 | 755 | util.nouideprecwarn(msg, b'5.3', stacklevel=2) |
|
756 | 756 | return self.index.nodemap |
|
757 | 757 | |
|
758 | 758 | def hasnode(self, node): |
|
759 | 759 | try: |
|
760 | 760 | self.rev(node) |
|
761 | 761 | return True |
|
762 | 762 | except KeyError: |
|
763 | 763 | return False |
|
764 | 764 | |
|
765 | 765 | def candelta(self, baserev, rev): |
|
766 | 766 | """whether two revisions (baserev, rev) can be delta-ed or not""" |
|
767 | 767 | # Disable delta if either rev requires a content-changing flag |
|
768 | 768 | # processor (ex. LFS). This is because such flag processor can alter |
|
769 | 769 | # the rawtext content that the delta will be based on, and two clients |
|
770 | 770 | # could have a same revlog node with different flags (i.e. different |
|
771 | 771 | # rawtext contents) and the delta could be incompatible. |
|
772 | 772 | if (self.flags(baserev) & REVIDX_RAWTEXT_CHANGING_FLAGS) or ( |
|
773 | 773 | self.flags(rev) & REVIDX_RAWTEXT_CHANGING_FLAGS |
|
774 | 774 | ): |
|
775 | 775 | return False |
|
776 | 776 | return True |
|
777 | 777 | |
|
778 | 778 | def update_caches(self, transaction): |
|
779 | 779 | if self.nodemap_file is not None: |
|
780 | 780 | if transaction is None: |
|
781 | 781 | nodemaputil.update_persistent_nodemap(self) |
|
782 | 782 | else: |
|
783 | 783 | nodemaputil.setup_persistent_nodemap(transaction, self) |
|
784 | 784 | |
|
785 | 785 | def clearcaches(self): |
|
786 | 786 | self._revisioncache = None |
|
787 | 787 | self._chainbasecache.clear() |
|
788 | 788 | self._chunkcache = (0, b'') |
|
789 | 789 | self._pcache = {} |
|
790 | 790 | self._nodemap_docket = None |
|
791 | 791 | self.index.clearcaches() |
|
792 | 792 | # The python code is the one responsible for validating the docket, we |
|
793 | 793 | # end up having to refresh it here. |
|
794 | 794 | use_nodemap = ( |
|
795 | 795 | not self._inline |
|
796 | 796 | and self.nodemap_file is not None |
|
797 | 797 | and util.safehasattr(self.index, 'update_nodemap_data') |
|
798 | 798 | ) |
|
799 | 799 | if use_nodemap: |
|
800 | 800 | nodemap_data = nodemaputil.persisted_data(self) |
|
801 | 801 | if nodemap_data is not None: |
|
802 | 802 | self._nodemap_docket = nodemap_data[0] |
|
803 | 803 | self.index.update_nodemap_data(*nodemap_data) |
|
804 | 804 | |
|
805 | 805 | def rev(self, node): |
|
806 | 806 | try: |
|
807 | 807 | return self.index.rev(node) |
|
808 | 808 | except TypeError: |
|
809 | 809 | raise |
|
810 | 810 | except error.RevlogError: |
|
811 | 811 | # parsers.c radix tree lookup failed |
|
812 | 812 | if node == wdirid or node in wdirfilenodeids: |
|
813 | 813 | raise error.WdirUnsupported |
|
814 | 814 | raise error.LookupError(node, self.indexfile, _(b'no node')) |
|
815 | 815 | |
|
816 | 816 | # Accessors for index entries. |
|
817 | 817 | |
|
818 | 818 | # First tuple entry is 8 bytes. First 6 bytes are offset. Last 2 bytes |
|
819 | 819 | # are flags. |
|
820 | 820 | def start(self, rev): |
|
821 | 821 | return int(self.index[rev][0] >> 16) |
|
822 | 822 | |
|
823 | 823 | def flags(self, rev): |
|
824 | 824 | return self.index[rev][0] & 0xFFFF |
|
825 | 825 | |
|
826 | 826 | def length(self, rev): |
|
827 | 827 | return self.index[rev][1] |
|
828 | 828 | |
|
829 | 829 | def rawsize(self, rev): |
|
830 | 830 | """return the length of the uncompressed text for a given revision""" |
|
831 | 831 | l = self.index[rev][2] |
|
832 | 832 | if l >= 0: |
|
833 | 833 | return l |
|
834 | 834 | |
|
835 | 835 | t = self.rawdata(rev) |
|
836 | 836 | return len(t) |
|
837 | 837 | |
|
838 | 838 | def size(self, rev): |
|
839 | 839 | """length of non-raw text (processed by a "read" flag processor)""" |
|
840 | 840 | # fast path: if no "read" flag processor could change the content, |
|
841 | 841 | # size is rawsize. note: ELLIPSIS is known to not change the content. |
|
842 | 842 | flags = self.flags(rev) |
|
843 | 843 | if flags & (flagutil.REVIDX_KNOWN_FLAGS ^ REVIDX_ELLIPSIS) == 0: |
|
844 | 844 | return self.rawsize(rev) |
|
845 | 845 | |
|
846 | 846 | return len(self.revision(rev, raw=False)) |
|
847 | 847 | |
|
848 | 848 | def chainbase(self, rev): |
|
849 | 849 | base = self._chainbasecache.get(rev) |
|
850 | 850 | if base is not None: |
|
851 | 851 | return base |
|
852 | 852 | |
|
853 | 853 | index = self.index |
|
854 | 854 | iterrev = rev |
|
855 | 855 | base = index[iterrev][3] |
|
856 | 856 | while base != iterrev: |
|
857 | 857 | iterrev = base |
|
858 | 858 | base = index[iterrev][3] |
|
859 | 859 | |
|
860 | 860 | self._chainbasecache[rev] = base |
|
861 | 861 | return base |
|
862 | 862 | |
|
863 | 863 | def linkrev(self, rev): |
|
864 | 864 | return self.index[rev][4] |
|
865 | 865 | |
|
866 | 866 | def parentrevs(self, rev): |
|
867 | 867 | try: |
|
868 | 868 | entry = self.index[rev] |
|
869 | 869 | except IndexError: |
|
870 | 870 | if rev == wdirrev: |
|
871 | 871 | raise error.WdirUnsupported |
|
872 | 872 | raise |
|
873 | 873 | |
|
874 | 874 | return entry[5], entry[6] |
|
875 | 875 | |
|
876 | 876 | # fast parentrevs(rev) where rev isn't filtered |
|
877 | 877 | _uncheckedparentrevs = parentrevs |
|
878 | 878 | |
|
879 | 879 | def node(self, rev): |
|
880 | 880 | try: |
|
881 | 881 | return self.index[rev][7] |
|
882 | 882 | except IndexError: |
|
883 | 883 | if rev == wdirrev: |
|
884 | 884 | raise error.WdirUnsupported |
|
885 | 885 | raise |
|
886 | 886 | |
|
887 | 887 | # Derived from index values. |
|
888 | 888 | |
|
889 | 889 | def end(self, rev): |
|
890 | 890 | return self.start(rev) + self.length(rev) |
|
891 | 891 | |
|
892 | 892 | def parents(self, node): |
|
893 | 893 | i = self.index |
|
894 | 894 | d = i[self.rev(node)] |
|
895 | 895 | return i[d[5]][7], i[d[6]][7] # map revisions to nodes inline |
|
896 | 896 | |
|
897 | 897 | def chainlen(self, rev): |
|
898 | 898 | return self._chaininfo(rev)[0] |
|
899 | 899 | |
|
900 | 900 | def _chaininfo(self, rev): |
|
901 | 901 | chaininfocache = self._chaininfocache |
|
902 | 902 | if rev in chaininfocache: |
|
903 | 903 | return chaininfocache[rev] |
|
904 | 904 | index = self.index |
|
905 | 905 | generaldelta = self._generaldelta |
|
906 | 906 | iterrev = rev |
|
907 | 907 | e = index[iterrev] |
|
908 | 908 | clen = 0 |
|
909 | 909 | compresseddeltalen = 0 |
|
910 | 910 | while iterrev != e[3]: |
|
911 | 911 | clen += 1 |
|
912 | 912 | compresseddeltalen += e[1] |
|
913 | 913 | if generaldelta: |
|
914 | 914 | iterrev = e[3] |
|
915 | 915 | else: |
|
916 | 916 | iterrev -= 1 |
|
917 | 917 | if iterrev in chaininfocache: |
|
918 | 918 | t = chaininfocache[iterrev] |
|
919 | 919 | clen += t[0] |
|
920 | 920 | compresseddeltalen += t[1] |
|
921 | 921 | break |
|
922 | 922 | e = index[iterrev] |
|
923 | 923 | else: |
|
924 | 924 | # Add text length of base since decompressing that also takes |
|
925 | 925 | # work. For cache hits the length is already included. |
|
926 | 926 | compresseddeltalen += e[1] |
|
927 | 927 | r = (clen, compresseddeltalen) |
|
928 | 928 | chaininfocache[rev] = r |
|
929 | 929 | return r |
|
930 | 930 | |
|
931 | 931 | def _deltachain(self, rev, stoprev=None): |
|
932 | 932 | """Obtain the delta chain for a revision. |
|
933 | 933 | |
|
934 | 934 | ``stoprev`` specifies a revision to stop at. If not specified, we |
|
935 | 935 | stop at the base of the chain. |
|
936 | 936 | |
|
937 | 937 | Returns a 2-tuple of (chain, stopped) where ``chain`` is a list of |
|
938 | 938 | revs in ascending order and ``stopped`` is a bool indicating whether |
|
939 | 939 | ``stoprev`` was hit. |
|
940 | 940 | """ |
|
941 | 941 | # Try C implementation. |
|
942 | 942 | try: |
|
943 | 943 | return self.index.deltachain(rev, stoprev, self._generaldelta) |
|
944 | 944 | except AttributeError: |
|
945 | 945 | pass |
|
946 | 946 | |
|
947 | 947 | chain = [] |
|
948 | 948 | |
|
949 | 949 | # Alias to prevent attribute lookup in tight loop. |
|
950 | 950 | index = self.index |
|
951 | 951 | generaldelta = self._generaldelta |
|
952 | 952 | |
|
953 | 953 | iterrev = rev |
|
954 | 954 | e = index[iterrev] |
|
955 | 955 | while iterrev != e[3] and iterrev != stoprev: |
|
956 | 956 | chain.append(iterrev) |
|
957 | 957 | if generaldelta: |
|
958 | 958 | iterrev = e[3] |
|
959 | 959 | else: |
|
960 | 960 | iterrev -= 1 |
|
961 | 961 | e = index[iterrev] |
|
962 | 962 | |
|
963 | 963 | if iterrev == stoprev: |
|
964 | 964 | stopped = True |
|
965 | 965 | else: |
|
966 | 966 | chain.append(iterrev) |
|
967 | 967 | stopped = False |
|
968 | 968 | |
|
969 | 969 | chain.reverse() |
|
970 | 970 | return chain, stopped |
|
971 | 971 | |
|
972 | 972 | def ancestors(self, revs, stoprev=0, inclusive=False): |
|
973 | 973 | """Generate the ancestors of 'revs' in reverse revision order. |
|
974 | 974 | Does not generate revs lower than stoprev. |
|
975 | 975 | |
|
976 | 976 | See the documentation for ancestor.lazyancestors for more details.""" |
|
977 | 977 | |
|
978 | 978 | # first, make sure start revisions aren't filtered |
|
979 | 979 | revs = list(revs) |
|
980 | 980 | checkrev = self.node |
|
981 | 981 | for r in revs: |
|
982 | 982 | checkrev(r) |
|
983 | 983 | # and we're sure ancestors aren't filtered as well |
|
984 | 984 | |
|
985 | 985 | if rustancestor is not None: |
|
986 | 986 | lazyancestors = rustancestor.LazyAncestors |
|
987 | 987 | arg = self.index |
|
988 | 988 | else: |
|
989 | 989 | lazyancestors = ancestor.lazyancestors |
|
990 | 990 | arg = self._uncheckedparentrevs |
|
991 | 991 | return lazyancestors(arg, revs, stoprev=stoprev, inclusive=inclusive) |
|
992 | 992 | |
|
993 | 993 | def descendants(self, revs): |
|
994 | 994 | return dagop.descendantrevs(revs, self.revs, self.parentrevs) |
|
995 | 995 | |
|
996 | 996 | def findcommonmissing(self, common=None, heads=None): |
|
997 | 997 | """Return a tuple of the ancestors of common and the ancestors of heads |
|
998 | 998 | that are not ancestors of common. In revset terminology, we return the |
|
999 | 999 | tuple: |
|
1000 | 1000 | |
|
1001 | 1001 | ::common, (::heads) - (::common) |
|
1002 | 1002 | |
|
1003 | 1003 | The list is sorted by revision number, meaning it is |
|
1004 | 1004 | topologically sorted. |
|
1005 | 1005 | |
|
1006 | 1006 | 'heads' and 'common' are both lists of node IDs. If heads is |
|
1007 | 1007 | not supplied, uses all of the revlog's heads. If common is not |
|
1008 | 1008 | supplied, uses nullid.""" |
|
1009 | 1009 | if common is None: |
|
1010 | 1010 | common = [nullid] |
|
1011 | 1011 | if heads is None: |
|
1012 | 1012 | heads = self.heads() |
|
1013 | 1013 | |
|
1014 | 1014 | common = [self.rev(n) for n in common] |
|
1015 | 1015 | heads = [self.rev(n) for n in heads] |
|
1016 | 1016 | |
|
1017 | 1017 | # we want the ancestors, but inclusive |
|
1018 | 1018 | class lazyset(object): |
|
1019 | 1019 | def __init__(self, lazyvalues): |
|
1020 | 1020 | self.addedvalues = set() |
|
1021 | 1021 | self.lazyvalues = lazyvalues |
|
1022 | 1022 | |
|
1023 | 1023 | def __contains__(self, value): |
|
1024 | 1024 | return value in self.addedvalues or value in self.lazyvalues |
|
1025 | 1025 | |
|
1026 | 1026 | def __iter__(self): |
|
1027 | 1027 | added = self.addedvalues |
|
1028 | 1028 | for r in added: |
|
1029 | 1029 | yield r |
|
1030 | 1030 | for r in self.lazyvalues: |
|
1031 | 1031 | if not r in added: |
|
1032 | 1032 | yield r |
|
1033 | 1033 | |
|
1034 | 1034 | def add(self, value): |
|
1035 | 1035 | self.addedvalues.add(value) |
|
1036 | 1036 | |
|
1037 | 1037 | def update(self, values): |
|
1038 | 1038 | self.addedvalues.update(values) |
|
1039 | 1039 | |
|
1040 | 1040 | has = lazyset(self.ancestors(common)) |
|
1041 | 1041 | has.add(nullrev) |
|
1042 | 1042 | has.update(common) |
|
1043 | 1043 | |
|
1044 | 1044 | # take all ancestors from heads that aren't in has |
|
1045 | 1045 | missing = set() |
|
1046 | 1046 | visit = collections.deque(r for r in heads if r not in has) |
|
1047 | 1047 | while visit: |
|
1048 | 1048 | r = visit.popleft() |
|
1049 | 1049 | if r in missing: |
|
1050 | 1050 | continue |
|
1051 | 1051 | else: |
|
1052 | 1052 | missing.add(r) |
|
1053 | 1053 | for p in self.parentrevs(r): |
|
1054 | 1054 | if p not in has: |
|
1055 | 1055 | visit.append(p) |
|
1056 | 1056 | missing = list(missing) |
|
1057 | 1057 | missing.sort() |
|
1058 | 1058 | return has, [self.node(miss) for miss in missing] |
|
1059 | 1059 | |
|
1060 | 1060 | def incrementalmissingrevs(self, common=None): |
|
1061 | 1061 | """Return an object that can be used to incrementally compute the |
|
1062 | 1062 | revision numbers of the ancestors of arbitrary sets that are not |
|
1063 | 1063 | ancestors of common. This is an ancestor.incrementalmissingancestors |
|
1064 | 1064 | object. |
|
1065 | 1065 | |
|
1066 | 1066 | 'common' is a list of revision numbers. If common is not supplied, uses |
|
1067 | 1067 | nullrev. |
|
1068 | 1068 | """ |
|
1069 | 1069 | if common is None: |
|
1070 | 1070 | common = [nullrev] |
|
1071 | 1071 | |
|
1072 | 1072 | if rustancestor is not None: |
|
1073 | 1073 | return rustancestor.MissingAncestors(self.index, common) |
|
1074 | 1074 | return ancestor.incrementalmissingancestors(self.parentrevs, common) |
|
1075 | 1075 | |
|
1076 | 1076 | def findmissingrevs(self, common=None, heads=None): |
|
1077 | 1077 | """Return the revision numbers of the ancestors of heads that |
|
1078 | 1078 | are not ancestors of common. |
|
1079 | 1079 | |
|
1080 | 1080 | More specifically, return a list of revision numbers corresponding to |
|
1081 | 1081 | nodes N such that every N satisfies the following constraints: |
|
1082 | 1082 | |
|
1083 | 1083 | 1. N is an ancestor of some node in 'heads' |
|
1084 | 1084 | 2. N is not an ancestor of any node in 'common' |
|
1085 | 1085 | |
|
1086 | 1086 | The list is sorted by revision number, meaning it is |
|
1087 | 1087 | topologically sorted. |
|
1088 | 1088 | |
|
1089 | 1089 | 'heads' and 'common' are both lists of revision numbers. If heads is |
|
1090 | 1090 | not supplied, uses all of the revlog's heads. If common is not |
|
1091 | 1091 | supplied, uses nullid.""" |
|
1092 | 1092 | if common is None: |
|
1093 | 1093 | common = [nullrev] |
|
1094 | 1094 | if heads is None: |
|
1095 | 1095 | heads = self.headrevs() |
|
1096 | 1096 | |
|
1097 | 1097 | inc = self.incrementalmissingrevs(common=common) |
|
1098 | 1098 | return inc.missingancestors(heads) |
|
1099 | 1099 | |
|
1100 | 1100 | def findmissing(self, common=None, heads=None): |
|
1101 | 1101 | """Return the ancestors of heads that are not ancestors of common. |
|
1102 | 1102 | |
|
1103 | 1103 | More specifically, return a list of nodes N such that every N |
|
1104 | 1104 | satisfies the following constraints: |
|
1105 | 1105 | |
|
1106 | 1106 | 1. N is an ancestor of some node in 'heads' |
|
1107 | 1107 | 2. N is not an ancestor of any node in 'common' |
|
1108 | 1108 | |
|
1109 | 1109 | The list is sorted by revision number, meaning it is |
|
1110 | 1110 | topologically sorted. |
|
1111 | 1111 | |
|
1112 | 1112 | 'heads' and 'common' are both lists of node IDs. If heads is |
|
1113 | 1113 | not supplied, uses all of the revlog's heads. If common is not |
|
1114 | 1114 | supplied, uses nullid.""" |
|
1115 | 1115 | if common is None: |
|
1116 | 1116 | common = [nullid] |
|
1117 | 1117 | if heads is None: |
|
1118 | 1118 | heads = self.heads() |
|
1119 | 1119 | |
|
1120 | 1120 | common = [self.rev(n) for n in common] |
|
1121 | 1121 | heads = [self.rev(n) for n in heads] |
|
1122 | 1122 | |
|
1123 | 1123 | inc = self.incrementalmissingrevs(common=common) |
|
1124 | 1124 | return [self.node(r) for r in inc.missingancestors(heads)] |
|
1125 | 1125 | |
|
1126 | 1126 | def nodesbetween(self, roots=None, heads=None): |
|
1127 | 1127 | """Return a topological path from 'roots' to 'heads'. |
|
1128 | 1128 | |
|
1129 | 1129 | Return a tuple (nodes, outroots, outheads) where 'nodes' is a |
|
1130 | 1130 | topologically sorted list of all nodes N that satisfy both of |
|
1131 | 1131 | these constraints: |
|
1132 | 1132 | |
|
1133 | 1133 | 1. N is a descendant of some node in 'roots' |
|
1134 | 1134 | 2. N is an ancestor of some node in 'heads' |
|
1135 | 1135 | |
|
1136 | 1136 | Every node is considered to be both a descendant and an ancestor |
|
1137 | 1137 | of itself, so every reachable node in 'roots' and 'heads' will be |
|
1138 | 1138 | included in 'nodes'. |
|
1139 | 1139 | |
|
1140 | 1140 | 'outroots' is the list of reachable nodes in 'roots', i.e., the |
|
1141 | 1141 | subset of 'roots' that is returned in 'nodes'. Likewise, |
|
1142 | 1142 | 'outheads' is the subset of 'heads' that is also in 'nodes'. |
|
1143 | 1143 | |
|
1144 | 1144 | 'roots' and 'heads' are both lists of node IDs. If 'roots' is |
|
1145 | 1145 | unspecified, uses nullid as the only root. If 'heads' is |
|
1146 | 1146 | unspecified, uses list of all of the revlog's heads.""" |
|
1147 | 1147 | nonodes = ([], [], []) |
|
1148 | 1148 | if roots is not None: |
|
1149 | 1149 | roots = list(roots) |
|
1150 | 1150 | if not roots: |
|
1151 | 1151 | return nonodes |
|
1152 | 1152 | lowestrev = min([self.rev(n) for n in roots]) |
|
1153 | 1153 | else: |
|
1154 | 1154 | roots = [nullid] # Everybody's a descendant of nullid |
|
1155 | 1155 | lowestrev = nullrev |
|
1156 | 1156 | if (lowestrev == nullrev) and (heads is None): |
|
1157 | 1157 | # We want _all_ the nodes! |
|
1158 | 1158 | return ([self.node(r) for r in self], [nullid], list(self.heads())) |
|
1159 | 1159 | if heads is None: |
|
1160 | 1160 | # All nodes are ancestors, so the latest ancestor is the last |
|
1161 | 1161 | # node. |
|
1162 | 1162 | highestrev = len(self) - 1 |
|
1163 | 1163 | # Set ancestors to None to signal that every node is an ancestor. |
|
1164 | 1164 | ancestors = None |
|
1165 | 1165 | # Set heads to an empty dictionary for later discovery of heads |
|
1166 | 1166 | heads = {} |
|
1167 | 1167 | else: |
|
1168 | 1168 | heads = list(heads) |
|
1169 | 1169 | if not heads: |
|
1170 | 1170 | return nonodes |
|
1171 | 1171 | ancestors = set() |
|
1172 | 1172 | # Turn heads into a dictionary so we can remove 'fake' heads. |
|
1173 | 1173 | # Also, later we will be using it to filter out the heads we can't |
|
1174 | 1174 | # find from roots. |
|
1175 | 1175 | heads = dict.fromkeys(heads, False) |
|
1176 | 1176 | # Start at the top and keep marking parents until we're done. |
|
1177 | 1177 | nodestotag = set(heads) |
|
1178 | 1178 | # Remember where the top was so we can use it as a limit later. |
|
1179 | 1179 | highestrev = max([self.rev(n) for n in nodestotag]) |
|
1180 | 1180 | while nodestotag: |
|
1181 | 1181 | # grab a node to tag |
|
1182 | 1182 | n = nodestotag.pop() |
|
1183 | 1183 | # Never tag nullid |
|
1184 | 1184 | if n == nullid: |
|
1185 | 1185 | continue |
|
1186 | 1186 | # A node's revision number represents its place in a |
|
1187 | 1187 | # topologically sorted list of nodes. |
|
1188 | 1188 | r = self.rev(n) |
|
1189 | 1189 | if r >= lowestrev: |
|
1190 | 1190 | if n not in ancestors: |
|
1191 | 1191 | # If we are possibly a descendant of one of the roots |
|
1192 | 1192 | # and we haven't already been marked as an ancestor |
|
1193 | 1193 | ancestors.add(n) # Mark as ancestor |
|
1194 | 1194 | # Add non-nullid parents to list of nodes to tag. |
|
1195 | 1195 | nodestotag.update( |
|
1196 | 1196 | [p for p in self.parents(n) if p != nullid] |
|
1197 | 1197 | ) |
|
1198 | 1198 | elif n in heads: # We've seen it before, is it a fake head? |
|
1199 | 1199 | # So it is, real heads should not be the ancestors of |
|
1200 | 1200 | # any other heads. |
|
1201 | 1201 | heads.pop(n) |
|
1202 | 1202 | if not ancestors: |
|
1203 | 1203 | return nonodes |
|
1204 | 1204 | # Now that we have our set of ancestors, we want to remove any |
|
1205 | 1205 | # roots that are not ancestors. |
|
1206 | 1206 | |
|
1207 | 1207 | # If one of the roots was nullid, everything is included anyway. |
|
1208 | 1208 | if lowestrev > nullrev: |
|
1209 | 1209 | # But, since we weren't, let's recompute the lowest rev to not |
|
1210 | 1210 | # include roots that aren't ancestors. |
|
1211 | 1211 | |
|
1212 | 1212 | # Filter out roots that aren't ancestors of heads |
|
1213 | 1213 | roots = [root for root in roots if root in ancestors] |
|
1214 | 1214 | # Recompute the lowest revision |
|
1215 | 1215 | if roots: |
|
1216 | 1216 | lowestrev = min([self.rev(root) for root in roots]) |
|
1217 | 1217 | else: |
|
1218 | 1218 | # No more roots? Return empty list |
|
1219 | 1219 | return nonodes |
|
1220 | 1220 | else: |
|
1221 | 1221 | # We are descending from nullid, and don't need to care about |
|
1222 | 1222 | # any other roots. |
|
1223 | 1223 | lowestrev = nullrev |
|
1224 | 1224 | roots = [nullid] |
|
1225 | 1225 | # Transform our roots list into a set. |
|
1226 | 1226 | descendants = set(roots) |
|
1227 | 1227 | # Also, keep the original roots so we can filter out roots that aren't |
|
1228 | 1228 | # 'real' roots (i.e. are descended from other roots). |
|
1229 | 1229 | roots = descendants.copy() |
|
1230 | 1230 | # Our topologically sorted list of output nodes. |
|
1231 | 1231 | orderedout = [] |
|
1232 | 1232 | # Don't start at nullid since we don't want nullid in our output list, |
|
1233 | 1233 | # and if nullid shows up in descendants, empty parents will look like |
|
1234 | 1234 | # they're descendants. |
|
1235 | 1235 | for r in self.revs(start=max(lowestrev, 0), stop=highestrev + 1): |
|
1236 | 1236 | n = self.node(r) |
|
1237 | 1237 | isdescendant = False |
|
1238 | 1238 | if lowestrev == nullrev: # Everybody is a descendant of nullid |
|
1239 | 1239 | isdescendant = True |
|
1240 | 1240 | elif n in descendants: |
|
1241 | 1241 | # n is already a descendant |
|
1242 | 1242 | isdescendant = True |
|
1243 | 1243 | # This check only needs to be done here because all the roots |
|
1244 | 1244 | # will start being marked is descendants before the loop. |
|
1245 | 1245 | if n in roots: |
|
1246 | 1246 | # If n was a root, check if it's a 'real' root. |
|
1247 | 1247 | p = tuple(self.parents(n)) |
|
1248 | 1248 | # If any of its parents are descendants, it's not a root. |
|
1249 | 1249 | if (p[0] in descendants) or (p[1] in descendants): |
|
1250 | 1250 | roots.remove(n) |
|
1251 | 1251 | else: |
|
1252 | 1252 | p = tuple(self.parents(n)) |
|
1253 | 1253 | # A node is a descendant if either of its parents are |
|
1254 | 1254 | # descendants. (We seeded the dependents list with the roots |
|
1255 | 1255 | # up there, remember?) |
|
1256 | 1256 | if (p[0] in descendants) or (p[1] in descendants): |
|
1257 | 1257 | descendants.add(n) |
|
1258 | 1258 | isdescendant = True |
|
1259 | 1259 | if isdescendant and ((ancestors is None) or (n in ancestors)): |
|
1260 | 1260 | # Only include nodes that are both descendants and ancestors. |
|
1261 | 1261 | orderedout.append(n) |
|
1262 | 1262 | if (ancestors is not None) and (n in heads): |
|
1263 | 1263 | # We're trying to figure out which heads are reachable |
|
1264 | 1264 | # from roots. |
|
1265 | 1265 | # Mark this head as having been reached |
|
1266 | 1266 | heads[n] = True |
|
1267 | 1267 | elif ancestors is None: |
|
1268 | 1268 | # Otherwise, we're trying to discover the heads. |
|
1269 | 1269 | # Assume this is a head because if it isn't, the next step |
|
1270 | 1270 | # will eventually remove it. |
|
1271 | 1271 | heads[n] = True |
|
1272 | 1272 | # But, obviously its parents aren't. |
|
1273 | 1273 | for p in self.parents(n): |
|
1274 | 1274 | heads.pop(p, None) |
|
1275 | 1275 | heads = [head for head, flag in pycompat.iteritems(heads) if flag] |
|
1276 | 1276 | roots = list(roots) |
|
1277 | 1277 | assert orderedout |
|
1278 | 1278 | assert roots |
|
1279 | 1279 | assert heads |
|
1280 | 1280 | return (orderedout, roots, heads) |
|
1281 | 1281 | |
|
1282 | 1282 | def headrevs(self, revs=None): |
|
1283 | 1283 | if revs is None: |
|
1284 | 1284 | try: |
|
1285 | 1285 | return self.index.headrevs() |
|
1286 | 1286 | except AttributeError: |
|
1287 | 1287 | return self._headrevs() |
|
1288 | 1288 | if rustdagop is not None: |
|
1289 | 1289 | return rustdagop.headrevs(self.index, revs) |
|
1290 | 1290 | return dagop.headrevs(revs, self._uncheckedparentrevs) |
|
1291 | 1291 | |
|
1292 | 1292 | def computephases(self, roots): |
|
1293 | 1293 | return self.index.computephasesmapsets(roots) |
|
1294 | 1294 | |
|
1295 | 1295 | def _headrevs(self): |
|
1296 | 1296 | count = len(self) |
|
1297 | 1297 | if not count: |
|
1298 | 1298 | return [nullrev] |
|
1299 | 1299 | # we won't iter over filtered rev so nobody is a head at start |
|
1300 | 1300 | ishead = [0] * (count + 1) |
|
1301 | 1301 | index = self.index |
|
1302 | 1302 | for r in self: |
|
1303 | 1303 | ishead[r] = 1 # I may be an head |
|
1304 | 1304 | e = index[r] |
|
1305 | 1305 | ishead[e[5]] = ishead[e[6]] = 0 # my parent are not |
|
1306 | 1306 | return [r for r, val in enumerate(ishead) if val] |
|
1307 | 1307 | |
|
1308 | 1308 | def heads(self, start=None, stop=None): |
|
1309 | 1309 | """return the list of all nodes that have no children |
|
1310 | 1310 | |
|
1311 | 1311 | if start is specified, only heads that are descendants of |
|
1312 | 1312 | start will be returned |
|
1313 | 1313 | if stop is specified, it will consider all the revs from stop |
|
1314 | 1314 | as if they had no children |
|
1315 | 1315 | """ |
|
1316 | 1316 | if start is None and stop is None: |
|
1317 | 1317 | if not len(self): |
|
1318 | 1318 | return [nullid] |
|
1319 | 1319 | return [self.node(r) for r in self.headrevs()] |
|
1320 | 1320 | |
|
1321 | 1321 | if start is None: |
|
1322 | 1322 | start = nullrev |
|
1323 | 1323 | else: |
|
1324 | 1324 | start = self.rev(start) |
|
1325 | 1325 | |
|
1326 | 1326 | stoprevs = {self.rev(n) for n in stop or []} |
|
1327 | 1327 | |
|
1328 | 1328 | revs = dagop.headrevssubset( |
|
1329 | 1329 | self.revs, self.parentrevs, startrev=start, stoprevs=stoprevs |
|
1330 | 1330 | ) |
|
1331 | 1331 | |
|
1332 | 1332 | return [self.node(rev) for rev in revs] |
|
1333 | 1333 | |
|
1334 | 1334 | def children(self, node): |
|
1335 | 1335 | """find the children of a given node""" |
|
1336 | 1336 | c = [] |
|
1337 | 1337 | p = self.rev(node) |
|
1338 | 1338 | for r in self.revs(start=p + 1): |
|
1339 | 1339 | prevs = [pr for pr in self.parentrevs(r) if pr != nullrev] |
|
1340 | 1340 | if prevs: |
|
1341 | 1341 | for pr in prevs: |
|
1342 | 1342 | if pr == p: |
|
1343 | 1343 | c.append(self.node(r)) |
|
1344 | 1344 | elif p == nullrev: |
|
1345 | 1345 | c.append(self.node(r)) |
|
1346 | 1346 | return c |
|
1347 | 1347 | |
|
1348 | 1348 | def commonancestorsheads(self, a, b): |
|
1349 | 1349 | """calculate all the heads of the common ancestors of nodes a and b""" |
|
1350 | 1350 | a, b = self.rev(a), self.rev(b) |
|
1351 | 1351 | ancs = self._commonancestorsheads(a, b) |
|
1352 | 1352 | return pycompat.maplist(self.node, ancs) |
|
1353 | 1353 | |
|
1354 | 1354 | def _commonancestorsheads(self, *revs): |
|
1355 | 1355 | """calculate all the heads of the common ancestors of revs""" |
|
1356 | 1356 | try: |
|
1357 | 1357 | ancs = self.index.commonancestorsheads(*revs) |
|
1358 | 1358 | except (AttributeError, OverflowError): # C implementation failed |
|
1359 | 1359 | ancs = ancestor.commonancestorsheads(self.parentrevs, *revs) |
|
1360 | 1360 | return ancs |
|
1361 | 1361 | |
|
1362 | 1362 | def isancestor(self, a, b): |
|
1363 | 1363 | """return True if node a is an ancestor of node b |
|
1364 | 1364 | |
|
1365 | 1365 | A revision is considered an ancestor of itself.""" |
|
1366 | 1366 | a, b = self.rev(a), self.rev(b) |
|
1367 | 1367 | return self.isancestorrev(a, b) |
|
1368 | 1368 | |
|
1369 | 1369 | def isancestorrev(self, a, b): |
|
1370 | 1370 | """return True if revision a is an ancestor of revision b |
|
1371 | 1371 | |
|
1372 | 1372 | A revision is considered an ancestor of itself. |
|
1373 | 1373 | |
|
1374 | 1374 | The implementation of this is trivial but the use of |
|
1375 | 1375 | reachableroots is not.""" |
|
1376 | 1376 | if a == nullrev: |
|
1377 | 1377 | return True |
|
1378 | 1378 | elif a == b: |
|
1379 | 1379 | return True |
|
1380 | 1380 | elif a > b: |
|
1381 | 1381 | return False |
|
1382 | 1382 | return bool(self.reachableroots(a, [b], [a], includepath=False)) |
|
1383 | 1383 | |
|
1384 | 1384 | def reachableroots(self, minroot, heads, roots, includepath=False): |
|
1385 | 1385 | """return (heads(::(<roots> and <roots>::<heads>))) |
|
1386 | 1386 | |
|
1387 | 1387 | If includepath is True, return (<roots>::<heads>).""" |
|
1388 | 1388 | try: |
|
1389 | 1389 | return self.index.reachableroots2( |
|
1390 | 1390 | minroot, heads, roots, includepath |
|
1391 | 1391 | ) |
|
1392 | 1392 | except AttributeError: |
|
1393 | 1393 | return dagop._reachablerootspure( |
|
1394 | 1394 | self.parentrevs, minroot, roots, heads, includepath |
|
1395 | 1395 | ) |
|
1396 | 1396 | |
|
1397 | 1397 | def ancestor(self, a, b): |
|
1398 | 1398 | """calculate the "best" common ancestor of nodes a and b""" |
|
1399 | 1399 | |
|
1400 | 1400 | a, b = self.rev(a), self.rev(b) |
|
1401 | 1401 | try: |
|
1402 | 1402 | ancs = self.index.ancestors(a, b) |
|
1403 | 1403 | except (AttributeError, OverflowError): |
|
1404 | 1404 | ancs = ancestor.ancestors(self.parentrevs, a, b) |
|
1405 | 1405 | if ancs: |
|
1406 | 1406 | # choose a consistent winner when there's a tie |
|
1407 | 1407 | return min(map(self.node, ancs)) |
|
1408 | 1408 | return nullid |
|
1409 | 1409 | |
|
1410 | 1410 | def _match(self, id): |
|
1411 | 1411 | if isinstance(id, int): |
|
1412 | 1412 | # rev |
|
1413 | 1413 | return self.node(id) |
|
1414 | 1414 | if len(id) == 20: |
|
1415 | 1415 | # possibly a binary node |
|
1416 | 1416 | # odds of a binary node being all hex in ASCII are 1 in 10**25 |
|
1417 | 1417 | try: |
|
1418 | 1418 | node = id |
|
1419 | 1419 | self.rev(node) # quick search the index |
|
1420 | 1420 | return node |
|
1421 | 1421 | except error.LookupError: |
|
1422 | 1422 | pass # may be partial hex id |
|
1423 | 1423 | try: |
|
1424 | 1424 | # str(rev) |
|
1425 | 1425 | rev = int(id) |
|
1426 | 1426 | if b"%d" % rev != id: |
|
1427 | 1427 | raise ValueError |
|
1428 | 1428 | if rev < 0: |
|
1429 | 1429 | rev = len(self) + rev |
|
1430 | 1430 | if rev < 0 or rev >= len(self): |
|
1431 | 1431 | raise ValueError |
|
1432 | 1432 | return self.node(rev) |
|
1433 | 1433 | except (ValueError, OverflowError): |
|
1434 | 1434 | pass |
|
1435 | 1435 | if len(id) == 40: |
|
1436 | 1436 | try: |
|
1437 | 1437 | # a full hex nodeid? |
|
1438 | 1438 | node = bin(id) |
|
1439 | 1439 | self.rev(node) |
|
1440 | 1440 | return node |
|
1441 | 1441 | except (TypeError, error.LookupError): |
|
1442 | 1442 | pass |
|
1443 | 1443 | |
|
1444 | 1444 | def _partialmatch(self, id): |
|
1445 | 1445 | # we don't care wdirfilenodeids as they should be always full hash |
|
1446 | 1446 | maybewdir = wdirhex.startswith(id) |
|
1447 | 1447 | try: |
|
1448 | 1448 | partial = self.index.partialmatch(id) |
|
1449 | 1449 | if partial and self.hasnode(partial): |
|
1450 | 1450 | if maybewdir: |
|
1451 | 1451 | # single 'ff...' match in radix tree, ambiguous with wdir |
|
1452 | 1452 | raise error.RevlogError |
|
1453 | 1453 | return partial |
|
1454 | 1454 | if maybewdir: |
|
1455 | 1455 | # no 'ff...' match in radix tree, wdir identified |
|
1456 | 1456 | raise error.WdirUnsupported |
|
1457 | 1457 | return None |
|
1458 | 1458 | except error.RevlogError: |
|
1459 | 1459 | # parsers.c radix tree lookup gave multiple matches |
|
1460 | 1460 | # fast path: for unfiltered changelog, radix tree is accurate |
|
1461 | 1461 | if not getattr(self, 'filteredrevs', None): |
|
1462 | 1462 | raise error.AmbiguousPrefixLookupError( |
|
1463 | 1463 | id, self.indexfile, _(b'ambiguous identifier') |
|
1464 | 1464 | ) |
|
1465 | 1465 | # fall through to slow path that filters hidden revisions |
|
1466 | 1466 | except (AttributeError, ValueError): |
|
1467 | 1467 | # we are pure python, or key was too short to search radix tree |
|
1468 | 1468 | pass |
|
1469 | 1469 | |
|
1470 | 1470 | if id in self._pcache: |
|
1471 | 1471 | return self._pcache[id] |
|
1472 | 1472 | |
|
1473 | 1473 | if len(id) <= 40: |
|
1474 | 1474 | try: |
|
1475 | 1475 | # hex(node)[:...] |
|
1476 | 1476 | l = len(id) // 2 # grab an even number of digits |
|
1477 | 1477 | prefix = bin(id[: l * 2]) |
|
1478 | 1478 | nl = [e[7] for e in self.index if e[7].startswith(prefix)] |
|
1479 | 1479 | nl = [ |
|
1480 | 1480 | n for n in nl if hex(n).startswith(id) and self.hasnode(n) |
|
1481 | 1481 | ] |
|
1482 | 1482 | if nullhex.startswith(id): |
|
1483 | 1483 | nl.append(nullid) |
|
1484 | 1484 | if len(nl) > 0: |
|
1485 | 1485 | if len(nl) == 1 and not maybewdir: |
|
1486 | 1486 | self._pcache[id] = nl[0] |
|
1487 | 1487 | return nl[0] |
|
1488 | 1488 | raise error.AmbiguousPrefixLookupError( |
|
1489 | 1489 | id, self.indexfile, _(b'ambiguous identifier') |
|
1490 | 1490 | ) |
|
1491 | 1491 | if maybewdir: |
|
1492 | 1492 | raise error.WdirUnsupported |
|
1493 | 1493 | return None |
|
1494 | 1494 | except TypeError: |
|
1495 | 1495 | pass |
|
1496 | 1496 | |
|
1497 | 1497 | def lookup(self, id): |
|
1498 | 1498 | """locate a node based on: |
|
1499 | 1499 | - revision number or str(revision number) |
|
1500 | 1500 | - nodeid or subset of hex nodeid |
|
1501 | 1501 | """ |
|
1502 | 1502 | n = self._match(id) |
|
1503 | 1503 | if n is not None: |
|
1504 | 1504 | return n |
|
1505 | 1505 | n = self._partialmatch(id) |
|
1506 | 1506 | if n: |
|
1507 | 1507 | return n |
|
1508 | 1508 | |
|
1509 | 1509 | raise error.LookupError(id, self.indexfile, _(b'no match found')) |
|
1510 | 1510 | |
|
1511 | 1511 | def shortest(self, node, minlength=1): |
|
1512 | 1512 | """Find the shortest unambiguous prefix that matches node.""" |
|
1513 | 1513 | |
|
1514 | 1514 | def isvalid(prefix): |
|
1515 | 1515 | try: |
|
1516 | 1516 | matchednode = self._partialmatch(prefix) |
|
1517 | 1517 | except error.AmbiguousPrefixLookupError: |
|
1518 | 1518 | return False |
|
1519 | 1519 | except error.WdirUnsupported: |
|
1520 | 1520 | # single 'ff...' match |
|
1521 | 1521 | return True |
|
1522 | 1522 | if matchednode is None: |
|
1523 | 1523 | raise error.LookupError(node, self.indexfile, _(b'no node')) |
|
1524 | 1524 | return True |
|
1525 | 1525 | |
|
1526 | 1526 | def maybewdir(prefix): |
|
1527 | 1527 | return all(c == b'f' for c in pycompat.iterbytestr(prefix)) |
|
1528 | 1528 | |
|
1529 | 1529 | hexnode = hex(node) |
|
1530 | 1530 | |
|
1531 | 1531 | def disambiguate(hexnode, minlength): |
|
1532 | 1532 | """Disambiguate against wdirid.""" |
|
1533 | 1533 | for length in range(minlength, len(hexnode) + 1): |
|
1534 | 1534 | prefix = hexnode[:length] |
|
1535 | 1535 | if not maybewdir(prefix): |
|
1536 | 1536 | return prefix |
|
1537 | 1537 | |
|
1538 | 1538 | if not getattr(self, 'filteredrevs', None): |
|
1539 | 1539 | try: |
|
1540 | 1540 | length = max(self.index.shortest(node), minlength) |
|
1541 | 1541 | return disambiguate(hexnode, length) |
|
1542 | 1542 | except error.RevlogError: |
|
1543 | 1543 | if node != wdirid: |
|
1544 | 1544 | raise error.LookupError(node, self.indexfile, _(b'no node')) |
|
1545 | 1545 | except AttributeError: |
|
1546 | 1546 | # Fall through to pure code |
|
1547 | 1547 | pass |
|
1548 | 1548 | |
|
1549 | 1549 | if node == wdirid: |
|
1550 | 1550 | for length in range(minlength, len(hexnode) + 1): |
|
1551 | 1551 | prefix = hexnode[:length] |
|
1552 | 1552 | if isvalid(prefix): |
|
1553 | 1553 | return prefix |
|
1554 | 1554 | |
|
1555 | 1555 | for length in range(minlength, len(hexnode) + 1): |
|
1556 | 1556 | prefix = hexnode[:length] |
|
1557 | 1557 | if isvalid(prefix): |
|
1558 | 1558 | return disambiguate(hexnode, length) |
|
1559 | 1559 | |
|
1560 | 1560 | def cmp(self, node, text): |
|
1561 | 1561 | """compare text with a given file revision |
|
1562 | 1562 | |
|
1563 | 1563 | returns True if text is different than what is stored. |
|
1564 | 1564 | """ |
|
1565 | 1565 | p1, p2 = self.parents(node) |
|
1566 | 1566 | return storageutil.hashrevisionsha1(text, p1, p2) != node |
|
1567 | 1567 | |
|
1568 | 1568 | def _cachesegment(self, offset, data): |
|
1569 | 1569 | """Add a segment to the revlog cache. |
|
1570 | 1570 | |
|
1571 | 1571 | Accepts an absolute offset and the data that is at that location. |
|
1572 | 1572 | """ |
|
1573 | 1573 | o, d = self._chunkcache |
|
1574 | 1574 | # try to add to existing cache |
|
1575 | 1575 | if o + len(d) == offset and len(d) + len(data) < _chunksize: |
|
1576 | 1576 | self._chunkcache = o, d + data |
|
1577 | 1577 | else: |
|
1578 | 1578 | self._chunkcache = offset, data |
|
1579 | 1579 | |
|
1580 | 1580 | def _readsegment(self, offset, length, df=None): |
|
1581 | 1581 | """Load a segment of raw data from the revlog. |
|
1582 | 1582 | |
|
1583 | 1583 | Accepts an absolute offset, length to read, and an optional existing |
|
1584 | 1584 | file handle to read from. |
|
1585 | 1585 | |
|
1586 | 1586 | If an existing file handle is passed, it will be seeked and the |
|
1587 | 1587 | original seek position will NOT be restored. |
|
1588 | 1588 | |
|
1589 | 1589 | Returns a str or buffer of raw byte data. |
|
1590 | 1590 | |
|
1591 | 1591 | Raises if the requested number of bytes could not be read. |
|
1592 | 1592 | """ |
|
1593 | 1593 | # Cache data both forward and backward around the requested |
|
1594 | 1594 | # data, in a fixed size window. This helps speed up operations |
|
1595 | 1595 | # involving reading the revlog backwards. |
|
1596 | 1596 | cachesize = self._chunkcachesize |
|
1597 | 1597 | realoffset = offset & ~(cachesize - 1) |
|
1598 | 1598 | reallength = ( |
|
1599 | 1599 | (offset + length + cachesize) & ~(cachesize - 1) |
|
1600 | 1600 | ) - realoffset |
|
1601 | 1601 | with self._datareadfp(df) as df: |
|
1602 | 1602 | df.seek(realoffset) |
|
1603 | 1603 | d = df.read(reallength) |
|
1604 | 1604 | |
|
1605 | 1605 | self._cachesegment(realoffset, d) |
|
1606 | 1606 | if offset != realoffset or reallength != length: |
|
1607 | 1607 | startoffset = offset - realoffset |
|
1608 | 1608 | if len(d) - startoffset < length: |
|
1609 | 1609 | raise error.RevlogError( |
|
1610 | 1610 | _( |
|
1611 | 1611 | b'partial read of revlog %s; expected %d bytes from ' |
|
1612 | 1612 | b'offset %d, got %d' |
|
1613 | 1613 | ) |
|
1614 | 1614 | % ( |
|
1615 | 1615 | self.indexfile if self._inline else self.datafile, |
|
1616 | 1616 | length, |
|
1617 | 1617 | realoffset, |
|
1618 | 1618 | len(d) - startoffset, |
|
1619 | 1619 | ) |
|
1620 | 1620 | ) |
|
1621 | 1621 | |
|
1622 | 1622 | return util.buffer(d, startoffset, length) |
|
1623 | 1623 | |
|
1624 | 1624 | if len(d) < length: |
|
1625 | 1625 | raise error.RevlogError( |
|
1626 | 1626 | _( |
|
1627 | 1627 | b'partial read of revlog %s; expected %d bytes from offset ' |
|
1628 | 1628 | b'%d, got %d' |
|
1629 | 1629 | ) |
|
1630 | 1630 | % ( |
|
1631 | 1631 | self.indexfile if self._inline else self.datafile, |
|
1632 | 1632 | length, |
|
1633 | 1633 | offset, |
|
1634 | 1634 | len(d), |
|
1635 | 1635 | ) |
|
1636 | 1636 | ) |
|
1637 | 1637 | |
|
1638 | 1638 | return d |
|
1639 | 1639 | |
|
1640 | 1640 | def _getsegment(self, offset, length, df=None): |
|
1641 | 1641 | """Obtain a segment of raw data from the revlog. |
|
1642 | 1642 | |
|
1643 | 1643 | Accepts an absolute offset, length of bytes to obtain, and an |
|
1644 | 1644 | optional file handle to the already-opened revlog. If the file |
|
1645 | 1645 | handle is used, it's original seek position will not be preserved. |
|
1646 | 1646 | |
|
1647 | 1647 | Requests for data may be returned from a cache. |
|
1648 | 1648 | |
|
1649 | 1649 | Returns a str or a buffer instance of raw byte data. |
|
1650 | 1650 | """ |
|
1651 | 1651 | o, d = self._chunkcache |
|
1652 | 1652 | l = len(d) |
|
1653 | 1653 | |
|
1654 | 1654 | # is it in the cache? |
|
1655 | 1655 | cachestart = offset - o |
|
1656 | 1656 | cacheend = cachestart + length |
|
1657 | 1657 | if cachestart >= 0 and cacheend <= l: |
|
1658 | 1658 | if cachestart == 0 and cacheend == l: |
|
1659 | 1659 | return d # avoid a copy |
|
1660 | 1660 | return util.buffer(d, cachestart, cacheend - cachestart) |
|
1661 | 1661 | |
|
1662 | 1662 | return self._readsegment(offset, length, df=df) |
|
1663 | 1663 | |
|
1664 | 1664 | def _getsegmentforrevs(self, startrev, endrev, df=None): |
|
1665 | 1665 | """Obtain a segment of raw data corresponding to a range of revisions. |
|
1666 | 1666 | |
|
1667 | 1667 | Accepts the start and end revisions and an optional already-open |
|
1668 | 1668 | file handle to be used for reading. If the file handle is read, its |
|
1669 | 1669 | seek position will not be preserved. |
|
1670 | 1670 | |
|
1671 | 1671 | Requests for data may be satisfied by a cache. |
|
1672 | 1672 | |
|
1673 | 1673 | Returns a 2-tuple of (offset, data) for the requested range of |
|
1674 | 1674 | revisions. Offset is the integer offset from the beginning of the |
|
1675 | 1675 | revlog and data is a str or buffer of the raw byte data. |
|
1676 | 1676 | |
|
1677 | 1677 | Callers will need to call ``self.start(rev)`` and ``self.length(rev)`` |
|
1678 | 1678 | to determine where each revision's data begins and ends. |
|
1679 | 1679 | """ |
|
1680 | 1680 | # Inlined self.start(startrev) & self.end(endrev) for perf reasons |
|
1681 | 1681 | # (functions are expensive). |
|
1682 | 1682 | index = self.index |
|
1683 | 1683 | istart = index[startrev] |
|
1684 | 1684 | start = int(istart[0] >> 16) |
|
1685 | 1685 | if startrev == endrev: |
|
1686 | 1686 | end = start + istart[1] |
|
1687 | 1687 | else: |
|
1688 | 1688 | iend = index[endrev] |
|
1689 | 1689 | end = int(iend[0] >> 16) + iend[1] |
|
1690 | 1690 | |
|
1691 | 1691 | if self._inline: |
|
1692 | 1692 | start += (startrev + 1) * self._io.size |
|
1693 | 1693 | end += (endrev + 1) * self._io.size |
|
1694 | 1694 | length = end - start |
|
1695 | 1695 | |
|
1696 | 1696 | return start, self._getsegment(start, length, df=df) |
|
1697 | 1697 | |
|
1698 | 1698 | def _chunk(self, rev, df=None): |
|
1699 | 1699 | """Obtain a single decompressed chunk for a revision. |
|
1700 | 1700 | |
|
1701 | 1701 | Accepts an integer revision and an optional already-open file handle |
|
1702 | 1702 | to be used for reading. If used, the seek position of the file will not |
|
1703 | 1703 | be preserved. |
|
1704 | 1704 | |
|
1705 | 1705 | Returns a str holding uncompressed data for the requested revision. |
|
1706 | 1706 | """ |
|
1707 | 1707 | return self.decompress(self._getsegmentforrevs(rev, rev, df=df)[1]) |
|
1708 | 1708 | |
|
1709 | 1709 | def _chunks(self, revs, df=None, targetsize=None): |
|
1710 | 1710 | """Obtain decompressed chunks for the specified revisions. |
|
1711 | 1711 | |
|
1712 | 1712 | Accepts an iterable of numeric revisions that are assumed to be in |
|
1713 | 1713 | ascending order. Also accepts an optional already-open file handle |
|
1714 | 1714 | to be used for reading. If used, the seek position of the file will |
|
1715 | 1715 | not be preserved. |
|
1716 | 1716 | |
|
1717 | 1717 | This function is similar to calling ``self._chunk()`` multiple times, |
|
1718 | 1718 | but is faster. |
|
1719 | 1719 | |
|
1720 | 1720 | Returns a list with decompressed data for each requested revision. |
|
1721 | 1721 | """ |
|
1722 | 1722 | if not revs: |
|
1723 | 1723 | return [] |
|
1724 | 1724 | start = self.start |
|
1725 | 1725 | length = self.length |
|
1726 | 1726 | inline = self._inline |
|
1727 | 1727 | iosize = self._io.size |
|
1728 | 1728 | buffer = util.buffer |
|
1729 | 1729 | |
|
1730 | 1730 | l = [] |
|
1731 | 1731 | ladd = l.append |
|
1732 | 1732 | |
|
1733 | 1733 | if not self._withsparseread: |
|
1734 | 1734 | slicedchunks = (revs,) |
|
1735 | 1735 | else: |
|
1736 | 1736 | slicedchunks = deltautil.slicechunk( |
|
1737 | 1737 | self, revs, targetsize=targetsize |
|
1738 | 1738 | ) |
|
1739 | 1739 | |
|
1740 | 1740 | for revschunk in slicedchunks: |
|
1741 | 1741 | firstrev = revschunk[0] |
|
1742 | 1742 | # Skip trailing revisions with empty diff |
|
1743 | 1743 | for lastrev in revschunk[::-1]: |
|
1744 | 1744 | if length(lastrev) != 0: |
|
1745 | 1745 | break |
|
1746 | 1746 | |
|
1747 | 1747 | try: |
|
1748 | 1748 | offset, data = self._getsegmentforrevs(firstrev, lastrev, df=df) |
|
1749 | 1749 | except OverflowError: |
|
1750 | 1750 | # issue4215 - we can't cache a run of chunks greater than |
|
1751 | 1751 | # 2G on Windows |
|
1752 | 1752 | return [self._chunk(rev, df=df) for rev in revschunk] |
|
1753 | 1753 | |
|
1754 | 1754 | decomp = self.decompress |
|
1755 | 1755 | for rev in revschunk: |
|
1756 | 1756 | chunkstart = start(rev) |
|
1757 | 1757 | if inline: |
|
1758 | 1758 | chunkstart += (rev + 1) * iosize |
|
1759 | 1759 | chunklength = length(rev) |
|
1760 | 1760 | ladd(decomp(buffer(data, chunkstart - offset, chunklength))) |
|
1761 | 1761 | |
|
1762 | 1762 | return l |
|
1763 | 1763 | |
|
1764 | 1764 | def _chunkclear(self): |
|
1765 | 1765 | """Clear the raw chunk cache.""" |
|
1766 | 1766 | self._chunkcache = (0, b'') |
|
1767 | 1767 | |
|
1768 | 1768 | def deltaparent(self, rev): |
|
1769 | 1769 | """return deltaparent of the given revision""" |
|
1770 | 1770 | base = self.index[rev][3] |
|
1771 | 1771 | if base == rev: |
|
1772 | 1772 | return nullrev |
|
1773 | 1773 | elif self._generaldelta: |
|
1774 | 1774 | return base |
|
1775 | 1775 | else: |
|
1776 | 1776 | return rev - 1 |
|
1777 | 1777 | |
|
1778 | 1778 | def issnapshot(self, rev): |
|
1779 | 1779 | """tells whether rev is a snapshot""" |
|
1780 | 1780 | if not self._sparserevlog: |
|
1781 | 1781 | return self.deltaparent(rev) == nullrev |
|
1782 | 1782 | elif util.safehasattr(self.index, b'issnapshot'): |
|
1783 | 1783 | # directly assign the method to cache the testing and access |
|
1784 | 1784 | self.issnapshot = self.index.issnapshot |
|
1785 | 1785 | return self.issnapshot(rev) |
|
1786 | 1786 | if rev == nullrev: |
|
1787 | 1787 | return True |
|
1788 | 1788 | entry = self.index[rev] |
|
1789 | 1789 | base = entry[3] |
|
1790 | 1790 | if base == rev: |
|
1791 | 1791 | return True |
|
1792 | 1792 | if base == nullrev: |
|
1793 | 1793 | return True |
|
1794 | 1794 | p1 = entry[5] |
|
1795 | 1795 | p2 = entry[6] |
|
1796 | 1796 | if base == p1 or base == p2: |
|
1797 | 1797 | return False |
|
1798 | 1798 | return self.issnapshot(base) |
|
1799 | 1799 | |
|
1800 | 1800 | def snapshotdepth(self, rev): |
|
1801 | 1801 | """number of snapshot in the chain before this one""" |
|
1802 | 1802 | if not self.issnapshot(rev): |
|
1803 | 1803 | raise error.ProgrammingError(b'revision %d not a snapshot') |
|
1804 | 1804 | return len(self._deltachain(rev)[0]) - 1 |
|
1805 | 1805 | |
|
1806 | 1806 | def revdiff(self, rev1, rev2): |
|
1807 | 1807 | """return or calculate a delta between two revisions |
|
1808 | 1808 | |
|
1809 | 1809 | The delta calculated is in binary form and is intended to be written to |
|
1810 | 1810 | revlog data directly. So this function needs raw revision data. |
|
1811 | 1811 | """ |
|
1812 | 1812 | if rev1 != nullrev and self.deltaparent(rev2) == rev1: |
|
1813 | 1813 | return bytes(self._chunk(rev2)) |
|
1814 | 1814 | |
|
1815 | 1815 | return mdiff.textdiff(self.rawdata(rev1), self.rawdata(rev2)) |
|
1816 | 1816 | |
|
1817 | 1817 | def _processflags(self, text, flags, operation, raw=False): |
|
1818 | 1818 | """deprecated entry point to access flag processors""" |
|
1819 | 1819 | msg = b'_processflag(...) use the specialized variant' |
|
1820 | 1820 | util.nouideprecwarn(msg, b'5.2', stacklevel=2) |
|
1821 | 1821 | if raw: |
|
1822 | 1822 | return text, flagutil.processflagsraw(self, text, flags) |
|
1823 | 1823 | elif operation == b'read': |
|
1824 | 1824 | return flagutil.processflagsread(self, text, flags) |
|
1825 | 1825 | else: # write operation |
|
1826 | 1826 | return flagutil.processflagswrite(self, text, flags, None) |
|
1827 | 1827 | |
|
1828 | 1828 | def revision(self, nodeorrev, _df=None, raw=False): |
|
1829 | 1829 | """return an uncompressed revision of a given node or revision |
|
1830 | 1830 | number. |
|
1831 | 1831 | |
|
1832 | 1832 | _df - an existing file handle to read from. (internal-only) |
|
1833 | 1833 | raw - an optional argument specifying if the revision data is to be |
|
1834 | 1834 | treated as raw data when applying flag transforms. 'raw' should be set |
|
1835 | 1835 | to True when generating changegroups or in debug commands. |
|
1836 | 1836 | """ |
|
1837 | 1837 | if raw: |
|
1838 | 1838 | msg = ( |
|
1839 | 1839 | b'revlog.revision(..., raw=True) is deprecated, ' |
|
1840 | 1840 | b'use revlog.rawdata(...)' |
|
1841 | 1841 | ) |
|
1842 | 1842 | util.nouideprecwarn(msg, b'5.2', stacklevel=2) |
|
1843 | 1843 | return self._revisiondata(nodeorrev, _df, raw=raw)[0] |
|
1844 | 1844 | |
|
1845 | 1845 | def sidedata(self, nodeorrev, _df=None): |
|
1846 | 1846 | """a map of extra data related to the changeset but not part of the hash |
|
1847 | 1847 | |
|
1848 | 1848 | This function currently return a dictionary. However, more advanced |
|
1849 | 1849 | mapping object will likely be used in the future for a more |
|
1850 | 1850 | efficient/lazy code. |
|
1851 | 1851 | """ |
|
1852 | 1852 | return self._revisiondata(nodeorrev, _df)[1] |
|
1853 | 1853 | |
|
1854 | 1854 | def _revisiondata(self, nodeorrev, _df=None, raw=False): |
|
1855 | 1855 | # deal with <nodeorrev> argument type |
|
1856 | 1856 | if isinstance(nodeorrev, int): |
|
1857 | 1857 | rev = nodeorrev |
|
1858 | 1858 | node = self.node(rev) |
|
1859 | 1859 | else: |
|
1860 | 1860 | node = nodeorrev |
|
1861 | 1861 | rev = None |
|
1862 | 1862 | |
|
1863 | 1863 | # fast path the special `nullid` rev |
|
1864 | 1864 | if node == nullid: |
|
1865 | 1865 | return b"", {} |
|
1866 | 1866 | |
|
1867 | 1867 | # ``rawtext`` is the text as stored inside the revlog. Might be the |
|
1868 | 1868 | # revision or might need to be processed to retrieve the revision. |
|
1869 | 1869 | rev, rawtext, validated = self._rawtext(node, rev, _df=_df) |
|
1870 | 1870 | |
|
1871 | 1871 | if raw and validated: |
|
1872 | 1872 | # if we don't want to process the raw text and that raw |
|
1873 | 1873 | # text is cached, we can exit early. |
|
1874 | 1874 | return rawtext, {} |
|
1875 | 1875 | if rev is None: |
|
1876 | 1876 | rev = self.rev(node) |
|
1877 | 1877 | # the revlog's flag for this revision |
|
1878 | 1878 | # (usually alter its state or content) |
|
1879 | 1879 | flags = self.flags(rev) |
|
1880 | 1880 | |
|
1881 | 1881 | if validated and flags == REVIDX_DEFAULT_FLAGS: |
|
1882 | 1882 | # no extra flags set, no flag processor runs, text = rawtext |
|
1883 | 1883 | return rawtext, {} |
|
1884 | 1884 | |
|
1885 | 1885 | sidedata = {} |
|
1886 | 1886 | if raw: |
|
1887 | 1887 | validatehash = flagutil.processflagsraw(self, rawtext, flags) |
|
1888 | 1888 | text = rawtext |
|
1889 | 1889 | else: |
|
1890 | 1890 | try: |
|
1891 | 1891 | r = flagutil.processflagsread(self, rawtext, flags) |
|
1892 | 1892 | except error.SidedataHashError as exc: |
|
1893 | 1893 | msg = _(b"integrity check failed on %s:%s sidedata key %d") |
|
1894 | 1894 | msg %= (self.indexfile, pycompat.bytestr(rev), exc.sidedatakey) |
|
1895 | 1895 | raise error.RevlogError(msg) |
|
1896 | 1896 | text, validatehash, sidedata = r |
|
1897 | 1897 | if validatehash: |
|
1898 | 1898 | self.checkhash(text, node, rev=rev) |
|
1899 | 1899 | if not validated: |
|
1900 | 1900 | self._revisioncache = (node, rev, rawtext) |
|
1901 | 1901 | |
|
1902 | 1902 | return text, sidedata |
|
1903 | 1903 | |
|
1904 | 1904 | def _rawtext(self, node, rev, _df=None): |
|
1905 | 1905 | """return the possibly unvalidated rawtext for a revision |
|
1906 | 1906 | |
|
1907 | 1907 | returns (rev, rawtext, validated) |
|
1908 | 1908 | """ |
|
1909 | 1909 | |
|
1910 | 1910 | # revision in the cache (could be useful to apply delta) |
|
1911 | 1911 | cachedrev = None |
|
1912 | 1912 | # An intermediate text to apply deltas to |
|
1913 | 1913 | basetext = None |
|
1914 | 1914 | |
|
1915 | 1915 | # Check if we have the entry in cache |
|
1916 | 1916 | # The cache entry looks like (node, rev, rawtext) |
|
1917 | 1917 | if self._revisioncache: |
|
1918 | 1918 | if self._revisioncache[0] == node: |
|
1919 | 1919 | return (rev, self._revisioncache[2], True) |
|
1920 | 1920 | cachedrev = self._revisioncache[1] |
|
1921 | 1921 | |
|
1922 | 1922 | if rev is None: |
|
1923 | 1923 | rev = self.rev(node) |
|
1924 | 1924 | |
|
1925 | 1925 | chain, stopped = self._deltachain(rev, stoprev=cachedrev) |
|
1926 | 1926 | if stopped: |
|
1927 | 1927 | basetext = self._revisioncache[2] |
|
1928 | 1928 | |
|
1929 | 1929 | # drop cache to save memory, the caller is expected to |
|
1930 | 1930 | # update self._revisioncache after validating the text |
|
1931 | 1931 | self._revisioncache = None |
|
1932 | 1932 | |
|
1933 | 1933 | targetsize = None |
|
1934 | 1934 | rawsize = self.index[rev][2] |
|
1935 | 1935 | if 0 <= rawsize: |
|
1936 | 1936 | targetsize = 4 * rawsize |
|
1937 | 1937 | |
|
1938 | 1938 | bins = self._chunks(chain, df=_df, targetsize=targetsize) |
|
1939 | 1939 | if basetext is None: |
|
1940 | 1940 | basetext = bytes(bins[0]) |
|
1941 | 1941 | bins = bins[1:] |
|
1942 | 1942 | |
|
1943 | 1943 | rawtext = mdiff.patches(basetext, bins) |
|
1944 | 1944 | del basetext # let us have a chance to free memory early |
|
1945 | 1945 | return (rev, rawtext, False) |
|
1946 | 1946 | |
|
1947 | 1947 | def rawdata(self, nodeorrev, _df=None): |
|
1948 | 1948 | """return an uncompressed raw data of a given node or revision number. |
|
1949 | 1949 | |
|
1950 | 1950 | _df - an existing file handle to read from. (internal-only) |
|
1951 | 1951 | """ |
|
1952 | 1952 | return self._revisiondata(nodeorrev, _df, raw=True)[0] |
|
1953 | 1953 | |
|
1954 | 1954 | def hash(self, text, p1, p2): |
|
1955 | 1955 | """Compute a node hash. |
|
1956 | 1956 | |
|
1957 | 1957 | Available as a function so that subclasses can replace the hash |
|
1958 | 1958 | as needed. |
|
1959 | 1959 | """ |
|
1960 | 1960 | return storageutil.hashrevisionsha1(text, p1, p2) |
|
1961 | 1961 | |
|
1962 | 1962 | def checkhash(self, text, node, p1=None, p2=None, rev=None): |
|
1963 | 1963 | """Check node hash integrity. |
|
1964 | 1964 | |
|
1965 | 1965 | Available as a function so that subclasses can extend hash mismatch |
|
1966 | 1966 | behaviors as needed. |
|
1967 | 1967 | """ |
|
1968 | 1968 | try: |
|
1969 | 1969 | if p1 is None and p2 is None: |
|
1970 | 1970 | p1, p2 = self.parents(node) |
|
1971 | 1971 | if node != self.hash(text, p1, p2): |
|
1972 | 1972 | # Clear the revision cache on hash failure. The revision cache |
|
1973 | 1973 | # only stores the raw revision and clearing the cache does have |
|
1974 | 1974 | # the side-effect that we won't have a cache hit when the raw |
|
1975 | 1975 | # revision data is accessed. But this case should be rare and |
|
1976 | 1976 | # it is extra work to teach the cache about the hash |
|
1977 | 1977 | # verification state. |
|
1978 | 1978 | if self._revisioncache and self._revisioncache[0] == node: |
|
1979 | 1979 | self._revisioncache = None |
|
1980 | 1980 | |
|
1981 | 1981 | revornode = rev |
|
1982 | 1982 | if revornode is None: |
|
1983 | 1983 | revornode = templatefilters.short(hex(node)) |
|
1984 | 1984 | raise error.RevlogError( |
|
1985 | 1985 | _(b"integrity check failed on %s:%s") |
|
1986 | 1986 | % (self.indexfile, pycompat.bytestr(revornode)) |
|
1987 | 1987 | ) |
|
1988 | 1988 | except error.RevlogError: |
|
1989 | 1989 | if self._censorable and storageutil.iscensoredtext(text): |
|
1990 | 1990 | raise error.CensoredNodeError(self.indexfile, node, text) |
|
1991 | 1991 | raise |
|
1992 | 1992 | |
|
1993 | 1993 | def _enforceinlinesize(self, tr, fp=None): |
|
1994 | 1994 | """Check if the revlog is too big for inline and convert if so. |
|
1995 | 1995 | |
|
1996 | 1996 | This should be called after revisions are added to the revlog. If the |
|
1997 | 1997 | revlog has grown too large to be an inline revlog, it will convert it |
|
1998 | 1998 | to use multiple index and data files. |
|
1999 | 1999 | """ |
|
2000 | 2000 | tiprev = len(self) - 1 |
|
2001 | 2001 | if ( |
|
2002 | 2002 | not self._inline |
|
2003 | 2003 | or (self.start(tiprev) + self.length(tiprev)) < _maxinline |
|
2004 | 2004 | ): |
|
2005 | 2005 | return |
|
2006 | 2006 | |
|
2007 | 2007 | troffset = tr.findoffset(self.indexfile) |
|
2008 | 2008 | if troffset is None: |
|
2009 | 2009 | raise error.RevlogError( |
|
2010 | 2010 | _(b"%s not found in the transaction") % self.indexfile |
|
2011 | 2011 | ) |
|
2012 | 2012 | trindex = 0 |
|
2013 | 2013 | tr.add(self.datafile, 0) |
|
2014 | 2014 | |
|
2015 | 2015 | if fp: |
|
2016 | 2016 | fp.flush() |
|
2017 | 2017 | fp.close() |
|
2018 | 2018 | # We can't use the cached file handle after close(). So prevent |
|
2019 | 2019 | # its usage. |
|
2020 | 2020 | self._writinghandles = None |
|
2021 | 2021 | |
|
2022 | 2022 | with self._indexfp(b'r') as ifh, self._datafp(b'w') as dfh: |
|
2023 | 2023 | for r in self: |
|
2024 | 2024 | dfh.write(self._getsegmentforrevs(r, r, df=ifh)[1]) |
|
2025 | 2025 | if troffset <= self.start(r): |
|
2026 | 2026 | trindex = r |
|
2027 | 2027 | |
|
2028 | 2028 | with self._indexfp(b'w') as fp: |
|
2029 | 2029 | self.version &= ~FLAG_INLINE_DATA |
|
2030 | 2030 | self._inline = False |
|
2031 | 2031 | io = self._io |
|
2032 | 2032 | for i in self: |
|
2033 | 2033 | e = io.packentry(self.index[i], self.node, self.version, i) |
|
2034 | 2034 | fp.write(e) |
|
2035 | 2035 | |
|
2036 | 2036 | # the temp file replace the real index when we exit the context |
|
2037 | 2037 | # manager |
|
2038 | 2038 | |
|
2039 | 2039 | tr.replace(self.indexfile, trindex * self._io.size) |
|
2040 | 2040 | nodemaputil.setup_persistent_nodemap(tr, self) |
|
2041 | 2041 | self._chunkclear() |
|
2042 | 2042 | |
|
2043 | 2043 | def _nodeduplicatecallback(self, transaction, node): |
|
2044 | 2044 | """called when trying to add a node already stored.""" |
|
2045 | 2045 | |
|
2046 | 2046 | def addrevision( |
|
2047 | 2047 | self, |
|
2048 | 2048 | text, |
|
2049 | 2049 | transaction, |
|
2050 | 2050 | link, |
|
2051 | 2051 | p1, |
|
2052 | 2052 | p2, |
|
2053 | 2053 | cachedelta=None, |
|
2054 | 2054 | node=None, |
|
2055 | 2055 | flags=REVIDX_DEFAULT_FLAGS, |
|
2056 | 2056 | deltacomputer=None, |
|
2057 | 2057 | sidedata=None, |
|
2058 | 2058 | ): |
|
2059 | 2059 | """add a revision to the log |
|
2060 | 2060 | |
|
2061 | 2061 | text - the revision data to add |
|
2062 | 2062 | transaction - the transaction object used for rollback |
|
2063 | 2063 | link - the linkrev data to add |
|
2064 | 2064 | p1, p2 - the parent nodeids of the revision |
|
2065 | 2065 | cachedelta - an optional precomputed delta |
|
2066 | 2066 | node - nodeid of revision; typically node is not specified, and it is |
|
2067 | 2067 | computed by default as hash(text, p1, p2), however subclasses might |
|
2068 | 2068 | use different hashing method (and override checkhash() in such case) |
|
2069 | 2069 | flags - the known flags to set on the revision |
|
2070 | 2070 | deltacomputer - an optional deltacomputer instance shared between |
|
2071 | 2071 | multiple calls |
|
2072 | 2072 | """ |
|
2073 | 2073 | if link == nullrev: |
|
2074 | 2074 | raise error.RevlogError( |
|
2075 | 2075 | _(b"attempted to add linkrev -1 to %s") % self.indexfile |
|
2076 | 2076 | ) |
|
2077 | 2077 | |
|
2078 | 2078 | if sidedata is None: |
|
2079 | 2079 | sidedata = {} |
|
2080 | 2080 | flags = flags & ~REVIDX_SIDEDATA |
|
2081 | 2081 | elif not self.hassidedata: |
|
2082 | 2082 | raise error.ProgrammingError( |
|
2083 | 2083 | _(b"trying to add sidedata to a revlog who don't support them") |
|
2084 | 2084 | ) |
|
2085 | 2085 | else: |
|
2086 | 2086 | flags |= REVIDX_SIDEDATA |
|
2087 | 2087 | |
|
2088 | 2088 | if flags: |
|
2089 | 2089 | node = node or self.hash(text, p1, p2) |
|
2090 | 2090 | |
|
2091 | 2091 | rawtext, validatehash = flagutil.processflagswrite( |
|
2092 | 2092 | self, text, flags, sidedata=sidedata |
|
2093 | 2093 | ) |
|
2094 | 2094 | |
|
2095 | 2095 | # If the flag processor modifies the revision data, ignore any provided |
|
2096 | 2096 | # cachedelta. |
|
2097 | 2097 | if rawtext != text: |
|
2098 | 2098 | cachedelta = None |
|
2099 | 2099 | |
|
2100 | 2100 | if len(rawtext) > _maxentrysize: |
|
2101 | 2101 | raise error.RevlogError( |
|
2102 | 2102 | _( |
|
2103 | 2103 | b"%s: size of %d bytes exceeds maximum revlog storage of 2GiB" |
|
2104 | 2104 | ) |
|
2105 | 2105 | % (self.indexfile, len(rawtext)) |
|
2106 | 2106 | ) |
|
2107 | 2107 | |
|
2108 | 2108 | node = node or self.hash(rawtext, p1, p2) |
|
2109 |
|
|
|
2110 |
|
|
|
2109 | rev = self.index.get_rev(node) | |
|
2110 | if rev is not None: | |
|
2111 | return rev | |
|
2111 | 2112 | |
|
2112 | 2113 | if validatehash: |
|
2113 | 2114 | self.checkhash(rawtext, node, p1=p1, p2=p2) |
|
2114 | 2115 | |
|
2115 |
re |
|
|
2116 | return self.addrawrevision( | |
|
2116 | 2117 | rawtext, |
|
2117 | 2118 | transaction, |
|
2118 | 2119 | link, |
|
2119 | 2120 | p1, |
|
2120 | 2121 | p2, |
|
2121 | 2122 | node, |
|
2122 | 2123 | flags, |
|
2123 | 2124 | cachedelta=cachedelta, |
|
2124 | 2125 | deltacomputer=deltacomputer, |
|
2125 | 2126 | ) |
|
2126 | return node | |
|
2127 | 2127 | |
|
2128 | 2128 | def addrawrevision( |
|
2129 | 2129 | self, |
|
2130 | 2130 | rawtext, |
|
2131 | 2131 | transaction, |
|
2132 | 2132 | link, |
|
2133 | 2133 | p1, |
|
2134 | 2134 | p2, |
|
2135 | 2135 | node, |
|
2136 | 2136 | flags, |
|
2137 | 2137 | cachedelta=None, |
|
2138 | 2138 | deltacomputer=None, |
|
2139 | 2139 | ): |
|
2140 | 2140 | """add a raw revision with known flags, node and parents |
|
2141 | 2141 | useful when reusing a revision not stored in this revlog (ex: received |
|
2142 | 2142 | over wire, or read from an external bundle). |
|
2143 | 2143 | """ |
|
2144 | 2144 | dfh = None |
|
2145 | 2145 | if not self._inline: |
|
2146 | 2146 | dfh = self._datafp(b"a+") |
|
2147 | 2147 | ifh = self._indexfp(b"a+") |
|
2148 | 2148 | try: |
|
2149 | 2149 | return self._addrevision( |
|
2150 | 2150 | node, |
|
2151 | 2151 | rawtext, |
|
2152 | 2152 | transaction, |
|
2153 | 2153 | link, |
|
2154 | 2154 | p1, |
|
2155 | 2155 | p2, |
|
2156 | 2156 | flags, |
|
2157 | 2157 | cachedelta, |
|
2158 | 2158 | ifh, |
|
2159 | 2159 | dfh, |
|
2160 | 2160 | deltacomputer=deltacomputer, |
|
2161 | 2161 | ) |
|
2162 | 2162 | finally: |
|
2163 | 2163 | if dfh: |
|
2164 | 2164 | dfh.close() |
|
2165 | 2165 | ifh.close() |
|
2166 | 2166 | |
|
2167 | 2167 | def compress(self, data): |
|
2168 | 2168 | """Generate a possibly-compressed representation of data.""" |
|
2169 | 2169 | if not data: |
|
2170 | 2170 | return b'', data |
|
2171 | 2171 | |
|
2172 | 2172 | compressed = self._compressor.compress(data) |
|
2173 | 2173 | |
|
2174 | 2174 | if compressed: |
|
2175 | 2175 | # The revlog compressor added the header in the returned data. |
|
2176 | 2176 | return b'', compressed |
|
2177 | 2177 | |
|
2178 | 2178 | if data[0:1] == b'\0': |
|
2179 | 2179 | return b'', data |
|
2180 | 2180 | return b'u', data |
|
2181 | 2181 | |
|
2182 | 2182 | def decompress(self, data): |
|
2183 | 2183 | """Decompress a revlog chunk. |
|
2184 | 2184 | |
|
2185 | 2185 | The chunk is expected to begin with a header identifying the |
|
2186 | 2186 | format type so it can be routed to an appropriate decompressor. |
|
2187 | 2187 | """ |
|
2188 | 2188 | if not data: |
|
2189 | 2189 | return data |
|
2190 | 2190 | |
|
2191 | 2191 | # Revlogs are read much more frequently than they are written and many |
|
2192 | 2192 | # chunks only take microseconds to decompress, so performance is |
|
2193 | 2193 | # important here. |
|
2194 | 2194 | # |
|
2195 | 2195 | # We can make a few assumptions about revlogs: |
|
2196 | 2196 | # |
|
2197 | 2197 | # 1) the majority of chunks will be compressed (as opposed to inline |
|
2198 | 2198 | # raw data). |
|
2199 | 2199 | # 2) decompressing *any* data will likely by at least 10x slower than |
|
2200 | 2200 | # returning raw inline data. |
|
2201 | 2201 | # 3) we want to prioritize common and officially supported compression |
|
2202 | 2202 | # engines |
|
2203 | 2203 | # |
|
2204 | 2204 | # It follows that we want to optimize for "decompress compressed data |
|
2205 | 2205 | # when encoded with common and officially supported compression engines" |
|
2206 | 2206 | # case over "raw data" and "data encoded by less common or non-official |
|
2207 | 2207 | # compression engines." That is why we have the inline lookup first |
|
2208 | 2208 | # followed by the compengines lookup. |
|
2209 | 2209 | # |
|
2210 | 2210 | # According to `hg perfrevlogchunks`, this is ~0.5% faster for zlib |
|
2211 | 2211 | # compressed chunks. And this matters for changelog and manifest reads. |
|
2212 | 2212 | t = data[0:1] |
|
2213 | 2213 | |
|
2214 | 2214 | if t == b'x': |
|
2215 | 2215 | try: |
|
2216 | 2216 | return _zlibdecompress(data) |
|
2217 | 2217 | except zlib.error as e: |
|
2218 | 2218 | raise error.RevlogError( |
|
2219 | 2219 | _(b'revlog decompress error: %s') |
|
2220 | 2220 | % stringutil.forcebytestr(e) |
|
2221 | 2221 | ) |
|
2222 | 2222 | # '\0' is more common than 'u' so it goes first. |
|
2223 | 2223 | elif t == b'\0': |
|
2224 | 2224 | return data |
|
2225 | 2225 | elif t == b'u': |
|
2226 | 2226 | return util.buffer(data, 1) |
|
2227 | 2227 | |
|
2228 | 2228 | try: |
|
2229 | 2229 | compressor = self._decompressors[t] |
|
2230 | 2230 | except KeyError: |
|
2231 | 2231 | try: |
|
2232 | 2232 | engine = util.compengines.forrevlogheader(t) |
|
2233 | 2233 | compressor = engine.revlogcompressor(self._compengineopts) |
|
2234 | 2234 | self._decompressors[t] = compressor |
|
2235 | 2235 | except KeyError: |
|
2236 | 2236 | raise error.RevlogError(_(b'unknown compression type %r') % t) |
|
2237 | 2237 | |
|
2238 | 2238 | return compressor.decompress(data) |
|
2239 | 2239 | |
|
2240 | 2240 | def _addrevision( |
|
2241 | 2241 | self, |
|
2242 | 2242 | node, |
|
2243 | 2243 | rawtext, |
|
2244 | 2244 | transaction, |
|
2245 | 2245 | link, |
|
2246 | 2246 | p1, |
|
2247 | 2247 | p2, |
|
2248 | 2248 | flags, |
|
2249 | 2249 | cachedelta, |
|
2250 | 2250 | ifh, |
|
2251 | 2251 | dfh, |
|
2252 | 2252 | alwayscache=False, |
|
2253 | 2253 | deltacomputer=None, |
|
2254 | 2254 | ): |
|
2255 | 2255 | """internal function to add revisions to the log |
|
2256 | 2256 | |
|
2257 | 2257 | see addrevision for argument descriptions. |
|
2258 | 2258 | |
|
2259 | 2259 | note: "addrevision" takes non-raw text, "_addrevision" takes raw text. |
|
2260 | 2260 | |
|
2261 | 2261 | if "deltacomputer" is not provided or None, a defaultdeltacomputer will |
|
2262 | 2262 | be used. |
|
2263 | 2263 | |
|
2264 | 2264 | invariants: |
|
2265 | 2265 | - rawtext is optional (can be None); if not set, cachedelta must be set. |
|
2266 | 2266 | if both are set, they must correspond to each other. |
|
2267 | 2267 | """ |
|
2268 | 2268 | if node == nullid: |
|
2269 | 2269 | raise error.RevlogError( |
|
2270 | 2270 | _(b"%s: attempt to add null revision") % self.indexfile |
|
2271 | 2271 | ) |
|
2272 | 2272 | if node == wdirid or node in wdirfilenodeids: |
|
2273 | 2273 | raise error.RevlogError( |
|
2274 | 2274 | _(b"%s: attempt to add wdir revision") % self.indexfile |
|
2275 | 2275 | ) |
|
2276 | 2276 | |
|
2277 | 2277 | if self._inline: |
|
2278 | 2278 | fh = ifh |
|
2279 | 2279 | else: |
|
2280 | 2280 | fh = dfh |
|
2281 | 2281 | |
|
2282 | 2282 | btext = [rawtext] |
|
2283 | 2283 | |
|
2284 | 2284 | curr = len(self) |
|
2285 | 2285 | prev = curr - 1 |
|
2286 | 2286 | offset = self.end(prev) |
|
2287 | 2287 | p1r, p2r = self.rev(p1), self.rev(p2) |
|
2288 | 2288 | |
|
2289 | 2289 | # full versions are inserted when the needed deltas |
|
2290 | 2290 | # become comparable to the uncompressed text |
|
2291 | 2291 | if rawtext is None: |
|
2292 | 2292 | # need rawtext size, before changed by flag processors, which is |
|
2293 | 2293 | # the non-raw size. use revlog explicitly to avoid filelog's extra |
|
2294 | 2294 | # logic that might remove metadata size. |
|
2295 | 2295 | textlen = mdiff.patchedsize( |
|
2296 | 2296 | revlog.size(self, cachedelta[0]), cachedelta[1] |
|
2297 | 2297 | ) |
|
2298 | 2298 | else: |
|
2299 | 2299 | textlen = len(rawtext) |
|
2300 | 2300 | |
|
2301 | 2301 | if deltacomputer is None: |
|
2302 | 2302 | deltacomputer = deltautil.deltacomputer(self) |
|
2303 | 2303 | |
|
2304 | 2304 | revinfo = _revisioninfo(node, p1, p2, btext, textlen, cachedelta, flags) |
|
2305 | 2305 | |
|
2306 | 2306 | deltainfo = deltacomputer.finddeltainfo(revinfo, fh) |
|
2307 | 2307 | |
|
2308 | 2308 | e = ( |
|
2309 | 2309 | offset_type(offset, flags), |
|
2310 | 2310 | deltainfo.deltalen, |
|
2311 | 2311 | textlen, |
|
2312 | 2312 | deltainfo.base, |
|
2313 | 2313 | link, |
|
2314 | 2314 | p1r, |
|
2315 | 2315 | p2r, |
|
2316 | 2316 | node, |
|
2317 | 2317 | ) |
|
2318 | 2318 | self.index.append(e) |
|
2319 | 2319 | |
|
2320 | 2320 | entry = self._io.packentry(e, self.node, self.version, curr) |
|
2321 | 2321 | self._writeentry( |
|
2322 | 2322 | transaction, ifh, dfh, entry, deltainfo.data, link, offset |
|
2323 | 2323 | ) |
|
2324 | 2324 | |
|
2325 | 2325 | rawtext = btext[0] |
|
2326 | 2326 | |
|
2327 | 2327 | if alwayscache and rawtext is None: |
|
2328 | 2328 | rawtext = deltacomputer.buildtext(revinfo, fh) |
|
2329 | 2329 | |
|
2330 | 2330 | if type(rawtext) == bytes: # only accept immutable objects |
|
2331 | 2331 | self._revisioncache = (node, curr, rawtext) |
|
2332 | 2332 | self._chainbasecache[curr] = deltainfo.chainbase |
|
2333 | 2333 | return curr |
|
2334 | 2334 | |
|
2335 | 2335 | def _writeentry(self, transaction, ifh, dfh, entry, data, link, offset): |
|
2336 | 2336 | # Files opened in a+ mode have inconsistent behavior on various |
|
2337 | 2337 | # platforms. Windows requires that a file positioning call be made |
|
2338 | 2338 | # when the file handle transitions between reads and writes. See |
|
2339 | 2339 | # 3686fa2b8eee and the mixedfilemodewrapper in windows.py. On other |
|
2340 | 2340 | # platforms, Python or the platform itself can be buggy. Some versions |
|
2341 | 2341 | # of Solaris have been observed to not append at the end of the file |
|
2342 | 2342 | # if the file was seeked to before the end. See issue4943 for more. |
|
2343 | 2343 | # |
|
2344 | 2344 | # We work around this issue by inserting a seek() before writing. |
|
2345 | 2345 | # Note: This is likely not necessary on Python 3. However, because |
|
2346 | 2346 | # the file handle is reused for reads and may be seeked there, we need |
|
2347 | 2347 | # to be careful before changing this. |
|
2348 | 2348 | ifh.seek(0, os.SEEK_END) |
|
2349 | 2349 | if dfh: |
|
2350 | 2350 | dfh.seek(0, os.SEEK_END) |
|
2351 | 2351 | |
|
2352 | 2352 | curr = len(self) - 1 |
|
2353 | 2353 | if not self._inline: |
|
2354 | 2354 | transaction.add(self.datafile, offset) |
|
2355 | 2355 | transaction.add(self.indexfile, curr * len(entry)) |
|
2356 | 2356 | if data[0]: |
|
2357 | 2357 | dfh.write(data[0]) |
|
2358 | 2358 | dfh.write(data[1]) |
|
2359 | 2359 | ifh.write(entry) |
|
2360 | 2360 | else: |
|
2361 | 2361 | offset += curr * self._io.size |
|
2362 | 2362 | transaction.add(self.indexfile, offset) |
|
2363 | 2363 | ifh.write(entry) |
|
2364 | 2364 | ifh.write(data[0]) |
|
2365 | 2365 | ifh.write(data[1]) |
|
2366 | 2366 | self._enforceinlinesize(transaction, ifh) |
|
2367 | 2367 | nodemaputil.setup_persistent_nodemap(transaction, self) |
|
2368 | 2368 | |
|
2369 | 2369 | def addgroup( |
|
2370 | 2370 | self, |
|
2371 | 2371 | deltas, |
|
2372 | 2372 | linkmapper, |
|
2373 | 2373 | transaction, |
|
2374 | 2374 | alwayscache=False, |
|
2375 | 2375 | addrevisioncb=None, |
|
2376 | 2376 | duplicaterevisioncb=None, |
|
2377 | 2377 | ): |
|
2378 | 2378 | """ |
|
2379 | 2379 | add a delta group |
|
2380 | 2380 | |
|
2381 | 2381 | given a set of deltas, add them to the revision log. the |
|
2382 | 2382 | first delta is against its parent, which should be in our |
|
2383 | 2383 | log, the rest are against the previous delta. |
|
2384 | 2384 | |
|
2385 | 2385 | If ``addrevisioncb`` is defined, it will be called with arguments of |
|
2386 | 2386 | this revlog and the node that was added. |
|
2387 | 2387 | """ |
|
2388 | 2388 | |
|
2389 | 2389 | if self._writinghandles: |
|
2390 | 2390 | raise error.ProgrammingError(b'cannot nest addgroup() calls') |
|
2391 | 2391 | |
|
2392 | 2392 | r = len(self) |
|
2393 | 2393 | end = 0 |
|
2394 | 2394 | if r: |
|
2395 | 2395 | end = self.end(r - 1) |
|
2396 | 2396 | ifh = self._indexfp(b"a+") |
|
2397 | 2397 | isize = r * self._io.size |
|
2398 | 2398 | if self._inline: |
|
2399 | 2399 | transaction.add(self.indexfile, end + isize) |
|
2400 | 2400 | dfh = None |
|
2401 | 2401 | else: |
|
2402 | 2402 | transaction.add(self.indexfile, isize) |
|
2403 | 2403 | transaction.add(self.datafile, end) |
|
2404 | 2404 | dfh = self._datafp(b"a+") |
|
2405 | 2405 | |
|
2406 | 2406 | def flush(): |
|
2407 | 2407 | if dfh: |
|
2408 | 2408 | dfh.flush() |
|
2409 | 2409 | ifh.flush() |
|
2410 | 2410 | |
|
2411 | 2411 | self._writinghandles = (ifh, dfh) |
|
2412 | 2412 | empty = True |
|
2413 | 2413 | |
|
2414 | 2414 | try: |
|
2415 | 2415 | deltacomputer = deltautil.deltacomputer(self) |
|
2416 | 2416 | # loop through our set of deltas |
|
2417 | 2417 | for data in deltas: |
|
2418 | 2418 | node, p1, p2, linknode, deltabase, delta, flags = data |
|
2419 | 2419 | link = linkmapper(linknode) |
|
2420 | 2420 | flags = flags or REVIDX_DEFAULT_FLAGS |
|
2421 | 2421 | |
|
2422 | 2422 | if self.index.has_node(node): |
|
2423 | 2423 | # this can happen if two branches make the same change |
|
2424 | 2424 | self._nodeduplicatecallback(transaction, node) |
|
2425 | 2425 | if duplicaterevisioncb: |
|
2426 | 2426 | duplicaterevisioncb(self, node) |
|
2427 | 2427 | empty = False |
|
2428 | 2428 | continue |
|
2429 | 2429 | |
|
2430 | 2430 | for p in (p1, p2): |
|
2431 | 2431 | if not self.index.has_node(p): |
|
2432 | 2432 | raise error.LookupError( |
|
2433 | 2433 | p, self.indexfile, _(b'unknown parent') |
|
2434 | 2434 | ) |
|
2435 | 2435 | |
|
2436 | 2436 | if not self.index.has_node(deltabase): |
|
2437 | 2437 | raise error.LookupError( |
|
2438 | 2438 | deltabase, self.indexfile, _(b'unknown delta base') |
|
2439 | 2439 | ) |
|
2440 | 2440 | |
|
2441 | 2441 | baserev = self.rev(deltabase) |
|
2442 | 2442 | |
|
2443 | 2443 | if baserev != nullrev and self.iscensored(baserev): |
|
2444 | 2444 | # if base is censored, delta must be full replacement in a |
|
2445 | 2445 | # single patch operation |
|
2446 | 2446 | hlen = struct.calcsize(b">lll") |
|
2447 | 2447 | oldlen = self.rawsize(baserev) |
|
2448 | 2448 | newlen = len(delta) - hlen |
|
2449 | 2449 | if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen): |
|
2450 | 2450 | raise error.CensoredBaseError( |
|
2451 | 2451 | self.indexfile, self.node(baserev) |
|
2452 | 2452 | ) |
|
2453 | 2453 | |
|
2454 | 2454 | if not flags and self._peek_iscensored(baserev, delta, flush): |
|
2455 | 2455 | flags |= REVIDX_ISCENSORED |
|
2456 | 2456 | |
|
2457 | 2457 | # We assume consumers of addrevisioncb will want to retrieve |
|
2458 | 2458 | # the added revision, which will require a call to |
|
2459 | 2459 | # revision(). revision() will fast path if there is a cache |
|
2460 | 2460 | # hit. So, we tell _addrevision() to always cache in this case. |
|
2461 | 2461 | # We're only using addgroup() in the context of changegroup |
|
2462 | 2462 | # generation so the revision data can always be handled as raw |
|
2463 | 2463 | # by the flagprocessor. |
|
2464 | 2464 | self._addrevision( |
|
2465 | 2465 | node, |
|
2466 | 2466 | None, |
|
2467 | 2467 | transaction, |
|
2468 | 2468 | link, |
|
2469 | 2469 | p1, |
|
2470 | 2470 | p2, |
|
2471 | 2471 | flags, |
|
2472 | 2472 | (baserev, delta), |
|
2473 | 2473 | ifh, |
|
2474 | 2474 | dfh, |
|
2475 | 2475 | alwayscache=alwayscache, |
|
2476 | 2476 | deltacomputer=deltacomputer, |
|
2477 | 2477 | ) |
|
2478 | 2478 | |
|
2479 | 2479 | if addrevisioncb: |
|
2480 | 2480 | addrevisioncb(self, node) |
|
2481 | 2481 | empty = False |
|
2482 | 2482 | |
|
2483 | 2483 | if not dfh and not self._inline: |
|
2484 | 2484 | # addrevision switched from inline to conventional |
|
2485 | 2485 | # reopen the index |
|
2486 | 2486 | ifh.close() |
|
2487 | 2487 | dfh = self._datafp(b"a+") |
|
2488 | 2488 | ifh = self._indexfp(b"a+") |
|
2489 | 2489 | self._writinghandles = (ifh, dfh) |
|
2490 | 2490 | finally: |
|
2491 | 2491 | self._writinghandles = None |
|
2492 | 2492 | |
|
2493 | 2493 | if dfh: |
|
2494 | 2494 | dfh.close() |
|
2495 | 2495 | ifh.close() |
|
2496 | 2496 | return not empty |
|
2497 | 2497 | |
|
2498 | 2498 | def iscensored(self, rev): |
|
2499 | 2499 | """Check if a file revision is censored.""" |
|
2500 | 2500 | if not self._censorable: |
|
2501 | 2501 | return False |
|
2502 | 2502 | |
|
2503 | 2503 | return self.flags(rev) & REVIDX_ISCENSORED |
|
2504 | 2504 | |
|
2505 | 2505 | def _peek_iscensored(self, baserev, delta, flush): |
|
2506 | 2506 | """Quickly check if a delta produces a censored revision.""" |
|
2507 | 2507 | if not self._censorable: |
|
2508 | 2508 | return False |
|
2509 | 2509 | |
|
2510 | 2510 | return storageutil.deltaiscensored(delta, baserev, self.rawsize) |
|
2511 | 2511 | |
|
2512 | 2512 | def getstrippoint(self, minlink): |
|
2513 | 2513 | """find the minimum rev that must be stripped to strip the linkrev |
|
2514 | 2514 | |
|
2515 | 2515 | Returns a tuple containing the minimum rev and a set of all revs that |
|
2516 | 2516 | have linkrevs that will be broken by this strip. |
|
2517 | 2517 | """ |
|
2518 | 2518 | return storageutil.resolvestripinfo( |
|
2519 | 2519 | minlink, |
|
2520 | 2520 | len(self) - 1, |
|
2521 | 2521 | self.headrevs(), |
|
2522 | 2522 | self.linkrev, |
|
2523 | 2523 | self.parentrevs, |
|
2524 | 2524 | ) |
|
2525 | 2525 | |
|
2526 | 2526 | def strip(self, minlink, transaction): |
|
2527 | 2527 | """truncate the revlog on the first revision with a linkrev >= minlink |
|
2528 | 2528 | |
|
2529 | 2529 | This function is called when we're stripping revision minlink and |
|
2530 | 2530 | its descendants from the repository. |
|
2531 | 2531 | |
|
2532 | 2532 | We have to remove all revisions with linkrev >= minlink, because |
|
2533 | 2533 | the equivalent changelog revisions will be renumbered after the |
|
2534 | 2534 | strip. |
|
2535 | 2535 | |
|
2536 | 2536 | So we truncate the revlog on the first of these revisions, and |
|
2537 | 2537 | trust that the caller has saved the revisions that shouldn't be |
|
2538 | 2538 | removed and that it'll re-add them after this truncation. |
|
2539 | 2539 | """ |
|
2540 | 2540 | if len(self) == 0: |
|
2541 | 2541 | return |
|
2542 | 2542 | |
|
2543 | 2543 | rev, _ = self.getstrippoint(minlink) |
|
2544 | 2544 | if rev == len(self): |
|
2545 | 2545 | return |
|
2546 | 2546 | |
|
2547 | 2547 | # first truncate the files on disk |
|
2548 | 2548 | end = self.start(rev) |
|
2549 | 2549 | if not self._inline: |
|
2550 | 2550 | transaction.add(self.datafile, end) |
|
2551 | 2551 | end = rev * self._io.size |
|
2552 | 2552 | else: |
|
2553 | 2553 | end += rev * self._io.size |
|
2554 | 2554 | |
|
2555 | 2555 | transaction.add(self.indexfile, end) |
|
2556 | 2556 | |
|
2557 | 2557 | # then reset internal state in memory to forget those revisions |
|
2558 | 2558 | self._revisioncache = None |
|
2559 | 2559 | self._chaininfocache = util.lrucachedict(500) |
|
2560 | 2560 | self._chunkclear() |
|
2561 | 2561 | |
|
2562 | 2562 | del self.index[rev:-1] |
|
2563 | 2563 | |
|
2564 | 2564 | def checksize(self): |
|
2565 | 2565 | """Check size of index and data files |
|
2566 | 2566 | |
|
2567 | 2567 | return a (dd, di) tuple. |
|
2568 | 2568 | - dd: extra bytes for the "data" file |
|
2569 | 2569 | - di: extra bytes for the "index" file |
|
2570 | 2570 | |
|
2571 | 2571 | A healthy revlog will return (0, 0). |
|
2572 | 2572 | """ |
|
2573 | 2573 | expected = 0 |
|
2574 | 2574 | if len(self): |
|
2575 | 2575 | expected = max(0, self.end(len(self) - 1)) |
|
2576 | 2576 | |
|
2577 | 2577 | try: |
|
2578 | 2578 | with self._datafp() as f: |
|
2579 | 2579 | f.seek(0, io.SEEK_END) |
|
2580 | 2580 | actual = f.tell() |
|
2581 | 2581 | dd = actual - expected |
|
2582 | 2582 | except IOError as inst: |
|
2583 | 2583 | if inst.errno != errno.ENOENT: |
|
2584 | 2584 | raise |
|
2585 | 2585 | dd = 0 |
|
2586 | 2586 | |
|
2587 | 2587 | try: |
|
2588 | 2588 | f = self.opener(self.indexfile) |
|
2589 | 2589 | f.seek(0, io.SEEK_END) |
|
2590 | 2590 | actual = f.tell() |
|
2591 | 2591 | f.close() |
|
2592 | 2592 | s = self._io.size |
|
2593 | 2593 | i = max(0, actual // s) |
|
2594 | 2594 | di = actual - (i * s) |
|
2595 | 2595 | if self._inline: |
|
2596 | 2596 | databytes = 0 |
|
2597 | 2597 | for r in self: |
|
2598 | 2598 | databytes += max(0, self.length(r)) |
|
2599 | 2599 | dd = 0 |
|
2600 | 2600 | di = actual - len(self) * s - databytes |
|
2601 | 2601 | except IOError as inst: |
|
2602 | 2602 | if inst.errno != errno.ENOENT: |
|
2603 | 2603 | raise |
|
2604 | 2604 | di = 0 |
|
2605 | 2605 | |
|
2606 | 2606 | return (dd, di) |
|
2607 | 2607 | |
|
2608 | 2608 | def files(self): |
|
2609 | 2609 | res = [self.indexfile] |
|
2610 | 2610 | if not self._inline: |
|
2611 | 2611 | res.append(self.datafile) |
|
2612 | 2612 | return res |
|
2613 | 2613 | |
|
2614 | 2614 | def emitrevisions( |
|
2615 | 2615 | self, |
|
2616 | 2616 | nodes, |
|
2617 | 2617 | nodesorder=None, |
|
2618 | 2618 | revisiondata=False, |
|
2619 | 2619 | assumehaveparentrevisions=False, |
|
2620 | 2620 | deltamode=repository.CG_DELTAMODE_STD, |
|
2621 | 2621 | ): |
|
2622 | 2622 | if nodesorder not in (b'nodes', b'storage', b'linear', None): |
|
2623 | 2623 | raise error.ProgrammingError( |
|
2624 | 2624 | b'unhandled value for nodesorder: %s' % nodesorder |
|
2625 | 2625 | ) |
|
2626 | 2626 | |
|
2627 | 2627 | if nodesorder is None and not self._generaldelta: |
|
2628 | 2628 | nodesorder = b'storage' |
|
2629 | 2629 | |
|
2630 | 2630 | if ( |
|
2631 | 2631 | not self._storedeltachains |
|
2632 | 2632 | and deltamode != repository.CG_DELTAMODE_PREV |
|
2633 | 2633 | ): |
|
2634 | 2634 | deltamode = repository.CG_DELTAMODE_FULL |
|
2635 | 2635 | |
|
2636 | 2636 | return storageutil.emitrevisions( |
|
2637 | 2637 | self, |
|
2638 | 2638 | nodes, |
|
2639 | 2639 | nodesorder, |
|
2640 | 2640 | revlogrevisiondelta, |
|
2641 | 2641 | deltaparentfn=self.deltaparent, |
|
2642 | 2642 | candeltafn=self.candelta, |
|
2643 | 2643 | rawsizefn=self.rawsize, |
|
2644 | 2644 | revdifffn=self.revdiff, |
|
2645 | 2645 | flagsfn=self.flags, |
|
2646 | 2646 | deltamode=deltamode, |
|
2647 | 2647 | revisiondata=revisiondata, |
|
2648 | 2648 | assumehaveparentrevisions=assumehaveparentrevisions, |
|
2649 | 2649 | ) |
|
2650 | 2650 | |
|
2651 | 2651 | DELTAREUSEALWAYS = b'always' |
|
2652 | 2652 | DELTAREUSESAMEREVS = b'samerevs' |
|
2653 | 2653 | DELTAREUSENEVER = b'never' |
|
2654 | 2654 | |
|
2655 | 2655 | DELTAREUSEFULLADD = b'fulladd' |
|
2656 | 2656 | |
|
2657 | 2657 | DELTAREUSEALL = {b'always', b'samerevs', b'never', b'fulladd'} |
|
2658 | 2658 | |
|
2659 | 2659 | def clone( |
|
2660 | 2660 | self, |
|
2661 | 2661 | tr, |
|
2662 | 2662 | destrevlog, |
|
2663 | 2663 | addrevisioncb=None, |
|
2664 | 2664 | deltareuse=DELTAREUSESAMEREVS, |
|
2665 | 2665 | forcedeltabothparents=None, |
|
2666 | 2666 | sidedatacompanion=None, |
|
2667 | 2667 | ): |
|
2668 | 2668 | """Copy this revlog to another, possibly with format changes. |
|
2669 | 2669 | |
|
2670 | 2670 | The destination revlog will contain the same revisions and nodes. |
|
2671 | 2671 | However, it may not be bit-for-bit identical due to e.g. delta encoding |
|
2672 | 2672 | differences. |
|
2673 | 2673 | |
|
2674 | 2674 | The ``deltareuse`` argument control how deltas from the existing revlog |
|
2675 | 2675 | are preserved in the destination revlog. The argument can have the |
|
2676 | 2676 | following values: |
|
2677 | 2677 | |
|
2678 | 2678 | DELTAREUSEALWAYS |
|
2679 | 2679 | Deltas will always be reused (if possible), even if the destination |
|
2680 | 2680 | revlog would not select the same revisions for the delta. This is the |
|
2681 | 2681 | fastest mode of operation. |
|
2682 | 2682 | DELTAREUSESAMEREVS |
|
2683 | 2683 | Deltas will be reused if the destination revlog would pick the same |
|
2684 | 2684 | revisions for the delta. This mode strikes a balance between speed |
|
2685 | 2685 | and optimization. |
|
2686 | 2686 | DELTAREUSENEVER |
|
2687 | 2687 | Deltas will never be reused. This is the slowest mode of execution. |
|
2688 | 2688 | This mode can be used to recompute deltas (e.g. if the diff/delta |
|
2689 | 2689 | algorithm changes). |
|
2690 | 2690 | DELTAREUSEFULLADD |
|
2691 | 2691 | Revision will be re-added as if their were new content. This is |
|
2692 | 2692 | slower than DELTAREUSEALWAYS but allow more mechanism to kicks in. |
|
2693 | 2693 | eg: large file detection and handling. |
|
2694 | 2694 | |
|
2695 | 2695 | Delta computation can be slow, so the choice of delta reuse policy can |
|
2696 | 2696 | significantly affect run time. |
|
2697 | 2697 | |
|
2698 | 2698 | The default policy (``DELTAREUSESAMEREVS``) strikes a balance between |
|
2699 | 2699 | two extremes. Deltas will be reused if they are appropriate. But if the |
|
2700 | 2700 | delta could choose a better revision, it will do so. This means if you |
|
2701 | 2701 | are converting a non-generaldelta revlog to a generaldelta revlog, |
|
2702 | 2702 | deltas will be recomputed if the delta's parent isn't a parent of the |
|
2703 | 2703 | revision. |
|
2704 | 2704 | |
|
2705 | 2705 | In addition to the delta policy, the ``forcedeltabothparents`` |
|
2706 | 2706 | argument controls whether to force compute deltas against both parents |
|
2707 | 2707 | for merges. By default, the current default is used. |
|
2708 | 2708 | |
|
2709 | 2709 | If not None, the `sidedatacompanion` is callable that accept two |
|
2710 | 2710 | arguments: |
|
2711 | 2711 | |
|
2712 | 2712 | (srcrevlog, rev) |
|
2713 | 2713 | |
|
2714 | 2714 | and return a quintet that control changes to sidedata content from the |
|
2715 | 2715 | old revision to the new clone result: |
|
2716 | 2716 | |
|
2717 | 2717 | (dropall, filterout, update, new_flags, dropped_flags) |
|
2718 | 2718 | |
|
2719 | 2719 | * if `dropall` is True, all sidedata should be dropped |
|
2720 | 2720 | * `filterout` is a set of sidedata keys that should be dropped |
|
2721 | 2721 | * `update` is a mapping of additionnal/new key -> value |
|
2722 | 2722 | * new_flags is a bitfields of new flags that the revision should get |
|
2723 | 2723 | * dropped_flags is a bitfields of new flags that the revision shoudl not longer have |
|
2724 | 2724 | """ |
|
2725 | 2725 | if deltareuse not in self.DELTAREUSEALL: |
|
2726 | 2726 | raise ValueError( |
|
2727 | 2727 | _(b'value for deltareuse invalid: %s') % deltareuse |
|
2728 | 2728 | ) |
|
2729 | 2729 | |
|
2730 | 2730 | if len(destrevlog): |
|
2731 | 2731 | raise ValueError(_(b'destination revlog is not empty')) |
|
2732 | 2732 | |
|
2733 | 2733 | if getattr(self, 'filteredrevs', None): |
|
2734 | 2734 | raise ValueError(_(b'source revlog has filtered revisions')) |
|
2735 | 2735 | if getattr(destrevlog, 'filteredrevs', None): |
|
2736 | 2736 | raise ValueError(_(b'destination revlog has filtered revisions')) |
|
2737 | 2737 | |
|
2738 | 2738 | # lazydelta and lazydeltabase controls whether to reuse a cached delta, |
|
2739 | 2739 | # if possible. |
|
2740 | 2740 | oldlazydelta = destrevlog._lazydelta |
|
2741 | 2741 | oldlazydeltabase = destrevlog._lazydeltabase |
|
2742 | 2742 | oldamd = destrevlog._deltabothparents |
|
2743 | 2743 | |
|
2744 | 2744 | try: |
|
2745 | 2745 | if deltareuse == self.DELTAREUSEALWAYS: |
|
2746 | 2746 | destrevlog._lazydeltabase = True |
|
2747 | 2747 | destrevlog._lazydelta = True |
|
2748 | 2748 | elif deltareuse == self.DELTAREUSESAMEREVS: |
|
2749 | 2749 | destrevlog._lazydeltabase = False |
|
2750 | 2750 | destrevlog._lazydelta = True |
|
2751 | 2751 | elif deltareuse == self.DELTAREUSENEVER: |
|
2752 | 2752 | destrevlog._lazydeltabase = False |
|
2753 | 2753 | destrevlog._lazydelta = False |
|
2754 | 2754 | |
|
2755 | 2755 | destrevlog._deltabothparents = forcedeltabothparents or oldamd |
|
2756 | 2756 | |
|
2757 | 2757 | self._clone( |
|
2758 | 2758 | tr, |
|
2759 | 2759 | destrevlog, |
|
2760 | 2760 | addrevisioncb, |
|
2761 | 2761 | deltareuse, |
|
2762 | 2762 | forcedeltabothparents, |
|
2763 | 2763 | sidedatacompanion, |
|
2764 | 2764 | ) |
|
2765 | 2765 | |
|
2766 | 2766 | finally: |
|
2767 | 2767 | destrevlog._lazydelta = oldlazydelta |
|
2768 | 2768 | destrevlog._lazydeltabase = oldlazydeltabase |
|
2769 | 2769 | destrevlog._deltabothparents = oldamd |
|
2770 | 2770 | |
|
2771 | 2771 | def _clone( |
|
2772 | 2772 | self, |
|
2773 | 2773 | tr, |
|
2774 | 2774 | destrevlog, |
|
2775 | 2775 | addrevisioncb, |
|
2776 | 2776 | deltareuse, |
|
2777 | 2777 | forcedeltabothparents, |
|
2778 | 2778 | sidedatacompanion, |
|
2779 | 2779 | ): |
|
2780 | 2780 | """perform the core duty of `revlog.clone` after parameter processing""" |
|
2781 | 2781 | deltacomputer = deltautil.deltacomputer(destrevlog) |
|
2782 | 2782 | index = self.index |
|
2783 | 2783 | for rev in self: |
|
2784 | 2784 | entry = index[rev] |
|
2785 | 2785 | |
|
2786 | 2786 | # Some classes override linkrev to take filtered revs into |
|
2787 | 2787 | # account. Use raw entry from index. |
|
2788 | 2788 | flags = entry[0] & 0xFFFF |
|
2789 | 2789 | linkrev = entry[4] |
|
2790 | 2790 | p1 = index[entry[5]][7] |
|
2791 | 2791 | p2 = index[entry[6]][7] |
|
2792 | 2792 | node = entry[7] |
|
2793 | 2793 | |
|
2794 | 2794 | sidedataactions = (False, [], {}, 0, 0) |
|
2795 | 2795 | if sidedatacompanion is not None: |
|
2796 | 2796 | sidedataactions = sidedatacompanion(self, rev) |
|
2797 | 2797 | |
|
2798 | 2798 | # (Possibly) reuse the delta from the revlog if allowed and |
|
2799 | 2799 | # the revlog chunk is a delta. |
|
2800 | 2800 | cachedelta = None |
|
2801 | 2801 | rawtext = None |
|
2802 | 2802 | if any(sidedataactions) or deltareuse == self.DELTAREUSEFULLADD: |
|
2803 | 2803 | dropall = sidedataactions[0] |
|
2804 | 2804 | filterout = sidedataactions[1] |
|
2805 | 2805 | update = sidedataactions[2] |
|
2806 | 2806 | new_flags = sidedataactions[3] |
|
2807 | 2807 | dropped_flags = sidedataactions[4] |
|
2808 | 2808 | text, sidedata = self._revisiondata(rev) |
|
2809 | 2809 | if dropall: |
|
2810 | 2810 | sidedata = {} |
|
2811 | 2811 | for key in filterout: |
|
2812 | 2812 | sidedata.pop(key, None) |
|
2813 | 2813 | sidedata.update(update) |
|
2814 | 2814 | if not sidedata: |
|
2815 | 2815 | sidedata = None |
|
2816 | 2816 | |
|
2817 | 2817 | flags |= new_flags |
|
2818 | 2818 | flags &= ~dropped_flags |
|
2819 | 2819 | |
|
2820 | 2820 | destrevlog.addrevision( |
|
2821 | 2821 | text, |
|
2822 | 2822 | tr, |
|
2823 | 2823 | linkrev, |
|
2824 | 2824 | p1, |
|
2825 | 2825 | p2, |
|
2826 | 2826 | cachedelta=cachedelta, |
|
2827 | 2827 | node=node, |
|
2828 | 2828 | flags=flags, |
|
2829 | 2829 | deltacomputer=deltacomputer, |
|
2830 | 2830 | sidedata=sidedata, |
|
2831 | 2831 | ) |
|
2832 | 2832 | else: |
|
2833 | 2833 | if destrevlog._lazydelta: |
|
2834 | 2834 | dp = self.deltaparent(rev) |
|
2835 | 2835 | if dp != nullrev: |
|
2836 | 2836 | cachedelta = (dp, bytes(self._chunk(rev))) |
|
2837 | 2837 | |
|
2838 | 2838 | if not cachedelta: |
|
2839 | 2839 | rawtext = self.rawdata(rev) |
|
2840 | 2840 | |
|
2841 | 2841 | ifh = destrevlog.opener( |
|
2842 | 2842 | destrevlog.indexfile, b'a+', checkambig=False |
|
2843 | 2843 | ) |
|
2844 | 2844 | dfh = None |
|
2845 | 2845 | if not destrevlog._inline: |
|
2846 | 2846 | dfh = destrevlog.opener(destrevlog.datafile, b'a+') |
|
2847 | 2847 | try: |
|
2848 | 2848 | destrevlog._addrevision( |
|
2849 | 2849 | node, |
|
2850 | 2850 | rawtext, |
|
2851 | 2851 | tr, |
|
2852 | 2852 | linkrev, |
|
2853 | 2853 | p1, |
|
2854 | 2854 | p2, |
|
2855 | 2855 | flags, |
|
2856 | 2856 | cachedelta, |
|
2857 | 2857 | ifh, |
|
2858 | 2858 | dfh, |
|
2859 | 2859 | deltacomputer=deltacomputer, |
|
2860 | 2860 | ) |
|
2861 | 2861 | finally: |
|
2862 | 2862 | if dfh: |
|
2863 | 2863 | dfh.close() |
|
2864 | 2864 | ifh.close() |
|
2865 | 2865 | |
|
2866 | 2866 | if addrevisioncb: |
|
2867 | 2867 | addrevisioncb(self, rev, node) |
|
2868 | 2868 | |
|
2869 | 2869 | def censorrevision(self, tr, censornode, tombstone=b''): |
|
2870 | 2870 | if (self.version & 0xFFFF) == REVLOGV0: |
|
2871 | 2871 | raise error.RevlogError( |
|
2872 | 2872 | _(b'cannot censor with version %d revlogs') % self.version |
|
2873 | 2873 | ) |
|
2874 | 2874 | |
|
2875 | 2875 | censorrev = self.rev(censornode) |
|
2876 | 2876 | tombstone = storageutil.packmeta({b'censored': tombstone}, b'') |
|
2877 | 2877 | |
|
2878 | 2878 | if len(tombstone) > self.rawsize(censorrev): |
|
2879 | 2879 | raise error.Abort( |
|
2880 | 2880 | _(b'censor tombstone must be no longer than censored data') |
|
2881 | 2881 | ) |
|
2882 | 2882 | |
|
2883 | 2883 | # Rewriting the revlog in place is hard. Our strategy for censoring is |
|
2884 | 2884 | # to create a new revlog, copy all revisions to it, then replace the |
|
2885 | 2885 | # revlogs on transaction close. |
|
2886 | 2886 | |
|
2887 | 2887 | newindexfile = self.indexfile + b'.tmpcensored' |
|
2888 | 2888 | newdatafile = self.datafile + b'.tmpcensored' |
|
2889 | 2889 | |
|
2890 | 2890 | # This is a bit dangerous. We could easily have a mismatch of state. |
|
2891 | 2891 | newrl = revlog(self.opener, newindexfile, newdatafile, censorable=True) |
|
2892 | 2892 | newrl.version = self.version |
|
2893 | 2893 | newrl._generaldelta = self._generaldelta |
|
2894 | 2894 | newrl._io = self._io |
|
2895 | 2895 | |
|
2896 | 2896 | for rev in self.revs(): |
|
2897 | 2897 | node = self.node(rev) |
|
2898 | 2898 | p1, p2 = self.parents(node) |
|
2899 | 2899 | |
|
2900 | 2900 | if rev == censorrev: |
|
2901 | 2901 | newrl.addrawrevision( |
|
2902 | 2902 | tombstone, |
|
2903 | 2903 | tr, |
|
2904 | 2904 | self.linkrev(censorrev), |
|
2905 | 2905 | p1, |
|
2906 | 2906 | p2, |
|
2907 | 2907 | censornode, |
|
2908 | 2908 | REVIDX_ISCENSORED, |
|
2909 | 2909 | ) |
|
2910 | 2910 | |
|
2911 | 2911 | if newrl.deltaparent(rev) != nullrev: |
|
2912 | 2912 | raise error.Abort( |
|
2913 | 2913 | _( |
|
2914 | 2914 | b'censored revision stored as delta; ' |
|
2915 | 2915 | b'cannot censor' |
|
2916 | 2916 | ), |
|
2917 | 2917 | hint=_( |
|
2918 | 2918 | b'censoring of revlogs is not ' |
|
2919 | 2919 | b'fully implemented; please report ' |
|
2920 | 2920 | b'this bug' |
|
2921 | 2921 | ), |
|
2922 | 2922 | ) |
|
2923 | 2923 | continue |
|
2924 | 2924 | |
|
2925 | 2925 | if self.iscensored(rev): |
|
2926 | 2926 | if self.deltaparent(rev) != nullrev: |
|
2927 | 2927 | raise error.Abort( |
|
2928 | 2928 | _( |
|
2929 | 2929 | b'cannot censor due to censored ' |
|
2930 | 2930 | b'revision having delta stored' |
|
2931 | 2931 | ) |
|
2932 | 2932 | ) |
|
2933 | 2933 | rawtext = self._chunk(rev) |
|
2934 | 2934 | else: |
|
2935 | 2935 | rawtext = self.rawdata(rev) |
|
2936 | 2936 | |
|
2937 | 2937 | newrl.addrawrevision( |
|
2938 | 2938 | rawtext, tr, self.linkrev(rev), p1, p2, node, self.flags(rev) |
|
2939 | 2939 | ) |
|
2940 | 2940 | |
|
2941 | 2941 | tr.addbackup(self.indexfile, location=b'store') |
|
2942 | 2942 | if not self._inline: |
|
2943 | 2943 | tr.addbackup(self.datafile, location=b'store') |
|
2944 | 2944 | |
|
2945 | 2945 | self.opener.rename(newrl.indexfile, self.indexfile) |
|
2946 | 2946 | if not self._inline: |
|
2947 | 2947 | self.opener.rename(newrl.datafile, self.datafile) |
|
2948 | 2948 | |
|
2949 | 2949 | self.clearcaches() |
|
2950 | 2950 | self._loadindex() |
|
2951 | 2951 | |
|
2952 | 2952 | def verifyintegrity(self, state): |
|
2953 | 2953 | """Verifies the integrity of the revlog. |
|
2954 | 2954 | |
|
2955 | 2955 | Yields ``revlogproblem`` instances describing problems that are |
|
2956 | 2956 | found. |
|
2957 | 2957 | """ |
|
2958 | 2958 | dd, di = self.checksize() |
|
2959 | 2959 | if dd: |
|
2960 | 2960 | yield revlogproblem(error=_(b'data length off by %d bytes') % dd) |
|
2961 | 2961 | if di: |
|
2962 | 2962 | yield revlogproblem(error=_(b'index contains %d extra bytes') % di) |
|
2963 | 2963 | |
|
2964 | 2964 | version = self.version & 0xFFFF |
|
2965 | 2965 | |
|
2966 | 2966 | # The verifier tells us what version revlog we should be. |
|
2967 | 2967 | if version != state[b'expectedversion']: |
|
2968 | 2968 | yield revlogproblem( |
|
2969 | 2969 | warning=_(b"warning: '%s' uses revlog format %d; expected %d") |
|
2970 | 2970 | % (self.indexfile, version, state[b'expectedversion']) |
|
2971 | 2971 | ) |
|
2972 | 2972 | |
|
2973 | 2973 | state[b'skipread'] = set() |
|
2974 | 2974 | state[b'safe_renamed'] = set() |
|
2975 | 2975 | |
|
2976 | 2976 | for rev in self: |
|
2977 | 2977 | node = self.node(rev) |
|
2978 | 2978 | |
|
2979 | 2979 | # Verify contents. 4 cases to care about: |
|
2980 | 2980 | # |
|
2981 | 2981 | # common: the most common case |
|
2982 | 2982 | # rename: with a rename |
|
2983 | 2983 | # meta: file content starts with b'\1\n', the metadata |
|
2984 | 2984 | # header defined in filelog.py, but without a rename |
|
2985 | 2985 | # ext: content stored externally |
|
2986 | 2986 | # |
|
2987 | 2987 | # More formally, their differences are shown below: |
|
2988 | 2988 | # |
|
2989 | 2989 | # | common | rename | meta | ext |
|
2990 | 2990 | # ------------------------------------------------------- |
|
2991 | 2991 | # flags() | 0 | 0 | 0 | not 0 |
|
2992 | 2992 | # renamed() | False | True | False | ? |
|
2993 | 2993 | # rawtext[0:2]=='\1\n'| False | True | True | ? |
|
2994 | 2994 | # |
|
2995 | 2995 | # "rawtext" means the raw text stored in revlog data, which |
|
2996 | 2996 | # could be retrieved by "rawdata(rev)". "text" |
|
2997 | 2997 | # mentioned below is "revision(rev)". |
|
2998 | 2998 | # |
|
2999 | 2999 | # There are 3 different lengths stored physically: |
|
3000 | 3000 | # 1. L1: rawsize, stored in revlog index |
|
3001 | 3001 | # 2. L2: len(rawtext), stored in revlog data |
|
3002 | 3002 | # 3. L3: len(text), stored in revlog data if flags==0, or |
|
3003 | 3003 | # possibly somewhere else if flags!=0 |
|
3004 | 3004 | # |
|
3005 | 3005 | # L1 should be equal to L2. L3 could be different from them. |
|
3006 | 3006 | # "text" may or may not affect commit hash depending on flag |
|
3007 | 3007 | # processors (see flagutil.addflagprocessor). |
|
3008 | 3008 | # |
|
3009 | 3009 | # | common | rename | meta | ext |
|
3010 | 3010 | # ------------------------------------------------- |
|
3011 | 3011 | # rawsize() | L1 | L1 | L1 | L1 |
|
3012 | 3012 | # size() | L1 | L2-LM | L1(*) | L1 (?) |
|
3013 | 3013 | # len(rawtext) | L2 | L2 | L2 | L2 |
|
3014 | 3014 | # len(text) | L2 | L2 | L2 | L3 |
|
3015 | 3015 | # len(read()) | L2 | L2-LM | L2-LM | L3 (?) |
|
3016 | 3016 | # |
|
3017 | 3017 | # LM: length of metadata, depending on rawtext |
|
3018 | 3018 | # (*): not ideal, see comment in filelog.size |
|
3019 | 3019 | # (?): could be "- len(meta)" if the resolved content has |
|
3020 | 3020 | # rename metadata |
|
3021 | 3021 | # |
|
3022 | 3022 | # Checks needed to be done: |
|
3023 | 3023 | # 1. length check: L1 == L2, in all cases. |
|
3024 | 3024 | # 2. hash check: depending on flag processor, we may need to |
|
3025 | 3025 | # use either "text" (external), or "rawtext" (in revlog). |
|
3026 | 3026 | |
|
3027 | 3027 | try: |
|
3028 | 3028 | skipflags = state.get(b'skipflags', 0) |
|
3029 | 3029 | if skipflags: |
|
3030 | 3030 | skipflags &= self.flags(rev) |
|
3031 | 3031 | |
|
3032 | 3032 | _verify_revision(self, skipflags, state, node) |
|
3033 | 3033 | |
|
3034 | 3034 | l1 = self.rawsize(rev) |
|
3035 | 3035 | l2 = len(self.rawdata(node)) |
|
3036 | 3036 | |
|
3037 | 3037 | if l1 != l2: |
|
3038 | 3038 | yield revlogproblem( |
|
3039 | 3039 | error=_(b'unpacked size is %d, %d expected') % (l2, l1), |
|
3040 | 3040 | node=node, |
|
3041 | 3041 | ) |
|
3042 | 3042 | |
|
3043 | 3043 | except error.CensoredNodeError: |
|
3044 | 3044 | if state[b'erroroncensored']: |
|
3045 | 3045 | yield revlogproblem( |
|
3046 | 3046 | error=_(b'censored file data'), node=node |
|
3047 | 3047 | ) |
|
3048 | 3048 | state[b'skipread'].add(node) |
|
3049 | 3049 | except Exception as e: |
|
3050 | 3050 | yield revlogproblem( |
|
3051 | 3051 | error=_(b'unpacking %s: %s') |
|
3052 | 3052 | % (short(node), stringutil.forcebytestr(e)), |
|
3053 | 3053 | node=node, |
|
3054 | 3054 | ) |
|
3055 | 3055 | state[b'skipread'].add(node) |
|
3056 | 3056 | |
|
3057 | 3057 | def storageinfo( |
|
3058 | 3058 | self, |
|
3059 | 3059 | exclusivefiles=False, |
|
3060 | 3060 | sharedfiles=False, |
|
3061 | 3061 | revisionscount=False, |
|
3062 | 3062 | trackedsize=False, |
|
3063 | 3063 | storedsize=False, |
|
3064 | 3064 | ): |
|
3065 | 3065 | d = {} |
|
3066 | 3066 | |
|
3067 | 3067 | if exclusivefiles: |
|
3068 | 3068 | d[b'exclusivefiles'] = [(self.opener, self.indexfile)] |
|
3069 | 3069 | if not self._inline: |
|
3070 | 3070 | d[b'exclusivefiles'].append((self.opener, self.datafile)) |
|
3071 | 3071 | |
|
3072 | 3072 | if sharedfiles: |
|
3073 | 3073 | d[b'sharedfiles'] = [] |
|
3074 | 3074 | |
|
3075 | 3075 | if revisionscount: |
|
3076 | 3076 | d[b'revisionscount'] = len(self) |
|
3077 | 3077 | |
|
3078 | 3078 | if trackedsize: |
|
3079 | 3079 | d[b'trackedsize'] = sum(map(self.rawsize, iter(self))) |
|
3080 | 3080 | |
|
3081 | 3081 | if storedsize: |
|
3082 | 3082 | d[b'storedsize'] = sum( |
|
3083 | 3083 | self.opener.stat(path).st_size for path in self.files() |
|
3084 | 3084 | ) |
|
3085 | 3085 | |
|
3086 | 3086 | return d |
General Comments 0
You need to be logged in to leave comments.
Login now