Show More
@@ -1,1310 +1,1315 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 | sidedata = attr.ib() | |
|
291 | 292 | linknode = attr.ib(default=None) |
|
292 | 293 | |
|
293 | 294 | |
|
294 | 295 | @interfaceutil.implementer(repository.iverifyproblem) |
|
295 | 296 | @attr.s(frozen=True) |
|
296 | 297 | class sqliteproblem(object): |
|
297 | 298 | warning = attr.ib(default=None) |
|
298 | 299 | error = attr.ib(default=None) |
|
299 | 300 | node = attr.ib(default=None) |
|
300 | 301 | |
|
301 | 302 | |
|
302 | 303 | @interfaceutil.implementer(repository.ifilestorage) |
|
303 | 304 | class sqlitefilestore(object): |
|
304 | 305 | """Implements storage for an individual tracked path.""" |
|
305 | 306 | |
|
306 | 307 | def __init__(self, db, path, compression): |
|
307 | 308 | self._db = db |
|
308 | 309 | self._path = path |
|
309 | 310 | |
|
310 | 311 | self._pathid = None |
|
311 | 312 | |
|
312 | 313 | # revnum -> node |
|
313 | 314 | self._revtonode = {} |
|
314 | 315 | # node -> revnum |
|
315 | 316 | self._nodetorev = {} |
|
316 | 317 | # node -> data structure |
|
317 | 318 | self._revisions = {} |
|
318 | 319 | |
|
319 | 320 | self._revisioncache = util.lrucachedict(10) |
|
320 | 321 | |
|
321 | 322 | self._compengine = compression |
|
322 | 323 | |
|
323 | 324 | if compression == b'zstd': |
|
324 | 325 | self._cctx = zstd.ZstdCompressor(level=3) |
|
325 | 326 | self._dctx = zstd.ZstdDecompressor() |
|
326 | 327 | else: |
|
327 | 328 | self._cctx = None |
|
328 | 329 | self._dctx = None |
|
329 | 330 | |
|
330 | 331 | self._refreshindex() |
|
331 | 332 | |
|
332 | 333 | def _refreshindex(self): |
|
333 | 334 | self._revtonode = {} |
|
334 | 335 | self._nodetorev = {} |
|
335 | 336 | self._revisions = {} |
|
336 | 337 | |
|
337 | 338 | res = list( |
|
338 | 339 | self._db.execute( |
|
339 | 340 | 'SELECT id FROM filepath WHERE path=?', (self._path,) |
|
340 | 341 | ) |
|
341 | 342 | ) |
|
342 | 343 | |
|
343 | 344 | if not res: |
|
344 | 345 | self._pathid = None |
|
345 | 346 | return |
|
346 | 347 | |
|
347 | 348 | self._pathid = res[0][0] |
|
348 | 349 | |
|
349 | 350 | res = self._db.execute( |
|
350 | 351 | 'SELECT id, revnum, node, p1rev, p2rev, linkrev, flags ' |
|
351 | 352 | 'FROM fileindex ' |
|
352 | 353 | 'WHERE pathid=? ' |
|
353 | 354 | 'ORDER BY revnum ASC', |
|
354 | 355 | (self._pathid,), |
|
355 | 356 | ) |
|
356 | 357 | |
|
357 | 358 | for i, row in enumerate(res): |
|
358 | 359 | rid, rev, node, p1rev, p2rev, linkrev, flags = row |
|
359 | 360 | |
|
360 | 361 | if i != rev: |
|
361 | 362 | raise SQLiteStoreError( |
|
362 | 363 | _(b'sqlite database has inconsistent revision numbers') |
|
363 | 364 | ) |
|
364 | 365 | |
|
365 | 366 | if p1rev == nullrev: |
|
366 | 367 | p1node = nullid |
|
367 | 368 | else: |
|
368 | 369 | p1node = self._revtonode[p1rev] |
|
369 | 370 | |
|
370 | 371 | if p2rev == nullrev: |
|
371 | 372 | p2node = nullid |
|
372 | 373 | else: |
|
373 | 374 | p2node = self._revtonode[p2rev] |
|
374 | 375 | |
|
375 | 376 | entry = revisionentry( |
|
376 | 377 | rid=rid, |
|
377 | 378 | rev=rev, |
|
378 | 379 | node=node, |
|
379 | 380 | p1rev=p1rev, |
|
380 | 381 | p2rev=p2rev, |
|
381 | 382 | p1node=p1node, |
|
382 | 383 | p2node=p2node, |
|
383 | 384 | linkrev=linkrev, |
|
384 | 385 | flags=flags, |
|
385 | 386 | ) |
|
386 | 387 | |
|
387 | 388 | self._revtonode[rev] = node |
|
388 | 389 | self._nodetorev[node] = rev |
|
389 | 390 | self._revisions[node] = entry |
|
390 | 391 | |
|
391 | 392 | # Start of ifileindex interface. |
|
392 | 393 | |
|
393 | 394 | def __len__(self): |
|
394 | 395 | return len(self._revisions) |
|
395 | 396 | |
|
396 | 397 | def __iter__(self): |
|
397 | 398 | return iter(pycompat.xrange(len(self._revisions))) |
|
398 | 399 | |
|
399 | 400 | def hasnode(self, node): |
|
400 | 401 | if node == nullid: |
|
401 | 402 | return False |
|
402 | 403 | |
|
403 | 404 | return node in self._nodetorev |
|
404 | 405 | |
|
405 | 406 | def revs(self, start=0, stop=None): |
|
406 | 407 | return storageutil.iterrevs( |
|
407 | 408 | len(self._revisions), start=start, stop=stop |
|
408 | 409 | ) |
|
409 | 410 | |
|
410 | 411 | def parents(self, node): |
|
411 | 412 | if node == nullid: |
|
412 | 413 | return nullid, nullid |
|
413 | 414 | |
|
414 | 415 | if node not in self._revisions: |
|
415 | 416 | raise error.LookupError(node, self._path, _(b'no node')) |
|
416 | 417 | |
|
417 | 418 | entry = self._revisions[node] |
|
418 | 419 | return entry.p1node, entry.p2node |
|
419 | 420 | |
|
420 | 421 | def parentrevs(self, rev): |
|
421 | 422 | if rev == nullrev: |
|
422 | 423 | return nullrev, nullrev |
|
423 | 424 | |
|
424 | 425 | if rev not in self._revtonode: |
|
425 | 426 | raise IndexError(rev) |
|
426 | 427 | |
|
427 | 428 | entry = self._revisions[self._revtonode[rev]] |
|
428 | 429 | return entry.p1rev, entry.p2rev |
|
429 | 430 | |
|
430 | 431 | def rev(self, node): |
|
431 | 432 | if node == nullid: |
|
432 | 433 | return nullrev |
|
433 | 434 | |
|
434 | 435 | if node not in self._nodetorev: |
|
435 | 436 | raise error.LookupError(node, self._path, _(b'no node')) |
|
436 | 437 | |
|
437 | 438 | return self._nodetorev[node] |
|
438 | 439 | |
|
439 | 440 | def node(self, rev): |
|
440 | 441 | if rev == nullrev: |
|
441 | 442 | return nullid |
|
442 | 443 | |
|
443 | 444 | if rev not in self._revtonode: |
|
444 | 445 | raise IndexError(rev) |
|
445 | 446 | |
|
446 | 447 | return self._revtonode[rev] |
|
447 | 448 | |
|
448 | 449 | def lookup(self, node): |
|
449 | 450 | return storageutil.fileidlookup(self, node, self._path) |
|
450 | 451 | |
|
451 | 452 | def linkrev(self, rev): |
|
452 | 453 | if rev == nullrev: |
|
453 | 454 | return nullrev |
|
454 | 455 | |
|
455 | 456 | if rev not in self._revtonode: |
|
456 | 457 | raise IndexError(rev) |
|
457 | 458 | |
|
458 | 459 | entry = self._revisions[self._revtonode[rev]] |
|
459 | 460 | return entry.linkrev |
|
460 | 461 | |
|
461 | 462 | def iscensored(self, rev): |
|
462 | 463 | if rev == nullrev: |
|
463 | 464 | return False |
|
464 | 465 | |
|
465 | 466 | if rev not in self._revtonode: |
|
466 | 467 | raise IndexError(rev) |
|
467 | 468 | |
|
468 | 469 | return self._revisions[self._revtonode[rev]].flags & FLAG_CENSORED |
|
469 | 470 | |
|
470 | 471 | def commonancestorsheads(self, node1, node2): |
|
471 | 472 | rev1 = self.rev(node1) |
|
472 | 473 | rev2 = self.rev(node2) |
|
473 | 474 | |
|
474 | 475 | ancestors = ancestor.commonancestorsheads(self.parentrevs, rev1, rev2) |
|
475 | 476 | return pycompat.maplist(self.node, ancestors) |
|
476 | 477 | |
|
477 | 478 | def descendants(self, revs): |
|
478 | 479 | # TODO we could implement this using a recursive SQL query, which |
|
479 | 480 | # might be faster. |
|
480 | 481 | return dagop.descendantrevs(revs, self.revs, self.parentrevs) |
|
481 | 482 | |
|
482 | 483 | def heads(self, start=None, stop=None): |
|
483 | 484 | if start is None and stop is None: |
|
484 | 485 | if not len(self): |
|
485 | 486 | return [nullid] |
|
486 | 487 | |
|
487 | 488 | startrev = self.rev(start) if start is not None else nullrev |
|
488 | 489 | stoprevs = {self.rev(n) for n in stop or []} |
|
489 | 490 | |
|
490 | 491 | revs = dagop.headrevssubset( |
|
491 | 492 | self.revs, self.parentrevs, startrev=startrev, stoprevs=stoprevs |
|
492 | 493 | ) |
|
493 | 494 | |
|
494 | 495 | return [self.node(rev) for rev in revs] |
|
495 | 496 | |
|
496 | 497 | def children(self, node): |
|
497 | 498 | rev = self.rev(node) |
|
498 | 499 | |
|
499 | 500 | res = self._db.execute( |
|
500 | 501 | 'SELECT' |
|
501 | 502 | ' node ' |
|
502 | 503 | ' FROM filedata ' |
|
503 | 504 | ' WHERE path=? AND (p1rev=? OR p2rev=?) ' |
|
504 | 505 | ' ORDER BY revnum ASC', |
|
505 | 506 | (self._path, rev, rev), |
|
506 | 507 | ) |
|
507 | 508 | |
|
508 | 509 | return [row[0] for row in res] |
|
509 | 510 | |
|
510 | 511 | # End of ifileindex interface. |
|
511 | 512 | |
|
512 | 513 | # Start of ifiledata interface. |
|
513 | 514 | |
|
514 | 515 | def size(self, rev): |
|
515 | 516 | if rev == nullrev: |
|
516 | 517 | return 0 |
|
517 | 518 | |
|
518 | 519 | if rev not in self._revtonode: |
|
519 | 520 | raise IndexError(rev) |
|
520 | 521 | |
|
521 | 522 | node = self._revtonode[rev] |
|
522 | 523 | |
|
523 | 524 | if self.renamed(node): |
|
524 | 525 | return len(self.read(node)) |
|
525 | 526 | |
|
526 | 527 | return len(self.revision(node)) |
|
527 | 528 | |
|
528 | 529 | def revision(self, node, raw=False, _verifyhash=True): |
|
529 | 530 | if node in (nullid, nullrev): |
|
530 | 531 | return b'' |
|
531 | 532 | |
|
532 | 533 | if isinstance(node, int): |
|
533 | 534 | node = self.node(node) |
|
534 | 535 | |
|
535 | 536 | if node not in self._nodetorev: |
|
536 | 537 | raise error.LookupError(node, self._path, _(b'no node')) |
|
537 | 538 | |
|
538 | 539 | if node in self._revisioncache: |
|
539 | 540 | return self._revisioncache[node] |
|
540 | 541 | |
|
541 | 542 | # Because we have a fulltext revision cache, we are able to |
|
542 | 543 | # short-circuit delta chain traversal and decompression as soon as |
|
543 | 544 | # we encounter a revision in the cache. |
|
544 | 545 | |
|
545 | 546 | stoprids = {self._revisions[n].rid: n for n in self._revisioncache} |
|
546 | 547 | |
|
547 | 548 | if not stoprids: |
|
548 | 549 | stoprids[-1] = None |
|
549 | 550 | |
|
550 | 551 | fulltext = resolvedeltachain( |
|
551 | 552 | self._db, |
|
552 | 553 | self._pathid, |
|
553 | 554 | node, |
|
554 | 555 | self._revisioncache, |
|
555 | 556 | stoprids, |
|
556 | 557 | zstddctx=self._dctx, |
|
557 | 558 | ) |
|
558 | 559 | |
|
559 | 560 | # Don't verify hashes if parent nodes were rewritten, as the hash |
|
560 | 561 | # wouldn't verify. |
|
561 | 562 | if self._revisions[node].flags & (FLAG_MISSING_P1 | FLAG_MISSING_P2): |
|
562 | 563 | _verifyhash = False |
|
563 | 564 | |
|
564 | 565 | if _verifyhash: |
|
565 | 566 | self._checkhash(fulltext, node) |
|
566 | 567 | self._revisioncache[node] = fulltext |
|
567 | 568 | |
|
568 | 569 | return fulltext |
|
569 | 570 | |
|
570 | 571 | def rawdata(self, *args, **kwargs): |
|
571 | 572 | return self.revision(*args, **kwargs) |
|
572 | 573 | |
|
573 | 574 | def read(self, node): |
|
574 | 575 | return storageutil.filtermetadata(self.revision(node)) |
|
575 | 576 | |
|
576 | 577 | def renamed(self, node): |
|
577 | 578 | return storageutil.filerevisioncopied(self, node) |
|
578 | 579 | |
|
579 | 580 | def cmp(self, node, fulltext): |
|
580 | 581 | return not storageutil.filedataequivalent(self, node, fulltext) |
|
581 | 582 | |
|
582 | 583 | def emitrevisions( |
|
583 | 584 | self, |
|
584 | 585 | nodes, |
|
585 | 586 | nodesorder=None, |
|
586 | 587 | revisiondata=False, |
|
587 | 588 | assumehaveparentrevisions=False, |
|
588 | 589 | deltamode=repository.CG_DELTAMODE_STD, |
|
589 | 590 | ): |
|
590 | 591 | if nodesorder not in (b'nodes', b'storage', b'linear', None): |
|
591 | 592 | raise error.ProgrammingError( |
|
592 | 593 | b'unhandled value for nodesorder: %s' % nodesorder |
|
593 | 594 | ) |
|
594 | 595 | |
|
595 | 596 | nodes = [n for n in nodes if n != nullid] |
|
596 | 597 | |
|
597 | 598 | if not nodes: |
|
598 | 599 | return |
|
599 | 600 | |
|
600 | 601 | # TODO perform in a single query. |
|
601 | 602 | res = self._db.execute( |
|
602 | 603 | 'SELECT revnum, deltaid FROM fileindex ' |
|
603 | 604 | 'WHERE pathid=? ' |
|
604 | 605 | ' AND node in (%s)' % (','.join(['?'] * len(nodes))), |
|
605 | 606 | tuple([self._pathid] + nodes), |
|
606 | 607 | ) |
|
607 | 608 | |
|
608 | 609 | deltabases = {} |
|
609 | 610 | |
|
610 | 611 | for rev, deltaid in res: |
|
611 | 612 | res = self._db.execute( |
|
612 | 613 | 'SELECT revnum from fileindex WHERE pathid=? AND deltaid=?', |
|
613 | 614 | (self._pathid, deltaid), |
|
614 | 615 | ) |
|
615 | 616 | deltabases[rev] = res.fetchone()[0] |
|
616 | 617 | |
|
617 | 618 | # TODO define revdifffn so we can use delta from storage. |
|
618 | 619 | for delta in storageutil.emitrevisions( |
|
619 | 620 | self, |
|
620 | 621 | nodes, |
|
621 | 622 | nodesorder, |
|
622 | 623 | sqliterevisiondelta, |
|
623 | 624 | deltaparentfn=deltabases.__getitem__, |
|
624 | 625 | revisiondata=revisiondata, |
|
625 | 626 | assumehaveparentrevisions=assumehaveparentrevisions, |
|
626 | 627 | deltamode=deltamode, |
|
627 | 628 | ): |
|
628 | 629 | |
|
629 | 630 | yield delta |
|
630 | 631 | |
|
631 | 632 | # End of ifiledata interface. |
|
632 | 633 | |
|
633 | 634 | # Start of ifilemutation interface. |
|
634 | 635 | |
|
635 | 636 | def add(self, filedata, meta, transaction, linkrev, p1, p2): |
|
636 | 637 | if meta or filedata.startswith(b'\x01\n'): |
|
637 | 638 | filedata = storageutil.packmeta(meta, filedata) |
|
638 | 639 | |
|
639 | 640 | rev = self.addrevision(filedata, transaction, linkrev, p1, p2) |
|
640 | 641 | return self.node(rev) |
|
641 | 642 | |
|
642 | 643 | def addrevision( |
|
643 | 644 | self, |
|
644 | 645 | revisiondata, |
|
645 | 646 | transaction, |
|
646 | 647 | linkrev, |
|
647 | 648 | p1, |
|
648 | 649 | p2, |
|
649 | 650 | node=None, |
|
650 | 651 | flags=0, |
|
651 | 652 | cachedelta=None, |
|
652 | 653 | ): |
|
653 | 654 | if flags: |
|
654 | 655 | raise SQLiteStoreError(_(b'flags not supported on revisions')) |
|
655 | 656 | |
|
656 | 657 | validatehash = node is not None |
|
657 | 658 | node = node or storageutil.hashrevisionsha1(revisiondata, p1, p2) |
|
658 | 659 | |
|
659 | 660 | if validatehash: |
|
660 | 661 | self._checkhash(revisiondata, node, p1, p2) |
|
661 | 662 | |
|
662 | 663 | rev = self._nodetorev.get(node) |
|
663 | 664 | if rev is not None: |
|
664 | 665 | return rev |
|
665 | 666 | |
|
666 | 667 | rev = self._addrawrevision( |
|
667 | 668 | node, revisiondata, transaction, linkrev, p1, p2 |
|
668 | 669 | ) |
|
669 | 670 | |
|
670 | 671 | self._revisioncache[node] = revisiondata |
|
671 | 672 | return rev |
|
672 | 673 | |
|
673 | 674 | def addgroup( |
|
674 | 675 | self, |
|
675 | 676 | deltas, |
|
676 | 677 | linkmapper, |
|
677 | 678 | transaction, |
|
678 | 679 | addrevisioncb=None, |
|
679 | 680 | duplicaterevisioncb=None, |
|
680 | 681 | maybemissingparents=False, |
|
681 | 682 | ): |
|
682 | 683 | empty = True |
|
683 | 684 | |
|
684 | 685 | for ( |
|
685 | 686 | node, |
|
686 | 687 | p1, |
|
687 | 688 | p2, |
|
688 | 689 | linknode, |
|
689 | 690 | deltabase, |
|
690 | 691 | delta, |
|
691 | 692 | wireflags, |
|
692 | 693 | sidedata, |
|
693 | 694 | ) in deltas: |
|
694 | 695 | storeflags = 0 |
|
695 | 696 | |
|
696 | 697 | if wireflags & repository.REVISION_FLAG_CENSORED: |
|
697 | 698 | storeflags |= FLAG_CENSORED |
|
698 | 699 | |
|
699 | 700 | if wireflags & ~repository.REVISION_FLAG_CENSORED: |
|
700 | 701 | raise SQLiteStoreError(b'unhandled revision flag') |
|
701 | 702 | |
|
702 | 703 | if maybemissingparents: |
|
703 | 704 | if p1 != nullid and not self.hasnode(p1): |
|
704 | 705 | p1 = nullid |
|
705 | 706 | storeflags |= FLAG_MISSING_P1 |
|
706 | 707 | |
|
707 | 708 | if p2 != nullid and not self.hasnode(p2): |
|
708 | 709 | p2 = nullid |
|
709 | 710 | storeflags |= FLAG_MISSING_P2 |
|
710 | 711 | |
|
711 | 712 | baserev = self.rev(deltabase) |
|
712 | 713 | |
|
713 | 714 | # If base is censored, delta must be full replacement in a single |
|
714 | 715 | # patch operation. |
|
715 | 716 | if baserev != nullrev and self.iscensored(baserev): |
|
716 | 717 | hlen = struct.calcsize(b'>lll') |
|
717 | 718 | oldlen = len(self.rawdata(deltabase, _verifyhash=False)) |
|
718 | 719 | newlen = len(delta) - hlen |
|
719 | 720 | |
|
720 | 721 | if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen): |
|
721 | 722 | raise error.CensoredBaseError(self._path, deltabase) |
|
722 | 723 | |
|
723 | 724 | if not (storeflags & FLAG_CENSORED) and storageutil.deltaiscensored( |
|
724 | 725 | delta, baserev, lambda x: len(self.rawdata(x)) |
|
725 | 726 | ): |
|
726 | 727 | storeflags |= FLAG_CENSORED |
|
727 | 728 | |
|
728 | 729 | linkrev = linkmapper(linknode) |
|
729 | 730 | |
|
730 | 731 | if node in self._revisions: |
|
731 | 732 | # Possibly reset parents to make them proper. |
|
732 | 733 | entry = self._revisions[node] |
|
733 | 734 | |
|
734 | 735 | if entry.flags & FLAG_MISSING_P1 and p1 != nullid: |
|
735 | 736 | entry.p1node = p1 |
|
736 | 737 | entry.p1rev = self._nodetorev[p1] |
|
737 | 738 | entry.flags &= ~FLAG_MISSING_P1 |
|
738 | 739 | |
|
739 | 740 | self._db.execute( |
|
740 | 741 | 'UPDATE fileindex SET p1rev=?, flags=? WHERE id=?', |
|
741 | 742 | (self._nodetorev[p1], entry.flags, entry.rid), |
|
742 | 743 | ) |
|
743 | 744 | |
|
744 | 745 | if entry.flags & FLAG_MISSING_P2 and p2 != nullid: |
|
745 | 746 | entry.p2node = p2 |
|
746 | 747 | entry.p2rev = self._nodetorev[p2] |
|
747 | 748 | entry.flags &= ~FLAG_MISSING_P2 |
|
748 | 749 | |
|
749 | 750 | self._db.execute( |
|
750 | 751 | 'UPDATE fileindex SET p2rev=?, flags=? WHERE id=?', |
|
751 | 752 | (self._nodetorev[p1], entry.flags, entry.rid), |
|
752 | 753 | ) |
|
753 | 754 | |
|
754 | 755 | if duplicaterevisioncb: |
|
755 | 756 | duplicaterevisioncb(self, self.rev(node)) |
|
756 | 757 | empty = False |
|
757 | 758 | continue |
|
758 | 759 | |
|
759 | 760 | if deltabase == nullid: |
|
760 | 761 | text = mdiff.patch(b'', delta) |
|
761 | 762 | storedelta = None |
|
762 | 763 | else: |
|
763 | 764 | text = None |
|
764 | 765 | storedelta = (deltabase, delta) |
|
765 | 766 | |
|
766 | 767 | rev = self._addrawrevision( |
|
767 | 768 | node, |
|
768 | 769 | text, |
|
769 | 770 | transaction, |
|
770 | 771 | linkrev, |
|
771 | 772 | p1, |
|
772 | 773 | p2, |
|
773 | 774 | storedelta=storedelta, |
|
774 | 775 | flags=storeflags, |
|
775 | 776 | ) |
|
776 | 777 | |
|
777 | 778 | if addrevisioncb: |
|
778 | 779 | addrevisioncb(self, rev) |
|
779 | 780 | empty = False |
|
780 | 781 | |
|
781 | 782 | return not empty |
|
782 | 783 | |
|
783 | 784 | def censorrevision(self, tr, censornode, tombstone=b''): |
|
784 | 785 | tombstone = storageutil.packmeta({b'censored': tombstone}, b'') |
|
785 | 786 | |
|
786 | 787 | # This restriction is cargo culted from revlogs and makes no sense for |
|
787 | 788 | # SQLite, since columns can be resized at will. |
|
788 | 789 | if len(tombstone) > len(self.rawdata(censornode)): |
|
789 | 790 | raise error.Abort( |
|
790 | 791 | _(b'censor tombstone must be no longer than censored data') |
|
791 | 792 | ) |
|
792 | 793 | |
|
793 | 794 | # We need to replace the censored revision's data with the tombstone. |
|
794 | 795 | # But replacing that data will have implications for delta chains that |
|
795 | 796 | # reference it. |
|
796 | 797 | # |
|
797 | 798 | # While "better," more complex strategies are possible, we do something |
|
798 | 799 | # simple: we find delta chain children of the censored revision and we |
|
799 | 800 | # replace those incremental deltas with fulltexts of their corresponding |
|
800 | 801 | # revision. Then we delete the now-unreferenced delta and original |
|
801 | 802 | # revision and insert a replacement. |
|
802 | 803 | |
|
803 | 804 | # Find the delta to be censored. |
|
804 | 805 | censoreddeltaid = self._db.execute( |
|
805 | 806 | 'SELECT deltaid FROM fileindex WHERE id=?', |
|
806 | 807 | (self._revisions[censornode].rid,), |
|
807 | 808 | ).fetchone()[0] |
|
808 | 809 | |
|
809 | 810 | # Find all its delta chain children. |
|
810 | 811 | # TODO once we support storing deltas for !files, we'll need to look |
|
811 | 812 | # for those delta chains too. |
|
812 | 813 | rows = list( |
|
813 | 814 | self._db.execute( |
|
814 | 815 | 'SELECT id, pathid, node FROM fileindex ' |
|
815 | 816 | 'WHERE deltabaseid=? OR deltaid=?', |
|
816 | 817 | (censoreddeltaid, censoreddeltaid), |
|
817 | 818 | ) |
|
818 | 819 | ) |
|
819 | 820 | |
|
820 | 821 | for row in rows: |
|
821 | 822 | rid, pathid, node = row |
|
822 | 823 | |
|
823 | 824 | fulltext = resolvedeltachain( |
|
824 | 825 | self._db, pathid, node, {}, {-1: None}, zstddctx=self._dctx |
|
825 | 826 | ) |
|
826 | 827 | |
|
827 | 828 | deltahash = hashutil.sha1(fulltext).digest() |
|
828 | 829 | |
|
829 | 830 | if self._compengine == b'zstd': |
|
830 | 831 | deltablob = self._cctx.compress(fulltext) |
|
831 | 832 | compression = COMPRESSION_ZSTD |
|
832 | 833 | elif self._compengine == b'zlib': |
|
833 | 834 | deltablob = zlib.compress(fulltext) |
|
834 | 835 | compression = COMPRESSION_ZLIB |
|
835 | 836 | elif self._compengine == b'none': |
|
836 | 837 | deltablob = fulltext |
|
837 | 838 | compression = COMPRESSION_NONE |
|
838 | 839 | else: |
|
839 | 840 | raise error.ProgrammingError( |
|
840 | 841 | b'unhandled compression engine: %s' % self._compengine |
|
841 | 842 | ) |
|
842 | 843 | |
|
843 | 844 | if len(deltablob) >= len(fulltext): |
|
844 | 845 | deltablob = fulltext |
|
845 | 846 | compression = COMPRESSION_NONE |
|
846 | 847 | |
|
847 | 848 | deltaid = insertdelta(self._db, compression, deltahash, deltablob) |
|
848 | 849 | |
|
849 | 850 | self._db.execute( |
|
850 | 851 | 'UPDATE fileindex SET deltaid=?, deltabaseid=NULL ' |
|
851 | 852 | 'WHERE id=?', |
|
852 | 853 | (deltaid, rid), |
|
853 | 854 | ) |
|
854 | 855 | |
|
855 | 856 | # Now create the tombstone delta and replace the delta on the censored |
|
856 | 857 | # node. |
|
857 | 858 | deltahash = hashutil.sha1(tombstone).digest() |
|
858 | 859 | tombstonedeltaid = insertdelta( |
|
859 | 860 | self._db, COMPRESSION_NONE, deltahash, tombstone |
|
860 | 861 | ) |
|
861 | 862 | |
|
862 | 863 | flags = self._revisions[censornode].flags |
|
863 | 864 | flags |= FLAG_CENSORED |
|
864 | 865 | |
|
865 | 866 | self._db.execute( |
|
866 | 867 | 'UPDATE fileindex SET flags=?, deltaid=?, deltabaseid=NULL ' |
|
867 | 868 | 'WHERE pathid=? AND node=?', |
|
868 | 869 | (flags, tombstonedeltaid, self._pathid, censornode), |
|
869 | 870 | ) |
|
870 | 871 | |
|
871 | 872 | self._db.execute('DELETE FROM delta WHERE id=?', (censoreddeltaid,)) |
|
872 | 873 | |
|
873 | 874 | self._refreshindex() |
|
874 | 875 | self._revisioncache.clear() |
|
875 | 876 | |
|
876 | 877 | def getstrippoint(self, minlink): |
|
877 | 878 | return storageutil.resolvestripinfo( |
|
878 | 879 | minlink, |
|
879 | 880 | len(self) - 1, |
|
880 | 881 | [self.rev(n) for n in self.heads()], |
|
881 | 882 | self.linkrev, |
|
882 | 883 | self.parentrevs, |
|
883 | 884 | ) |
|
884 | 885 | |
|
885 | 886 | def strip(self, minlink, transaction): |
|
886 | 887 | if not len(self): |
|
887 | 888 | return |
|
888 | 889 | |
|
889 | 890 | rev, _ignored = self.getstrippoint(minlink) |
|
890 | 891 | |
|
891 | 892 | if rev == len(self): |
|
892 | 893 | return |
|
893 | 894 | |
|
894 | 895 | for rev in self.revs(rev): |
|
895 | 896 | self._db.execute( |
|
896 | 897 | 'DELETE FROM fileindex WHERE pathid=? AND node=?', |
|
897 | 898 | (self._pathid, self.node(rev)), |
|
898 | 899 | ) |
|
899 | 900 | |
|
900 | 901 | # TODO how should we garbage collect data in delta table? |
|
901 | 902 | |
|
902 | 903 | self._refreshindex() |
|
903 | 904 | |
|
904 | 905 | # End of ifilemutation interface. |
|
905 | 906 | |
|
906 | 907 | # Start of ifilestorage interface. |
|
907 | 908 | |
|
908 | 909 | def files(self): |
|
909 | 910 | return [] |
|
910 | 911 | |
|
912 | def sidedata(self, nodeorrev, _df=None): | |
|
913 | # Not supported for now | |
|
914 | return {} | |
|
915 | ||
|
911 | 916 | def storageinfo( |
|
912 | 917 | self, |
|
913 | 918 | exclusivefiles=False, |
|
914 | 919 | sharedfiles=False, |
|
915 | 920 | revisionscount=False, |
|
916 | 921 | trackedsize=False, |
|
917 | 922 | storedsize=False, |
|
918 | 923 | ): |
|
919 | 924 | d = {} |
|
920 | 925 | |
|
921 | 926 | if exclusivefiles: |
|
922 | 927 | d[b'exclusivefiles'] = [] |
|
923 | 928 | |
|
924 | 929 | if sharedfiles: |
|
925 | 930 | # TODO list sqlite file(s) here. |
|
926 | 931 | d[b'sharedfiles'] = [] |
|
927 | 932 | |
|
928 | 933 | if revisionscount: |
|
929 | 934 | d[b'revisionscount'] = len(self) |
|
930 | 935 | |
|
931 | 936 | if trackedsize: |
|
932 | 937 | d[b'trackedsize'] = sum( |
|
933 | 938 | len(self.revision(node)) for node in self._nodetorev |
|
934 | 939 | ) |
|
935 | 940 | |
|
936 | 941 | if storedsize: |
|
937 | 942 | # TODO implement this? |
|
938 | 943 | d[b'storedsize'] = None |
|
939 | 944 | |
|
940 | 945 | return d |
|
941 | 946 | |
|
942 | 947 | def verifyintegrity(self, state): |
|
943 | 948 | state[b'skipread'] = set() |
|
944 | 949 | |
|
945 | 950 | for rev in self: |
|
946 | 951 | node = self.node(rev) |
|
947 | 952 | |
|
948 | 953 | try: |
|
949 | 954 | self.revision(node) |
|
950 | 955 | except Exception as e: |
|
951 | 956 | yield sqliteproblem( |
|
952 | 957 | error=_(b'unpacking %s: %s') % (short(node), e), node=node |
|
953 | 958 | ) |
|
954 | 959 | |
|
955 | 960 | state[b'skipread'].add(node) |
|
956 | 961 | |
|
957 | 962 | # End of ifilestorage interface. |
|
958 | 963 | |
|
959 | 964 | def _checkhash(self, fulltext, node, p1=None, p2=None): |
|
960 | 965 | if p1 is None and p2 is None: |
|
961 | 966 | p1, p2 = self.parents(node) |
|
962 | 967 | |
|
963 | 968 | if node == storageutil.hashrevisionsha1(fulltext, p1, p2): |
|
964 | 969 | return |
|
965 | 970 | |
|
966 | 971 | try: |
|
967 | 972 | del self._revisioncache[node] |
|
968 | 973 | except KeyError: |
|
969 | 974 | pass |
|
970 | 975 | |
|
971 | 976 | if storageutil.iscensoredtext(fulltext): |
|
972 | 977 | raise error.CensoredNodeError(self._path, node, fulltext) |
|
973 | 978 | |
|
974 | 979 | raise SQLiteStoreError(_(b'integrity check failed on %s') % self._path) |
|
975 | 980 | |
|
976 | 981 | def _addrawrevision( |
|
977 | 982 | self, |
|
978 | 983 | node, |
|
979 | 984 | revisiondata, |
|
980 | 985 | transaction, |
|
981 | 986 | linkrev, |
|
982 | 987 | p1, |
|
983 | 988 | p2, |
|
984 | 989 | storedelta=None, |
|
985 | 990 | flags=0, |
|
986 | 991 | ): |
|
987 | 992 | if self._pathid is None: |
|
988 | 993 | res = self._db.execute( |
|
989 | 994 | 'INSERT INTO filepath (path) VALUES (?)', (self._path,) |
|
990 | 995 | ) |
|
991 | 996 | self._pathid = res.lastrowid |
|
992 | 997 | |
|
993 | 998 | # For simplicity, always store a delta against p1. |
|
994 | 999 | # TODO we need a lot more logic here to make behavior reasonable. |
|
995 | 1000 | |
|
996 | 1001 | if storedelta: |
|
997 | 1002 | deltabase, delta = storedelta |
|
998 | 1003 | |
|
999 | 1004 | if isinstance(deltabase, int): |
|
1000 | 1005 | deltabase = self.node(deltabase) |
|
1001 | 1006 | |
|
1002 | 1007 | else: |
|
1003 | 1008 | assert revisiondata is not None |
|
1004 | 1009 | deltabase = p1 |
|
1005 | 1010 | |
|
1006 | 1011 | if deltabase == nullid: |
|
1007 | 1012 | delta = revisiondata |
|
1008 | 1013 | else: |
|
1009 | 1014 | delta = mdiff.textdiff( |
|
1010 | 1015 | self.revision(self.rev(deltabase)), revisiondata |
|
1011 | 1016 | ) |
|
1012 | 1017 | |
|
1013 | 1018 | # File index stores a pointer to its delta and the parent delta. |
|
1014 | 1019 | # The parent delta is stored via a pointer to the fileindex PK. |
|
1015 | 1020 | if deltabase == nullid: |
|
1016 | 1021 | baseid = None |
|
1017 | 1022 | else: |
|
1018 | 1023 | baseid = self._revisions[deltabase].rid |
|
1019 | 1024 | |
|
1020 | 1025 | # Deltas are stored with a hash of their content. This allows |
|
1021 | 1026 | # us to de-duplicate. The table is configured to ignore conflicts |
|
1022 | 1027 | # and it is faster to just insert and silently noop than to look |
|
1023 | 1028 | # first. |
|
1024 | 1029 | deltahash = hashutil.sha1(delta).digest() |
|
1025 | 1030 | |
|
1026 | 1031 | if self._compengine == b'zstd': |
|
1027 | 1032 | deltablob = self._cctx.compress(delta) |
|
1028 | 1033 | compression = COMPRESSION_ZSTD |
|
1029 | 1034 | elif self._compengine == b'zlib': |
|
1030 | 1035 | deltablob = zlib.compress(delta) |
|
1031 | 1036 | compression = COMPRESSION_ZLIB |
|
1032 | 1037 | elif self._compengine == b'none': |
|
1033 | 1038 | deltablob = delta |
|
1034 | 1039 | compression = COMPRESSION_NONE |
|
1035 | 1040 | else: |
|
1036 | 1041 | raise error.ProgrammingError( |
|
1037 | 1042 | b'unhandled compression engine: %s' % self._compengine |
|
1038 | 1043 | ) |
|
1039 | 1044 | |
|
1040 | 1045 | # Don't store compressed data if it isn't practical. |
|
1041 | 1046 | if len(deltablob) >= len(delta): |
|
1042 | 1047 | deltablob = delta |
|
1043 | 1048 | compression = COMPRESSION_NONE |
|
1044 | 1049 | |
|
1045 | 1050 | deltaid = insertdelta(self._db, compression, deltahash, deltablob) |
|
1046 | 1051 | |
|
1047 | 1052 | rev = len(self) |
|
1048 | 1053 | |
|
1049 | 1054 | if p1 == nullid: |
|
1050 | 1055 | p1rev = nullrev |
|
1051 | 1056 | else: |
|
1052 | 1057 | p1rev = self._nodetorev[p1] |
|
1053 | 1058 | |
|
1054 | 1059 | if p2 == nullid: |
|
1055 | 1060 | p2rev = nullrev |
|
1056 | 1061 | else: |
|
1057 | 1062 | p2rev = self._nodetorev[p2] |
|
1058 | 1063 | |
|
1059 | 1064 | rid = self._db.execute( |
|
1060 | 1065 | 'INSERT INTO fileindex (' |
|
1061 | 1066 | ' pathid, revnum, node, p1rev, p2rev, linkrev, flags, ' |
|
1062 | 1067 | ' deltaid, deltabaseid) ' |
|
1063 | 1068 | ' VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)', |
|
1064 | 1069 | ( |
|
1065 | 1070 | self._pathid, |
|
1066 | 1071 | rev, |
|
1067 | 1072 | node, |
|
1068 | 1073 | p1rev, |
|
1069 | 1074 | p2rev, |
|
1070 | 1075 | linkrev, |
|
1071 | 1076 | flags, |
|
1072 | 1077 | deltaid, |
|
1073 | 1078 | baseid, |
|
1074 | 1079 | ), |
|
1075 | 1080 | ).lastrowid |
|
1076 | 1081 | |
|
1077 | 1082 | entry = revisionentry( |
|
1078 | 1083 | rid=rid, |
|
1079 | 1084 | rev=rev, |
|
1080 | 1085 | node=node, |
|
1081 | 1086 | p1rev=p1rev, |
|
1082 | 1087 | p2rev=p2rev, |
|
1083 | 1088 | p1node=p1, |
|
1084 | 1089 | p2node=p2, |
|
1085 | 1090 | linkrev=linkrev, |
|
1086 | 1091 | flags=flags, |
|
1087 | 1092 | ) |
|
1088 | 1093 | |
|
1089 | 1094 | self._nodetorev[node] = rev |
|
1090 | 1095 | self._revtonode[rev] = node |
|
1091 | 1096 | self._revisions[node] = entry |
|
1092 | 1097 | |
|
1093 | 1098 | return rev |
|
1094 | 1099 | |
|
1095 | 1100 | |
|
1096 | 1101 | class sqliterepository(localrepo.localrepository): |
|
1097 | 1102 | def cancopy(self): |
|
1098 | 1103 | return False |
|
1099 | 1104 | |
|
1100 | 1105 | def transaction(self, *args, **kwargs): |
|
1101 | 1106 | current = self.currenttransaction() |
|
1102 | 1107 | |
|
1103 | 1108 | tr = super(sqliterepository, self).transaction(*args, **kwargs) |
|
1104 | 1109 | |
|
1105 | 1110 | if current: |
|
1106 | 1111 | return tr |
|
1107 | 1112 | |
|
1108 | 1113 | self._dbconn.execute('BEGIN TRANSACTION') |
|
1109 | 1114 | |
|
1110 | 1115 | def committransaction(_): |
|
1111 | 1116 | self._dbconn.commit() |
|
1112 | 1117 | |
|
1113 | 1118 | tr.addfinalize(b'sqlitestore', committransaction) |
|
1114 | 1119 | |
|
1115 | 1120 | return tr |
|
1116 | 1121 | |
|
1117 | 1122 | @property |
|
1118 | 1123 | def _dbconn(self): |
|
1119 | 1124 | # SQLite connections can only be used on the thread that created |
|
1120 | 1125 | # them. In most cases, this "just works." However, hgweb uses |
|
1121 | 1126 | # multiple threads. |
|
1122 | 1127 | tid = threading.current_thread().ident |
|
1123 | 1128 | |
|
1124 | 1129 | if self._db: |
|
1125 | 1130 | if self._db[0] == tid: |
|
1126 | 1131 | return self._db[1] |
|
1127 | 1132 | |
|
1128 | 1133 | db = makedb(self.svfs.join(b'db.sqlite')) |
|
1129 | 1134 | self._db = (tid, db) |
|
1130 | 1135 | |
|
1131 | 1136 | return db |
|
1132 | 1137 | |
|
1133 | 1138 | |
|
1134 | 1139 | def makedb(path): |
|
1135 | 1140 | """Construct a database handle for a database at path.""" |
|
1136 | 1141 | |
|
1137 | 1142 | db = sqlite3.connect(encoding.strfromlocal(path)) |
|
1138 | 1143 | db.text_factory = bytes |
|
1139 | 1144 | |
|
1140 | 1145 | res = db.execute('PRAGMA user_version').fetchone()[0] |
|
1141 | 1146 | |
|
1142 | 1147 | # New database. |
|
1143 | 1148 | if res == 0: |
|
1144 | 1149 | for statement in CREATE_SCHEMA: |
|
1145 | 1150 | db.execute(statement) |
|
1146 | 1151 | |
|
1147 | 1152 | db.commit() |
|
1148 | 1153 | |
|
1149 | 1154 | elif res == CURRENT_SCHEMA_VERSION: |
|
1150 | 1155 | pass |
|
1151 | 1156 | |
|
1152 | 1157 | else: |
|
1153 | 1158 | raise error.Abort(_(b'sqlite database has unrecognized version')) |
|
1154 | 1159 | |
|
1155 | 1160 | db.execute('PRAGMA journal_mode=WAL') |
|
1156 | 1161 | |
|
1157 | 1162 | return db |
|
1158 | 1163 | |
|
1159 | 1164 | |
|
1160 | 1165 | def featuresetup(ui, supported): |
|
1161 | 1166 | supported.add(REQUIREMENT) |
|
1162 | 1167 | |
|
1163 | 1168 | if zstd: |
|
1164 | 1169 | supported.add(REQUIREMENT_ZSTD) |
|
1165 | 1170 | |
|
1166 | 1171 | supported.add(REQUIREMENT_ZLIB) |
|
1167 | 1172 | supported.add(REQUIREMENT_NONE) |
|
1168 | 1173 | supported.add(REQUIREMENT_SHALLOW_FILES) |
|
1169 | 1174 | supported.add(requirements.NARROW_REQUIREMENT) |
|
1170 | 1175 | |
|
1171 | 1176 | |
|
1172 | 1177 | def newreporequirements(orig, ui, createopts): |
|
1173 | 1178 | if createopts[b'backend'] != b'sqlite': |
|
1174 | 1179 | return orig(ui, createopts) |
|
1175 | 1180 | |
|
1176 | 1181 | # This restriction can be lifted once we have more confidence. |
|
1177 | 1182 | if b'sharedrepo' in createopts: |
|
1178 | 1183 | raise error.Abort( |
|
1179 | 1184 | _(b'shared repositories not supported with SQLite store') |
|
1180 | 1185 | ) |
|
1181 | 1186 | |
|
1182 | 1187 | # This filtering is out of an abundance of caution: we want to ensure |
|
1183 | 1188 | # we honor creation options and we do that by annotating exactly the |
|
1184 | 1189 | # creation options we recognize. |
|
1185 | 1190 | known = { |
|
1186 | 1191 | b'narrowfiles', |
|
1187 | 1192 | b'backend', |
|
1188 | 1193 | b'shallowfilestore', |
|
1189 | 1194 | } |
|
1190 | 1195 | |
|
1191 | 1196 | unsupported = set(createopts) - known |
|
1192 | 1197 | if unsupported: |
|
1193 | 1198 | raise error.Abort( |
|
1194 | 1199 | _(b'SQLite store does not support repo creation option: %s') |
|
1195 | 1200 | % b', '.join(sorted(unsupported)) |
|
1196 | 1201 | ) |
|
1197 | 1202 | |
|
1198 | 1203 | # Since we're a hybrid store that still relies on revlogs, we fall back |
|
1199 | 1204 | # to using the revlogv1 backend's storage requirements then adding our |
|
1200 | 1205 | # own requirement. |
|
1201 | 1206 | createopts[b'backend'] = b'revlogv1' |
|
1202 | 1207 | requirements = orig(ui, createopts) |
|
1203 | 1208 | requirements.add(REQUIREMENT) |
|
1204 | 1209 | |
|
1205 | 1210 | compression = ui.config(b'storage', b'sqlite.compression') |
|
1206 | 1211 | |
|
1207 | 1212 | if compression == b'zstd' and not zstd: |
|
1208 | 1213 | raise error.Abort( |
|
1209 | 1214 | _( |
|
1210 | 1215 | b'storage.sqlite.compression set to "zstd" but ' |
|
1211 | 1216 | b'zstandard compression not available to this ' |
|
1212 | 1217 | b'Mercurial install' |
|
1213 | 1218 | ) |
|
1214 | 1219 | ) |
|
1215 | 1220 | |
|
1216 | 1221 | if compression == b'zstd': |
|
1217 | 1222 | requirements.add(REQUIREMENT_ZSTD) |
|
1218 | 1223 | elif compression == b'zlib': |
|
1219 | 1224 | requirements.add(REQUIREMENT_ZLIB) |
|
1220 | 1225 | elif compression == b'none': |
|
1221 | 1226 | requirements.add(REQUIREMENT_NONE) |
|
1222 | 1227 | else: |
|
1223 | 1228 | raise error.Abort( |
|
1224 | 1229 | _( |
|
1225 | 1230 | b'unknown compression engine defined in ' |
|
1226 | 1231 | b'storage.sqlite.compression: %s' |
|
1227 | 1232 | ) |
|
1228 | 1233 | % compression |
|
1229 | 1234 | ) |
|
1230 | 1235 | |
|
1231 | 1236 | if createopts.get(b'shallowfilestore'): |
|
1232 | 1237 | requirements.add(REQUIREMENT_SHALLOW_FILES) |
|
1233 | 1238 | |
|
1234 | 1239 | return requirements |
|
1235 | 1240 | |
|
1236 | 1241 | |
|
1237 | 1242 | @interfaceutil.implementer(repository.ilocalrepositoryfilestorage) |
|
1238 | 1243 | class sqlitefilestorage(object): |
|
1239 | 1244 | """Repository file storage backed by SQLite.""" |
|
1240 | 1245 | |
|
1241 | 1246 | def file(self, path): |
|
1242 | 1247 | if path[0] == b'/': |
|
1243 | 1248 | path = path[1:] |
|
1244 | 1249 | |
|
1245 | 1250 | if REQUIREMENT_ZSTD in self.requirements: |
|
1246 | 1251 | compression = b'zstd' |
|
1247 | 1252 | elif REQUIREMENT_ZLIB in self.requirements: |
|
1248 | 1253 | compression = b'zlib' |
|
1249 | 1254 | elif REQUIREMENT_NONE in self.requirements: |
|
1250 | 1255 | compression = b'none' |
|
1251 | 1256 | else: |
|
1252 | 1257 | raise error.Abort( |
|
1253 | 1258 | _( |
|
1254 | 1259 | b'unable to determine what compression engine ' |
|
1255 | 1260 | b'to use for SQLite storage' |
|
1256 | 1261 | ) |
|
1257 | 1262 | ) |
|
1258 | 1263 | |
|
1259 | 1264 | return sqlitefilestore(self._dbconn, path, compression) |
|
1260 | 1265 | |
|
1261 | 1266 | |
|
1262 | 1267 | def makefilestorage(orig, requirements, features, **kwargs): |
|
1263 | 1268 | """Produce a type conforming to ``ilocalrepositoryfilestorage``.""" |
|
1264 | 1269 | if REQUIREMENT in requirements: |
|
1265 | 1270 | if REQUIREMENT_SHALLOW_FILES in requirements: |
|
1266 | 1271 | features.add(repository.REPO_FEATURE_SHALLOW_FILE_STORAGE) |
|
1267 | 1272 | |
|
1268 | 1273 | return sqlitefilestorage |
|
1269 | 1274 | else: |
|
1270 | 1275 | return orig(requirements=requirements, features=features, **kwargs) |
|
1271 | 1276 | |
|
1272 | 1277 | |
|
1273 | 1278 | def makemain(orig, ui, requirements, **kwargs): |
|
1274 | 1279 | if REQUIREMENT in requirements: |
|
1275 | 1280 | if REQUIREMENT_ZSTD in requirements and not zstd: |
|
1276 | 1281 | raise error.Abort( |
|
1277 | 1282 | _( |
|
1278 | 1283 | b'repository uses zstandard compression, which ' |
|
1279 | 1284 | b'is not available to this Mercurial install' |
|
1280 | 1285 | ) |
|
1281 | 1286 | ) |
|
1282 | 1287 | |
|
1283 | 1288 | return sqliterepository |
|
1284 | 1289 | |
|
1285 | 1290 | return orig(requirements=requirements, **kwargs) |
|
1286 | 1291 | |
|
1287 | 1292 | |
|
1288 | 1293 | def verifierinit(orig, self, *args, **kwargs): |
|
1289 | 1294 | orig(self, *args, **kwargs) |
|
1290 | 1295 | |
|
1291 | 1296 | # We don't care that files in the store don't align with what is |
|
1292 | 1297 | # advertised. So suppress these warnings. |
|
1293 | 1298 | self.warnorphanstorefiles = False |
|
1294 | 1299 | |
|
1295 | 1300 | |
|
1296 | 1301 | def extsetup(ui): |
|
1297 | 1302 | localrepo.featuresetupfuncs.add(featuresetup) |
|
1298 | 1303 | extensions.wrapfunction( |
|
1299 | 1304 | localrepo, b'newreporequirements', newreporequirements |
|
1300 | 1305 | ) |
|
1301 | 1306 | extensions.wrapfunction(localrepo, b'makefilestorage', makefilestorage) |
|
1302 | 1307 | extensions.wrapfunction(localrepo, b'makemain', makemain) |
|
1303 | 1308 | extensions.wrapfunction(verify.verifier, b'__init__', verifierinit) |
|
1304 | 1309 | |
|
1305 | 1310 | |
|
1306 | 1311 | def reposetup(ui, repo): |
|
1307 | 1312 | if isinstance(repo, sqliterepository): |
|
1308 | 1313 | repo._db = None |
|
1309 | 1314 | |
|
1310 | 1315 | # TODO check for bundlerepository? |
@@ -1,1784 +1,1791 b'' | |||
|
1 | 1 | # changegroup.py - Mercurial changegroup manipulation functions |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2006 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 os |
|
11 | 11 | import struct |
|
12 | 12 | import weakref |
|
13 | 13 | |
|
14 | 14 | from .i18n import _ |
|
15 | 15 | from .node import ( |
|
16 | 16 | hex, |
|
17 | 17 | nullid, |
|
18 | 18 | nullrev, |
|
19 | 19 | short, |
|
20 | 20 | ) |
|
21 | 21 | from .pycompat import open |
|
22 | 22 | |
|
23 | 23 | from . import ( |
|
24 | 24 | error, |
|
25 | 25 | match as matchmod, |
|
26 | 26 | mdiff, |
|
27 | 27 | phases, |
|
28 | 28 | pycompat, |
|
29 | 29 | requirements, |
|
30 | 30 | scmutil, |
|
31 | 31 | util, |
|
32 | 32 | ) |
|
33 | 33 | |
|
34 | 34 | from .interfaces import repository |
|
35 | 35 | from .revlogutils import sidedata as sidedatamod |
|
36 | 36 | |
|
37 | 37 | _CHANGEGROUPV1_DELTA_HEADER = struct.Struct(b"20s20s20s20s") |
|
38 | 38 | _CHANGEGROUPV2_DELTA_HEADER = struct.Struct(b"20s20s20s20s20s") |
|
39 | 39 | _CHANGEGROUPV3_DELTA_HEADER = struct.Struct(b">20s20s20s20s20sH") |
|
40 | 40 | |
|
41 | 41 | LFS_REQUIREMENT = b'lfs' |
|
42 | 42 | |
|
43 | 43 | readexactly = util.readexactly |
|
44 | 44 | |
|
45 | 45 | |
|
46 | 46 | def getchunk(stream): |
|
47 | 47 | """return the next chunk from stream as a string""" |
|
48 | 48 | d = readexactly(stream, 4) |
|
49 | 49 | l = struct.unpack(b">l", d)[0] |
|
50 | 50 | if l <= 4: |
|
51 | 51 | if l: |
|
52 | 52 | raise error.Abort(_(b"invalid chunk length %d") % l) |
|
53 | 53 | return b"" |
|
54 | 54 | return readexactly(stream, l - 4) |
|
55 | 55 | |
|
56 | 56 | |
|
57 | 57 | def chunkheader(length): |
|
58 | 58 | """return a changegroup chunk header (string)""" |
|
59 | 59 | return struct.pack(b">l", length + 4) |
|
60 | 60 | |
|
61 | 61 | |
|
62 | 62 | def closechunk(): |
|
63 | 63 | """return a changegroup chunk header (string) for a zero-length chunk""" |
|
64 | 64 | return struct.pack(b">l", 0) |
|
65 | 65 | |
|
66 | 66 | |
|
67 | 67 | def _fileheader(path): |
|
68 | 68 | """Obtain a changegroup chunk header for a named path.""" |
|
69 | 69 | return chunkheader(len(path)) + path |
|
70 | 70 | |
|
71 | 71 | |
|
72 | 72 | def writechunks(ui, chunks, filename, vfs=None): |
|
73 | 73 | """Write chunks to a file and return its filename. |
|
74 | 74 | |
|
75 | 75 | The stream is assumed to be a bundle file. |
|
76 | 76 | Existing files will not be overwritten. |
|
77 | 77 | If no filename is specified, a temporary file is created. |
|
78 | 78 | """ |
|
79 | 79 | fh = None |
|
80 | 80 | cleanup = None |
|
81 | 81 | try: |
|
82 | 82 | if filename: |
|
83 | 83 | if vfs: |
|
84 | 84 | fh = vfs.open(filename, b"wb") |
|
85 | 85 | else: |
|
86 | 86 | # Increase default buffer size because default is usually |
|
87 | 87 | # small (4k is common on Linux). |
|
88 | 88 | fh = open(filename, b"wb", 131072) |
|
89 | 89 | else: |
|
90 | 90 | fd, filename = pycompat.mkstemp(prefix=b"hg-bundle-", suffix=b".hg") |
|
91 | 91 | fh = os.fdopen(fd, "wb") |
|
92 | 92 | cleanup = filename |
|
93 | 93 | for c in chunks: |
|
94 | 94 | fh.write(c) |
|
95 | 95 | cleanup = None |
|
96 | 96 | return filename |
|
97 | 97 | finally: |
|
98 | 98 | if fh is not None: |
|
99 | 99 | fh.close() |
|
100 | 100 | if cleanup is not None: |
|
101 | 101 | if filename and vfs: |
|
102 | 102 | vfs.unlink(cleanup) |
|
103 | 103 | else: |
|
104 | 104 | os.unlink(cleanup) |
|
105 | 105 | |
|
106 | 106 | |
|
107 | 107 | class cg1unpacker(object): |
|
108 | 108 | """Unpacker for cg1 changegroup streams. |
|
109 | 109 | |
|
110 | 110 | A changegroup unpacker handles the framing of the revision data in |
|
111 | 111 | the wire format. Most consumers will want to use the apply() |
|
112 | 112 | method to add the changes from the changegroup to a repository. |
|
113 | 113 | |
|
114 | 114 | If you're forwarding a changegroup unmodified to another consumer, |
|
115 | 115 | use getchunks(), which returns an iterator of changegroup |
|
116 | 116 | chunks. This is mostly useful for cases where you need to know the |
|
117 | 117 | data stream has ended by observing the end of the changegroup. |
|
118 | 118 | |
|
119 | 119 | deltachunk() is useful only if you're applying delta data. Most |
|
120 | 120 | consumers should prefer apply() instead. |
|
121 | 121 | |
|
122 | 122 | A few other public methods exist. Those are used only for |
|
123 | 123 | bundlerepo and some debug commands - their use is discouraged. |
|
124 | 124 | """ |
|
125 | 125 | |
|
126 | 126 | deltaheader = _CHANGEGROUPV1_DELTA_HEADER |
|
127 | 127 | deltaheadersize = deltaheader.size |
|
128 | 128 | version = b'01' |
|
129 | 129 | _grouplistcount = 1 # One list of files after the manifests |
|
130 | 130 | |
|
131 | 131 | def __init__(self, fh, alg, extras=None): |
|
132 | 132 | if alg is None: |
|
133 | 133 | alg = b'UN' |
|
134 | 134 | if alg not in util.compengines.supportedbundletypes: |
|
135 | 135 | raise error.Abort(_(b'unknown stream compression type: %s') % alg) |
|
136 | 136 | if alg == b'BZ': |
|
137 | 137 | alg = b'_truncatedBZ' |
|
138 | 138 | |
|
139 | 139 | compengine = util.compengines.forbundletype(alg) |
|
140 | 140 | self._stream = compengine.decompressorreader(fh) |
|
141 | 141 | self._type = alg |
|
142 | 142 | self.extras = extras or {} |
|
143 | 143 | self.callback = None |
|
144 | 144 | |
|
145 | 145 | # These methods (compressed, read, seek, tell) all appear to only |
|
146 | 146 | # be used by bundlerepo, but it's a little hard to tell. |
|
147 | 147 | def compressed(self): |
|
148 | 148 | return self._type is not None and self._type != b'UN' |
|
149 | 149 | |
|
150 | 150 | def read(self, l): |
|
151 | 151 | return self._stream.read(l) |
|
152 | 152 | |
|
153 | 153 | def seek(self, pos): |
|
154 | 154 | return self._stream.seek(pos) |
|
155 | 155 | |
|
156 | 156 | def tell(self): |
|
157 | 157 | return self._stream.tell() |
|
158 | 158 | |
|
159 | 159 | def close(self): |
|
160 | 160 | return self._stream.close() |
|
161 | 161 | |
|
162 | 162 | def _chunklength(self): |
|
163 | 163 | d = readexactly(self._stream, 4) |
|
164 | 164 | l = struct.unpack(b">l", d)[0] |
|
165 | 165 | if l <= 4: |
|
166 | 166 | if l: |
|
167 | 167 | raise error.Abort(_(b"invalid chunk length %d") % l) |
|
168 | 168 | return 0 |
|
169 | 169 | if self.callback: |
|
170 | 170 | self.callback() |
|
171 | 171 | return l - 4 |
|
172 | 172 | |
|
173 | 173 | def changelogheader(self): |
|
174 | 174 | """v10 does not have a changelog header chunk""" |
|
175 | 175 | return {} |
|
176 | 176 | |
|
177 | 177 | def manifestheader(self): |
|
178 | 178 | """v10 does not have a manifest header chunk""" |
|
179 | 179 | return {} |
|
180 | 180 | |
|
181 | 181 | def filelogheader(self): |
|
182 | 182 | """return the header of the filelogs chunk, v10 only has the filename""" |
|
183 | 183 | l = self._chunklength() |
|
184 | 184 | if not l: |
|
185 | 185 | return {} |
|
186 | 186 | fname = readexactly(self._stream, l) |
|
187 | 187 | return {b'filename': fname} |
|
188 | 188 | |
|
189 | 189 | def _deltaheader(self, headertuple, prevnode): |
|
190 | 190 | node, p1, p2, cs = headertuple |
|
191 | 191 | if prevnode is None: |
|
192 | 192 | deltabase = p1 |
|
193 | 193 | else: |
|
194 | 194 | deltabase = prevnode |
|
195 | 195 | flags = 0 |
|
196 | 196 | return node, p1, p2, deltabase, cs, flags |
|
197 | 197 | |
|
198 | 198 | def deltachunk(self, prevnode): |
|
199 | 199 | l = self._chunklength() |
|
200 | 200 | if not l: |
|
201 | 201 | return {} |
|
202 | 202 | headerdata = readexactly(self._stream, self.deltaheadersize) |
|
203 | 203 | header = self.deltaheader.unpack(headerdata) |
|
204 | 204 | delta = readexactly(self._stream, l - self.deltaheadersize) |
|
205 | 205 | node, p1, p2, deltabase, cs, flags = self._deltaheader(header, prevnode) |
|
206 | 206 | # cg4 forward-compat |
|
207 | 207 | sidedata = {} |
|
208 | 208 | return (node, p1, p2, cs, deltabase, delta, flags, sidedata) |
|
209 | 209 | |
|
210 | 210 | def getchunks(self): |
|
211 | 211 | """returns all the chunks contains in the bundle |
|
212 | 212 | |
|
213 | 213 | Used when you need to forward the binary stream to a file or another |
|
214 | 214 | network API. To do so, it parse the changegroup data, otherwise it will |
|
215 | 215 | block in case of sshrepo because it don't know the end of the stream. |
|
216 | 216 | """ |
|
217 | 217 | # For changegroup 1 and 2, we expect 3 parts: changelog, manifestlog, |
|
218 | 218 | # and a list of filelogs. For changegroup 3, we expect 4 parts: |
|
219 | 219 | # changelog, manifestlog, a list of tree manifestlogs, and a list of |
|
220 | 220 | # filelogs. |
|
221 | 221 | # |
|
222 | 222 | # Changelog and manifestlog parts are terminated with empty chunks. The |
|
223 | 223 | # tree and file parts are a list of entry sections. Each entry section |
|
224 | 224 | # is a series of chunks terminating in an empty chunk. The list of these |
|
225 | 225 | # entry sections is terminated in yet another empty chunk, so we know |
|
226 | 226 | # we've reached the end of the tree/file list when we reach an empty |
|
227 | 227 | # chunk that was proceeded by no non-empty chunks. |
|
228 | 228 | |
|
229 | 229 | parts = 0 |
|
230 | 230 | while parts < 2 + self._grouplistcount: |
|
231 | 231 | noentries = True |
|
232 | 232 | while True: |
|
233 | 233 | chunk = getchunk(self) |
|
234 | 234 | if not chunk: |
|
235 | 235 | # The first two empty chunks represent the end of the |
|
236 | 236 | # changelog and the manifestlog portions. The remaining |
|
237 | 237 | # empty chunks represent either A) the end of individual |
|
238 | 238 | # tree or file entries in the file list, or B) the end of |
|
239 | 239 | # the entire list. It's the end of the entire list if there |
|
240 | 240 | # were no entries (i.e. noentries is True). |
|
241 | 241 | if parts < 2: |
|
242 | 242 | parts += 1 |
|
243 | 243 | elif noentries: |
|
244 | 244 | parts += 1 |
|
245 | 245 | break |
|
246 | 246 | noentries = False |
|
247 | 247 | yield chunkheader(len(chunk)) |
|
248 | 248 | pos = 0 |
|
249 | 249 | while pos < len(chunk): |
|
250 | 250 | next = pos + 2 ** 20 |
|
251 | 251 | yield chunk[pos:next] |
|
252 | 252 | pos = next |
|
253 | 253 | yield closechunk() |
|
254 | 254 | |
|
255 | 255 | def _unpackmanifests(self, repo, revmap, trp, prog): |
|
256 | 256 | self.callback = prog.increment |
|
257 | 257 | # no need to check for empty manifest group here: |
|
258 | 258 | # if the result of the merge of 1 and 2 is the same in 3 and 4, |
|
259 | 259 | # no new manifest will be created and the manifest group will |
|
260 | 260 | # be empty during the pull |
|
261 | 261 | self.manifestheader() |
|
262 | 262 | deltas = self.deltaiter() |
|
263 | 263 | repo.manifestlog.getstorage(b'').addgroup(deltas, revmap, trp) |
|
264 | 264 | prog.complete() |
|
265 | 265 | self.callback = None |
|
266 | 266 | |
|
267 | 267 | def apply( |
|
268 | 268 | self, |
|
269 | 269 | repo, |
|
270 | 270 | tr, |
|
271 | 271 | srctype, |
|
272 | 272 | url, |
|
273 | 273 | targetphase=phases.draft, |
|
274 | 274 | expectedtotal=None, |
|
275 | 275 | ): |
|
276 | 276 | """Add the changegroup returned by source.read() to this repo. |
|
277 | 277 | srctype is a string like 'push', 'pull', or 'unbundle'. url is |
|
278 | 278 | the URL of the repo where this changegroup is coming from. |
|
279 | 279 | |
|
280 | 280 | Return an integer summarizing the change to this repo: |
|
281 | 281 | - nothing changed or no source: 0 |
|
282 | 282 | - more heads than before: 1+added heads (2..n) |
|
283 | 283 | - fewer heads than before: -1-removed heads (-2..-n) |
|
284 | 284 | - number of heads stays the same: 1 |
|
285 | 285 | """ |
|
286 | 286 | repo = repo.unfiltered() |
|
287 | 287 | |
|
288 | 288 | def csmap(x): |
|
289 | 289 | repo.ui.debug(b"add changeset %s\n" % short(x)) |
|
290 | 290 | return len(cl) |
|
291 | 291 | |
|
292 | 292 | def revmap(x): |
|
293 | 293 | return cl.rev(x) |
|
294 | 294 | |
|
295 | 295 | try: |
|
296 | 296 | # The transaction may already carry source information. In this |
|
297 | 297 | # case we use the top level data. We overwrite the argument |
|
298 | 298 | # because we need to use the top level value (if they exist) |
|
299 | 299 | # in this function. |
|
300 | 300 | srctype = tr.hookargs.setdefault(b'source', srctype) |
|
301 | 301 | tr.hookargs.setdefault(b'url', url) |
|
302 | 302 | repo.hook( |
|
303 | 303 | b'prechangegroup', throw=True, **pycompat.strkwargs(tr.hookargs) |
|
304 | 304 | ) |
|
305 | 305 | |
|
306 | 306 | # write changelog data to temp files so concurrent readers |
|
307 | 307 | # will not see an inconsistent view |
|
308 | 308 | cl = repo.changelog |
|
309 | 309 | cl.delayupdate(tr) |
|
310 | 310 | oldheads = set(cl.heads()) |
|
311 | 311 | |
|
312 | 312 | trp = weakref.proxy(tr) |
|
313 | 313 | # pull off the changeset group |
|
314 | 314 | repo.ui.status(_(b"adding changesets\n")) |
|
315 | 315 | clstart = len(cl) |
|
316 | 316 | progress = repo.ui.makeprogress( |
|
317 | 317 | _(b'changesets'), unit=_(b'chunks'), total=expectedtotal |
|
318 | 318 | ) |
|
319 | 319 | self.callback = progress.increment |
|
320 | 320 | |
|
321 | 321 | efilesset = set() |
|
322 | 322 | duprevs = [] |
|
323 | 323 | |
|
324 | 324 | def ondupchangelog(cl, rev): |
|
325 | 325 | if rev < clstart: |
|
326 | 326 | duprevs.append(rev) |
|
327 | 327 | |
|
328 | 328 | def onchangelog(cl, rev): |
|
329 | 329 | ctx = cl.changelogrevision(rev) |
|
330 | 330 | efilesset.update(ctx.files) |
|
331 | 331 | repo.register_changeset(rev, ctx) |
|
332 | 332 | |
|
333 | 333 | self.changelogheader() |
|
334 | 334 | deltas = self.deltaiter() |
|
335 | 335 | if not cl.addgroup( |
|
336 | 336 | deltas, |
|
337 | 337 | csmap, |
|
338 | 338 | trp, |
|
339 | 339 | alwayscache=True, |
|
340 | 340 | addrevisioncb=onchangelog, |
|
341 | 341 | duplicaterevisioncb=ondupchangelog, |
|
342 | 342 | ): |
|
343 | 343 | repo.ui.develwarn( |
|
344 | 344 | b'applied empty changelog from changegroup', |
|
345 | 345 | config=b'warn-empty-changegroup', |
|
346 | 346 | ) |
|
347 | 347 | efiles = len(efilesset) |
|
348 | 348 | clend = len(cl) |
|
349 | 349 | changesets = clend - clstart |
|
350 | 350 | progress.complete() |
|
351 | 351 | del deltas |
|
352 | 352 | # TODO Python 2.7 removal |
|
353 | 353 | # del efilesset |
|
354 | 354 | efilesset = None |
|
355 | 355 | self.callback = None |
|
356 | 356 | |
|
357 | 357 | # pull off the manifest group |
|
358 | 358 | repo.ui.status(_(b"adding manifests\n")) |
|
359 | 359 | # We know that we'll never have more manifests than we had |
|
360 | 360 | # changesets. |
|
361 | 361 | progress = repo.ui.makeprogress( |
|
362 | 362 | _(b'manifests'), unit=_(b'chunks'), total=changesets |
|
363 | 363 | ) |
|
364 | 364 | self._unpackmanifests(repo, revmap, trp, progress) |
|
365 | 365 | |
|
366 | 366 | needfiles = {} |
|
367 | 367 | if repo.ui.configbool(b'server', b'validate'): |
|
368 | 368 | cl = repo.changelog |
|
369 | 369 | ml = repo.manifestlog |
|
370 | 370 | # validate incoming csets have their manifests |
|
371 | 371 | for cset in pycompat.xrange(clstart, clend): |
|
372 | 372 | mfnode = cl.changelogrevision(cset).manifest |
|
373 | 373 | mfest = ml[mfnode].readdelta() |
|
374 | 374 | # store file nodes we must see |
|
375 | 375 | for f, n in pycompat.iteritems(mfest): |
|
376 | 376 | needfiles.setdefault(f, set()).add(n) |
|
377 | 377 | |
|
378 | 378 | # process the files |
|
379 | 379 | repo.ui.status(_(b"adding file changes\n")) |
|
380 | 380 | newrevs, newfiles = _addchangegroupfiles( |
|
381 | 381 | repo, self, revmap, trp, efiles, needfiles |
|
382 | 382 | ) |
|
383 | 383 | |
|
384 | 384 | # making sure the value exists |
|
385 | 385 | tr.changes.setdefault(b'changegroup-count-changesets', 0) |
|
386 | 386 | tr.changes.setdefault(b'changegroup-count-revisions', 0) |
|
387 | 387 | tr.changes.setdefault(b'changegroup-count-files', 0) |
|
388 | 388 | tr.changes.setdefault(b'changegroup-count-heads', 0) |
|
389 | 389 | |
|
390 | 390 | # some code use bundle operation for internal purpose. They usually |
|
391 | 391 | # set `ui.quiet` to do this outside of user sight. Size the report |
|
392 | 392 | # of such operation now happens at the end of the transaction, that |
|
393 | 393 | # ui.quiet has not direct effect on the output. |
|
394 | 394 | # |
|
395 | 395 | # To preserve this intend use an inelegant hack, we fail to report |
|
396 | 396 | # the change if `quiet` is set. We should probably move to |
|
397 | 397 | # something better, but this is a good first step to allow the "end |
|
398 | 398 | # of transaction report" to pass tests. |
|
399 | 399 | if not repo.ui.quiet: |
|
400 | 400 | tr.changes[b'changegroup-count-changesets'] += changesets |
|
401 | 401 | tr.changes[b'changegroup-count-revisions'] += newrevs |
|
402 | 402 | tr.changes[b'changegroup-count-files'] += newfiles |
|
403 | 403 | |
|
404 | 404 | deltaheads = 0 |
|
405 | 405 | if oldheads: |
|
406 | 406 | heads = cl.heads() |
|
407 | 407 | deltaheads += len(heads) - len(oldheads) |
|
408 | 408 | for h in heads: |
|
409 | 409 | if h not in oldheads and repo[h].closesbranch(): |
|
410 | 410 | deltaheads -= 1 |
|
411 | 411 | |
|
412 | 412 | # see previous comment about checking ui.quiet |
|
413 | 413 | if not repo.ui.quiet: |
|
414 | 414 | tr.changes[b'changegroup-count-heads'] += deltaheads |
|
415 | 415 | repo.invalidatevolatilesets() |
|
416 | 416 | |
|
417 | 417 | if changesets > 0: |
|
418 | 418 | if b'node' not in tr.hookargs: |
|
419 | 419 | tr.hookargs[b'node'] = hex(cl.node(clstart)) |
|
420 | 420 | tr.hookargs[b'node_last'] = hex(cl.node(clend - 1)) |
|
421 | 421 | hookargs = dict(tr.hookargs) |
|
422 | 422 | else: |
|
423 | 423 | hookargs = dict(tr.hookargs) |
|
424 | 424 | hookargs[b'node'] = hex(cl.node(clstart)) |
|
425 | 425 | hookargs[b'node_last'] = hex(cl.node(clend - 1)) |
|
426 | 426 | repo.hook( |
|
427 | 427 | b'pretxnchangegroup', |
|
428 | 428 | throw=True, |
|
429 | 429 | **pycompat.strkwargs(hookargs) |
|
430 | 430 | ) |
|
431 | 431 | |
|
432 | 432 | added = pycompat.xrange(clstart, clend) |
|
433 | 433 | phaseall = None |
|
434 | 434 | if srctype in (b'push', b'serve'): |
|
435 | 435 | # Old servers can not push the boundary themselves. |
|
436 | 436 | # New servers won't push the boundary if changeset already |
|
437 | 437 | # exists locally as secret |
|
438 | 438 | # |
|
439 | 439 | # We should not use added here but the list of all change in |
|
440 | 440 | # the bundle |
|
441 | 441 | if repo.publishing(): |
|
442 | 442 | targetphase = phaseall = phases.public |
|
443 | 443 | else: |
|
444 | 444 | # closer target phase computation |
|
445 | 445 | |
|
446 | 446 | # Those changesets have been pushed from the |
|
447 | 447 | # outside, their phases are going to be pushed |
|
448 | 448 | # alongside. Therefor `targetphase` is |
|
449 | 449 | # ignored. |
|
450 | 450 | targetphase = phaseall = phases.draft |
|
451 | 451 | if added: |
|
452 | 452 | phases.registernew(repo, tr, targetphase, added) |
|
453 | 453 | if phaseall is not None: |
|
454 | 454 | if duprevs: |
|
455 | 455 | duprevs.extend(added) |
|
456 | 456 | else: |
|
457 | 457 | duprevs = added |
|
458 | 458 | phases.advanceboundary(repo, tr, phaseall, [], revs=duprevs) |
|
459 | 459 | duprevs = [] |
|
460 | 460 | |
|
461 | 461 | if changesets > 0: |
|
462 | 462 | |
|
463 | 463 | def runhooks(unused_success): |
|
464 | 464 | # These hooks run when the lock releases, not when the |
|
465 | 465 | # transaction closes. So it's possible for the changelog |
|
466 | 466 | # to have changed since we last saw it. |
|
467 | 467 | if clstart >= len(repo): |
|
468 | 468 | return |
|
469 | 469 | |
|
470 | 470 | repo.hook(b"changegroup", **pycompat.strkwargs(hookargs)) |
|
471 | 471 | |
|
472 | 472 | for rev in added: |
|
473 | 473 | args = hookargs.copy() |
|
474 | 474 | args[b'node'] = hex(cl.node(rev)) |
|
475 | 475 | del args[b'node_last'] |
|
476 | 476 | repo.hook(b"incoming", **pycompat.strkwargs(args)) |
|
477 | 477 | |
|
478 | 478 | newheads = [h for h in repo.heads() if h not in oldheads] |
|
479 | 479 | repo.ui.log( |
|
480 | 480 | b"incoming", |
|
481 | 481 | b"%d incoming changes - new heads: %s\n", |
|
482 | 482 | len(added), |
|
483 | 483 | b', '.join([hex(c[:6]) for c in newheads]), |
|
484 | 484 | ) |
|
485 | 485 | |
|
486 | 486 | tr.addpostclose( |
|
487 | 487 | b'changegroup-runhooks-%020i' % clstart, |
|
488 | 488 | lambda tr: repo._afterlock(runhooks), |
|
489 | 489 | ) |
|
490 | 490 | finally: |
|
491 | 491 | repo.ui.flush() |
|
492 | 492 | # never return 0 here: |
|
493 | 493 | if deltaheads < 0: |
|
494 | 494 | ret = deltaheads - 1 |
|
495 | 495 | else: |
|
496 | 496 | ret = deltaheads + 1 |
|
497 | 497 | return ret |
|
498 | 498 | |
|
499 | 499 | def deltaiter(self): |
|
500 | 500 | """ |
|
501 | 501 | returns an iterator of the deltas in this changegroup |
|
502 | 502 | |
|
503 | 503 | Useful for passing to the underlying storage system to be stored. |
|
504 | 504 | """ |
|
505 | 505 | chain = None |
|
506 | 506 | for chunkdata in iter(lambda: self.deltachunk(chain), {}): |
|
507 | 507 | # Chunkdata: (node, p1, p2, cs, deltabase, delta, flags) |
|
508 | 508 | yield chunkdata |
|
509 | 509 | chain = chunkdata[0] |
|
510 | 510 | |
|
511 | 511 | |
|
512 | 512 | class cg2unpacker(cg1unpacker): |
|
513 | 513 | """Unpacker for cg2 streams. |
|
514 | 514 | |
|
515 | 515 | cg2 streams add support for generaldelta, so the delta header |
|
516 | 516 | format is slightly different. All other features about the data |
|
517 | 517 | remain the same. |
|
518 | 518 | """ |
|
519 | 519 | |
|
520 | 520 | deltaheader = _CHANGEGROUPV2_DELTA_HEADER |
|
521 | 521 | deltaheadersize = deltaheader.size |
|
522 | 522 | version = b'02' |
|
523 | 523 | |
|
524 | 524 | def _deltaheader(self, headertuple, prevnode): |
|
525 | 525 | node, p1, p2, deltabase, cs = headertuple |
|
526 | 526 | flags = 0 |
|
527 | 527 | return node, p1, p2, deltabase, cs, flags |
|
528 | 528 | |
|
529 | 529 | |
|
530 | 530 | class cg3unpacker(cg2unpacker): |
|
531 | 531 | """Unpacker for cg3 streams. |
|
532 | 532 | |
|
533 | 533 | cg3 streams add support for exchanging treemanifests and revlog |
|
534 | 534 | flags. It adds the revlog flags to the delta header and an empty chunk |
|
535 | 535 | separating manifests and files. |
|
536 | 536 | """ |
|
537 | 537 | |
|
538 | 538 | deltaheader = _CHANGEGROUPV3_DELTA_HEADER |
|
539 | 539 | deltaheadersize = deltaheader.size |
|
540 | 540 | version = b'03' |
|
541 | 541 | _grouplistcount = 2 # One list of manifests and one list of files |
|
542 | 542 | |
|
543 | 543 | def _deltaheader(self, headertuple, prevnode): |
|
544 | 544 | node, p1, p2, deltabase, cs, flags = headertuple |
|
545 | 545 | return node, p1, p2, deltabase, cs, flags |
|
546 | 546 | |
|
547 | 547 | def _unpackmanifests(self, repo, revmap, trp, prog): |
|
548 | 548 | super(cg3unpacker, self)._unpackmanifests(repo, revmap, trp, prog) |
|
549 | 549 | for chunkdata in iter(self.filelogheader, {}): |
|
550 | 550 | # If we get here, there are directory manifests in the changegroup |
|
551 | 551 | d = chunkdata[b"filename"] |
|
552 | 552 | repo.ui.debug(b"adding %s revisions\n" % d) |
|
553 | 553 | deltas = self.deltaiter() |
|
554 | 554 | if not repo.manifestlog.getstorage(d).addgroup(deltas, revmap, trp): |
|
555 | 555 | raise error.Abort(_(b"received dir revlog group is empty")) |
|
556 | 556 | |
|
557 | 557 | |
|
558 | 558 | class cg4unpacker(cg3unpacker): |
|
559 | 559 | """Unpacker for cg4 streams. |
|
560 | 560 | |
|
561 | 561 | cg4 streams add support for exchanging sidedata. |
|
562 | 562 | """ |
|
563 | 563 | |
|
564 | 564 | version = b'04' |
|
565 | 565 | |
|
566 | 566 | def deltachunk(self, prevnode): |
|
567 | 567 | res = super(cg4unpacker, self).deltachunk(prevnode) |
|
568 | 568 | if not res: |
|
569 | 569 | return res |
|
570 | 570 | |
|
571 | 571 | (node, p1, p2, cs, deltabase, delta, flags, _sidedata) = res |
|
572 | 572 | |
|
573 | 573 | sidedata_raw = getchunk(self._stream) |
|
574 | 574 | sidedata = {} |
|
575 | 575 | if len(sidedata_raw) > 0: |
|
576 | 576 | sidedata = sidedatamod.deserialize_sidedata(sidedata_raw) |
|
577 | 577 | |
|
578 | 578 | return node, p1, p2, cs, deltabase, delta, flags, sidedata |
|
579 | 579 | |
|
580 | 580 | |
|
581 | 581 | class headerlessfixup(object): |
|
582 | 582 | def __init__(self, fh, h): |
|
583 | 583 | self._h = h |
|
584 | 584 | self._fh = fh |
|
585 | 585 | |
|
586 | 586 | def read(self, n): |
|
587 | 587 | if self._h: |
|
588 | 588 | d, self._h = self._h[:n], self._h[n:] |
|
589 | 589 | if len(d) < n: |
|
590 | 590 | d += readexactly(self._fh, n - len(d)) |
|
591 | 591 | return d |
|
592 | 592 | return readexactly(self._fh, n) |
|
593 | 593 | |
|
594 | 594 | |
|
595 | 595 | def _revisiondeltatochunks(delta, headerfn): |
|
596 | 596 | """Serialize a revisiondelta to changegroup chunks.""" |
|
597 | 597 | |
|
598 | 598 | # The captured revision delta may be encoded as a delta against |
|
599 | 599 | # a base revision or as a full revision. The changegroup format |
|
600 | 600 | # requires that everything on the wire be deltas. So for full |
|
601 | 601 | # revisions, we need to invent a header that says to rewrite |
|
602 | 602 | # data. |
|
603 | 603 | |
|
604 | 604 | if delta.delta is not None: |
|
605 | 605 | prefix, data = b'', delta.delta |
|
606 | 606 | elif delta.basenode == nullid: |
|
607 | 607 | data = delta.revision |
|
608 | 608 | prefix = mdiff.trivialdiffheader(len(data)) |
|
609 | 609 | else: |
|
610 | 610 | data = delta.revision |
|
611 | 611 | prefix = mdiff.replacediffheader(delta.baserevisionsize, len(data)) |
|
612 | 612 | |
|
613 | 613 | meta = headerfn(delta) |
|
614 | 614 | |
|
615 | 615 | yield chunkheader(len(meta) + len(prefix) + len(data)) |
|
616 | 616 | yield meta |
|
617 | 617 | if prefix: |
|
618 | 618 | yield prefix |
|
619 | 619 | yield data |
|
620 | 620 | |
|
621 | sidedata = delta.sidedata | |
|
622 | if sidedata is not None: | |
|
623 | # Need a separate chunk for sidedata to be able to differentiate | |
|
624 | # "raw delta" length and sidedata length | |
|
625 | yield chunkheader(len(sidedata)) | |
|
626 | yield sidedata | |
|
627 | ||
|
621 | 628 | |
|
622 | 629 | def _sortnodesellipsis(store, nodes, cl, lookup): |
|
623 | 630 | """Sort nodes for changegroup generation.""" |
|
624 | 631 | # Ellipses serving mode. |
|
625 | 632 | # |
|
626 | 633 | # In a perfect world, we'd generate better ellipsis-ified graphs |
|
627 | 634 | # for non-changelog revlogs. In practice, we haven't started doing |
|
628 | 635 | # that yet, so the resulting DAGs for the manifestlog and filelogs |
|
629 | 636 | # are actually full of bogus parentage on all the ellipsis |
|
630 | 637 | # nodes. This has the side effect that, while the contents are |
|
631 | 638 | # correct, the individual DAGs might be completely out of whack in |
|
632 | 639 | # a case like 882681bc3166 and its ancestors (back about 10 |
|
633 | 640 | # revisions or so) in the main hg repo. |
|
634 | 641 | # |
|
635 | 642 | # The one invariant we *know* holds is that the new (potentially |
|
636 | 643 | # bogus) DAG shape will be valid if we order the nodes in the |
|
637 | 644 | # order that they're introduced in dramatis personae by the |
|
638 | 645 | # changelog, so what we do is we sort the non-changelog histories |
|
639 | 646 | # by the order in which they are used by the changelog. |
|
640 | 647 | key = lambda n: cl.rev(lookup(n)) |
|
641 | 648 | return sorted(nodes, key=key) |
|
642 | 649 | |
|
643 | 650 | |
|
644 | 651 | def _resolvenarrowrevisioninfo( |
|
645 | 652 | cl, |
|
646 | 653 | store, |
|
647 | 654 | ischangelog, |
|
648 | 655 | rev, |
|
649 | 656 | linkrev, |
|
650 | 657 | linknode, |
|
651 | 658 | clrevtolocalrev, |
|
652 | 659 | fullclnodes, |
|
653 | 660 | precomputedellipsis, |
|
654 | 661 | ): |
|
655 | 662 | linkparents = precomputedellipsis[linkrev] |
|
656 | 663 | |
|
657 | 664 | def local(clrev): |
|
658 | 665 | """Turn a changelog revnum into a local revnum. |
|
659 | 666 | |
|
660 | 667 | The ellipsis dag is stored as revnums on the changelog, |
|
661 | 668 | but when we're producing ellipsis entries for |
|
662 | 669 | non-changelog revlogs, we need to turn those numbers into |
|
663 | 670 | something local. This does that for us, and during the |
|
664 | 671 | changelog sending phase will also expand the stored |
|
665 | 672 | mappings as needed. |
|
666 | 673 | """ |
|
667 | 674 | if clrev == nullrev: |
|
668 | 675 | return nullrev |
|
669 | 676 | |
|
670 | 677 | if ischangelog: |
|
671 | 678 | return clrev |
|
672 | 679 | |
|
673 | 680 | # Walk the ellipsis-ized changelog breadth-first looking for a |
|
674 | 681 | # change that has been linked from the current revlog. |
|
675 | 682 | # |
|
676 | 683 | # For a flat manifest revlog only a single step should be necessary |
|
677 | 684 | # as all relevant changelog entries are relevant to the flat |
|
678 | 685 | # manifest. |
|
679 | 686 | # |
|
680 | 687 | # For a filelog or tree manifest dirlog however not every changelog |
|
681 | 688 | # entry will have been relevant, so we need to skip some changelog |
|
682 | 689 | # nodes even after ellipsis-izing. |
|
683 | 690 | walk = [clrev] |
|
684 | 691 | while walk: |
|
685 | 692 | p = walk[0] |
|
686 | 693 | walk = walk[1:] |
|
687 | 694 | if p in clrevtolocalrev: |
|
688 | 695 | return clrevtolocalrev[p] |
|
689 | 696 | elif p in fullclnodes: |
|
690 | 697 | walk.extend([pp for pp in cl.parentrevs(p) if pp != nullrev]) |
|
691 | 698 | elif p in precomputedellipsis: |
|
692 | 699 | walk.extend( |
|
693 | 700 | [pp for pp in precomputedellipsis[p] if pp != nullrev] |
|
694 | 701 | ) |
|
695 | 702 | else: |
|
696 | 703 | # In this case, we've got an ellipsis with parents |
|
697 | 704 | # outside the current bundle (likely an |
|
698 | 705 | # incremental pull). We "know" that we can use the |
|
699 | 706 | # value of this same revlog at whatever revision |
|
700 | 707 | # is pointed to by linknode. "Know" is in scare |
|
701 | 708 | # quotes because I haven't done enough examination |
|
702 | 709 | # of edge cases to convince myself this is really |
|
703 | 710 | # a fact - it works for all the (admittedly |
|
704 | 711 | # thorough) cases in our testsuite, but I would be |
|
705 | 712 | # somewhat unsurprised to find a case in the wild |
|
706 | 713 | # where this breaks down a bit. That said, I don't |
|
707 | 714 | # know if it would hurt anything. |
|
708 | 715 | for i in pycompat.xrange(rev, 0, -1): |
|
709 | 716 | if store.linkrev(i) == clrev: |
|
710 | 717 | return i |
|
711 | 718 | # We failed to resolve a parent for this node, so |
|
712 | 719 | # we crash the changegroup construction. |
|
713 | 720 | raise error.Abort( |
|
714 | 721 | b"unable to resolve parent while packing '%s' %r" |
|
715 | 722 | b' for changeset %r' % (store.indexfile, rev, clrev) |
|
716 | 723 | ) |
|
717 | 724 | |
|
718 | 725 | return nullrev |
|
719 | 726 | |
|
720 | 727 | if not linkparents or (store.parentrevs(rev) == (nullrev, nullrev)): |
|
721 | 728 | p1, p2 = nullrev, nullrev |
|
722 | 729 | elif len(linkparents) == 1: |
|
723 | 730 | (p1,) = sorted(local(p) for p in linkparents) |
|
724 | 731 | p2 = nullrev |
|
725 | 732 | else: |
|
726 | 733 | p1, p2 = sorted(local(p) for p in linkparents) |
|
727 | 734 | |
|
728 | 735 | p1node, p2node = store.node(p1), store.node(p2) |
|
729 | 736 | |
|
730 | 737 | return p1node, p2node, linknode |
|
731 | 738 | |
|
732 | 739 | |
|
733 | 740 | def deltagroup( |
|
734 | 741 | repo, |
|
735 | 742 | store, |
|
736 | 743 | nodes, |
|
737 | 744 | ischangelog, |
|
738 | 745 | lookup, |
|
739 | 746 | forcedeltaparentprev, |
|
740 | 747 | topic=None, |
|
741 | 748 | ellipses=False, |
|
742 | 749 | clrevtolocalrev=None, |
|
743 | 750 | fullclnodes=None, |
|
744 | 751 | precomputedellipsis=None, |
|
745 | 752 | ): |
|
746 | 753 | """Calculate deltas for a set of revisions. |
|
747 | 754 | |
|
748 | 755 | Is a generator of ``revisiondelta`` instances. |
|
749 | 756 | |
|
750 | 757 | If topic is not None, progress detail will be generated using this |
|
751 | 758 | topic name (e.g. changesets, manifests, etc). |
|
752 | 759 | """ |
|
753 | 760 | if not nodes: |
|
754 | 761 | return |
|
755 | 762 | |
|
756 | 763 | cl = repo.changelog |
|
757 | 764 | |
|
758 | 765 | if ischangelog: |
|
759 | 766 | # `hg log` shows changesets in storage order. To preserve order |
|
760 | 767 | # across clones, send out changesets in storage order. |
|
761 | 768 | nodesorder = b'storage' |
|
762 | 769 | elif ellipses: |
|
763 | 770 | nodes = _sortnodesellipsis(store, nodes, cl, lookup) |
|
764 | 771 | nodesorder = b'nodes' |
|
765 | 772 | else: |
|
766 | 773 | nodesorder = None |
|
767 | 774 | |
|
768 | 775 | # Perform ellipses filtering and revision massaging. We do this before |
|
769 | 776 | # emitrevisions() because a) filtering out revisions creates less work |
|
770 | 777 | # for emitrevisions() b) dropping revisions would break emitrevisions()'s |
|
771 | 778 | # assumptions about delta choices and we would possibly send a delta |
|
772 | 779 | # referencing a missing base revision. |
|
773 | 780 | # |
|
774 | 781 | # Also, calling lookup() has side-effects with regards to populating |
|
775 | 782 | # data structures. If we don't call lookup() for each node or if we call |
|
776 | 783 | # lookup() after the first pass through each node, things can break - |
|
777 | 784 | # possibly intermittently depending on the python hash seed! For that |
|
778 | 785 | # reason, we store a mapping of all linknodes during the initial node |
|
779 | 786 | # pass rather than use lookup() on the output side. |
|
780 | 787 | if ellipses: |
|
781 | 788 | filtered = [] |
|
782 | 789 | adjustedparents = {} |
|
783 | 790 | linknodes = {} |
|
784 | 791 | |
|
785 | 792 | for node in nodes: |
|
786 | 793 | rev = store.rev(node) |
|
787 | 794 | linknode = lookup(node) |
|
788 | 795 | linkrev = cl.rev(linknode) |
|
789 | 796 | clrevtolocalrev[linkrev] = rev |
|
790 | 797 | |
|
791 | 798 | # If linknode is in fullclnodes, it means the corresponding |
|
792 | 799 | # changeset was a full changeset and is being sent unaltered. |
|
793 | 800 | if linknode in fullclnodes: |
|
794 | 801 | linknodes[node] = linknode |
|
795 | 802 | |
|
796 | 803 | # If the corresponding changeset wasn't in the set computed |
|
797 | 804 | # as relevant to us, it should be dropped outright. |
|
798 | 805 | elif linkrev not in precomputedellipsis: |
|
799 | 806 | continue |
|
800 | 807 | |
|
801 | 808 | else: |
|
802 | 809 | # We could probably do this later and avoid the dict |
|
803 | 810 | # holding state. But it likely doesn't matter. |
|
804 | 811 | p1node, p2node, linknode = _resolvenarrowrevisioninfo( |
|
805 | 812 | cl, |
|
806 | 813 | store, |
|
807 | 814 | ischangelog, |
|
808 | 815 | rev, |
|
809 | 816 | linkrev, |
|
810 | 817 | linknode, |
|
811 | 818 | clrevtolocalrev, |
|
812 | 819 | fullclnodes, |
|
813 | 820 | precomputedellipsis, |
|
814 | 821 | ) |
|
815 | 822 | |
|
816 | 823 | adjustedparents[node] = (p1node, p2node) |
|
817 | 824 | linknodes[node] = linknode |
|
818 | 825 | |
|
819 | 826 | filtered.append(node) |
|
820 | 827 | |
|
821 | 828 | nodes = filtered |
|
822 | 829 | |
|
823 | 830 | # We expect the first pass to be fast, so we only engage the progress |
|
824 | 831 | # meter for constructing the revision deltas. |
|
825 | 832 | progress = None |
|
826 | 833 | if topic is not None: |
|
827 | 834 | progress = repo.ui.makeprogress( |
|
828 | 835 | topic, unit=_(b'chunks'), total=len(nodes) |
|
829 | 836 | ) |
|
830 | 837 | |
|
831 | 838 | configtarget = repo.ui.config(b'devel', b'bundle.delta') |
|
832 | 839 | if configtarget not in (b'', b'p1', b'full'): |
|
833 | 840 | msg = _("""config "devel.bundle.delta" as unknown value: %s""") |
|
834 | 841 | repo.ui.warn(msg % configtarget) |
|
835 | 842 | |
|
836 | 843 | deltamode = repository.CG_DELTAMODE_STD |
|
837 | 844 | if forcedeltaparentprev: |
|
838 | 845 | deltamode = repository.CG_DELTAMODE_PREV |
|
839 | 846 | elif configtarget == b'p1': |
|
840 | 847 | deltamode = repository.CG_DELTAMODE_P1 |
|
841 | 848 | elif configtarget == b'full': |
|
842 | 849 | deltamode = repository.CG_DELTAMODE_FULL |
|
843 | 850 | |
|
844 | 851 | revisions = store.emitrevisions( |
|
845 | 852 | nodes, |
|
846 | 853 | nodesorder=nodesorder, |
|
847 | 854 | revisiondata=True, |
|
848 | 855 | assumehaveparentrevisions=not ellipses, |
|
849 | 856 | deltamode=deltamode, |
|
850 | 857 | ) |
|
851 | 858 | |
|
852 | 859 | for i, revision in enumerate(revisions): |
|
853 | 860 | if progress: |
|
854 | 861 | progress.update(i + 1) |
|
855 | 862 | |
|
856 | 863 | if ellipses: |
|
857 | 864 | linknode = linknodes[revision.node] |
|
858 | 865 | |
|
859 | 866 | if revision.node in adjustedparents: |
|
860 | 867 | p1node, p2node = adjustedparents[revision.node] |
|
861 | 868 | revision.p1node = p1node |
|
862 | 869 | revision.p2node = p2node |
|
863 | 870 | revision.flags |= repository.REVISION_FLAG_ELLIPSIS |
|
864 | 871 | |
|
865 | 872 | else: |
|
866 | 873 | linknode = lookup(revision.node) |
|
867 | 874 | |
|
868 | 875 | revision.linknode = linknode |
|
869 | 876 | yield revision |
|
870 | 877 | |
|
871 | 878 | if progress: |
|
872 | 879 | progress.complete() |
|
873 | 880 | |
|
874 | 881 | |
|
875 | 882 | class cgpacker(object): |
|
876 | 883 | def __init__( |
|
877 | 884 | self, |
|
878 | 885 | repo, |
|
879 | 886 | oldmatcher, |
|
880 | 887 | matcher, |
|
881 | 888 | version, |
|
882 | 889 | builddeltaheader, |
|
883 | 890 | manifestsend, |
|
884 | 891 | forcedeltaparentprev=False, |
|
885 | 892 | bundlecaps=None, |
|
886 | 893 | ellipses=False, |
|
887 | 894 | shallow=False, |
|
888 | 895 | ellipsisroots=None, |
|
889 | 896 | fullnodes=None, |
|
890 | 897 | remote_sidedata=None, |
|
891 | 898 | ): |
|
892 | 899 | """Given a source repo, construct a bundler. |
|
893 | 900 | |
|
894 | 901 | oldmatcher is a matcher that matches on files the client already has. |
|
895 | 902 | These will not be included in the changegroup. |
|
896 | 903 | |
|
897 | 904 | matcher is a matcher that matches on files to include in the |
|
898 | 905 | changegroup. Used to facilitate sparse changegroups. |
|
899 | 906 | |
|
900 | 907 | forcedeltaparentprev indicates whether delta parents must be against |
|
901 | 908 | the previous revision in a delta group. This should only be used for |
|
902 | 909 | compatibility with changegroup version 1. |
|
903 | 910 | |
|
904 | 911 | builddeltaheader is a callable that constructs the header for a group |
|
905 | 912 | delta. |
|
906 | 913 | |
|
907 | 914 | manifestsend is a chunk to send after manifests have been fully emitted. |
|
908 | 915 | |
|
909 | 916 | ellipses indicates whether ellipsis serving mode is enabled. |
|
910 | 917 | |
|
911 | 918 | bundlecaps is optional and can be used to specify the set of |
|
912 | 919 | capabilities which can be used to build the bundle. While bundlecaps is |
|
913 | 920 | unused in core Mercurial, extensions rely on this feature to communicate |
|
914 | 921 | capabilities to customize the changegroup packer. |
|
915 | 922 | |
|
916 | 923 | shallow indicates whether shallow data might be sent. The packer may |
|
917 | 924 | need to pack file contents not introduced by the changes being packed. |
|
918 | 925 | |
|
919 | 926 | fullnodes is the set of changelog nodes which should not be ellipsis |
|
920 | 927 | nodes. We store this rather than the set of nodes that should be |
|
921 | 928 | ellipsis because for very large histories we expect this to be |
|
922 | 929 | significantly smaller. |
|
923 | 930 | |
|
924 | 931 | remote_sidedata is the set of sidedata categories wanted by the remote. |
|
925 | 932 | """ |
|
926 | 933 | assert oldmatcher |
|
927 | 934 | assert matcher |
|
928 | 935 | self._oldmatcher = oldmatcher |
|
929 | 936 | self._matcher = matcher |
|
930 | 937 | |
|
931 | 938 | self.version = version |
|
932 | 939 | self._forcedeltaparentprev = forcedeltaparentprev |
|
933 | 940 | self._builddeltaheader = builddeltaheader |
|
934 | 941 | self._manifestsend = manifestsend |
|
935 | 942 | self._ellipses = ellipses |
|
936 | 943 | |
|
937 | 944 | # Set of capabilities we can use to build the bundle. |
|
938 | 945 | if bundlecaps is None: |
|
939 | 946 | bundlecaps = set() |
|
940 | 947 | self._bundlecaps = bundlecaps |
|
941 | 948 | self._isshallow = shallow |
|
942 | 949 | self._fullclnodes = fullnodes |
|
943 | 950 | |
|
944 | 951 | # Maps ellipsis revs to their roots at the changelog level. |
|
945 | 952 | self._precomputedellipsis = ellipsisroots |
|
946 | 953 | |
|
947 | 954 | self._repo = repo |
|
948 | 955 | |
|
949 | 956 | if self._repo.ui.verbose and not self._repo.ui.debugflag: |
|
950 | 957 | self._verbosenote = self._repo.ui.note |
|
951 | 958 | else: |
|
952 | 959 | self._verbosenote = lambda s: None |
|
953 | 960 | |
|
954 | 961 | def generate( |
|
955 | 962 | self, commonrevs, clnodes, fastpathlinkrev, source, changelog=True |
|
956 | 963 | ): |
|
957 | 964 | """Yield a sequence of changegroup byte chunks. |
|
958 | 965 | If changelog is False, changelog data won't be added to changegroup |
|
959 | 966 | """ |
|
960 | 967 | |
|
961 | 968 | repo = self._repo |
|
962 | 969 | cl = repo.changelog |
|
963 | 970 | |
|
964 | 971 | self._verbosenote(_(b'uncompressed size of bundle content:\n')) |
|
965 | 972 | size = 0 |
|
966 | 973 | |
|
967 | 974 | clstate, deltas = self._generatechangelog( |
|
968 | 975 | cl, clnodes, generate=changelog |
|
969 | 976 | ) |
|
970 | 977 | for delta in deltas: |
|
971 | 978 | for chunk in _revisiondeltatochunks(delta, self._builddeltaheader): |
|
972 | 979 | size += len(chunk) |
|
973 | 980 | yield chunk |
|
974 | 981 | |
|
975 | 982 | close = closechunk() |
|
976 | 983 | size += len(close) |
|
977 | 984 | yield closechunk() |
|
978 | 985 | |
|
979 | 986 | self._verbosenote(_(b'%8.i (changelog)\n') % size) |
|
980 | 987 | |
|
981 | 988 | clrevorder = clstate[b'clrevorder'] |
|
982 | 989 | manifests = clstate[b'manifests'] |
|
983 | 990 | changedfiles = clstate[b'changedfiles'] |
|
984 | 991 | |
|
985 | 992 | # We need to make sure that the linkrev in the changegroup refers to |
|
986 | 993 | # the first changeset that introduced the manifest or file revision. |
|
987 | 994 | # The fastpath is usually safer than the slowpath, because the filelogs |
|
988 | 995 | # are walked in revlog order. |
|
989 | 996 | # |
|
990 | 997 | # When taking the slowpath when the manifest revlog uses generaldelta, |
|
991 | 998 | # the manifest may be walked in the "wrong" order. Without 'clrevorder', |
|
992 | 999 | # we would get an incorrect linkrev (see fix in cc0ff93d0c0c). |
|
993 | 1000 | # |
|
994 | 1001 | # When taking the fastpath, we are only vulnerable to reordering |
|
995 | 1002 | # of the changelog itself. The changelog never uses generaldelta and is |
|
996 | 1003 | # never reordered. To handle this case, we simply take the slowpath, |
|
997 | 1004 | # which already has the 'clrevorder' logic. This was also fixed in |
|
998 | 1005 | # cc0ff93d0c0c. |
|
999 | 1006 | |
|
1000 | 1007 | # Treemanifests don't work correctly with fastpathlinkrev |
|
1001 | 1008 | # either, because we don't discover which directory nodes to |
|
1002 | 1009 | # send along with files. This could probably be fixed. |
|
1003 | 1010 | fastpathlinkrev = fastpathlinkrev and not scmutil.istreemanifest(repo) |
|
1004 | 1011 | |
|
1005 | 1012 | fnodes = {} # needed file nodes |
|
1006 | 1013 | |
|
1007 | 1014 | size = 0 |
|
1008 | 1015 | it = self.generatemanifests( |
|
1009 | 1016 | commonrevs, |
|
1010 | 1017 | clrevorder, |
|
1011 | 1018 | fastpathlinkrev, |
|
1012 | 1019 | manifests, |
|
1013 | 1020 | fnodes, |
|
1014 | 1021 | source, |
|
1015 | 1022 | clstate[b'clrevtomanifestrev'], |
|
1016 | 1023 | ) |
|
1017 | 1024 | |
|
1018 | 1025 | for tree, deltas in it: |
|
1019 | 1026 | if tree: |
|
1020 | 1027 | assert self.version in (b'03', b'04') |
|
1021 | 1028 | chunk = _fileheader(tree) |
|
1022 | 1029 | size += len(chunk) |
|
1023 | 1030 | yield chunk |
|
1024 | 1031 | |
|
1025 | 1032 | for delta in deltas: |
|
1026 | 1033 | chunks = _revisiondeltatochunks(delta, self._builddeltaheader) |
|
1027 | 1034 | for chunk in chunks: |
|
1028 | 1035 | size += len(chunk) |
|
1029 | 1036 | yield chunk |
|
1030 | 1037 | |
|
1031 | 1038 | close = closechunk() |
|
1032 | 1039 | size += len(close) |
|
1033 | 1040 | yield close |
|
1034 | 1041 | |
|
1035 | 1042 | self._verbosenote(_(b'%8.i (manifests)\n') % size) |
|
1036 | 1043 | yield self._manifestsend |
|
1037 | 1044 | |
|
1038 | 1045 | mfdicts = None |
|
1039 | 1046 | if self._ellipses and self._isshallow: |
|
1040 | 1047 | mfdicts = [ |
|
1041 | 1048 | (repo.manifestlog[n].read(), lr) |
|
1042 | 1049 | for (n, lr) in pycompat.iteritems(manifests) |
|
1043 | 1050 | ] |
|
1044 | 1051 | |
|
1045 | 1052 | manifests.clear() |
|
1046 | 1053 | clrevs = {cl.rev(x) for x in clnodes} |
|
1047 | 1054 | |
|
1048 | 1055 | it = self.generatefiles( |
|
1049 | 1056 | changedfiles, |
|
1050 | 1057 | commonrevs, |
|
1051 | 1058 | source, |
|
1052 | 1059 | mfdicts, |
|
1053 | 1060 | fastpathlinkrev, |
|
1054 | 1061 | fnodes, |
|
1055 | 1062 | clrevs, |
|
1056 | 1063 | ) |
|
1057 | 1064 | |
|
1058 | 1065 | for path, deltas in it: |
|
1059 | 1066 | h = _fileheader(path) |
|
1060 | 1067 | size = len(h) |
|
1061 | 1068 | yield h |
|
1062 | 1069 | |
|
1063 | 1070 | for delta in deltas: |
|
1064 | 1071 | chunks = _revisiondeltatochunks(delta, self._builddeltaheader) |
|
1065 | 1072 | for chunk in chunks: |
|
1066 | 1073 | size += len(chunk) |
|
1067 | 1074 | yield chunk |
|
1068 | 1075 | |
|
1069 | 1076 | close = closechunk() |
|
1070 | 1077 | size += len(close) |
|
1071 | 1078 | yield close |
|
1072 | 1079 | |
|
1073 | 1080 | self._verbosenote(_(b'%8.i %s\n') % (size, path)) |
|
1074 | 1081 | |
|
1075 | 1082 | yield closechunk() |
|
1076 | 1083 | |
|
1077 | 1084 | if clnodes: |
|
1078 | 1085 | repo.hook(b'outgoing', node=hex(clnodes[0]), source=source) |
|
1079 | 1086 | |
|
1080 | 1087 | def _generatechangelog(self, cl, nodes, generate=True): |
|
1081 | 1088 | """Generate data for changelog chunks. |
|
1082 | 1089 | |
|
1083 | 1090 | Returns a 2-tuple of a dict containing state and an iterable of |
|
1084 | 1091 | byte chunks. The state will not be fully populated until the |
|
1085 | 1092 | chunk stream has been fully consumed. |
|
1086 | 1093 | |
|
1087 | 1094 | if generate is False, the state will be fully populated and no chunk |
|
1088 | 1095 | stream will be yielded |
|
1089 | 1096 | """ |
|
1090 | 1097 | clrevorder = {} |
|
1091 | 1098 | manifests = {} |
|
1092 | 1099 | mfl = self._repo.manifestlog |
|
1093 | 1100 | changedfiles = set() |
|
1094 | 1101 | clrevtomanifestrev = {} |
|
1095 | 1102 | |
|
1096 | 1103 | state = { |
|
1097 | 1104 | b'clrevorder': clrevorder, |
|
1098 | 1105 | b'manifests': manifests, |
|
1099 | 1106 | b'changedfiles': changedfiles, |
|
1100 | 1107 | b'clrevtomanifestrev': clrevtomanifestrev, |
|
1101 | 1108 | } |
|
1102 | 1109 | |
|
1103 | 1110 | if not (generate or self._ellipses): |
|
1104 | 1111 | # sort the nodes in storage order |
|
1105 | 1112 | nodes = sorted(nodes, key=cl.rev) |
|
1106 | 1113 | for node in nodes: |
|
1107 | 1114 | c = cl.changelogrevision(node) |
|
1108 | 1115 | clrevorder[node] = len(clrevorder) |
|
1109 | 1116 | # record the first changeset introducing this manifest version |
|
1110 | 1117 | manifests.setdefault(c.manifest, node) |
|
1111 | 1118 | # Record a complete list of potentially-changed files in |
|
1112 | 1119 | # this manifest. |
|
1113 | 1120 | changedfiles.update(c.files) |
|
1114 | 1121 | |
|
1115 | 1122 | return state, () |
|
1116 | 1123 | |
|
1117 | 1124 | # Callback for the changelog, used to collect changed files and |
|
1118 | 1125 | # manifest nodes. |
|
1119 | 1126 | # Returns the linkrev node (identity in the changelog case). |
|
1120 | 1127 | def lookupcl(x): |
|
1121 | 1128 | c = cl.changelogrevision(x) |
|
1122 | 1129 | clrevorder[x] = len(clrevorder) |
|
1123 | 1130 | |
|
1124 | 1131 | if self._ellipses: |
|
1125 | 1132 | # Only update manifests if x is going to be sent. Otherwise we |
|
1126 | 1133 | # end up with bogus linkrevs specified for manifests and |
|
1127 | 1134 | # we skip some manifest nodes that we should otherwise |
|
1128 | 1135 | # have sent. |
|
1129 | 1136 | if ( |
|
1130 | 1137 | x in self._fullclnodes |
|
1131 | 1138 | or cl.rev(x) in self._precomputedellipsis |
|
1132 | 1139 | ): |
|
1133 | 1140 | |
|
1134 | 1141 | manifestnode = c.manifest |
|
1135 | 1142 | # Record the first changeset introducing this manifest |
|
1136 | 1143 | # version. |
|
1137 | 1144 | manifests.setdefault(manifestnode, x) |
|
1138 | 1145 | # Set this narrow-specific dict so we have the lowest |
|
1139 | 1146 | # manifest revnum to look up for this cl revnum. (Part of |
|
1140 | 1147 | # mapping changelog ellipsis parents to manifest ellipsis |
|
1141 | 1148 | # parents) |
|
1142 | 1149 | clrevtomanifestrev.setdefault( |
|
1143 | 1150 | cl.rev(x), mfl.rev(manifestnode) |
|
1144 | 1151 | ) |
|
1145 | 1152 | # We can't trust the changed files list in the changeset if the |
|
1146 | 1153 | # client requested a shallow clone. |
|
1147 | 1154 | if self._isshallow: |
|
1148 | 1155 | changedfiles.update(mfl[c.manifest].read().keys()) |
|
1149 | 1156 | else: |
|
1150 | 1157 | changedfiles.update(c.files) |
|
1151 | 1158 | else: |
|
1152 | 1159 | # record the first changeset introducing this manifest version |
|
1153 | 1160 | manifests.setdefault(c.manifest, x) |
|
1154 | 1161 | # Record a complete list of potentially-changed files in |
|
1155 | 1162 | # this manifest. |
|
1156 | 1163 | changedfiles.update(c.files) |
|
1157 | 1164 | |
|
1158 | 1165 | return x |
|
1159 | 1166 | |
|
1160 | 1167 | gen = deltagroup( |
|
1161 | 1168 | self._repo, |
|
1162 | 1169 | cl, |
|
1163 | 1170 | nodes, |
|
1164 | 1171 | True, |
|
1165 | 1172 | lookupcl, |
|
1166 | 1173 | self._forcedeltaparentprev, |
|
1167 | 1174 | ellipses=self._ellipses, |
|
1168 | 1175 | topic=_(b'changesets'), |
|
1169 | 1176 | clrevtolocalrev={}, |
|
1170 | 1177 | fullclnodes=self._fullclnodes, |
|
1171 | 1178 | precomputedellipsis=self._precomputedellipsis, |
|
1172 | 1179 | ) |
|
1173 | 1180 | |
|
1174 | 1181 | return state, gen |
|
1175 | 1182 | |
|
1176 | 1183 | def generatemanifests( |
|
1177 | 1184 | self, |
|
1178 | 1185 | commonrevs, |
|
1179 | 1186 | clrevorder, |
|
1180 | 1187 | fastpathlinkrev, |
|
1181 | 1188 | manifests, |
|
1182 | 1189 | fnodes, |
|
1183 | 1190 | source, |
|
1184 | 1191 | clrevtolocalrev, |
|
1185 | 1192 | ): |
|
1186 | 1193 | """Returns an iterator of changegroup chunks containing manifests. |
|
1187 | 1194 | |
|
1188 | 1195 | `source` is unused here, but is used by extensions like remotefilelog to |
|
1189 | 1196 | change what is sent based in pulls vs pushes, etc. |
|
1190 | 1197 | """ |
|
1191 | 1198 | repo = self._repo |
|
1192 | 1199 | mfl = repo.manifestlog |
|
1193 | 1200 | tmfnodes = {b'': manifests} |
|
1194 | 1201 | |
|
1195 | 1202 | # Callback for the manifest, used to collect linkrevs for filelog |
|
1196 | 1203 | # revisions. |
|
1197 | 1204 | # Returns the linkrev node (collected in lookupcl). |
|
1198 | 1205 | def makelookupmflinknode(tree, nodes): |
|
1199 | 1206 | if fastpathlinkrev: |
|
1200 | 1207 | assert not tree |
|
1201 | 1208 | return ( |
|
1202 | 1209 | manifests.__getitem__ |
|
1203 | 1210 | ) # pytype: disable=unsupported-operands |
|
1204 | 1211 | |
|
1205 | 1212 | def lookupmflinknode(x): |
|
1206 | 1213 | """Callback for looking up the linknode for manifests. |
|
1207 | 1214 | |
|
1208 | 1215 | Returns the linkrev node for the specified manifest. |
|
1209 | 1216 | |
|
1210 | 1217 | SIDE EFFECT: |
|
1211 | 1218 | |
|
1212 | 1219 | 1) fclnodes gets populated with the list of relevant |
|
1213 | 1220 | file nodes if we're not using fastpathlinkrev |
|
1214 | 1221 | 2) When treemanifests are in use, collects treemanifest nodes |
|
1215 | 1222 | to send |
|
1216 | 1223 | |
|
1217 | 1224 | Note that this means manifests must be completely sent to |
|
1218 | 1225 | the client before you can trust the list of files and |
|
1219 | 1226 | treemanifests to send. |
|
1220 | 1227 | """ |
|
1221 | 1228 | clnode = nodes[x] |
|
1222 | 1229 | mdata = mfl.get(tree, x).readfast(shallow=True) |
|
1223 | 1230 | for p, n, fl in mdata.iterentries(): |
|
1224 | 1231 | if fl == b't': # subdirectory manifest |
|
1225 | 1232 | subtree = tree + p + b'/' |
|
1226 | 1233 | tmfclnodes = tmfnodes.setdefault(subtree, {}) |
|
1227 | 1234 | tmfclnode = tmfclnodes.setdefault(n, clnode) |
|
1228 | 1235 | if clrevorder[clnode] < clrevorder[tmfclnode]: |
|
1229 | 1236 | tmfclnodes[n] = clnode |
|
1230 | 1237 | else: |
|
1231 | 1238 | f = tree + p |
|
1232 | 1239 | fclnodes = fnodes.setdefault(f, {}) |
|
1233 | 1240 | fclnode = fclnodes.setdefault(n, clnode) |
|
1234 | 1241 | if clrevorder[clnode] < clrevorder[fclnode]: |
|
1235 | 1242 | fclnodes[n] = clnode |
|
1236 | 1243 | return clnode |
|
1237 | 1244 | |
|
1238 | 1245 | return lookupmflinknode |
|
1239 | 1246 | |
|
1240 | 1247 | while tmfnodes: |
|
1241 | 1248 | tree, nodes = tmfnodes.popitem() |
|
1242 | 1249 | |
|
1243 | 1250 | should_visit = self._matcher.visitdir(tree[:-1]) |
|
1244 | 1251 | if tree and not should_visit: |
|
1245 | 1252 | continue |
|
1246 | 1253 | |
|
1247 | 1254 | store = mfl.getstorage(tree) |
|
1248 | 1255 | |
|
1249 | 1256 | if not should_visit: |
|
1250 | 1257 | # No nodes to send because this directory is out of |
|
1251 | 1258 | # the client's view of the repository (probably |
|
1252 | 1259 | # because of narrow clones). Do this even for the root |
|
1253 | 1260 | # directory (tree=='') |
|
1254 | 1261 | prunednodes = [] |
|
1255 | 1262 | else: |
|
1256 | 1263 | # Avoid sending any manifest nodes we can prove the |
|
1257 | 1264 | # client already has by checking linkrevs. See the |
|
1258 | 1265 | # related comment in generatefiles(). |
|
1259 | 1266 | prunednodes = self._prunemanifests(store, nodes, commonrevs) |
|
1260 | 1267 | |
|
1261 | 1268 | if tree and not prunednodes: |
|
1262 | 1269 | continue |
|
1263 | 1270 | |
|
1264 | 1271 | lookupfn = makelookupmflinknode(tree, nodes) |
|
1265 | 1272 | |
|
1266 | 1273 | deltas = deltagroup( |
|
1267 | 1274 | self._repo, |
|
1268 | 1275 | store, |
|
1269 | 1276 | prunednodes, |
|
1270 | 1277 | False, |
|
1271 | 1278 | lookupfn, |
|
1272 | 1279 | self._forcedeltaparentprev, |
|
1273 | 1280 | ellipses=self._ellipses, |
|
1274 | 1281 | topic=_(b'manifests'), |
|
1275 | 1282 | clrevtolocalrev=clrevtolocalrev, |
|
1276 | 1283 | fullclnodes=self._fullclnodes, |
|
1277 | 1284 | precomputedellipsis=self._precomputedellipsis, |
|
1278 | 1285 | ) |
|
1279 | 1286 | |
|
1280 | 1287 | if not self._oldmatcher.visitdir(store.tree[:-1]): |
|
1281 | 1288 | yield tree, deltas |
|
1282 | 1289 | else: |
|
1283 | 1290 | # 'deltas' is a generator and we need to consume it even if |
|
1284 | 1291 | # we are not going to send it because a side-effect is that |
|
1285 | 1292 | # it updates tmdnodes (via lookupfn) |
|
1286 | 1293 | for d in deltas: |
|
1287 | 1294 | pass |
|
1288 | 1295 | if not tree: |
|
1289 | 1296 | yield tree, [] |
|
1290 | 1297 | |
|
1291 | 1298 | def _prunemanifests(self, store, nodes, commonrevs): |
|
1292 | 1299 | if not self._ellipses: |
|
1293 | 1300 | # In non-ellipses case and large repositories, it is better to |
|
1294 | 1301 | # prevent calling of store.rev and store.linkrev on a lot of |
|
1295 | 1302 | # nodes as compared to sending some extra data |
|
1296 | 1303 | return nodes.copy() |
|
1297 | 1304 | # This is split out as a separate method to allow filtering |
|
1298 | 1305 | # commonrevs in extension code. |
|
1299 | 1306 | # |
|
1300 | 1307 | # TODO(augie): this shouldn't be required, instead we should |
|
1301 | 1308 | # make filtering of revisions to send delegated to the store |
|
1302 | 1309 | # layer. |
|
1303 | 1310 | frev, flr = store.rev, store.linkrev |
|
1304 | 1311 | return [n for n in nodes if flr(frev(n)) not in commonrevs] |
|
1305 | 1312 | |
|
1306 | 1313 | # The 'source' parameter is useful for extensions |
|
1307 | 1314 | def generatefiles( |
|
1308 | 1315 | self, |
|
1309 | 1316 | changedfiles, |
|
1310 | 1317 | commonrevs, |
|
1311 | 1318 | source, |
|
1312 | 1319 | mfdicts, |
|
1313 | 1320 | fastpathlinkrev, |
|
1314 | 1321 | fnodes, |
|
1315 | 1322 | clrevs, |
|
1316 | 1323 | ): |
|
1317 | 1324 | changedfiles = [ |
|
1318 | 1325 | f |
|
1319 | 1326 | for f in changedfiles |
|
1320 | 1327 | if self._matcher(f) and not self._oldmatcher(f) |
|
1321 | 1328 | ] |
|
1322 | 1329 | |
|
1323 | 1330 | if not fastpathlinkrev: |
|
1324 | 1331 | |
|
1325 | 1332 | def normallinknodes(unused, fname): |
|
1326 | 1333 | return fnodes.get(fname, {}) |
|
1327 | 1334 | |
|
1328 | 1335 | else: |
|
1329 | 1336 | cln = self._repo.changelog.node |
|
1330 | 1337 | |
|
1331 | 1338 | def normallinknodes(store, fname): |
|
1332 | 1339 | flinkrev = store.linkrev |
|
1333 | 1340 | fnode = store.node |
|
1334 | 1341 | revs = ((r, flinkrev(r)) for r in store) |
|
1335 | 1342 | return {fnode(r): cln(lr) for r, lr in revs if lr in clrevs} |
|
1336 | 1343 | |
|
1337 | 1344 | clrevtolocalrev = {} |
|
1338 | 1345 | |
|
1339 | 1346 | if self._isshallow: |
|
1340 | 1347 | # In a shallow clone, the linknodes callback needs to also include |
|
1341 | 1348 | # those file nodes that are in the manifests we sent but weren't |
|
1342 | 1349 | # introduced by those manifests. |
|
1343 | 1350 | commonctxs = [self._repo[c] for c in commonrevs] |
|
1344 | 1351 | clrev = self._repo.changelog.rev |
|
1345 | 1352 | |
|
1346 | 1353 | def linknodes(flog, fname): |
|
1347 | 1354 | for c in commonctxs: |
|
1348 | 1355 | try: |
|
1349 | 1356 | fnode = c.filenode(fname) |
|
1350 | 1357 | clrevtolocalrev[c.rev()] = flog.rev(fnode) |
|
1351 | 1358 | except error.ManifestLookupError: |
|
1352 | 1359 | pass |
|
1353 | 1360 | links = normallinknodes(flog, fname) |
|
1354 | 1361 | if len(links) != len(mfdicts): |
|
1355 | 1362 | for mf, lr in mfdicts: |
|
1356 | 1363 | fnode = mf.get(fname, None) |
|
1357 | 1364 | if fnode in links: |
|
1358 | 1365 | links[fnode] = min(links[fnode], lr, key=clrev) |
|
1359 | 1366 | elif fnode: |
|
1360 | 1367 | links[fnode] = lr |
|
1361 | 1368 | return links |
|
1362 | 1369 | |
|
1363 | 1370 | else: |
|
1364 | 1371 | linknodes = normallinknodes |
|
1365 | 1372 | |
|
1366 | 1373 | repo = self._repo |
|
1367 | 1374 | progress = repo.ui.makeprogress( |
|
1368 | 1375 | _(b'files'), unit=_(b'files'), total=len(changedfiles) |
|
1369 | 1376 | ) |
|
1370 | 1377 | for i, fname in enumerate(sorted(changedfiles)): |
|
1371 | 1378 | filerevlog = repo.file(fname) |
|
1372 | 1379 | if not filerevlog: |
|
1373 | 1380 | raise error.Abort( |
|
1374 | 1381 | _(b"empty or missing file data for %s") % fname |
|
1375 | 1382 | ) |
|
1376 | 1383 | |
|
1377 | 1384 | clrevtolocalrev.clear() |
|
1378 | 1385 | |
|
1379 | 1386 | linkrevnodes = linknodes(filerevlog, fname) |
|
1380 | 1387 | # Lookup for filenodes, we collected the linkrev nodes above in the |
|
1381 | 1388 | # fastpath case and with lookupmf in the slowpath case. |
|
1382 | 1389 | def lookupfilelog(x): |
|
1383 | 1390 | return linkrevnodes[x] |
|
1384 | 1391 | |
|
1385 | 1392 | frev, flr = filerevlog.rev, filerevlog.linkrev |
|
1386 | 1393 | # Skip sending any filenode we know the client already |
|
1387 | 1394 | # has. This avoids over-sending files relatively |
|
1388 | 1395 | # inexpensively, so it's not a problem if we under-filter |
|
1389 | 1396 | # here. |
|
1390 | 1397 | filenodes = [ |
|
1391 | 1398 | n for n in linkrevnodes if flr(frev(n)) not in commonrevs |
|
1392 | 1399 | ] |
|
1393 | 1400 | |
|
1394 | 1401 | if not filenodes: |
|
1395 | 1402 | continue |
|
1396 | 1403 | |
|
1397 | 1404 | progress.update(i + 1, item=fname) |
|
1398 | 1405 | |
|
1399 | 1406 | deltas = deltagroup( |
|
1400 | 1407 | self._repo, |
|
1401 | 1408 | filerevlog, |
|
1402 | 1409 | filenodes, |
|
1403 | 1410 | False, |
|
1404 | 1411 | lookupfilelog, |
|
1405 | 1412 | self._forcedeltaparentprev, |
|
1406 | 1413 | ellipses=self._ellipses, |
|
1407 | 1414 | clrevtolocalrev=clrevtolocalrev, |
|
1408 | 1415 | fullclnodes=self._fullclnodes, |
|
1409 | 1416 | precomputedellipsis=self._precomputedellipsis, |
|
1410 | 1417 | ) |
|
1411 | 1418 | |
|
1412 | 1419 | yield fname, deltas |
|
1413 | 1420 | |
|
1414 | 1421 | progress.complete() |
|
1415 | 1422 | |
|
1416 | 1423 | |
|
1417 | 1424 | def _makecg1packer( |
|
1418 | 1425 | repo, |
|
1419 | 1426 | oldmatcher, |
|
1420 | 1427 | matcher, |
|
1421 | 1428 | bundlecaps, |
|
1422 | 1429 | ellipses=False, |
|
1423 | 1430 | shallow=False, |
|
1424 | 1431 | ellipsisroots=None, |
|
1425 | 1432 | fullnodes=None, |
|
1426 | 1433 | remote_sidedata=None, |
|
1427 | 1434 | ): |
|
1428 | 1435 | builddeltaheader = lambda d: _CHANGEGROUPV1_DELTA_HEADER.pack( |
|
1429 | 1436 | d.node, d.p1node, d.p2node, d.linknode |
|
1430 | 1437 | ) |
|
1431 | 1438 | |
|
1432 | 1439 | return cgpacker( |
|
1433 | 1440 | repo, |
|
1434 | 1441 | oldmatcher, |
|
1435 | 1442 | matcher, |
|
1436 | 1443 | b'01', |
|
1437 | 1444 | builddeltaheader=builddeltaheader, |
|
1438 | 1445 | manifestsend=b'', |
|
1439 | 1446 | forcedeltaparentprev=True, |
|
1440 | 1447 | bundlecaps=bundlecaps, |
|
1441 | 1448 | ellipses=ellipses, |
|
1442 | 1449 | shallow=shallow, |
|
1443 | 1450 | ellipsisroots=ellipsisroots, |
|
1444 | 1451 | fullnodes=fullnodes, |
|
1445 | 1452 | ) |
|
1446 | 1453 | |
|
1447 | 1454 | |
|
1448 | 1455 | def _makecg2packer( |
|
1449 | 1456 | repo, |
|
1450 | 1457 | oldmatcher, |
|
1451 | 1458 | matcher, |
|
1452 | 1459 | bundlecaps, |
|
1453 | 1460 | ellipses=False, |
|
1454 | 1461 | shallow=False, |
|
1455 | 1462 | ellipsisroots=None, |
|
1456 | 1463 | fullnodes=None, |
|
1457 | 1464 | remote_sidedata=None, |
|
1458 | 1465 | ): |
|
1459 | 1466 | builddeltaheader = lambda d: _CHANGEGROUPV2_DELTA_HEADER.pack( |
|
1460 | 1467 | d.node, d.p1node, d.p2node, d.basenode, d.linknode |
|
1461 | 1468 | ) |
|
1462 | 1469 | |
|
1463 | 1470 | return cgpacker( |
|
1464 | 1471 | repo, |
|
1465 | 1472 | oldmatcher, |
|
1466 | 1473 | matcher, |
|
1467 | 1474 | b'02', |
|
1468 | 1475 | builddeltaheader=builddeltaheader, |
|
1469 | 1476 | manifestsend=b'', |
|
1470 | 1477 | bundlecaps=bundlecaps, |
|
1471 | 1478 | ellipses=ellipses, |
|
1472 | 1479 | shallow=shallow, |
|
1473 | 1480 | ellipsisroots=ellipsisroots, |
|
1474 | 1481 | fullnodes=fullnodes, |
|
1475 | 1482 | ) |
|
1476 | 1483 | |
|
1477 | 1484 | |
|
1478 | 1485 | def _makecg3packer( |
|
1479 | 1486 | repo, |
|
1480 | 1487 | oldmatcher, |
|
1481 | 1488 | matcher, |
|
1482 | 1489 | bundlecaps, |
|
1483 | 1490 | ellipses=False, |
|
1484 | 1491 | shallow=False, |
|
1485 | 1492 | ellipsisroots=None, |
|
1486 | 1493 | fullnodes=None, |
|
1487 | 1494 | remote_sidedata=None, |
|
1488 | 1495 | ): |
|
1489 | 1496 | builddeltaheader = lambda d: _CHANGEGROUPV3_DELTA_HEADER.pack( |
|
1490 | 1497 | d.node, d.p1node, d.p2node, d.basenode, d.linknode, d.flags |
|
1491 | 1498 | ) |
|
1492 | 1499 | |
|
1493 | 1500 | return cgpacker( |
|
1494 | 1501 | repo, |
|
1495 | 1502 | oldmatcher, |
|
1496 | 1503 | matcher, |
|
1497 | 1504 | b'03', |
|
1498 | 1505 | builddeltaheader=builddeltaheader, |
|
1499 | 1506 | manifestsend=closechunk(), |
|
1500 | 1507 | bundlecaps=bundlecaps, |
|
1501 | 1508 | ellipses=ellipses, |
|
1502 | 1509 | shallow=shallow, |
|
1503 | 1510 | ellipsisroots=ellipsisroots, |
|
1504 | 1511 | fullnodes=fullnodes, |
|
1505 | 1512 | ) |
|
1506 | 1513 | |
|
1507 | 1514 | |
|
1508 | 1515 | def _makecg4packer( |
|
1509 | 1516 | repo, |
|
1510 | 1517 | oldmatcher, |
|
1511 | 1518 | matcher, |
|
1512 | 1519 | bundlecaps, |
|
1513 | 1520 | ellipses=False, |
|
1514 | 1521 | shallow=False, |
|
1515 | 1522 | ellipsisroots=None, |
|
1516 | 1523 | fullnodes=None, |
|
1517 | 1524 | remote_sidedata=None, |
|
1518 | 1525 | ): |
|
1519 | 1526 | # Same header func as cg3. Sidedata is in a separate chunk from the delta to |
|
1520 | 1527 | # differenciate "raw delta" and sidedata. |
|
1521 | 1528 | builddeltaheader = lambda d: _CHANGEGROUPV3_DELTA_HEADER.pack( |
|
1522 | 1529 | d.node, d.p1node, d.p2node, d.basenode, d.linknode, d.flags |
|
1523 | 1530 | ) |
|
1524 | 1531 | |
|
1525 | 1532 | return cgpacker( |
|
1526 | 1533 | repo, |
|
1527 | 1534 | oldmatcher, |
|
1528 | 1535 | matcher, |
|
1529 | 1536 | b'04', |
|
1530 | 1537 | builddeltaheader=builddeltaheader, |
|
1531 | 1538 | manifestsend=closechunk(), |
|
1532 | 1539 | bundlecaps=bundlecaps, |
|
1533 | 1540 | ellipses=ellipses, |
|
1534 | 1541 | shallow=shallow, |
|
1535 | 1542 | ellipsisroots=ellipsisroots, |
|
1536 | 1543 | fullnodes=fullnodes, |
|
1537 | 1544 | remote_sidedata=remote_sidedata, |
|
1538 | 1545 | ) |
|
1539 | 1546 | |
|
1540 | 1547 | |
|
1541 | 1548 | _packermap = { |
|
1542 | 1549 | b'01': (_makecg1packer, cg1unpacker), |
|
1543 | 1550 | # cg2 adds support for exchanging generaldelta |
|
1544 | 1551 | b'02': (_makecg2packer, cg2unpacker), |
|
1545 | 1552 | # cg3 adds support for exchanging revlog flags and treemanifests |
|
1546 | 1553 | b'03': (_makecg3packer, cg3unpacker), |
|
1547 | 1554 | # ch4 adds support for exchanging sidedata |
|
1548 | 1555 | b'04': (_makecg4packer, cg4unpacker), |
|
1549 | 1556 | } |
|
1550 | 1557 | |
|
1551 | 1558 | |
|
1552 | 1559 | def allsupportedversions(repo): |
|
1553 | 1560 | versions = set(_packermap.keys()) |
|
1554 | 1561 | needv03 = False |
|
1555 | 1562 | if ( |
|
1556 | 1563 | repo.ui.configbool(b'experimental', b'changegroup3') |
|
1557 | 1564 | or repo.ui.configbool(b'experimental', b'treemanifest') |
|
1558 | 1565 | or scmutil.istreemanifest(repo) |
|
1559 | 1566 | ): |
|
1560 | 1567 | # we keep version 03 because we need to to exchange treemanifest data |
|
1561 | 1568 | # |
|
1562 | 1569 | # we also keep vresion 01 and 02, because it is possible for repo to |
|
1563 | 1570 | # contains both normal and tree manifest at the same time. so using |
|
1564 | 1571 | # older version to pull data is viable |
|
1565 | 1572 | # |
|
1566 | 1573 | # (or even to push subset of history) |
|
1567 | 1574 | needv03 = True |
|
1568 | 1575 | has_revlogv2 = requirements.REVLOGV2_REQUIREMENT in repo.requirements |
|
1569 | 1576 | if not has_revlogv2: |
|
1570 | 1577 | versions.discard(b'04') |
|
1571 | 1578 | if not needv03: |
|
1572 | 1579 | versions.discard(b'03') |
|
1573 | 1580 | return versions |
|
1574 | 1581 | |
|
1575 | 1582 | |
|
1576 | 1583 | # Changegroup versions that can be applied to the repo |
|
1577 | 1584 | def supportedincomingversions(repo): |
|
1578 | 1585 | return allsupportedversions(repo) |
|
1579 | 1586 | |
|
1580 | 1587 | |
|
1581 | 1588 | # Changegroup versions that can be created from the repo |
|
1582 | 1589 | def supportedoutgoingversions(repo): |
|
1583 | 1590 | versions = allsupportedversions(repo) |
|
1584 | 1591 | if scmutil.istreemanifest(repo): |
|
1585 | 1592 | # Versions 01 and 02 support only flat manifests and it's just too |
|
1586 | 1593 | # expensive to convert between the flat manifest and tree manifest on |
|
1587 | 1594 | # the fly. Since tree manifests are hashed differently, all of history |
|
1588 | 1595 | # would have to be converted. Instead, we simply don't even pretend to |
|
1589 | 1596 | # support versions 01 and 02. |
|
1590 | 1597 | versions.discard(b'01') |
|
1591 | 1598 | versions.discard(b'02') |
|
1592 | 1599 | if requirements.NARROW_REQUIREMENT in repo.requirements: |
|
1593 | 1600 | # Versions 01 and 02 don't support revlog flags, and we need to |
|
1594 | 1601 | # support that for stripping and unbundling to work. |
|
1595 | 1602 | versions.discard(b'01') |
|
1596 | 1603 | versions.discard(b'02') |
|
1597 | 1604 | if LFS_REQUIREMENT in repo.requirements: |
|
1598 | 1605 | # Versions 01 and 02 don't support revlog flags, and we need to |
|
1599 | 1606 | # mark LFS entries with REVIDX_EXTSTORED. |
|
1600 | 1607 | versions.discard(b'01') |
|
1601 | 1608 | versions.discard(b'02') |
|
1602 | 1609 | |
|
1603 | 1610 | return versions |
|
1604 | 1611 | |
|
1605 | 1612 | |
|
1606 | 1613 | def localversion(repo): |
|
1607 | 1614 | # Finds the best version to use for bundles that are meant to be used |
|
1608 | 1615 | # locally, such as those from strip and shelve, and temporary bundles. |
|
1609 | 1616 | return max(supportedoutgoingversions(repo)) |
|
1610 | 1617 | |
|
1611 | 1618 | |
|
1612 | 1619 | def safeversion(repo): |
|
1613 | 1620 | # Finds the smallest version that it's safe to assume clients of the repo |
|
1614 | 1621 | # will support. For example, all hg versions that support generaldelta also |
|
1615 | 1622 | # support changegroup 02. |
|
1616 | 1623 | versions = supportedoutgoingversions(repo) |
|
1617 | 1624 | if requirements.GENERALDELTA_REQUIREMENT in repo.requirements: |
|
1618 | 1625 | versions.discard(b'01') |
|
1619 | 1626 | assert versions |
|
1620 | 1627 | return min(versions) |
|
1621 | 1628 | |
|
1622 | 1629 | |
|
1623 | 1630 | def getbundler( |
|
1624 | 1631 | version, |
|
1625 | 1632 | repo, |
|
1626 | 1633 | bundlecaps=None, |
|
1627 | 1634 | oldmatcher=None, |
|
1628 | 1635 | matcher=None, |
|
1629 | 1636 | ellipses=False, |
|
1630 | 1637 | shallow=False, |
|
1631 | 1638 | ellipsisroots=None, |
|
1632 | 1639 | fullnodes=None, |
|
1633 | 1640 | remote_sidedata=None, |
|
1634 | 1641 | ): |
|
1635 | 1642 | assert version in supportedoutgoingversions(repo) |
|
1636 | 1643 | |
|
1637 | 1644 | if matcher is None: |
|
1638 | 1645 | matcher = matchmod.always() |
|
1639 | 1646 | if oldmatcher is None: |
|
1640 | 1647 | oldmatcher = matchmod.never() |
|
1641 | 1648 | |
|
1642 | 1649 | if version == b'01' and not matcher.always(): |
|
1643 | 1650 | raise error.ProgrammingError( |
|
1644 | 1651 | b'version 01 changegroups do not support sparse file matchers' |
|
1645 | 1652 | ) |
|
1646 | 1653 | |
|
1647 | 1654 | if ellipses and version in (b'01', b'02'): |
|
1648 | 1655 | raise error.Abort( |
|
1649 | 1656 | _( |
|
1650 | 1657 | b'ellipsis nodes require at least cg3 on client and server, ' |
|
1651 | 1658 | b'but negotiated version %s' |
|
1652 | 1659 | ) |
|
1653 | 1660 | % version |
|
1654 | 1661 | ) |
|
1655 | 1662 | |
|
1656 | 1663 | # Requested files could include files not in the local store. So |
|
1657 | 1664 | # filter those out. |
|
1658 | 1665 | matcher = repo.narrowmatch(matcher) |
|
1659 | 1666 | |
|
1660 | 1667 | fn = _packermap[version][0] |
|
1661 | 1668 | return fn( |
|
1662 | 1669 | repo, |
|
1663 | 1670 | oldmatcher, |
|
1664 | 1671 | matcher, |
|
1665 | 1672 | bundlecaps, |
|
1666 | 1673 | ellipses=ellipses, |
|
1667 | 1674 | shallow=shallow, |
|
1668 | 1675 | ellipsisroots=ellipsisroots, |
|
1669 | 1676 | fullnodes=fullnodes, |
|
1670 | 1677 | remote_sidedata=remote_sidedata, |
|
1671 | 1678 | ) |
|
1672 | 1679 | |
|
1673 | 1680 | |
|
1674 | 1681 | def getunbundler(version, fh, alg, extras=None): |
|
1675 | 1682 | return _packermap[version][1](fh, alg, extras=extras) |
|
1676 | 1683 | |
|
1677 | 1684 | |
|
1678 | 1685 | def _changegroupinfo(repo, nodes, source): |
|
1679 | 1686 | if repo.ui.verbose or source == b'bundle': |
|
1680 | 1687 | repo.ui.status(_(b"%d changesets found\n") % len(nodes)) |
|
1681 | 1688 | if repo.ui.debugflag: |
|
1682 | 1689 | repo.ui.debug(b"list of changesets:\n") |
|
1683 | 1690 | for node in nodes: |
|
1684 | 1691 | repo.ui.debug(b"%s\n" % hex(node)) |
|
1685 | 1692 | |
|
1686 | 1693 | |
|
1687 | 1694 | def makechangegroup( |
|
1688 | 1695 | repo, outgoing, version, source, fastpath=False, bundlecaps=None |
|
1689 | 1696 | ): |
|
1690 | 1697 | cgstream = makestream( |
|
1691 | 1698 | repo, |
|
1692 | 1699 | outgoing, |
|
1693 | 1700 | version, |
|
1694 | 1701 | source, |
|
1695 | 1702 | fastpath=fastpath, |
|
1696 | 1703 | bundlecaps=bundlecaps, |
|
1697 | 1704 | ) |
|
1698 | 1705 | return getunbundler( |
|
1699 | 1706 | version, |
|
1700 | 1707 | util.chunkbuffer(cgstream), |
|
1701 | 1708 | None, |
|
1702 | 1709 | {b'clcount': len(outgoing.missing)}, |
|
1703 | 1710 | ) |
|
1704 | 1711 | |
|
1705 | 1712 | |
|
1706 | 1713 | def makestream( |
|
1707 | 1714 | repo, |
|
1708 | 1715 | outgoing, |
|
1709 | 1716 | version, |
|
1710 | 1717 | source, |
|
1711 | 1718 | fastpath=False, |
|
1712 | 1719 | bundlecaps=None, |
|
1713 | 1720 | matcher=None, |
|
1714 | 1721 | remote_sidedata=None, |
|
1715 | 1722 | ): |
|
1716 | 1723 | bundler = getbundler( |
|
1717 | 1724 | version, |
|
1718 | 1725 | repo, |
|
1719 | 1726 | bundlecaps=bundlecaps, |
|
1720 | 1727 | matcher=matcher, |
|
1721 | 1728 | remote_sidedata=remote_sidedata, |
|
1722 | 1729 | ) |
|
1723 | 1730 | |
|
1724 | 1731 | repo = repo.unfiltered() |
|
1725 | 1732 | commonrevs = outgoing.common |
|
1726 | 1733 | csets = outgoing.missing |
|
1727 | 1734 | heads = outgoing.ancestorsof |
|
1728 | 1735 | # We go through the fast path if we get told to, or if all (unfiltered |
|
1729 | 1736 | # heads have been requested (since we then know there all linkrevs will |
|
1730 | 1737 | # be pulled by the client). |
|
1731 | 1738 | heads.sort() |
|
1732 | 1739 | fastpathlinkrev = fastpath or ( |
|
1733 | 1740 | repo.filtername is None and heads == sorted(repo.heads()) |
|
1734 | 1741 | ) |
|
1735 | 1742 | |
|
1736 | 1743 | repo.hook(b'preoutgoing', throw=True, source=source) |
|
1737 | 1744 | _changegroupinfo(repo, csets, source) |
|
1738 | 1745 | return bundler.generate(commonrevs, csets, fastpathlinkrev, source) |
|
1739 | 1746 | |
|
1740 | 1747 | |
|
1741 | 1748 | def _addchangegroupfiles(repo, source, revmap, trp, expectedfiles, needfiles): |
|
1742 | 1749 | revisions = 0 |
|
1743 | 1750 | files = 0 |
|
1744 | 1751 | progress = repo.ui.makeprogress( |
|
1745 | 1752 | _(b'files'), unit=_(b'files'), total=expectedfiles |
|
1746 | 1753 | ) |
|
1747 | 1754 | for chunkdata in iter(source.filelogheader, {}): |
|
1748 | 1755 | files += 1 |
|
1749 | 1756 | f = chunkdata[b"filename"] |
|
1750 | 1757 | repo.ui.debug(b"adding %s revisions\n" % f) |
|
1751 | 1758 | progress.increment() |
|
1752 | 1759 | fl = repo.file(f) |
|
1753 | 1760 | o = len(fl) |
|
1754 | 1761 | try: |
|
1755 | 1762 | deltas = source.deltaiter() |
|
1756 | 1763 | if not fl.addgroup(deltas, revmap, trp): |
|
1757 | 1764 | raise error.Abort(_(b"received file revlog group is empty")) |
|
1758 | 1765 | except error.CensoredBaseError as e: |
|
1759 | 1766 | raise error.Abort(_(b"received delta base is censored: %s") % e) |
|
1760 | 1767 | revisions += len(fl) - o |
|
1761 | 1768 | if f in needfiles: |
|
1762 | 1769 | needs = needfiles[f] |
|
1763 | 1770 | for new in pycompat.xrange(o, len(fl)): |
|
1764 | 1771 | n = fl.node(new) |
|
1765 | 1772 | if n in needs: |
|
1766 | 1773 | needs.remove(n) |
|
1767 | 1774 | else: |
|
1768 | 1775 | raise error.Abort(_(b"received spurious file revlog entry")) |
|
1769 | 1776 | if not needs: |
|
1770 | 1777 | del needfiles[f] |
|
1771 | 1778 | progress.complete() |
|
1772 | 1779 | |
|
1773 | 1780 | for f, needs in pycompat.iteritems(needfiles): |
|
1774 | 1781 | fl = repo.file(f) |
|
1775 | 1782 | for n in needs: |
|
1776 | 1783 | try: |
|
1777 | 1784 | fl.rev(n) |
|
1778 | 1785 | except error.LookupError: |
|
1779 | 1786 | raise error.Abort( |
|
1780 | 1787 | _(b'missing file data for %s:%s - run hg verify') |
|
1781 | 1788 | % (f, hex(n)) |
|
1782 | 1789 | ) |
|
1783 | 1790 | |
|
1784 | 1791 | return revisions, files |
@@ -1,1994 +1,1998 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 | sidedata = interfaceutil.Attribute( | |
|
457 | """Raw sidedata bytes for the given revision.""" | |
|
458 | ) | |
|
459 | ||
|
456 | 460 | |
|
457 | 461 | class ifilerevisionssequence(interfaceutil.Interface): |
|
458 | 462 | """Contains index data for all revisions of a file. |
|
459 | 463 | |
|
460 | 464 | Types implementing this behave like lists of tuples. The index |
|
461 | 465 | in the list corresponds to the revision number. The values contain |
|
462 | 466 | index metadata. |
|
463 | 467 | |
|
464 | 468 | The *null* revision (revision number -1) is always the last item |
|
465 | 469 | in the index. |
|
466 | 470 | """ |
|
467 | 471 | |
|
468 | 472 | def __len__(): |
|
469 | 473 | """The total number of revisions.""" |
|
470 | 474 | |
|
471 | 475 | def __getitem__(rev): |
|
472 | 476 | """Returns the object having a specific revision number. |
|
473 | 477 | |
|
474 | 478 | Returns an 8-tuple with the following fields: |
|
475 | 479 | |
|
476 | 480 | offset+flags |
|
477 | 481 | Contains the offset and flags for the revision. 64-bit unsigned |
|
478 | 482 | integer where first 6 bytes are the offset and the next 2 bytes |
|
479 | 483 | are flags. The offset can be 0 if it is not used by the store. |
|
480 | 484 | compressed size |
|
481 | 485 | Size of the revision data in the store. It can be 0 if it isn't |
|
482 | 486 | needed by the store. |
|
483 | 487 | uncompressed size |
|
484 | 488 | Fulltext size. It can be 0 if it isn't needed by the store. |
|
485 | 489 | base revision |
|
486 | 490 | Revision number of revision the delta for storage is encoded |
|
487 | 491 | against. -1 indicates not encoded against a base revision. |
|
488 | 492 | link revision |
|
489 | 493 | Revision number of changelog revision this entry is related to. |
|
490 | 494 | p1 revision |
|
491 | 495 | Revision number of 1st parent. -1 if no 1st parent. |
|
492 | 496 | p2 revision |
|
493 | 497 | Revision number of 2nd parent. -1 if no 1st parent. |
|
494 | 498 | node |
|
495 | 499 | Binary node value for this revision number. |
|
496 | 500 | |
|
497 | 501 | Negative values should index off the end of the sequence. ``-1`` |
|
498 | 502 | should return the null revision. ``-2`` should return the most |
|
499 | 503 | recent revision. |
|
500 | 504 | """ |
|
501 | 505 | |
|
502 | 506 | def __contains__(rev): |
|
503 | 507 | """Whether a revision number exists.""" |
|
504 | 508 | |
|
505 | 509 | def insert(self, i, entry): |
|
506 | 510 | """Add an item to the index at specific revision.""" |
|
507 | 511 | |
|
508 | 512 | |
|
509 | 513 | class ifileindex(interfaceutil.Interface): |
|
510 | 514 | """Storage interface for index data of a single file. |
|
511 | 515 | |
|
512 | 516 | File storage data is divided into index metadata and data storage. |
|
513 | 517 | This interface defines the index portion of the interface. |
|
514 | 518 | |
|
515 | 519 | The index logically consists of: |
|
516 | 520 | |
|
517 | 521 | * A mapping between revision numbers and nodes. |
|
518 | 522 | * DAG data (storing and querying the relationship between nodes). |
|
519 | 523 | * Metadata to facilitate storage. |
|
520 | 524 | """ |
|
521 | 525 | |
|
522 | 526 | def __len__(): |
|
523 | 527 | """Obtain the number of revisions stored for this file.""" |
|
524 | 528 | |
|
525 | 529 | def __iter__(): |
|
526 | 530 | """Iterate over revision numbers for this file.""" |
|
527 | 531 | |
|
528 | 532 | def hasnode(node): |
|
529 | 533 | """Returns a bool indicating if a node is known to this store. |
|
530 | 534 | |
|
531 | 535 | Implementations must only return True for full, binary node values: |
|
532 | 536 | hex nodes, revision numbers, and partial node matches must be |
|
533 | 537 | rejected. |
|
534 | 538 | |
|
535 | 539 | The null node is never present. |
|
536 | 540 | """ |
|
537 | 541 | |
|
538 | 542 | def revs(start=0, stop=None): |
|
539 | 543 | """Iterate over revision numbers for this file, with control.""" |
|
540 | 544 | |
|
541 | 545 | def parents(node): |
|
542 | 546 | """Returns a 2-tuple of parent nodes for a revision. |
|
543 | 547 | |
|
544 | 548 | Values will be ``nullid`` if the parent is empty. |
|
545 | 549 | """ |
|
546 | 550 | |
|
547 | 551 | def parentrevs(rev): |
|
548 | 552 | """Like parents() but operates on revision numbers.""" |
|
549 | 553 | |
|
550 | 554 | def rev(node): |
|
551 | 555 | """Obtain the revision number given a node. |
|
552 | 556 | |
|
553 | 557 | Raises ``error.LookupError`` if the node is not known. |
|
554 | 558 | """ |
|
555 | 559 | |
|
556 | 560 | def node(rev): |
|
557 | 561 | """Obtain the node value given a revision number. |
|
558 | 562 | |
|
559 | 563 | Raises ``IndexError`` if the node is not known. |
|
560 | 564 | """ |
|
561 | 565 | |
|
562 | 566 | def lookup(node): |
|
563 | 567 | """Attempt to resolve a value to a node. |
|
564 | 568 | |
|
565 | 569 | Value can be a binary node, hex node, revision number, or a string |
|
566 | 570 | that can be converted to an integer. |
|
567 | 571 | |
|
568 | 572 | Raises ``error.LookupError`` if a node could not be resolved. |
|
569 | 573 | """ |
|
570 | 574 | |
|
571 | 575 | def linkrev(rev): |
|
572 | 576 | """Obtain the changeset revision number a revision is linked to.""" |
|
573 | 577 | |
|
574 | 578 | def iscensored(rev): |
|
575 | 579 | """Return whether a revision's content has been censored.""" |
|
576 | 580 | |
|
577 | 581 | def commonancestorsheads(node1, node2): |
|
578 | 582 | """Obtain an iterable of nodes containing heads of common ancestors. |
|
579 | 583 | |
|
580 | 584 | See ``ancestor.commonancestorsheads()``. |
|
581 | 585 | """ |
|
582 | 586 | |
|
583 | 587 | def descendants(revs): |
|
584 | 588 | """Obtain descendant revision numbers for a set of revision numbers. |
|
585 | 589 | |
|
586 | 590 | If ``nullrev`` is in the set, this is equivalent to ``revs()``. |
|
587 | 591 | """ |
|
588 | 592 | |
|
589 | 593 | def heads(start=None, stop=None): |
|
590 | 594 | """Obtain a list of nodes that are DAG heads, with control. |
|
591 | 595 | |
|
592 | 596 | The set of revisions examined can be limited by specifying |
|
593 | 597 | ``start`` and ``stop``. ``start`` is a node. ``stop`` is an |
|
594 | 598 | iterable of nodes. DAG traversal starts at earlier revision |
|
595 | 599 | ``start`` and iterates forward until any node in ``stop`` is |
|
596 | 600 | encountered. |
|
597 | 601 | """ |
|
598 | 602 | |
|
599 | 603 | def children(node): |
|
600 | 604 | """Obtain nodes that are children of a node. |
|
601 | 605 | |
|
602 | 606 | Returns a list of nodes. |
|
603 | 607 | """ |
|
604 | 608 | |
|
605 | 609 | |
|
606 | 610 | class ifiledata(interfaceutil.Interface): |
|
607 | 611 | """Storage interface for data storage of a specific file. |
|
608 | 612 | |
|
609 | 613 | This complements ``ifileindex`` and provides an interface for accessing |
|
610 | 614 | data for a tracked file. |
|
611 | 615 | """ |
|
612 | 616 | |
|
613 | 617 | def size(rev): |
|
614 | 618 | """Obtain the fulltext size of file data. |
|
615 | 619 | |
|
616 | 620 | Any metadata is excluded from size measurements. |
|
617 | 621 | """ |
|
618 | 622 | |
|
619 | 623 | def revision(node, raw=False): |
|
620 | 624 | """Obtain fulltext data for a node. |
|
621 | 625 | |
|
622 | 626 | By default, any storage transformations are applied before the data |
|
623 | 627 | is returned. If ``raw`` is True, non-raw storage transformations |
|
624 | 628 | are not applied. |
|
625 | 629 | |
|
626 | 630 | The fulltext data may contain a header containing metadata. Most |
|
627 | 631 | consumers should use ``read()`` to obtain the actual file data. |
|
628 | 632 | """ |
|
629 | 633 | |
|
630 | 634 | def rawdata(node): |
|
631 | 635 | """Obtain raw data for a node.""" |
|
632 | 636 | |
|
633 | 637 | def read(node): |
|
634 | 638 | """Resolve file fulltext data. |
|
635 | 639 | |
|
636 | 640 | This is similar to ``revision()`` except any metadata in the data |
|
637 | 641 | headers is stripped. |
|
638 | 642 | """ |
|
639 | 643 | |
|
640 | 644 | def renamed(node): |
|
641 | 645 | """Obtain copy metadata for a node. |
|
642 | 646 | |
|
643 | 647 | Returns ``False`` if no copy metadata is stored or a 2-tuple of |
|
644 | 648 | (path, node) from which this revision was copied. |
|
645 | 649 | """ |
|
646 | 650 | |
|
647 | 651 | def cmp(node, fulltext): |
|
648 | 652 | """Compare fulltext to another revision. |
|
649 | 653 | |
|
650 | 654 | Returns True if the fulltext is different from what is stored. |
|
651 | 655 | |
|
652 | 656 | This takes copy metadata into account. |
|
653 | 657 | |
|
654 | 658 | TODO better document the copy metadata and censoring logic. |
|
655 | 659 | """ |
|
656 | 660 | |
|
657 | 661 | def emitrevisions( |
|
658 | 662 | nodes, |
|
659 | 663 | nodesorder=None, |
|
660 | 664 | revisiondata=False, |
|
661 | 665 | assumehaveparentrevisions=False, |
|
662 | 666 | deltamode=CG_DELTAMODE_STD, |
|
663 | 667 | ): |
|
664 | 668 | """Produce ``irevisiondelta`` for revisions. |
|
665 | 669 | |
|
666 | 670 | Given an iterable of nodes, emits objects conforming to the |
|
667 | 671 | ``irevisiondelta`` interface that describe revisions in storage. |
|
668 | 672 | |
|
669 | 673 | This method is a generator. |
|
670 | 674 | |
|
671 | 675 | The input nodes may be unordered. Implementations must ensure that a |
|
672 | 676 | node's parents are emitted before the node itself. Transitively, this |
|
673 | 677 | means that a node may only be emitted once all its ancestors in |
|
674 | 678 | ``nodes`` have also been emitted. |
|
675 | 679 | |
|
676 | 680 | By default, emits "index" data (the ``node``, ``p1node``, and |
|
677 | 681 | ``p2node`` attributes). If ``revisiondata`` is set, revision data |
|
678 | 682 | will also be present on the emitted objects. |
|
679 | 683 | |
|
680 | 684 | With default argument values, implementations can choose to emit |
|
681 | 685 | either fulltext revision data or a delta. When emitting deltas, |
|
682 | 686 | implementations must consider whether the delta's base revision |
|
683 | 687 | fulltext is available to the receiver. |
|
684 | 688 | |
|
685 | 689 | The base revision fulltext is guaranteed to be available if any of |
|
686 | 690 | the following are met: |
|
687 | 691 | |
|
688 | 692 | * Its fulltext revision was emitted by this method call. |
|
689 | 693 | * A delta for that revision was emitted by this method call. |
|
690 | 694 | * ``assumehaveparentrevisions`` is True and the base revision is a |
|
691 | 695 | parent of the node. |
|
692 | 696 | |
|
693 | 697 | ``nodesorder`` can be used to control the order that revisions are |
|
694 | 698 | emitted. By default, revisions can be reordered as long as they are |
|
695 | 699 | in DAG topological order (see above). If the value is ``nodes``, |
|
696 | 700 | the iteration order from ``nodes`` should be used. If the value is |
|
697 | 701 | ``storage``, then the native order from the backing storage layer |
|
698 | 702 | is used. (Not all storage layers will have strong ordering and behavior |
|
699 | 703 | of this mode is storage-dependent.) ``nodes`` ordering can force |
|
700 | 704 | revisions to be emitted before their ancestors, so consumers should |
|
701 | 705 | use it with care. |
|
702 | 706 | |
|
703 | 707 | The ``linknode`` attribute on the returned ``irevisiondelta`` may not |
|
704 | 708 | be set and it is the caller's responsibility to resolve it, if needed. |
|
705 | 709 | |
|
706 | 710 | If ``deltamode`` is CG_DELTAMODE_PREV and revision data is requested, |
|
707 | 711 | all revision data should be emitted as deltas against the revision |
|
708 | 712 | emitted just prior. The initial revision should be a delta against its |
|
709 | 713 | 1st parent. |
|
710 | 714 | """ |
|
711 | 715 | |
|
712 | 716 | |
|
713 | 717 | class ifilemutation(interfaceutil.Interface): |
|
714 | 718 | """Storage interface for mutation events of a tracked file.""" |
|
715 | 719 | |
|
716 | 720 | def add(filedata, meta, transaction, linkrev, p1, p2): |
|
717 | 721 | """Add a new revision to the store. |
|
718 | 722 | |
|
719 | 723 | Takes file data, dictionary of metadata, a transaction, linkrev, |
|
720 | 724 | and parent nodes. |
|
721 | 725 | |
|
722 | 726 | Returns the node that was added. |
|
723 | 727 | |
|
724 | 728 | May no-op if a revision matching the supplied data is already stored. |
|
725 | 729 | """ |
|
726 | 730 | |
|
727 | 731 | def addrevision( |
|
728 | 732 | revisiondata, |
|
729 | 733 | transaction, |
|
730 | 734 | linkrev, |
|
731 | 735 | p1, |
|
732 | 736 | p2, |
|
733 | 737 | node=None, |
|
734 | 738 | flags=0, |
|
735 | 739 | cachedelta=None, |
|
736 | 740 | ): |
|
737 | 741 | """Add a new revision to the store and return its number. |
|
738 | 742 | |
|
739 | 743 | This is similar to ``add()`` except it operates at a lower level. |
|
740 | 744 | |
|
741 | 745 | The data passed in already contains a metadata header, if any. |
|
742 | 746 | |
|
743 | 747 | ``node`` and ``flags`` can be used to define the expected node and |
|
744 | 748 | the flags to use with storage. ``flags`` is a bitwise value composed |
|
745 | 749 | of the various ``REVISION_FLAG_*`` constants. |
|
746 | 750 | |
|
747 | 751 | ``add()`` is usually called when adding files from e.g. the working |
|
748 | 752 | directory. ``addrevision()`` is often called by ``add()`` and for |
|
749 | 753 | scenarios where revision data has already been computed, such as when |
|
750 | 754 | applying raw data from a peer repo. |
|
751 | 755 | """ |
|
752 | 756 | |
|
753 | 757 | def addgroup( |
|
754 | 758 | deltas, |
|
755 | 759 | linkmapper, |
|
756 | 760 | transaction, |
|
757 | 761 | addrevisioncb=None, |
|
758 | 762 | duplicaterevisioncb=None, |
|
759 | 763 | maybemissingparents=False, |
|
760 | 764 | ): |
|
761 | 765 | """Process a series of deltas for storage. |
|
762 | 766 | |
|
763 | 767 | ``deltas`` is an iterable of 7-tuples of |
|
764 | 768 | (node, p1, p2, linknode, deltabase, delta, flags) defining revisions |
|
765 | 769 | to add. |
|
766 | 770 | |
|
767 | 771 | The ``delta`` field contains ``mpatch`` data to apply to a base |
|
768 | 772 | revision, identified by ``deltabase``. The base node can be |
|
769 | 773 | ``nullid``, in which case the header from the delta can be ignored |
|
770 | 774 | and the delta used as the fulltext. |
|
771 | 775 | |
|
772 | 776 | ``alwayscache`` instructs the lower layers to cache the content of the |
|
773 | 777 | newly added revision, even if it needs to be explicitly computed. |
|
774 | 778 | This used to be the default when ``addrevisioncb`` was provided up to |
|
775 | 779 | Mercurial 5.8. |
|
776 | 780 | |
|
777 | 781 | ``addrevisioncb`` should be called for each new rev as it is committed. |
|
778 | 782 | ``duplicaterevisioncb`` should be called for all revs with a |
|
779 | 783 | pre-existing node. |
|
780 | 784 | |
|
781 | 785 | ``maybemissingparents`` is a bool indicating whether the incoming |
|
782 | 786 | data may reference parents/ancestor revisions that aren't present. |
|
783 | 787 | This flag is set when receiving data into a "shallow" store that |
|
784 | 788 | doesn't hold all history. |
|
785 | 789 | |
|
786 | 790 | Returns a list of nodes that were processed. A node will be in the list |
|
787 | 791 | even if it existed in the store previously. |
|
788 | 792 | """ |
|
789 | 793 | |
|
790 | 794 | def censorrevision(tr, node, tombstone=b''): |
|
791 | 795 | """Remove the content of a single revision. |
|
792 | 796 | |
|
793 | 797 | The specified ``node`` will have its content purged from storage. |
|
794 | 798 | Future attempts to access the revision data for this node will |
|
795 | 799 | result in failure. |
|
796 | 800 | |
|
797 | 801 | A ``tombstone`` message can optionally be stored. This message may be |
|
798 | 802 | displayed to users when they attempt to access the missing revision |
|
799 | 803 | data. |
|
800 | 804 | |
|
801 | 805 | Storage backends may have stored deltas against the previous content |
|
802 | 806 | in this revision. As part of censoring a revision, these storage |
|
803 | 807 | backends are expected to rewrite any internally stored deltas such |
|
804 | 808 | that they no longer reference the deleted content. |
|
805 | 809 | """ |
|
806 | 810 | |
|
807 | 811 | def getstrippoint(minlink): |
|
808 | 812 | """Find the minimum revision that must be stripped to strip a linkrev. |
|
809 | 813 | |
|
810 | 814 | Returns a 2-tuple containing the minimum revision number and a set |
|
811 | 815 | of all revisions numbers that would be broken by this strip. |
|
812 | 816 | |
|
813 | 817 | TODO this is highly revlog centric and should be abstracted into |
|
814 | 818 | a higher-level deletion API. ``repair.strip()`` relies on this. |
|
815 | 819 | """ |
|
816 | 820 | |
|
817 | 821 | def strip(minlink, transaction): |
|
818 | 822 | """Remove storage of items starting at a linkrev. |
|
819 | 823 | |
|
820 | 824 | This uses ``getstrippoint()`` to determine the first node to remove. |
|
821 | 825 | Then it effectively truncates storage for all revisions after that. |
|
822 | 826 | |
|
823 | 827 | TODO this is highly revlog centric and should be abstracted into a |
|
824 | 828 | higher-level deletion API. |
|
825 | 829 | """ |
|
826 | 830 | |
|
827 | 831 | |
|
828 | 832 | class ifilestorage(ifileindex, ifiledata, ifilemutation): |
|
829 | 833 | """Complete storage interface for a single tracked file.""" |
|
830 | 834 | |
|
831 | 835 | def files(): |
|
832 | 836 | """Obtain paths that are backing storage for this file. |
|
833 | 837 | |
|
834 | 838 | TODO this is used heavily by verify code and there should probably |
|
835 | 839 | be a better API for that. |
|
836 | 840 | """ |
|
837 | 841 | |
|
838 | 842 | def storageinfo( |
|
839 | 843 | exclusivefiles=False, |
|
840 | 844 | sharedfiles=False, |
|
841 | 845 | revisionscount=False, |
|
842 | 846 | trackedsize=False, |
|
843 | 847 | storedsize=False, |
|
844 | 848 | ): |
|
845 | 849 | """Obtain information about storage for this file's data. |
|
846 | 850 | |
|
847 | 851 | Returns a dict describing storage for this tracked path. The keys |
|
848 | 852 | in the dict map to arguments of the same. The arguments are bools |
|
849 | 853 | indicating whether to calculate and obtain that data. |
|
850 | 854 | |
|
851 | 855 | exclusivefiles |
|
852 | 856 | Iterable of (vfs, path) describing files that are exclusively |
|
853 | 857 | used to back storage for this tracked path. |
|
854 | 858 | |
|
855 | 859 | sharedfiles |
|
856 | 860 | Iterable of (vfs, path) describing files that are used to back |
|
857 | 861 | storage for this tracked path. Those files may also provide storage |
|
858 | 862 | for other stored entities. |
|
859 | 863 | |
|
860 | 864 | revisionscount |
|
861 | 865 | Number of revisions available for retrieval. |
|
862 | 866 | |
|
863 | 867 | trackedsize |
|
864 | 868 | Total size in bytes of all tracked revisions. This is a sum of the |
|
865 | 869 | length of the fulltext of all revisions. |
|
866 | 870 | |
|
867 | 871 | storedsize |
|
868 | 872 | Total size in bytes used to store data for all tracked revisions. |
|
869 | 873 | This is commonly less than ``trackedsize`` due to internal usage |
|
870 | 874 | of deltas rather than fulltext revisions. |
|
871 | 875 | |
|
872 | 876 | Not all storage backends may support all queries are have a reasonable |
|
873 | 877 | value to use. In that case, the value should be set to ``None`` and |
|
874 | 878 | callers are expected to handle this special value. |
|
875 | 879 | """ |
|
876 | 880 | |
|
877 | 881 | def verifyintegrity(state): |
|
878 | 882 | """Verifies the integrity of file storage. |
|
879 | 883 | |
|
880 | 884 | ``state`` is a dict holding state of the verifier process. It can be |
|
881 | 885 | used to communicate data between invocations of multiple storage |
|
882 | 886 | primitives. |
|
883 | 887 | |
|
884 | 888 | If individual revisions cannot have their revision content resolved, |
|
885 | 889 | the method is expected to set the ``skipread`` key to a set of nodes |
|
886 | 890 | that encountered problems. If set, the method can also add the node(s) |
|
887 | 891 | to ``safe_renamed`` in order to indicate nodes that may perform the |
|
888 | 892 | rename checks with currently accessible data. |
|
889 | 893 | |
|
890 | 894 | The method yields objects conforming to the ``iverifyproblem`` |
|
891 | 895 | interface. |
|
892 | 896 | """ |
|
893 | 897 | |
|
894 | 898 | |
|
895 | 899 | class idirs(interfaceutil.Interface): |
|
896 | 900 | """Interface representing a collection of directories from paths. |
|
897 | 901 | |
|
898 | 902 | This interface is essentially a derived data structure representing |
|
899 | 903 | directories from a collection of paths. |
|
900 | 904 | """ |
|
901 | 905 | |
|
902 | 906 | def addpath(path): |
|
903 | 907 | """Add a path to the collection. |
|
904 | 908 | |
|
905 | 909 | All directories in the path will be added to the collection. |
|
906 | 910 | """ |
|
907 | 911 | |
|
908 | 912 | def delpath(path): |
|
909 | 913 | """Remove a path from the collection. |
|
910 | 914 | |
|
911 | 915 | If the removal was the last path in a particular directory, the |
|
912 | 916 | directory is removed from the collection. |
|
913 | 917 | """ |
|
914 | 918 | |
|
915 | 919 | def __iter__(): |
|
916 | 920 | """Iterate over the directories in this collection of paths.""" |
|
917 | 921 | |
|
918 | 922 | def __contains__(path): |
|
919 | 923 | """Whether a specific directory is in this collection.""" |
|
920 | 924 | |
|
921 | 925 | |
|
922 | 926 | class imanifestdict(interfaceutil.Interface): |
|
923 | 927 | """Interface representing a manifest data structure. |
|
924 | 928 | |
|
925 | 929 | A manifest is effectively a dict mapping paths to entries. Each entry |
|
926 | 930 | consists of a binary node and extra flags affecting that entry. |
|
927 | 931 | """ |
|
928 | 932 | |
|
929 | 933 | def __getitem__(path): |
|
930 | 934 | """Returns the binary node value for a path in the manifest. |
|
931 | 935 | |
|
932 | 936 | Raises ``KeyError`` if the path does not exist in the manifest. |
|
933 | 937 | |
|
934 | 938 | Equivalent to ``self.find(path)[0]``. |
|
935 | 939 | """ |
|
936 | 940 | |
|
937 | 941 | def find(path): |
|
938 | 942 | """Returns the entry for a path in the manifest. |
|
939 | 943 | |
|
940 | 944 | Returns a 2-tuple of (node, flags). |
|
941 | 945 | |
|
942 | 946 | Raises ``KeyError`` if the path does not exist in the manifest. |
|
943 | 947 | """ |
|
944 | 948 | |
|
945 | 949 | def __len__(): |
|
946 | 950 | """Return the number of entries in the manifest.""" |
|
947 | 951 | |
|
948 | 952 | def __nonzero__(): |
|
949 | 953 | """Returns True if the manifest has entries, False otherwise.""" |
|
950 | 954 | |
|
951 | 955 | __bool__ = __nonzero__ |
|
952 | 956 | |
|
953 | 957 | def __setitem__(path, node): |
|
954 | 958 | """Define the node value for a path in the manifest. |
|
955 | 959 | |
|
956 | 960 | If the path is already in the manifest, its flags will be copied to |
|
957 | 961 | the new entry. |
|
958 | 962 | """ |
|
959 | 963 | |
|
960 | 964 | def __contains__(path): |
|
961 | 965 | """Whether a path exists in the manifest.""" |
|
962 | 966 | |
|
963 | 967 | def __delitem__(path): |
|
964 | 968 | """Remove a path from the manifest. |
|
965 | 969 | |
|
966 | 970 | Raises ``KeyError`` if the path is not in the manifest. |
|
967 | 971 | """ |
|
968 | 972 | |
|
969 | 973 | def __iter__(): |
|
970 | 974 | """Iterate over paths in the manifest.""" |
|
971 | 975 | |
|
972 | 976 | def iterkeys(): |
|
973 | 977 | """Iterate over paths in the manifest.""" |
|
974 | 978 | |
|
975 | 979 | def keys(): |
|
976 | 980 | """Obtain a list of paths in the manifest.""" |
|
977 | 981 | |
|
978 | 982 | def filesnotin(other, match=None): |
|
979 | 983 | """Obtain the set of paths in this manifest but not in another. |
|
980 | 984 | |
|
981 | 985 | ``match`` is an optional matcher function to be applied to both |
|
982 | 986 | manifests. |
|
983 | 987 | |
|
984 | 988 | Returns a set of paths. |
|
985 | 989 | """ |
|
986 | 990 | |
|
987 | 991 | def dirs(): |
|
988 | 992 | """Returns an object implementing the ``idirs`` interface.""" |
|
989 | 993 | |
|
990 | 994 | def hasdir(dir): |
|
991 | 995 | """Returns a bool indicating if a directory is in this manifest.""" |
|
992 | 996 | |
|
993 | 997 | def walk(match): |
|
994 | 998 | """Generator of paths in manifest satisfying a matcher. |
|
995 | 999 | |
|
996 | 1000 | If the matcher has explicit files listed and they don't exist in |
|
997 | 1001 | the manifest, ``match.bad()`` is called for each missing file. |
|
998 | 1002 | """ |
|
999 | 1003 | |
|
1000 | 1004 | def diff(other, match=None, clean=False): |
|
1001 | 1005 | """Find differences between this manifest and another. |
|
1002 | 1006 | |
|
1003 | 1007 | This manifest is compared to ``other``. |
|
1004 | 1008 | |
|
1005 | 1009 | If ``match`` is provided, the two manifests are filtered against this |
|
1006 | 1010 | matcher and only entries satisfying the matcher are compared. |
|
1007 | 1011 | |
|
1008 | 1012 | If ``clean`` is True, unchanged files are included in the returned |
|
1009 | 1013 | object. |
|
1010 | 1014 | |
|
1011 | 1015 | Returns a dict with paths as keys and values of 2-tuples of 2-tuples of |
|
1012 | 1016 | the form ``((node1, flag1), (node2, flag2))`` where ``(node1, flag1)`` |
|
1013 | 1017 | represents the node and flags for this manifest and ``(node2, flag2)`` |
|
1014 | 1018 | are the same for the other manifest. |
|
1015 | 1019 | """ |
|
1016 | 1020 | |
|
1017 | 1021 | def setflag(path, flag): |
|
1018 | 1022 | """Set the flag value for a given path. |
|
1019 | 1023 | |
|
1020 | 1024 | Raises ``KeyError`` if the path is not already in the manifest. |
|
1021 | 1025 | """ |
|
1022 | 1026 | |
|
1023 | 1027 | def get(path, default=None): |
|
1024 | 1028 | """Obtain the node value for a path or a default value if missing.""" |
|
1025 | 1029 | |
|
1026 | 1030 | def flags(path): |
|
1027 | 1031 | """Return the flags value for a path (default: empty bytestring).""" |
|
1028 | 1032 | |
|
1029 | 1033 | def copy(): |
|
1030 | 1034 | """Return a copy of this manifest.""" |
|
1031 | 1035 | |
|
1032 | 1036 | def items(): |
|
1033 | 1037 | """Returns an iterable of (path, node) for items in this manifest.""" |
|
1034 | 1038 | |
|
1035 | 1039 | def iteritems(): |
|
1036 | 1040 | """Identical to items().""" |
|
1037 | 1041 | |
|
1038 | 1042 | def iterentries(): |
|
1039 | 1043 | """Returns an iterable of (path, node, flags) for this manifest. |
|
1040 | 1044 | |
|
1041 | 1045 | Similar to ``iteritems()`` except items are a 3-tuple and include |
|
1042 | 1046 | flags. |
|
1043 | 1047 | """ |
|
1044 | 1048 | |
|
1045 | 1049 | def text(): |
|
1046 | 1050 | """Obtain the raw data representation for this manifest. |
|
1047 | 1051 | |
|
1048 | 1052 | Result is used to create a manifest revision. |
|
1049 | 1053 | """ |
|
1050 | 1054 | |
|
1051 | 1055 | def fastdelta(base, changes): |
|
1052 | 1056 | """Obtain a delta between this manifest and another given changes. |
|
1053 | 1057 | |
|
1054 | 1058 | ``base`` in the raw data representation for another manifest. |
|
1055 | 1059 | |
|
1056 | 1060 | ``changes`` is an iterable of ``(path, to_delete)``. |
|
1057 | 1061 | |
|
1058 | 1062 | Returns a 2-tuple containing ``bytearray(self.text())`` and the |
|
1059 | 1063 | delta between ``base`` and this manifest. |
|
1060 | 1064 | |
|
1061 | 1065 | If this manifest implementation can't support ``fastdelta()``, |
|
1062 | 1066 | raise ``mercurial.manifest.FastdeltaUnavailable``. |
|
1063 | 1067 | """ |
|
1064 | 1068 | |
|
1065 | 1069 | |
|
1066 | 1070 | class imanifestrevisionbase(interfaceutil.Interface): |
|
1067 | 1071 | """Base interface representing a single revision of a manifest. |
|
1068 | 1072 | |
|
1069 | 1073 | Should not be used as a primary interface: should always be inherited |
|
1070 | 1074 | as part of a larger interface. |
|
1071 | 1075 | """ |
|
1072 | 1076 | |
|
1073 | 1077 | def copy(): |
|
1074 | 1078 | """Obtain a copy of this manifest instance. |
|
1075 | 1079 | |
|
1076 | 1080 | Returns an object conforming to the ``imanifestrevisionwritable`` |
|
1077 | 1081 | interface. The instance will be associated with the same |
|
1078 | 1082 | ``imanifestlog`` collection as this instance. |
|
1079 | 1083 | """ |
|
1080 | 1084 | |
|
1081 | 1085 | def read(): |
|
1082 | 1086 | """Obtain the parsed manifest data structure. |
|
1083 | 1087 | |
|
1084 | 1088 | The returned object conforms to the ``imanifestdict`` interface. |
|
1085 | 1089 | """ |
|
1086 | 1090 | |
|
1087 | 1091 | |
|
1088 | 1092 | class imanifestrevisionstored(imanifestrevisionbase): |
|
1089 | 1093 | """Interface representing a manifest revision committed to storage.""" |
|
1090 | 1094 | |
|
1091 | 1095 | def node(): |
|
1092 | 1096 | """The binary node for this manifest.""" |
|
1093 | 1097 | |
|
1094 | 1098 | parents = interfaceutil.Attribute( |
|
1095 | 1099 | """List of binary nodes that are parents for this manifest revision.""" |
|
1096 | 1100 | ) |
|
1097 | 1101 | |
|
1098 | 1102 | def readdelta(shallow=False): |
|
1099 | 1103 | """Obtain the manifest data structure representing changes from parent. |
|
1100 | 1104 | |
|
1101 | 1105 | This manifest is compared to its 1st parent. A new manifest representing |
|
1102 | 1106 | those differences is constructed. |
|
1103 | 1107 | |
|
1104 | 1108 | The returned object conforms to the ``imanifestdict`` interface. |
|
1105 | 1109 | """ |
|
1106 | 1110 | |
|
1107 | 1111 | def readfast(shallow=False): |
|
1108 | 1112 | """Calls either ``read()`` or ``readdelta()``. |
|
1109 | 1113 | |
|
1110 | 1114 | The faster of the two options is called. |
|
1111 | 1115 | """ |
|
1112 | 1116 | |
|
1113 | 1117 | def find(key): |
|
1114 | 1118 | """Calls self.read().find(key)``. |
|
1115 | 1119 | |
|
1116 | 1120 | Returns a 2-tuple of ``(node, flags)`` or raises ``KeyError``. |
|
1117 | 1121 | """ |
|
1118 | 1122 | |
|
1119 | 1123 | |
|
1120 | 1124 | class imanifestrevisionwritable(imanifestrevisionbase): |
|
1121 | 1125 | """Interface representing a manifest revision that can be committed.""" |
|
1122 | 1126 | |
|
1123 | 1127 | def write(transaction, linkrev, p1node, p2node, added, removed, match=None): |
|
1124 | 1128 | """Add this revision to storage. |
|
1125 | 1129 | |
|
1126 | 1130 | Takes a transaction object, the changeset revision number it will |
|
1127 | 1131 | be associated with, its parent nodes, and lists of added and |
|
1128 | 1132 | removed paths. |
|
1129 | 1133 | |
|
1130 | 1134 | If match is provided, storage can choose not to inspect or write out |
|
1131 | 1135 | items that do not match. Storage is still required to be able to provide |
|
1132 | 1136 | the full manifest in the future for any directories written (these |
|
1133 | 1137 | manifests should not be "narrowed on disk"). |
|
1134 | 1138 | |
|
1135 | 1139 | Returns the binary node of the created revision. |
|
1136 | 1140 | """ |
|
1137 | 1141 | |
|
1138 | 1142 | |
|
1139 | 1143 | class imanifeststorage(interfaceutil.Interface): |
|
1140 | 1144 | """Storage interface for manifest data.""" |
|
1141 | 1145 | |
|
1142 | 1146 | tree = interfaceutil.Attribute( |
|
1143 | 1147 | """The path to the directory this manifest tracks. |
|
1144 | 1148 | |
|
1145 | 1149 | The empty bytestring represents the root manifest. |
|
1146 | 1150 | """ |
|
1147 | 1151 | ) |
|
1148 | 1152 | |
|
1149 | 1153 | index = interfaceutil.Attribute( |
|
1150 | 1154 | """An ``ifilerevisionssequence`` instance.""" |
|
1151 | 1155 | ) |
|
1152 | 1156 | |
|
1153 | 1157 | indexfile = interfaceutil.Attribute( |
|
1154 | 1158 | """Path of revlog index file. |
|
1155 | 1159 | |
|
1156 | 1160 | TODO this is revlog specific and should not be exposed. |
|
1157 | 1161 | """ |
|
1158 | 1162 | ) |
|
1159 | 1163 | |
|
1160 | 1164 | opener = interfaceutil.Attribute( |
|
1161 | 1165 | """VFS opener to use to access underlying files used for storage. |
|
1162 | 1166 | |
|
1163 | 1167 | TODO this is revlog specific and should not be exposed. |
|
1164 | 1168 | """ |
|
1165 | 1169 | ) |
|
1166 | 1170 | |
|
1167 | 1171 | version = interfaceutil.Attribute( |
|
1168 | 1172 | """Revlog version number. |
|
1169 | 1173 | |
|
1170 | 1174 | TODO this is revlog specific and should not be exposed. |
|
1171 | 1175 | """ |
|
1172 | 1176 | ) |
|
1173 | 1177 | |
|
1174 | 1178 | _generaldelta = interfaceutil.Attribute( |
|
1175 | 1179 | """Whether generaldelta storage is being used. |
|
1176 | 1180 | |
|
1177 | 1181 | TODO this is revlog specific and should not be exposed. |
|
1178 | 1182 | """ |
|
1179 | 1183 | ) |
|
1180 | 1184 | |
|
1181 | 1185 | fulltextcache = interfaceutil.Attribute( |
|
1182 | 1186 | """Dict with cache of fulltexts. |
|
1183 | 1187 | |
|
1184 | 1188 | TODO this doesn't feel appropriate for the storage interface. |
|
1185 | 1189 | """ |
|
1186 | 1190 | ) |
|
1187 | 1191 | |
|
1188 | 1192 | def __len__(): |
|
1189 | 1193 | """Obtain the number of revisions stored for this manifest.""" |
|
1190 | 1194 | |
|
1191 | 1195 | def __iter__(): |
|
1192 | 1196 | """Iterate over revision numbers for this manifest.""" |
|
1193 | 1197 | |
|
1194 | 1198 | def rev(node): |
|
1195 | 1199 | """Obtain the revision number given a binary node. |
|
1196 | 1200 | |
|
1197 | 1201 | Raises ``error.LookupError`` if the node is not known. |
|
1198 | 1202 | """ |
|
1199 | 1203 | |
|
1200 | 1204 | def node(rev): |
|
1201 | 1205 | """Obtain the node value given a revision number. |
|
1202 | 1206 | |
|
1203 | 1207 | Raises ``error.LookupError`` if the revision is not known. |
|
1204 | 1208 | """ |
|
1205 | 1209 | |
|
1206 | 1210 | def lookup(value): |
|
1207 | 1211 | """Attempt to resolve a value to a node. |
|
1208 | 1212 | |
|
1209 | 1213 | Value can be a binary node, hex node, revision number, or a bytes |
|
1210 | 1214 | that can be converted to an integer. |
|
1211 | 1215 | |
|
1212 | 1216 | Raises ``error.LookupError`` if a ndoe could not be resolved. |
|
1213 | 1217 | """ |
|
1214 | 1218 | |
|
1215 | 1219 | def parents(node): |
|
1216 | 1220 | """Returns a 2-tuple of parent nodes for a node. |
|
1217 | 1221 | |
|
1218 | 1222 | Values will be ``nullid`` if the parent is empty. |
|
1219 | 1223 | """ |
|
1220 | 1224 | |
|
1221 | 1225 | def parentrevs(rev): |
|
1222 | 1226 | """Like parents() but operates on revision numbers.""" |
|
1223 | 1227 | |
|
1224 | 1228 | def linkrev(rev): |
|
1225 | 1229 | """Obtain the changeset revision number a revision is linked to.""" |
|
1226 | 1230 | |
|
1227 | 1231 | def revision(node, _df=None, raw=False): |
|
1228 | 1232 | """Obtain fulltext data for a node.""" |
|
1229 | 1233 | |
|
1230 | 1234 | def rawdata(node, _df=None): |
|
1231 | 1235 | """Obtain raw data for a node.""" |
|
1232 | 1236 | |
|
1233 | 1237 | def revdiff(rev1, rev2): |
|
1234 | 1238 | """Obtain a delta between two revision numbers. |
|
1235 | 1239 | |
|
1236 | 1240 | The returned data is the result of ``bdiff.bdiff()`` on the raw |
|
1237 | 1241 | revision data. |
|
1238 | 1242 | """ |
|
1239 | 1243 | |
|
1240 | 1244 | def cmp(node, fulltext): |
|
1241 | 1245 | """Compare fulltext to another revision. |
|
1242 | 1246 | |
|
1243 | 1247 | Returns True if the fulltext is different from what is stored. |
|
1244 | 1248 | """ |
|
1245 | 1249 | |
|
1246 | 1250 | def emitrevisions( |
|
1247 | 1251 | nodes, |
|
1248 | 1252 | nodesorder=None, |
|
1249 | 1253 | revisiondata=False, |
|
1250 | 1254 | assumehaveparentrevisions=False, |
|
1251 | 1255 | ): |
|
1252 | 1256 | """Produce ``irevisiondelta`` describing revisions. |
|
1253 | 1257 | |
|
1254 | 1258 | See the documentation for ``ifiledata`` for more. |
|
1255 | 1259 | """ |
|
1256 | 1260 | |
|
1257 | 1261 | def addgroup( |
|
1258 | 1262 | deltas, |
|
1259 | 1263 | linkmapper, |
|
1260 | 1264 | transaction, |
|
1261 | 1265 | addrevisioncb=None, |
|
1262 | 1266 | duplicaterevisioncb=None, |
|
1263 | 1267 | ): |
|
1264 | 1268 | """Process a series of deltas for storage. |
|
1265 | 1269 | |
|
1266 | 1270 | See the documentation in ``ifilemutation`` for more. |
|
1267 | 1271 | """ |
|
1268 | 1272 | |
|
1269 | 1273 | def rawsize(rev): |
|
1270 | 1274 | """Obtain the size of tracked data. |
|
1271 | 1275 | |
|
1272 | 1276 | Is equivalent to ``len(m.rawdata(node))``. |
|
1273 | 1277 | |
|
1274 | 1278 | TODO this method is only used by upgrade code and may be removed. |
|
1275 | 1279 | """ |
|
1276 | 1280 | |
|
1277 | 1281 | def getstrippoint(minlink): |
|
1278 | 1282 | """Find minimum revision that must be stripped to strip a linkrev. |
|
1279 | 1283 | |
|
1280 | 1284 | See the documentation in ``ifilemutation`` for more. |
|
1281 | 1285 | """ |
|
1282 | 1286 | |
|
1283 | 1287 | def strip(minlink, transaction): |
|
1284 | 1288 | """Remove storage of items starting at a linkrev. |
|
1285 | 1289 | |
|
1286 | 1290 | See the documentation in ``ifilemutation`` for more. |
|
1287 | 1291 | """ |
|
1288 | 1292 | |
|
1289 | 1293 | def checksize(): |
|
1290 | 1294 | """Obtain the expected sizes of backing files. |
|
1291 | 1295 | |
|
1292 | 1296 | TODO this is used by verify and it should not be part of the interface. |
|
1293 | 1297 | """ |
|
1294 | 1298 | |
|
1295 | 1299 | def files(): |
|
1296 | 1300 | """Obtain paths that are backing storage for this manifest. |
|
1297 | 1301 | |
|
1298 | 1302 | TODO this is used by verify and there should probably be a better API |
|
1299 | 1303 | for this functionality. |
|
1300 | 1304 | """ |
|
1301 | 1305 | |
|
1302 | 1306 | def deltaparent(rev): |
|
1303 | 1307 | """Obtain the revision that a revision is delta'd against. |
|
1304 | 1308 | |
|
1305 | 1309 | TODO delta encoding is an implementation detail of storage and should |
|
1306 | 1310 | not be exposed to the storage interface. |
|
1307 | 1311 | """ |
|
1308 | 1312 | |
|
1309 | 1313 | def clone(tr, dest, **kwargs): |
|
1310 | 1314 | """Clone this instance to another.""" |
|
1311 | 1315 | |
|
1312 | 1316 | def clearcaches(clear_persisted_data=False): |
|
1313 | 1317 | """Clear any caches associated with this instance.""" |
|
1314 | 1318 | |
|
1315 | 1319 | def dirlog(d): |
|
1316 | 1320 | """Obtain a manifest storage instance for a tree.""" |
|
1317 | 1321 | |
|
1318 | 1322 | def add( |
|
1319 | 1323 | m, transaction, link, p1, p2, added, removed, readtree=None, match=None |
|
1320 | 1324 | ): |
|
1321 | 1325 | """Add a revision to storage. |
|
1322 | 1326 | |
|
1323 | 1327 | ``m`` is an object conforming to ``imanifestdict``. |
|
1324 | 1328 | |
|
1325 | 1329 | ``link`` is the linkrev revision number. |
|
1326 | 1330 | |
|
1327 | 1331 | ``p1`` and ``p2`` are the parent revision numbers. |
|
1328 | 1332 | |
|
1329 | 1333 | ``added`` and ``removed`` are iterables of added and removed paths, |
|
1330 | 1334 | respectively. |
|
1331 | 1335 | |
|
1332 | 1336 | ``readtree`` is a function that can be used to read the child tree(s) |
|
1333 | 1337 | when recursively writing the full tree structure when using |
|
1334 | 1338 | treemanifets. |
|
1335 | 1339 | |
|
1336 | 1340 | ``match`` is a matcher that can be used to hint to storage that not all |
|
1337 | 1341 | paths must be inspected; this is an optimization and can be safely |
|
1338 | 1342 | ignored. Note that the storage must still be able to reproduce a full |
|
1339 | 1343 | manifest including files that did not match. |
|
1340 | 1344 | """ |
|
1341 | 1345 | |
|
1342 | 1346 | def storageinfo( |
|
1343 | 1347 | exclusivefiles=False, |
|
1344 | 1348 | sharedfiles=False, |
|
1345 | 1349 | revisionscount=False, |
|
1346 | 1350 | trackedsize=False, |
|
1347 | 1351 | storedsize=False, |
|
1348 | 1352 | ): |
|
1349 | 1353 | """Obtain information about storage for this manifest's data. |
|
1350 | 1354 | |
|
1351 | 1355 | See ``ifilestorage.storageinfo()`` for a description of this method. |
|
1352 | 1356 | This one behaves the same way, except for manifest data. |
|
1353 | 1357 | """ |
|
1354 | 1358 | |
|
1355 | 1359 | |
|
1356 | 1360 | class imanifestlog(interfaceutil.Interface): |
|
1357 | 1361 | """Interface representing a collection of manifest snapshots. |
|
1358 | 1362 | |
|
1359 | 1363 | Represents the root manifest in a repository. |
|
1360 | 1364 | |
|
1361 | 1365 | Also serves as a means to access nested tree manifests and to cache |
|
1362 | 1366 | tree manifests. |
|
1363 | 1367 | """ |
|
1364 | 1368 | |
|
1365 | 1369 | def __getitem__(node): |
|
1366 | 1370 | """Obtain a manifest instance for a given binary node. |
|
1367 | 1371 | |
|
1368 | 1372 | Equivalent to calling ``self.get('', node)``. |
|
1369 | 1373 | |
|
1370 | 1374 | The returned object conforms to the ``imanifestrevisionstored`` |
|
1371 | 1375 | interface. |
|
1372 | 1376 | """ |
|
1373 | 1377 | |
|
1374 | 1378 | def get(tree, node, verify=True): |
|
1375 | 1379 | """Retrieve the manifest instance for a given directory and binary node. |
|
1376 | 1380 | |
|
1377 | 1381 | ``node`` always refers to the node of the root manifest (which will be |
|
1378 | 1382 | the only manifest if flat manifests are being used). |
|
1379 | 1383 | |
|
1380 | 1384 | If ``tree`` is the empty string, the root manifest is returned. |
|
1381 | 1385 | Otherwise the manifest for the specified directory will be returned |
|
1382 | 1386 | (requires tree manifests). |
|
1383 | 1387 | |
|
1384 | 1388 | If ``verify`` is True, ``LookupError`` is raised if the node is not |
|
1385 | 1389 | known. |
|
1386 | 1390 | |
|
1387 | 1391 | The returned object conforms to the ``imanifestrevisionstored`` |
|
1388 | 1392 | interface. |
|
1389 | 1393 | """ |
|
1390 | 1394 | |
|
1391 | 1395 | def getstorage(tree): |
|
1392 | 1396 | """Retrieve an interface to storage for a particular tree. |
|
1393 | 1397 | |
|
1394 | 1398 | If ``tree`` is the empty bytestring, storage for the root manifest will |
|
1395 | 1399 | be returned. Otherwise storage for a tree manifest is returned. |
|
1396 | 1400 | |
|
1397 | 1401 | TODO formalize interface for returned object. |
|
1398 | 1402 | """ |
|
1399 | 1403 | |
|
1400 | 1404 | def clearcaches(): |
|
1401 | 1405 | """Clear caches associated with this collection.""" |
|
1402 | 1406 | |
|
1403 | 1407 | def rev(node): |
|
1404 | 1408 | """Obtain the revision number for a binary node. |
|
1405 | 1409 | |
|
1406 | 1410 | Raises ``error.LookupError`` if the node is not known. |
|
1407 | 1411 | """ |
|
1408 | 1412 | |
|
1409 | 1413 | def update_caches(transaction): |
|
1410 | 1414 | """update whatever cache are relevant for the used storage.""" |
|
1411 | 1415 | |
|
1412 | 1416 | |
|
1413 | 1417 | class ilocalrepositoryfilestorage(interfaceutil.Interface): |
|
1414 | 1418 | """Local repository sub-interface providing access to tracked file storage. |
|
1415 | 1419 | |
|
1416 | 1420 | This interface defines how a repository accesses storage for a single |
|
1417 | 1421 | tracked file path. |
|
1418 | 1422 | """ |
|
1419 | 1423 | |
|
1420 | 1424 | def file(f): |
|
1421 | 1425 | """Obtain a filelog for a tracked path. |
|
1422 | 1426 | |
|
1423 | 1427 | The returned type conforms to the ``ifilestorage`` interface. |
|
1424 | 1428 | """ |
|
1425 | 1429 | |
|
1426 | 1430 | |
|
1427 | 1431 | class ilocalrepositorymain(interfaceutil.Interface): |
|
1428 | 1432 | """Main interface for local repositories. |
|
1429 | 1433 | |
|
1430 | 1434 | This currently captures the reality of things - not how things should be. |
|
1431 | 1435 | """ |
|
1432 | 1436 | |
|
1433 | 1437 | supportedformats = interfaceutil.Attribute( |
|
1434 | 1438 | """Set of requirements that apply to stream clone. |
|
1435 | 1439 | |
|
1436 | 1440 | This is actually a class attribute and is shared among all instances. |
|
1437 | 1441 | """ |
|
1438 | 1442 | ) |
|
1439 | 1443 | |
|
1440 | 1444 | supported = interfaceutil.Attribute( |
|
1441 | 1445 | """Set of requirements that this repo is capable of opening.""" |
|
1442 | 1446 | ) |
|
1443 | 1447 | |
|
1444 | 1448 | requirements = interfaceutil.Attribute( |
|
1445 | 1449 | """Set of requirements this repo uses.""" |
|
1446 | 1450 | ) |
|
1447 | 1451 | |
|
1448 | 1452 | features = interfaceutil.Attribute( |
|
1449 | 1453 | """Set of "features" this repository supports. |
|
1450 | 1454 | |
|
1451 | 1455 | A "feature" is a loosely-defined term. It can refer to a feature |
|
1452 | 1456 | in the classical sense or can describe an implementation detail |
|
1453 | 1457 | of the repository. For example, a ``readonly`` feature may denote |
|
1454 | 1458 | the repository as read-only. Or a ``revlogfilestore`` feature may |
|
1455 | 1459 | denote that the repository is using revlogs for file storage. |
|
1456 | 1460 | |
|
1457 | 1461 | The intent of features is to provide a machine-queryable mechanism |
|
1458 | 1462 | for repo consumers to test for various repository characteristics. |
|
1459 | 1463 | |
|
1460 | 1464 | Features are similar to ``requirements``. The main difference is that |
|
1461 | 1465 | requirements are stored on-disk and represent requirements to open the |
|
1462 | 1466 | repository. Features are more run-time capabilities of the repository |
|
1463 | 1467 | and more granular capabilities (which may be derived from requirements). |
|
1464 | 1468 | """ |
|
1465 | 1469 | ) |
|
1466 | 1470 | |
|
1467 | 1471 | filtername = interfaceutil.Attribute( |
|
1468 | 1472 | """Name of the repoview that is active on this repo.""" |
|
1469 | 1473 | ) |
|
1470 | 1474 | |
|
1471 | 1475 | wvfs = interfaceutil.Attribute( |
|
1472 | 1476 | """VFS used to access the working directory.""" |
|
1473 | 1477 | ) |
|
1474 | 1478 | |
|
1475 | 1479 | vfs = interfaceutil.Attribute( |
|
1476 | 1480 | """VFS rooted at the .hg directory. |
|
1477 | 1481 | |
|
1478 | 1482 | Used to access repository data not in the store. |
|
1479 | 1483 | """ |
|
1480 | 1484 | ) |
|
1481 | 1485 | |
|
1482 | 1486 | svfs = interfaceutil.Attribute( |
|
1483 | 1487 | """VFS rooted at the store. |
|
1484 | 1488 | |
|
1485 | 1489 | Used to access repository data in the store. Typically .hg/store. |
|
1486 | 1490 | But can point elsewhere if the store is shared. |
|
1487 | 1491 | """ |
|
1488 | 1492 | ) |
|
1489 | 1493 | |
|
1490 | 1494 | root = interfaceutil.Attribute( |
|
1491 | 1495 | """Path to the root of the working directory.""" |
|
1492 | 1496 | ) |
|
1493 | 1497 | |
|
1494 | 1498 | path = interfaceutil.Attribute("""Path to the .hg directory.""") |
|
1495 | 1499 | |
|
1496 | 1500 | origroot = interfaceutil.Attribute( |
|
1497 | 1501 | """The filesystem path that was used to construct the repo.""" |
|
1498 | 1502 | ) |
|
1499 | 1503 | |
|
1500 | 1504 | auditor = interfaceutil.Attribute( |
|
1501 | 1505 | """A pathauditor for the working directory. |
|
1502 | 1506 | |
|
1503 | 1507 | This checks if a path refers to a nested repository. |
|
1504 | 1508 | |
|
1505 | 1509 | Operates on the filesystem. |
|
1506 | 1510 | """ |
|
1507 | 1511 | ) |
|
1508 | 1512 | |
|
1509 | 1513 | nofsauditor = interfaceutil.Attribute( |
|
1510 | 1514 | """A pathauditor for the working directory. |
|
1511 | 1515 | |
|
1512 | 1516 | This is like ``auditor`` except it doesn't do filesystem checks. |
|
1513 | 1517 | """ |
|
1514 | 1518 | ) |
|
1515 | 1519 | |
|
1516 | 1520 | baseui = interfaceutil.Attribute( |
|
1517 | 1521 | """Original ui instance passed into constructor.""" |
|
1518 | 1522 | ) |
|
1519 | 1523 | |
|
1520 | 1524 | ui = interfaceutil.Attribute("""Main ui instance for this instance.""") |
|
1521 | 1525 | |
|
1522 | 1526 | sharedpath = interfaceutil.Attribute( |
|
1523 | 1527 | """Path to the .hg directory of the repo this repo was shared from.""" |
|
1524 | 1528 | ) |
|
1525 | 1529 | |
|
1526 | 1530 | store = interfaceutil.Attribute("""A store instance.""") |
|
1527 | 1531 | |
|
1528 | 1532 | spath = interfaceutil.Attribute("""Path to the store.""") |
|
1529 | 1533 | |
|
1530 | 1534 | sjoin = interfaceutil.Attribute("""Alias to self.store.join.""") |
|
1531 | 1535 | |
|
1532 | 1536 | cachevfs = interfaceutil.Attribute( |
|
1533 | 1537 | """A VFS used to access the cache directory. |
|
1534 | 1538 | |
|
1535 | 1539 | Typically .hg/cache. |
|
1536 | 1540 | """ |
|
1537 | 1541 | ) |
|
1538 | 1542 | |
|
1539 | 1543 | wcachevfs = interfaceutil.Attribute( |
|
1540 | 1544 | """A VFS used to access the cache directory dedicated to working copy |
|
1541 | 1545 | |
|
1542 | 1546 | Typically .hg/wcache. |
|
1543 | 1547 | """ |
|
1544 | 1548 | ) |
|
1545 | 1549 | |
|
1546 | 1550 | filteredrevcache = interfaceutil.Attribute( |
|
1547 | 1551 | """Holds sets of revisions to be filtered.""" |
|
1548 | 1552 | ) |
|
1549 | 1553 | |
|
1550 | 1554 | names = interfaceutil.Attribute("""A ``namespaces`` instance.""") |
|
1551 | 1555 | |
|
1552 | 1556 | filecopiesmode = interfaceutil.Attribute( |
|
1553 | 1557 | """The way files copies should be dealt with in this repo.""" |
|
1554 | 1558 | ) |
|
1555 | 1559 | |
|
1556 | 1560 | def close(): |
|
1557 | 1561 | """Close the handle on this repository.""" |
|
1558 | 1562 | |
|
1559 | 1563 | def peer(): |
|
1560 | 1564 | """Obtain an object conforming to the ``peer`` interface.""" |
|
1561 | 1565 | |
|
1562 | 1566 | def unfiltered(): |
|
1563 | 1567 | """Obtain an unfiltered/raw view of this repo.""" |
|
1564 | 1568 | |
|
1565 | 1569 | def filtered(name, visibilityexceptions=None): |
|
1566 | 1570 | """Obtain a named view of this repository.""" |
|
1567 | 1571 | |
|
1568 | 1572 | obsstore = interfaceutil.Attribute("""A store of obsolescence data.""") |
|
1569 | 1573 | |
|
1570 | 1574 | changelog = interfaceutil.Attribute("""A handle on the changelog revlog.""") |
|
1571 | 1575 | |
|
1572 | 1576 | manifestlog = interfaceutil.Attribute( |
|
1573 | 1577 | """An instance conforming to the ``imanifestlog`` interface. |
|
1574 | 1578 | |
|
1575 | 1579 | Provides access to manifests for the repository. |
|
1576 | 1580 | """ |
|
1577 | 1581 | ) |
|
1578 | 1582 | |
|
1579 | 1583 | dirstate = interfaceutil.Attribute("""Working directory state.""") |
|
1580 | 1584 | |
|
1581 | 1585 | narrowpats = interfaceutil.Attribute( |
|
1582 | 1586 | """Matcher patterns for this repository's narrowspec.""" |
|
1583 | 1587 | ) |
|
1584 | 1588 | |
|
1585 | 1589 | def narrowmatch(match=None, includeexact=False): |
|
1586 | 1590 | """Obtain a matcher for the narrowspec.""" |
|
1587 | 1591 | |
|
1588 | 1592 | def setnarrowpats(newincludes, newexcludes): |
|
1589 | 1593 | """Define the narrowspec for this repository.""" |
|
1590 | 1594 | |
|
1591 | 1595 | def __getitem__(changeid): |
|
1592 | 1596 | """Try to resolve a changectx.""" |
|
1593 | 1597 | |
|
1594 | 1598 | def __contains__(changeid): |
|
1595 | 1599 | """Whether a changeset exists.""" |
|
1596 | 1600 | |
|
1597 | 1601 | def __nonzero__(): |
|
1598 | 1602 | """Always returns True.""" |
|
1599 | 1603 | return True |
|
1600 | 1604 | |
|
1601 | 1605 | __bool__ = __nonzero__ |
|
1602 | 1606 | |
|
1603 | 1607 | def __len__(): |
|
1604 | 1608 | """Returns the number of changesets in the repo.""" |
|
1605 | 1609 | |
|
1606 | 1610 | def __iter__(): |
|
1607 | 1611 | """Iterate over revisions in the changelog.""" |
|
1608 | 1612 | |
|
1609 | 1613 | def revs(expr, *args): |
|
1610 | 1614 | """Evaluate a revset. |
|
1611 | 1615 | |
|
1612 | 1616 | Emits revisions. |
|
1613 | 1617 | """ |
|
1614 | 1618 | |
|
1615 | 1619 | def set(expr, *args): |
|
1616 | 1620 | """Evaluate a revset. |
|
1617 | 1621 | |
|
1618 | 1622 | Emits changectx instances. |
|
1619 | 1623 | """ |
|
1620 | 1624 | |
|
1621 | 1625 | def anyrevs(specs, user=False, localalias=None): |
|
1622 | 1626 | """Find revisions matching one of the given revsets.""" |
|
1623 | 1627 | |
|
1624 | 1628 | def url(): |
|
1625 | 1629 | """Returns a string representing the location of this repo.""" |
|
1626 | 1630 | |
|
1627 | 1631 | def hook(name, throw=False, **args): |
|
1628 | 1632 | """Call a hook.""" |
|
1629 | 1633 | |
|
1630 | 1634 | def tags(): |
|
1631 | 1635 | """Return a mapping of tag to node.""" |
|
1632 | 1636 | |
|
1633 | 1637 | def tagtype(tagname): |
|
1634 | 1638 | """Return the type of a given tag.""" |
|
1635 | 1639 | |
|
1636 | 1640 | def tagslist(): |
|
1637 | 1641 | """Return a list of tags ordered by revision.""" |
|
1638 | 1642 | |
|
1639 | 1643 | def nodetags(node): |
|
1640 | 1644 | """Return the tags associated with a node.""" |
|
1641 | 1645 | |
|
1642 | 1646 | def nodebookmarks(node): |
|
1643 | 1647 | """Return the list of bookmarks pointing to the specified node.""" |
|
1644 | 1648 | |
|
1645 | 1649 | def branchmap(): |
|
1646 | 1650 | """Return a mapping of branch to heads in that branch.""" |
|
1647 | 1651 | |
|
1648 | 1652 | def revbranchcache(): |
|
1649 | 1653 | pass |
|
1650 | 1654 | |
|
1651 | 1655 | def register_changeset(rev, changelogrevision): |
|
1652 | 1656 | """Extension point for caches for new nodes. |
|
1653 | 1657 | |
|
1654 | 1658 | Multiple consumers are expected to need parts of the changelogrevision, |
|
1655 | 1659 | so it is provided as optimization to avoid duplicate lookups. A simple |
|
1656 | 1660 | cache would be fragile when other revisions are accessed, too.""" |
|
1657 | 1661 | pass |
|
1658 | 1662 | |
|
1659 | 1663 | def branchtip(branchtip, ignoremissing=False): |
|
1660 | 1664 | """Return the tip node for a given branch.""" |
|
1661 | 1665 | |
|
1662 | 1666 | def lookup(key): |
|
1663 | 1667 | """Resolve the node for a revision.""" |
|
1664 | 1668 | |
|
1665 | 1669 | def lookupbranch(key): |
|
1666 | 1670 | """Look up the branch name of the given revision or branch name.""" |
|
1667 | 1671 | |
|
1668 | 1672 | def known(nodes): |
|
1669 | 1673 | """Determine whether a series of nodes is known. |
|
1670 | 1674 | |
|
1671 | 1675 | Returns a list of bools. |
|
1672 | 1676 | """ |
|
1673 | 1677 | |
|
1674 | 1678 | def local(): |
|
1675 | 1679 | """Whether the repository is local.""" |
|
1676 | 1680 | return True |
|
1677 | 1681 | |
|
1678 | 1682 | def publishing(): |
|
1679 | 1683 | """Whether the repository is a publishing repository.""" |
|
1680 | 1684 | |
|
1681 | 1685 | def cancopy(): |
|
1682 | 1686 | pass |
|
1683 | 1687 | |
|
1684 | 1688 | def shared(): |
|
1685 | 1689 | """The type of shared repository or None.""" |
|
1686 | 1690 | |
|
1687 | 1691 | def wjoin(f, *insidef): |
|
1688 | 1692 | """Calls self.vfs.reljoin(self.root, f, *insidef)""" |
|
1689 | 1693 | |
|
1690 | 1694 | def setparents(p1, p2): |
|
1691 | 1695 | """Set the parent nodes of the working directory.""" |
|
1692 | 1696 | |
|
1693 | 1697 | def filectx(path, changeid=None, fileid=None): |
|
1694 | 1698 | """Obtain a filectx for the given file revision.""" |
|
1695 | 1699 | |
|
1696 | 1700 | def getcwd(): |
|
1697 | 1701 | """Obtain the current working directory from the dirstate.""" |
|
1698 | 1702 | |
|
1699 | 1703 | def pathto(f, cwd=None): |
|
1700 | 1704 | """Obtain the relative path to a file.""" |
|
1701 | 1705 | |
|
1702 | 1706 | def adddatafilter(name, fltr): |
|
1703 | 1707 | pass |
|
1704 | 1708 | |
|
1705 | 1709 | def wread(filename): |
|
1706 | 1710 | """Read a file from wvfs, using data filters.""" |
|
1707 | 1711 | |
|
1708 | 1712 | def wwrite(filename, data, flags, backgroundclose=False, **kwargs): |
|
1709 | 1713 | """Write data to a file in the wvfs, using data filters.""" |
|
1710 | 1714 | |
|
1711 | 1715 | def wwritedata(filename, data): |
|
1712 | 1716 | """Resolve data for writing to the wvfs, using data filters.""" |
|
1713 | 1717 | |
|
1714 | 1718 | def currenttransaction(): |
|
1715 | 1719 | """Obtain the current transaction instance or None.""" |
|
1716 | 1720 | |
|
1717 | 1721 | def transaction(desc, report=None): |
|
1718 | 1722 | """Open a new transaction to write to the repository.""" |
|
1719 | 1723 | |
|
1720 | 1724 | def undofiles(): |
|
1721 | 1725 | """Returns a list of (vfs, path) for files to undo transactions.""" |
|
1722 | 1726 | |
|
1723 | 1727 | def recover(): |
|
1724 | 1728 | """Roll back an interrupted transaction.""" |
|
1725 | 1729 | |
|
1726 | 1730 | def rollback(dryrun=False, force=False): |
|
1727 | 1731 | """Undo the last transaction. |
|
1728 | 1732 | |
|
1729 | 1733 | DANGEROUS. |
|
1730 | 1734 | """ |
|
1731 | 1735 | |
|
1732 | 1736 | def updatecaches(tr=None, full=False): |
|
1733 | 1737 | """Warm repo caches.""" |
|
1734 | 1738 | |
|
1735 | 1739 | def invalidatecaches(): |
|
1736 | 1740 | """Invalidate cached data due to the repository mutating.""" |
|
1737 | 1741 | |
|
1738 | 1742 | def invalidatevolatilesets(): |
|
1739 | 1743 | pass |
|
1740 | 1744 | |
|
1741 | 1745 | def invalidatedirstate(): |
|
1742 | 1746 | """Invalidate the dirstate.""" |
|
1743 | 1747 | |
|
1744 | 1748 | def invalidate(clearfilecache=False): |
|
1745 | 1749 | pass |
|
1746 | 1750 | |
|
1747 | 1751 | def invalidateall(): |
|
1748 | 1752 | pass |
|
1749 | 1753 | |
|
1750 | 1754 | def lock(wait=True): |
|
1751 | 1755 | """Lock the repository store and return a lock instance.""" |
|
1752 | 1756 | |
|
1753 | 1757 | def wlock(wait=True): |
|
1754 | 1758 | """Lock the non-store parts of the repository.""" |
|
1755 | 1759 | |
|
1756 | 1760 | def currentwlock(): |
|
1757 | 1761 | """Return the wlock if it's held or None.""" |
|
1758 | 1762 | |
|
1759 | 1763 | def checkcommitpatterns(wctx, match, status, fail): |
|
1760 | 1764 | pass |
|
1761 | 1765 | |
|
1762 | 1766 | def commit( |
|
1763 | 1767 | text=b'', |
|
1764 | 1768 | user=None, |
|
1765 | 1769 | date=None, |
|
1766 | 1770 | match=None, |
|
1767 | 1771 | force=False, |
|
1768 | 1772 | editor=False, |
|
1769 | 1773 | extra=None, |
|
1770 | 1774 | ): |
|
1771 | 1775 | """Add a new revision to the repository.""" |
|
1772 | 1776 | |
|
1773 | 1777 | def commitctx(ctx, error=False, origctx=None): |
|
1774 | 1778 | """Commit a commitctx instance to the repository.""" |
|
1775 | 1779 | |
|
1776 | 1780 | def destroying(): |
|
1777 | 1781 | """Inform the repository that nodes are about to be destroyed.""" |
|
1778 | 1782 | |
|
1779 | 1783 | def destroyed(): |
|
1780 | 1784 | """Inform the repository that nodes have been destroyed.""" |
|
1781 | 1785 | |
|
1782 | 1786 | def status( |
|
1783 | 1787 | node1=b'.', |
|
1784 | 1788 | node2=None, |
|
1785 | 1789 | match=None, |
|
1786 | 1790 | ignored=False, |
|
1787 | 1791 | clean=False, |
|
1788 | 1792 | unknown=False, |
|
1789 | 1793 | listsubrepos=False, |
|
1790 | 1794 | ): |
|
1791 | 1795 | """Convenience method to call repo[x].status().""" |
|
1792 | 1796 | |
|
1793 | 1797 | def addpostdsstatus(ps): |
|
1794 | 1798 | pass |
|
1795 | 1799 | |
|
1796 | 1800 | def postdsstatus(): |
|
1797 | 1801 | pass |
|
1798 | 1802 | |
|
1799 | 1803 | def clearpostdsstatus(): |
|
1800 | 1804 | pass |
|
1801 | 1805 | |
|
1802 | 1806 | def heads(start=None): |
|
1803 | 1807 | """Obtain list of nodes that are DAG heads.""" |
|
1804 | 1808 | |
|
1805 | 1809 | def branchheads(branch=None, start=None, closed=False): |
|
1806 | 1810 | pass |
|
1807 | 1811 | |
|
1808 | 1812 | def branches(nodes): |
|
1809 | 1813 | pass |
|
1810 | 1814 | |
|
1811 | 1815 | def between(pairs): |
|
1812 | 1816 | pass |
|
1813 | 1817 | |
|
1814 | 1818 | def checkpush(pushop): |
|
1815 | 1819 | pass |
|
1816 | 1820 | |
|
1817 | 1821 | prepushoutgoinghooks = interfaceutil.Attribute("""util.hooks instance.""") |
|
1818 | 1822 | |
|
1819 | 1823 | def pushkey(namespace, key, old, new): |
|
1820 | 1824 | pass |
|
1821 | 1825 | |
|
1822 | 1826 | def listkeys(namespace): |
|
1823 | 1827 | pass |
|
1824 | 1828 | |
|
1825 | 1829 | def debugwireargs(one, two, three=None, four=None, five=None): |
|
1826 | 1830 | pass |
|
1827 | 1831 | |
|
1828 | 1832 | def savecommitmessage(text): |
|
1829 | 1833 | pass |
|
1830 | 1834 | |
|
1831 | 1835 | |
|
1832 | 1836 | class completelocalrepository( |
|
1833 | 1837 | ilocalrepositorymain, ilocalrepositoryfilestorage |
|
1834 | 1838 | ): |
|
1835 | 1839 | """Complete interface for a local repository.""" |
|
1836 | 1840 | |
|
1837 | 1841 | |
|
1838 | 1842 | class iwireprotocolcommandcacher(interfaceutil.Interface): |
|
1839 | 1843 | """Represents a caching backend for wire protocol commands. |
|
1840 | 1844 | |
|
1841 | 1845 | Wire protocol version 2 supports transparent caching of many commands. |
|
1842 | 1846 | To leverage this caching, servers can activate objects that cache |
|
1843 | 1847 | command responses. Objects handle both cache writing and reading. |
|
1844 | 1848 | This interface defines how that response caching mechanism works. |
|
1845 | 1849 | |
|
1846 | 1850 | Wire protocol version 2 commands emit a series of objects that are |
|
1847 | 1851 | serialized and sent to the client. The caching layer exists between |
|
1848 | 1852 | the invocation of the command function and the sending of its output |
|
1849 | 1853 | objects to an output layer. |
|
1850 | 1854 | |
|
1851 | 1855 | Instances of this interface represent a binding to a cache that |
|
1852 | 1856 | can serve a response (in place of calling a command function) and/or |
|
1853 | 1857 | write responses to a cache for subsequent use. |
|
1854 | 1858 | |
|
1855 | 1859 | When a command request arrives, the following happens with regards |
|
1856 | 1860 | to this interface: |
|
1857 | 1861 | |
|
1858 | 1862 | 1. The server determines whether the command request is cacheable. |
|
1859 | 1863 | 2. If it is, an instance of this interface is spawned. |
|
1860 | 1864 | 3. The cacher is activated in a context manager (``__enter__`` is called). |
|
1861 | 1865 | 4. A cache *key* for that request is derived. This will call the |
|
1862 | 1866 | instance's ``adjustcachekeystate()`` method so the derivation |
|
1863 | 1867 | can be influenced. |
|
1864 | 1868 | 5. The cacher is informed of the derived cache key via a call to |
|
1865 | 1869 | ``setcachekey()``. |
|
1866 | 1870 | 6. The cacher's ``lookup()`` method is called to test for presence of |
|
1867 | 1871 | the derived key in the cache. |
|
1868 | 1872 | 7. If ``lookup()`` returns a hit, that cached result is used in place |
|
1869 | 1873 | of invoking the command function. ``__exit__`` is called and the instance |
|
1870 | 1874 | is discarded. |
|
1871 | 1875 | 8. The command function is invoked. |
|
1872 | 1876 | 9. ``onobject()`` is called for each object emitted by the command |
|
1873 | 1877 | function. |
|
1874 | 1878 | 10. After the final object is seen, ``onfinished()`` is called. |
|
1875 | 1879 | 11. ``__exit__`` is called to signal the end of use of the instance. |
|
1876 | 1880 | |
|
1877 | 1881 | Cache *key* derivation can be influenced by the instance. |
|
1878 | 1882 | |
|
1879 | 1883 | Cache keys are initially derived by a deterministic representation of |
|
1880 | 1884 | the command request. This includes the command name, arguments, protocol |
|
1881 | 1885 | version, etc. This initial key derivation is performed by CBOR-encoding a |
|
1882 | 1886 | data structure and feeding that output into a hasher. |
|
1883 | 1887 | |
|
1884 | 1888 | Instances of this interface can influence this initial key derivation |
|
1885 | 1889 | via ``adjustcachekeystate()``. |
|
1886 | 1890 | |
|
1887 | 1891 | The instance is informed of the derived cache key via a call to |
|
1888 | 1892 | ``setcachekey()``. The instance must store the key locally so it can |
|
1889 | 1893 | be consulted on subsequent operations that may require it. |
|
1890 | 1894 | |
|
1891 | 1895 | When constructed, the instance has access to a callable that can be used |
|
1892 | 1896 | for encoding response objects. This callable receives as its single |
|
1893 | 1897 | argument an object emitted by a command function. It returns an iterable |
|
1894 | 1898 | of bytes chunks representing the encoded object. Unless the cacher is |
|
1895 | 1899 | caching native Python objects in memory or has a way of reconstructing |
|
1896 | 1900 | the original Python objects, implementations typically call this function |
|
1897 | 1901 | to produce bytes from the output objects and then store those bytes in |
|
1898 | 1902 | the cache. When it comes time to re-emit those bytes, they are wrapped |
|
1899 | 1903 | in a ``wireprototypes.encodedresponse`` instance to tell the output |
|
1900 | 1904 | layer that they are pre-encoded. |
|
1901 | 1905 | |
|
1902 | 1906 | When receiving the objects emitted by the command function, instances |
|
1903 | 1907 | can choose what to do with those objects. The simplest thing to do is |
|
1904 | 1908 | re-emit the original objects. They will be forwarded to the output |
|
1905 | 1909 | layer and will be processed as if the cacher did not exist. |
|
1906 | 1910 | |
|
1907 | 1911 | Implementations could also choose to not emit objects - instead locally |
|
1908 | 1912 | buffering objects or their encoded representation. They could then emit |
|
1909 | 1913 | a single "coalesced" object when ``onfinished()`` is called. In |
|
1910 | 1914 | this way, the implementation would function as a filtering layer of |
|
1911 | 1915 | sorts. |
|
1912 | 1916 | |
|
1913 | 1917 | When caching objects, typically the encoded form of the object will |
|
1914 | 1918 | be stored. Keep in mind that if the original object is forwarded to |
|
1915 | 1919 | the output layer, it will need to be encoded there as well. For large |
|
1916 | 1920 | output, this redundant encoding could add overhead. Implementations |
|
1917 | 1921 | could wrap the encoded object data in ``wireprototypes.encodedresponse`` |
|
1918 | 1922 | instances to avoid this overhead. |
|
1919 | 1923 | """ |
|
1920 | 1924 | |
|
1921 | 1925 | def __enter__(): |
|
1922 | 1926 | """Marks the instance as active. |
|
1923 | 1927 | |
|
1924 | 1928 | Should return self. |
|
1925 | 1929 | """ |
|
1926 | 1930 | |
|
1927 | 1931 | def __exit__(exctype, excvalue, exctb): |
|
1928 | 1932 | """Called when cacher is no longer used. |
|
1929 | 1933 | |
|
1930 | 1934 | This can be used by implementations to perform cleanup actions (e.g. |
|
1931 | 1935 | disconnecting network sockets, aborting a partially cached response. |
|
1932 | 1936 | """ |
|
1933 | 1937 | |
|
1934 | 1938 | def adjustcachekeystate(state): |
|
1935 | 1939 | """Influences cache key derivation by adjusting state to derive key. |
|
1936 | 1940 | |
|
1937 | 1941 | A dict defining the state used to derive the cache key is passed. |
|
1938 | 1942 | |
|
1939 | 1943 | Implementations can modify this dict to record additional state that |
|
1940 | 1944 | is wanted to influence key derivation. |
|
1941 | 1945 | |
|
1942 | 1946 | Implementations are *highly* encouraged to not modify or delete |
|
1943 | 1947 | existing keys. |
|
1944 | 1948 | """ |
|
1945 | 1949 | |
|
1946 | 1950 | def setcachekey(key): |
|
1947 | 1951 | """Record the derived cache key for this request. |
|
1948 | 1952 | |
|
1949 | 1953 | Instances may mutate the key for internal usage, as desired. e.g. |
|
1950 | 1954 | instances may wish to prepend the repo name, introduce path |
|
1951 | 1955 | components for filesystem or URL addressing, etc. Behavior is up to |
|
1952 | 1956 | the cache. |
|
1953 | 1957 | |
|
1954 | 1958 | Returns a bool indicating if the request is cacheable by this |
|
1955 | 1959 | instance. |
|
1956 | 1960 | """ |
|
1957 | 1961 | |
|
1958 | 1962 | def lookup(): |
|
1959 | 1963 | """Attempt to resolve an entry in the cache. |
|
1960 | 1964 | |
|
1961 | 1965 | The instance is instructed to look for the cache key that it was |
|
1962 | 1966 | informed about via the call to ``setcachekey()``. |
|
1963 | 1967 | |
|
1964 | 1968 | If there's no cache hit or the cacher doesn't wish to use the cached |
|
1965 | 1969 | entry, ``None`` should be returned. |
|
1966 | 1970 | |
|
1967 | 1971 | Else, a dict defining the cached result should be returned. The |
|
1968 | 1972 | dict may have the following keys: |
|
1969 | 1973 | |
|
1970 | 1974 | objs |
|
1971 | 1975 | An iterable of objects that should be sent to the client. That |
|
1972 | 1976 | iterable of objects is expected to be what the command function |
|
1973 | 1977 | would return if invoked or an equivalent representation thereof. |
|
1974 | 1978 | """ |
|
1975 | 1979 | |
|
1976 | 1980 | def onobject(obj): |
|
1977 | 1981 | """Called when a new object is emitted from the command function. |
|
1978 | 1982 | |
|
1979 | 1983 | Receives as its argument the object that was emitted from the |
|
1980 | 1984 | command function. |
|
1981 | 1985 | |
|
1982 | 1986 | This method returns an iterator of objects to forward to the output |
|
1983 | 1987 | layer. The easiest implementation is a generator that just |
|
1984 | 1988 | ``yield obj``. |
|
1985 | 1989 | """ |
|
1986 | 1990 | |
|
1987 | 1991 | def onfinished(): |
|
1988 | 1992 | """Called after all objects have been emitted from the command function. |
|
1989 | 1993 | |
|
1990 | 1994 | Implementations should return an iterator of objects to forward to |
|
1991 | 1995 | the output layer. |
|
1992 | 1996 | |
|
1993 | 1997 | This method can be a generator. |
|
1994 | 1998 | """ |
@@ -1,3199 +1,3201 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 | from .pure import parsers as pureparsers |
|
87 | 87 | |
|
88 | 88 | # blanked usage of all the name to prevent pyflakes constraints |
|
89 | 89 | # We need these name available in the module for extensions. |
|
90 | 90 | REVLOGV0 |
|
91 | 91 | REVLOGV1 |
|
92 | 92 | REVLOGV2 |
|
93 | 93 | FLAG_INLINE_DATA |
|
94 | 94 | FLAG_GENERALDELTA |
|
95 | 95 | REVLOG_DEFAULT_FLAGS |
|
96 | 96 | REVLOG_DEFAULT_FORMAT |
|
97 | 97 | REVLOG_DEFAULT_VERSION |
|
98 | 98 | REVLOGV1_FLAGS |
|
99 | 99 | REVLOGV2_FLAGS |
|
100 | 100 | REVIDX_ISCENSORED |
|
101 | 101 | REVIDX_ELLIPSIS |
|
102 | 102 | REVIDX_SIDEDATA |
|
103 | 103 | REVIDX_HASCOPIESINFO |
|
104 | 104 | REVIDX_EXTSTORED |
|
105 | 105 | REVIDX_DEFAULT_FLAGS |
|
106 | 106 | REVIDX_FLAGS_ORDER |
|
107 | 107 | REVIDX_RAWTEXT_CHANGING_FLAGS |
|
108 | 108 | |
|
109 | 109 | parsers = policy.importmod('parsers') |
|
110 | 110 | rustancestor = policy.importrust('ancestor') |
|
111 | 111 | rustdagop = policy.importrust('dagop') |
|
112 | 112 | rustrevlog = policy.importrust('revlog') |
|
113 | 113 | |
|
114 | 114 | # Aliased for performance. |
|
115 | 115 | _zlibdecompress = zlib.decompress |
|
116 | 116 | |
|
117 | 117 | # max size of revlog with inline data |
|
118 | 118 | _maxinline = 131072 |
|
119 | 119 | _chunksize = 1048576 |
|
120 | 120 | |
|
121 | 121 | # Flag processors for REVIDX_ELLIPSIS. |
|
122 | 122 | def ellipsisreadprocessor(rl, text): |
|
123 | 123 | return text, False |
|
124 | 124 | |
|
125 | 125 | |
|
126 | 126 | def ellipsiswriteprocessor(rl, text): |
|
127 | 127 | return text, False |
|
128 | 128 | |
|
129 | 129 | |
|
130 | 130 | def ellipsisrawprocessor(rl, text): |
|
131 | 131 | return False |
|
132 | 132 | |
|
133 | 133 | |
|
134 | 134 | ellipsisprocessor = ( |
|
135 | 135 | ellipsisreadprocessor, |
|
136 | 136 | ellipsiswriteprocessor, |
|
137 | 137 | ellipsisrawprocessor, |
|
138 | 138 | ) |
|
139 | 139 | |
|
140 | 140 | |
|
141 | 141 | def getoffset(q): |
|
142 | 142 | return int(q >> 16) |
|
143 | 143 | |
|
144 | 144 | |
|
145 | 145 | def gettype(q): |
|
146 | 146 | return int(q & 0xFFFF) |
|
147 | 147 | |
|
148 | 148 | |
|
149 | 149 | def offset_type(offset, type): |
|
150 | 150 | if (type & ~flagutil.REVIDX_KNOWN_FLAGS) != 0: |
|
151 | 151 | raise ValueError(b'unknown revlog index flags') |
|
152 | 152 | return int(int(offset) << 16 | type) |
|
153 | 153 | |
|
154 | 154 | |
|
155 | 155 | def _verify_revision(rl, skipflags, state, node): |
|
156 | 156 | """Verify the integrity of the given revlog ``node`` while providing a hook |
|
157 | 157 | point for extensions to influence the operation.""" |
|
158 | 158 | if skipflags: |
|
159 | 159 | state[b'skipread'].add(node) |
|
160 | 160 | else: |
|
161 | 161 | # Side-effect: read content and verify hash. |
|
162 | 162 | rl.revision(node) |
|
163 | 163 | |
|
164 | 164 | |
|
165 | 165 | # True if a fast implementation for persistent-nodemap is available |
|
166 | 166 | # |
|
167 | 167 | # We also consider we have a "fast" implementation in "pure" python because |
|
168 | 168 | # people using pure don't really have performance consideration (and a |
|
169 | 169 | # wheelbarrow of other slowness source) |
|
170 | 170 | HAS_FAST_PERSISTENT_NODEMAP = rustrevlog is not None or util.safehasattr( |
|
171 | 171 | parsers, 'BaseIndexObject' |
|
172 | 172 | ) |
|
173 | 173 | |
|
174 | 174 | |
|
175 | 175 | @attr.s(slots=True, frozen=True) |
|
176 | 176 | class _revisioninfo(object): |
|
177 | 177 | """Information about a revision that allows building its fulltext |
|
178 | 178 | node: expected hash of the revision |
|
179 | 179 | p1, p2: parent revs of the revision |
|
180 | 180 | btext: built text cache consisting of a one-element list |
|
181 | 181 | cachedelta: (baserev, uncompressed_delta) or None |
|
182 | 182 | flags: flags associated to the revision storage |
|
183 | 183 | |
|
184 | 184 | One of btext[0] or cachedelta must be set. |
|
185 | 185 | """ |
|
186 | 186 | |
|
187 | 187 | node = attr.ib() |
|
188 | 188 | p1 = attr.ib() |
|
189 | 189 | p2 = attr.ib() |
|
190 | 190 | btext = attr.ib() |
|
191 | 191 | textlen = attr.ib() |
|
192 | 192 | cachedelta = attr.ib() |
|
193 | 193 | flags = attr.ib() |
|
194 | 194 | |
|
195 | 195 | |
|
196 | 196 | @interfaceutil.implementer(repository.irevisiondelta) |
|
197 | 197 | @attr.s(slots=True) |
|
198 | 198 | class revlogrevisiondelta(object): |
|
199 | 199 | node = attr.ib() |
|
200 | 200 | p1node = attr.ib() |
|
201 | 201 | p2node = attr.ib() |
|
202 | 202 | basenode = attr.ib() |
|
203 | 203 | flags = attr.ib() |
|
204 | 204 | baserevisionsize = attr.ib() |
|
205 | 205 | revision = attr.ib() |
|
206 | 206 | delta = attr.ib() |
|
207 | sidedata = attr.ib() | |
|
207 | 208 | linknode = attr.ib(default=None) |
|
208 | 209 | |
|
209 | 210 | |
|
210 | 211 | @interfaceutil.implementer(repository.iverifyproblem) |
|
211 | 212 | @attr.s(frozen=True) |
|
212 | 213 | class revlogproblem(object): |
|
213 | 214 | warning = attr.ib(default=None) |
|
214 | 215 | error = attr.ib(default=None) |
|
215 | 216 | node = attr.ib(default=None) |
|
216 | 217 | |
|
217 | 218 | |
|
218 | 219 | # index v0: |
|
219 | 220 | # 4 bytes: offset |
|
220 | 221 | # 4 bytes: compressed length |
|
221 | 222 | # 4 bytes: base rev |
|
222 | 223 | # 4 bytes: link rev |
|
223 | 224 | # 20 bytes: parent 1 nodeid |
|
224 | 225 | # 20 bytes: parent 2 nodeid |
|
225 | 226 | # 20 bytes: nodeid |
|
226 | 227 | indexformatv0 = struct.Struct(b">4l20s20s20s") |
|
227 | 228 | indexformatv0_pack = indexformatv0.pack |
|
228 | 229 | indexformatv0_unpack = indexformatv0.unpack |
|
229 | 230 | |
|
230 | 231 | |
|
231 | 232 | class revlogoldindex(list): |
|
232 | 233 | @property |
|
233 | 234 | def nodemap(self): |
|
234 | 235 | msg = b"index.nodemap is deprecated, use index.[has_node|rev|get_rev]" |
|
235 | 236 | util.nouideprecwarn(msg, b'5.3', stacklevel=2) |
|
236 | 237 | return self._nodemap |
|
237 | 238 | |
|
238 | 239 | @util.propertycache |
|
239 | 240 | def _nodemap(self): |
|
240 | 241 | nodemap = nodemaputil.NodeMap({nullid: nullrev}) |
|
241 | 242 | for r in range(0, len(self)): |
|
242 | 243 | n = self[r][7] |
|
243 | 244 | nodemap[n] = r |
|
244 | 245 | return nodemap |
|
245 | 246 | |
|
246 | 247 | def has_node(self, node): |
|
247 | 248 | """return True if the node exist in the index""" |
|
248 | 249 | return node in self._nodemap |
|
249 | 250 | |
|
250 | 251 | def rev(self, node): |
|
251 | 252 | """return a revision for a node |
|
252 | 253 | |
|
253 | 254 | If the node is unknown, raise a RevlogError""" |
|
254 | 255 | return self._nodemap[node] |
|
255 | 256 | |
|
256 | 257 | def get_rev(self, node): |
|
257 | 258 | """return a revision for a node |
|
258 | 259 | |
|
259 | 260 | If the node is unknown, return None""" |
|
260 | 261 | return self._nodemap.get(node) |
|
261 | 262 | |
|
262 | 263 | def append(self, tup): |
|
263 | 264 | self._nodemap[tup[7]] = len(self) |
|
264 | 265 | super(revlogoldindex, self).append(tup) |
|
265 | 266 | |
|
266 | 267 | def __delitem__(self, i): |
|
267 | 268 | if not isinstance(i, slice) or not i.stop == -1 or i.step is not None: |
|
268 | 269 | raise ValueError(b"deleting slices only supports a:-1 with step 1") |
|
269 | 270 | for r in pycompat.xrange(i.start, len(self)): |
|
270 | 271 | del self._nodemap[self[r][7]] |
|
271 | 272 | super(revlogoldindex, self).__delitem__(i) |
|
272 | 273 | |
|
273 | 274 | def clearcaches(self): |
|
274 | 275 | self.__dict__.pop('_nodemap', None) |
|
275 | 276 | |
|
276 | 277 | def __getitem__(self, i): |
|
277 | 278 | if i == -1: |
|
278 | 279 | return (0, 0, 0, -1, -1, -1, -1, nullid) |
|
279 | 280 | return list.__getitem__(self, i) |
|
280 | 281 | |
|
281 | 282 | |
|
282 | 283 | class revlogoldio(object): |
|
283 | 284 | def __init__(self): |
|
284 | 285 | self.size = indexformatv0.size |
|
285 | 286 | |
|
286 | 287 | def parseindex(self, data, inline): |
|
287 | 288 | s = self.size |
|
288 | 289 | index = [] |
|
289 | 290 | nodemap = nodemaputil.NodeMap({nullid: nullrev}) |
|
290 | 291 | n = off = 0 |
|
291 | 292 | l = len(data) |
|
292 | 293 | while off + s <= l: |
|
293 | 294 | cur = data[off : off + s] |
|
294 | 295 | off += s |
|
295 | 296 | e = indexformatv0_unpack(cur) |
|
296 | 297 | # transform to revlogv1 format |
|
297 | 298 | e2 = ( |
|
298 | 299 | offset_type(e[0], 0), |
|
299 | 300 | e[1], |
|
300 | 301 | -1, |
|
301 | 302 | e[2], |
|
302 | 303 | e[3], |
|
303 | 304 | nodemap.get(e[4], nullrev), |
|
304 | 305 | nodemap.get(e[5], nullrev), |
|
305 | 306 | e[6], |
|
306 | 307 | ) |
|
307 | 308 | index.append(e2) |
|
308 | 309 | nodemap[e[6]] = n |
|
309 | 310 | n += 1 |
|
310 | 311 | |
|
311 | 312 | index = revlogoldindex(index) |
|
312 | 313 | return index, None |
|
313 | 314 | |
|
314 | 315 | def packentry(self, entry, node, version, rev): |
|
315 | 316 | if gettype(entry[0]): |
|
316 | 317 | raise error.RevlogError( |
|
317 | 318 | _(b'index entry flags need revlog version 1') |
|
318 | 319 | ) |
|
319 | 320 | e2 = ( |
|
320 | 321 | getoffset(entry[0]), |
|
321 | 322 | entry[1], |
|
322 | 323 | entry[3], |
|
323 | 324 | entry[4], |
|
324 | 325 | node(entry[5]), |
|
325 | 326 | node(entry[6]), |
|
326 | 327 | entry[7], |
|
327 | 328 | ) |
|
328 | 329 | return indexformatv0_pack(*e2) |
|
329 | 330 | |
|
330 | 331 | |
|
331 | 332 | # index ng: |
|
332 | 333 | # 6 bytes: offset |
|
333 | 334 | # 2 bytes: flags |
|
334 | 335 | # 4 bytes: compressed length |
|
335 | 336 | # 4 bytes: uncompressed length |
|
336 | 337 | # 4 bytes: base rev |
|
337 | 338 | # 4 bytes: link rev |
|
338 | 339 | # 4 bytes: parent 1 rev |
|
339 | 340 | # 4 bytes: parent 2 rev |
|
340 | 341 | # 32 bytes: nodeid |
|
341 | 342 | indexformatng = struct.Struct(b">Qiiiiii20s12x") |
|
342 | 343 | indexformatng_pack = indexformatng.pack |
|
343 | 344 | versionformat = struct.Struct(b">I") |
|
344 | 345 | versionformat_pack = versionformat.pack |
|
345 | 346 | versionformat_unpack = versionformat.unpack |
|
346 | 347 | |
|
347 | 348 | # corresponds to uncompressed length of indexformatng (2 gigs, 4-byte |
|
348 | 349 | # signed integer) |
|
349 | 350 | _maxentrysize = 0x7FFFFFFF |
|
350 | 351 | |
|
351 | 352 | |
|
352 | 353 | class revlogio(object): |
|
353 | 354 | def __init__(self): |
|
354 | 355 | self.size = indexformatng.size |
|
355 | 356 | |
|
356 | 357 | def parseindex(self, data, inline): |
|
357 | 358 | # call the C implementation to parse the index data |
|
358 | 359 | index, cache = parsers.parse_index2(data, inline) |
|
359 | 360 | return index, cache |
|
360 | 361 | |
|
361 | 362 | def packentry(self, entry, node, version, rev): |
|
362 | 363 | p = indexformatng_pack(*entry) |
|
363 | 364 | if rev == 0: |
|
364 | 365 | p = versionformat_pack(version) + p[4:] |
|
365 | 366 | return p |
|
366 | 367 | |
|
367 | 368 | |
|
368 | 369 | indexformatv2 = struct.Struct(pureparsers.Index2Mixin.index_format) |
|
369 | 370 | indexformatv2_pack = indexformatv2.pack |
|
370 | 371 | |
|
371 | 372 | |
|
372 | 373 | class revlogv2io(object): |
|
373 | 374 | def __init__(self): |
|
374 | 375 | self.size = indexformatv2.size |
|
375 | 376 | |
|
376 | 377 | def parseindex(self, data, inline): |
|
377 | 378 | index, cache = parsers.parse_index2(data, inline, revlogv2=True) |
|
378 | 379 | return index, cache |
|
379 | 380 | |
|
380 | 381 | def packentry(self, entry, node, version, rev): |
|
381 | 382 | p = indexformatv2_pack(*entry) |
|
382 | 383 | if rev == 0: |
|
383 | 384 | p = versionformat_pack(version) + p[4:] |
|
384 | 385 | return p |
|
385 | 386 | |
|
386 | 387 | |
|
387 | 388 | NodemapRevlogIO = None |
|
388 | 389 | |
|
389 | 390 | if util.safehasattr(parsers, 'parse_index_devel_nodemap'): |
|
390 | 391 | |
|
391 | 392 | class NodemapRevlogIO(revlogio): |
|
392 | 393 | """A debug oriented IO class that return a PersistentNodeMapIndexObject |
|
393 | 394 | |
|
394 | 395 | The PersistentNodeMapIndexObject object is meant to test the persistent nodemap feature. |
|
395 | 396 | """ |
|
396 | 397 | |
|
397 | 398 | def parseindex(self, data, inline): |
|
398 | 399 | index, cache = parsers.parse_index_devel_nodemap(data, inline) |
|
399 | 400 | return index, cache |
|
400 | 401 | |
|
401 | 402 | |
|
402 | 403 | class rustrevlogio(revlogio): |
|
403 | 404 | def parseindex(self, data, inline): |
|
404 | 405 | index, cache = super(rustrevlogio, self).parseindex(data, inline) |
|
405 | 406 | return rustrevlog.MixedIndex(index), cache |
|
406 | 407 | |
|
407 | 408 | |
|
408 | 409 | class revlog(object): |
|
409 | 410 | """ |
|
410 | 411 | the underlying revision storage object |
|
411 | 412 | |
|
412 | 413 | A revlog consists of two parts, an index and the revision data. |
|
413 | 414 | |
|
414 | 415 | The index is a file with a fixed record size containing |
|
415 | 416 | information on each revision, including its nodeid (hash), the |
|
416 | 417 | nodeids of its parents, the position and offset of its data within |
|
417 | 418 | the data file, and the revision it's based on. Finally, each entry |
|
418 | 419 | contains a linkrev entry that can serve as a pointer to external |
|
419 | 420 | data. |
|
420 | 421 | |
|
421 | 422 | The revision data itself is a linear collection of data chunks. |
|
422 | 423 | Each chunk represents a revision and is usually represented as a |
|
423 | 424 | delta against the previous chunk. To bound lookup time, runs of |
|
424 | 425 | deltas are limited to about 2 times the length of the original |
|
425 | 426 | version data. This makes retrieval of a version proportional to |
|
426 | 427 | its size, or O(1) relative to the number of revisions. |
|
427 | 428 | |
|
428 | 429 | Both pieces of the revlog are written to in an append-only |
|
429 | 430 | fashion, which means we never need to rewrite a file to insert or |
|
430 | 431 | remove data, and can use some simple techniques to avoid the need |
|
431 | 432 | for locking while reading. |
|
432 | 433 | |
|
433 | 434 | If checkambig, indexfile is opened with checkambig=True at |
|
434 | 435 | writing, to avoid file stat ambiguity. |
|
435 | 436 | |
|
436 | 437 | If mmaplargeindex is True, and an mmapindexthreshold is set, the |
|
437 | 438 | index will be mmapped rather than read if it is larger than the |
|
438 | 439 | configured threshold. |
|
439 | 440 | |
|
440 | 441 | If censorable is True, the revlog can have censored revisions. |
|
441 | 442 | |
|
442 | 443 | If `upperboundcomp` is not None, this is the expected maximal gain from |
|
443 | 444 | compression for the data content. |
|
444 | 445 | |
|
445 | 446 | `concurrencychecker` is an optional function that receives 3 arguments: a |
|
446 | 447 | file handle, a filename, and an expected position. It should check whether |
|
447 | 448 | the current position in the file handle is valid, and log/warn/fail (by |
|
448 | 449 | raising). |
|
449 | 450 | """ |
|
450 | 451 | |
|
451 | 452 | _flagserrorclass = error.RevlogError |
|
452 | 453 | |
|
453 | 454 | def __init__( |
|
454 | 455 | self, |
|
455 | 456 | opener, |
|
456 | 457 | indexfile, |
|
457 | 458 | datafile=None, |
|
458 | 459 | checkambig=False, |
|
459 | 460 | mmaplargeindex=False, |
|
460 | 461 | censorable=False, |
|
461 | 462 | upperboundcomp=None, |
|
462 | 463 | persistentnodemap=False, |
|
463 | 464 | concurrencychecker=None, |
|
464 | 465 | ): |
|
465 | 466 | """ |
|
466 | 467 | create a revlog object |
|
467 | 468 | |
|
468 | 469 | opener is a function that abstracts the file opening operation |
|
469 | 470 | and can be used to implement COW semantics or the like. |
|
470 | 471 | |
|
471 | 472 | """ |
|
472 | 473 | self.upperboundcomp = upperboundcomp |
|
473 | 474 | self.indexfile = indexfile |
|
474 | 475 | self.datafile = datafile or (indexfile[:-2] + b".d") |
|
475 | 476 | self.nodemap_file = None |
|
476 | 477 | if persistentnodemap: |
|
477 | 478 | self.nodemap_file = nodemaputil.get_nodemap_file( |
|
478 | 479 | opener, self.indexfile |
|
479 | 480 | ) |
|
480 | 481 | |
|
481 | 482 | self.opener = opener |
|
482 | 483 | # When True, indexfile is opened with checkambig=True at writing, to |
|
483 | 484 | # avoid file stat ambiguity. |
|
484 | 485 | self._checkambig = checkambig |
|
485 | 486 | self._mmaplargeindex = mmaplargeindex |
|
486 | 487 | self._censorable = censorable |
|
487 | 488 | # 3-tuple of (node, rev, text) for a raw revision. |
|
488 | 489 | self._revisioncache = None |
|
489 | 490 | # Maps rev to chain base rev. |
|
490 | 491 | self._chainbasecache = util.lrucachedict(100) |
|
491 | 492 | # 2-tuple of (offset, data) of raw data from the revlog at an offset. |
|
492 | 493 | self._chunkcache = (0, b'') |
|
493 | 494 | # How much data to read and cache into the raw revlog data cache. |
|
494 | 495 | self._chunkcachesize = 65536 |
|
495 | 496 | self._maxchainlen = None |
|
496 | 497 | self._deltabothparents = True |
|
497 | 498 | self.index = None |
|
498 | 499 | self._nodemap_docket = None |
|
499 | 500 | # Mapping of partial identifiers to full nodes. |
|
500 | 501 | self._pcache = {} |
|
501 | 502 | # Mapping of revision integer to full node. |
|
502 | 503 | self._compengine = b'zlib' |
|
503 | 504 | self._compengineopts = {} |
|
504 | 505 | self._maxdeltachainspan = -1 |
|
505 | 506 | self._withsparseread = False |
|
506 | 507 | self._sparserevlog = False |
|
507 | 508 | self._srdensitythreshold = 0.50 |
|
508 | 509 | self._srmingapsize = 262144 |
|
509 | 510 | |
|
510 | 511 | # Make copy of flag processors so each revlog instance can support |
|
511 | 512 | # custom flags. |
|
512 | 513 | self._flagprocessors = dict(flagutil.flagprocessors) |
|
513 | 514 | |
|
514 | 515 | # 2-tuple of file handles being used for active writing. |
|
515 | 516 | self._writinghandles = None |
|
516 | 517 | |
|
517 | 518 | self._loadindex() |
|
518 | 519 | |
|
519 | 520 | self._concurrencychecker = concurrencychecker |
|
520 | 521 | |
|
521 | 522 | def _loadindex(self): |
|
522 | 523 | mmapindexthreshold = None |
|
523 | 524 | opts = self.opener.options |
|
524 | 525 | |
|
525 | 526 | if b'revlogv2' in opts: |
|
526 | 527 | newversionflags = REVLOGV2 | FLAG_INLINE_DATA |
|
527 | 528 | elif b'revlogv1' in opts: |
|
528 | 529 | newversionflags = REVLOGV1 | FLAG_INLINE_DATA |
|
529 | 530 | if b'generaldelta' in opts: |
|
530 | 531 | newversionflags |= FLAG_GENERALDELTA |
|
531 | 532 | elif b'revlogv0' in self.opener.options: |
|
532 | 533 | newversionflags = REVLOGV0 |
|
533 | 534 | else: |
|
534 | 535 | newversionflags = REVLOG_DEFAULT_VERSION |
|
535 | 536 | |
|
536 | 537 | if b'chunkcachesize' in opts: |
|
537 | 538 | self._chunkcachesize = opts[b'chunkcachesize'] |
|
538 | 539 | if b'maxchainlen' in opts: |
|
539 | 540 | self._maxchainlen = opts[b'maxchainlen'] |
|
540 | 541 | if b'deltabothparents' in opts: |
|
541 | 542 | self._deltabothparents = opts[b'deltabothparents'] |
|
542 | 543 | self._lazydelta = bool(opts.get(b'lazydelta', True)) |
|
543 | 544 | self._lazydeltabase = False |
|
544 | 545 | if self._lazydelta: |
|
545 | 546 | self._lazydeltabase = bool(opts.get(b'lazydeltabase', False)) |
|
546 | 547 | if b'compengine' in opts: |
|
547 | 548 | self._compengine = opts[b'compengine'] |
|
548 | 549 | if b'zlib.level' in opts: |
|
549 | 550 | self._compengineopts[b'zlib.level'] = opts[b'zlib.level'] |
|
550 | 551 | if b'zstd.level' in opts: |
|
551 | 552 | self._compengineopts[b'zstd.level'] = opts[b'zstd.level'] |
|
552 | 553 | if b'maxdeltachainspan' in opts: |
|
553 | 554 | self._maxdeltachainspan = opts[b'maxdeltachainspan'] |
|
554 | 555 | if self._mmaplargeindex and b'mmapindexthreshold' in opts: |
|
555 | 556 | mmapindexthreshold = opts[b'mmapindexthreshold'] |
|
556 | 557 | self.hassidedata = bool(opts.get(b'side-data', False)) |
|
557 | 558 | self._sparserevlog = bool(opts.get(b'sparse-revlog', False)) |
|
558 | 559 | withsparseread = bool(opts.get(b'with-sparse-read', False)) |
|
559 | 560 | # sparse-revlog forces sparse-read |
|
560 | 561 | self._withsparseread = self._sparserevlog or withsparseread |
|
561 | 562 | if b'sparse-read-density-threshold' in opts: |
|
562 | 563 | self._srdensitythreshold = opts[b'sparse-read-density-threshold'] |
|
563 | 564 | if b'sparse-read-min-gap-size' in opts: |
|
564 | 565 | self._srmingapsize = opts[b'sparse-read-min-gap-size'] |
|
565 | 566 | if opts.get(b'enableellipsis'): |
|
566 | 567 | self._flagprocessors[REVIDX_ELLIPSIS] = ellipsisprocessor |
|
567 | 568 | |
|
568 | 569 | # revlog v0 doesn't have flag processors |
|
569 | 570 | for flag, processor in pycompat.iteritems( |
|
570 | 571 | opts.get(b'flagprocessors', {}) |
|
571 | 572 | ): |
|
572 | 573 | flagutil.insertflagprocessor(flag, processor, self._flagprocessors) |
|
573 | 574 | |
|
574 | 575 | if self._chunkcachesize <= 0: |
|
575 | 576 | raise error.RevlogError( |
|
576 | 577 | _(b'revlog chunk cache size %r is not greater than 0') |
|
577 | 578 | % self._chunkcachesize |
|
578 | 579 | ) |
|
579 | 580 | elif self._chunkcachesize & (self._chunkcachesize - 1): |
|
580 | 581 | raise error.RevlogError( |
|
581 | 582 | _(b'revlog chunk cache size %r is not a power of 2') |
|
582 | 583 | % self._chunkcachesize |
|
583 | 584 | ) |
|
584 | 585 | |
|
585 | 586 | indexdata = b'' |
|
586 | 587 | self._initempty = True |
|
587 | 588 | try: |
|
588 | 589 | with self._indexfp() as f: |
|
589 | 590 | if ( |
|
590 | 591 | mmapindexthreshold is not None |
|
591 | 592 | and self.opener.fstat(f).st_size >= mmapindexthreshold |
|
592 | 593 | ): |
|
593 | 594 | # TODO: should .close() to release resources without |
|
594 | 595 | # relying on Python GC |
|
595 | 596 | indexdata = util.buffer(util.mmapread(f)) |
|
596 | 597 | else: |
|
597 | 598 | indexdata = f.read() |
|
598 | 599 | if len(indexdata) > 0: |
|
599 | 600 | versionflags = versionformat_unpack(indexdata[:4])[0] |
|
600 | 601 | self._initempty = False |
|
601 | 602 | else: |
|
602 | 603 | versionflags = newversionflags |
|
603 | 604 | except IOError as inst: |
|
604 | 605 | if inst.errno != errno.ENOENT: |
|
605 | 606 | raise |
|
606 | 607 | |
|
607 | 608 | versionflags = newversionflags |
|
608 | 609 | |
|
609 | 610 | self.version = versionflags |
|
610 | 611 | |
|
611 | 612 | flags = versionflags & ~0xFFFF |
|
612 | 613 | fmt = versionflags & 0xFFFF |
|
613 | 614 | |
|
614 | 615 | if fmt == REVLOGV0: |
|
615 | 616 | if flags: |
|
616 | 617 | raise error.RevlogError( |
|
617 | 618 | _(b'unknown flags (%#04x) in version %d revlog %s') |
|
618 | 619 | % (flags >> 16, fmt, self.indexfile) |
|
619 | 620 | ) |
|
620 | 621 | |
|
621 | 622 | self._inline = False |
|
622 | 623 | self._generaldelta = False |
|
623 | 624 | |
|
624 | 625 | elif fmt == REVLOGV1: |
|
625 | 626 | if flags & ~REVLOGV1_FLAGS: |
|
626 | 627 | raise error.RevlogError( |
|
627 | 628 | _(b'unknown flags (%#04x) in version %d revlog %s') |
|
628 | 629 | % (flags >> 16, fmt, self.indexfile) |
|
629 | 630 | ) |
|
630 | 631 | |
|
631 | 632 | self._inline = versionflags & FLAG_INLINE_DATA |
|
632 | 633 | self._generaldelta = versionflags & FLAG_GENERALDELTA |
|
633 | 634 | |
|
634 | 635 | elif fmt == REVLOGV2: |
|
635 | 636 | if flags & ~REVLOGV2_FLAGS: |
|
636 | 637 | raise error.RevlogError( |
|
637 | 638 | _(b'unknown flags (%#04x) in version %d revlog %s') |
|
638 | 639 | % (flags >> 16, fmt, self.indexfile) |
|
639 | 640 | ) |
|
640 | 641 | |
|
641 | 642 | self._inline = versionflags & FLAG_INLINE_DATA |
|
642 | 643 | # generaldelta implied by version 2 revlogs. |
|
643 | 644 | self._generaldelta = True |
|
644 | 645 | |
|
645 | 646 | else: |
|
646 | 647 | raise error.RevlogError( |
|
647 | 648 | _(b'unknown version (%d) in revlog %s') % (fmt, self.indexfile) |
|
648 | 649 | ) |
|
649 | 650 | # sparse-revlog can't be on without general-delta (issue6056) |
|
650 | 651 | if not self._generaldelta: |
|
651 | 652 | self._sparserevlog = False |
|
652 | 653 | |
|
653 | 654 | self._storedeltachains = True |
|
654 | 655 | |
|
655 | 656 | devel_nodemap = ( |
|
656 | 657 | self.nodemap_file |
|
657 | 658 | and opts.get(b'devel-force-nodemap', False) |
|
658 | 659 | and NodemapRevlogIO is not None |
|
659 | 660 | ) |
|
660 | 661 | |
|
661 | 662 | use_rust_index = False |
|
662 | 663 | if rustrevlog is not None: |
|
663 | 664 | if self.nodemap_file is not None: |
|
664 | 665 | use_rust_index = True |
|
665 | 666 | else: |
|
666 | 667 | use_rust_index = self.opener.options.get(b'rust.index') |
|
667 | 668 | |
|
668 | 669 | self._io = revlogio() |
|
669 | 670 | if self.version == REVLOGV0: |
|
670 | 671 | self._io = revlogoldio() |
|
671 | 672 | elif fmt == REVLOGV2: |
|
672 | 673 | self._io = revlogv2io() |
|
673 | 674 | elif devel_nodemap: |
|
674 | 675 | self._io = NodemapRevlogIO() |
|
675 | 676 | elif use_rust_index: |
|
676 | 677 | self._io = rustrevlogio() |
|
677 | 678 | try: |
|
678 | 679 | d = self._io.parseindex(indexdata, self._inline) |
|
679 | 680 | index, _chunkcache = d |
|
680 | 681 | use_nodemap = ( |
|
681 | 682 | not self._inline |
|
682 | 683 | and self.nodemap_file is not None |
|
683 | 684 | and util.safehasattr(index, 'update_nodemap_data') |
|
684 | 685 | ) |
|
685 | 686 | if use_nodemap: |
|
686 | 687 | nodemap_data = nodemaputil.persisted_data(self) |
|
687 | 688 | if nodemap_data is not None: |
|
688 | 689 | docket = nodemap_data[0] |
|
689 | 690 | if ( |
|
690 | 691 | len(d[0]) > docket.tip_rev |
|
691 | 692 | and d[0][docket.tip_rev][7] == docket.tip_node |
|
692 | 693 | ): |
|
693 | 694 | # no changelog tampering |
|
694 | 695 | self._nodemap_docket = docket |
|
695 | 696 | index.update_nodemap_data(*nodemap_data) |
|
696 | 697 | except (ValueError, IndexError): |
|
697 | 698 | raise error.RevlogError( |
|
698 | 699 | _(b"index %s is corrupted") % self.indexfile |
|
699 | 700 | ) |
|
700 | 701 | self.index, self._chunkcache = d |
|
701 | 702 | if not self._chunkcache: |
|
702 | 703 | self._chunkclear() |
|
703 | 704 | # revnum -> (chain-length, sum-delta-length) |
|
704 | 705 | self._chaininfocache = util.lrucachedict(500) |
|
705 | 706 | # revlog header -> revlog compressor |
|
706 | 707 | self._decompressors = {} |
|
707 | 708 | |
|
708 | 709 | @util.propertycache |
|
709 | 710 | def _compressor(self): |
|
710 | 711 | engine = util.compengines[self._compengine] |
|
711 | 712 | return engine.revlogcompressor(self._compengineopts) |
|
712 | 713 | |
|
713 | 714 | def _indexfp(self, mode=b'r'): |
|
714 | 715 | """file object for the revlog's index file""" |
|
715 | 716 | args = {'mode': mode} |
|
716 | 717 | if mode != b'r': |
|
717 | 718 | args['checkambig'] = self._checkambig |
|
718 | 719 | if mode == b'w': |
|
719 | 720 | args['atomictemp'] = True |
|
720 | 721 | return self.opener(self.indexfile, **args) |
|
721 | 722 | |
|
722 | 723 | def _datafp(self, mode=b'r'): |
|
723 | 724 | """file object for the revlog's data file""" |
|
724 | 725 | return self.opener(self.datafile, mode=mode) |
|
725 | 726 | |
|
726 | 727 | @contextlib.contextmanager |
|
727 | 728 | def _datareadfp(self, existingfp=None): |
|
728 | 729 | """file object suitable to read data""" |
|
729 | 730 | # Use explicit file handle, if given. |
|
730 | 731 | if existingfp is not None: |
|
731 | 732 | yield existingfp |
|
732 | 733 | |
|
733 | 734 | # Use a file handle being actively used for writes, if available. |
|
734 | 735 | # There is some danger to doing this because reads will seek the |
|
735 | 736 | # file. However, _writeentry() performs a SEEK_END before all writes, |
|
736 | 737 | # so we should be safe. |
|
737 | 738 | elif self._writinghandles: |
|
738 | 739 | if self._inline: |
|
739 | 740 | yield self._writinghandles[0] |
|
740 | 741 | else: |
|
741 | 742 | yield self._writinghandles[1] |
|
742 | 743 | |
|
743 | 744 | # Otherwise open a new file handle. |
|
744 | 745 | else: |
|
745 | 746 | if self._inline: |
|
746 | 747 | func = self._indexfp |
|
747 | 748 | else: |
|
748 | 749 | func = self._datafp |
|
749 | 750 | with func() as fp: |
|
750 | 751 | yield fp |
|
751 | 752 | |
|
752 | 753 | def tiprev(self): |
|
753 | 754 | return len(self.index) - 1 |
|
754 | 755 | |
|
755 | 756 | def tip(self): |
|
756 | 757 | return self.node(self.tiprev()) |
|
757 | 758 | |
|
758 | 759 | def __contains__(self, rev): |
|
759 | 760 | return 0 <= rev < len(self) |
|
760 | 761 | |
|
761 | 762 | def __len__(self): |
|
762 | 763 | return len(self.index) |
|
763 | 764 | |
|
764 | 765 | def __iter__(self): |
|
765 | 766 | return iter(pycompat.xrange(len(self))) |
|
766 | 767 | |
|
767 | 768 | def revs(self, start=0, stop=None): |
|
768 | 769 | """iterate over all rev in this revlog (from start to stop)""" |
|
769 | 770 | return storageutil.iterrevs(len(self), start=start, stop=stop) |
|
770 | 771 | |
|
771 | 772 | @property |
|
772 | 773 | def nodemap(self): |
|
773 | 774 | msg = ( |
|
774 | 775 | b"revlog.nodemap is deprecated, " |
|
775 | 776 | b"use revlog.index.[has_node|rev|get_rev]" |
|
776 | 777 | ) |
|
777 | 778 | util.nouideprecwarn(msg, b'5.3', stacklevel=2) |
|
778 | 779 | return self.index.nodemap |
|
779 | 780 | |
|
780 | 781 | @property |
|
781 | 782 | def _nodecache(self): |
|
782 | 783 | msg = b"revlog._nodecache is deprecated, use revlog.index.nodemap" |
|
783 | 784 | util.nouideprecwarn(msg, b'5.3', stacklevel=2) |
|
784 | 785 | return self.index.nodemap |
|
785 | 786 | |
|
786 | 787 | def hasnode(self, node): |
|
787 | 788 | try: |
|
788 | 789 | self.rev(node) |
|
789 | 790 | return True |
|
790 | 791 | except KeyError: |
|
791 | 792 | return False |
|
792 | 793 | |
|
793 | 794 | def candelta(self, baserev, rev): |
|
794 | 795 | """whether two revisions (baserev, rev) can be delta-ed or not""" |
|
795 | 796 | # Disable delta if either rev requires a content-changing flag |
|
796 | 797 | # processor (ex. LFS). This is because such flag processor can alter |
|
797 | 798 | # the rawtext content that the delta will be based on, and two clients |
|
798 | 799 | # could have a same revlog node with different flags (i.e. different |
|
799 | 800 | # rawtext contents) and the delta could be incompatible. |
|
800 | 801 | if (self.flags(baserev) & REVIDX_RAWTEXT_CHANGING_FLAGS) or ( |
|
801 | 802 | self.flags(rev) & REVIDX_RAWTEXT_CHANGING_FLAGS |
|
802 | 803 | ): |
|
803 | 804 | return False |
|
804 | 805 | return True |
|
805 | 806 | |
|
806 | 807 | def update_caches(self, transaction): |
|
807 | 808 | if self.nodemap_file is not None: |
|
808 | 809 | if transaction is None: |
|
809 | 810 | nodemaputil.update_persistent_nodemap(self) |
|
810 | 811 | else: |
|
811 | 812 | nodemaputil.setup_persistent_nodemap(transaction, self) |
|
812 | 813 | |
|
813 | 814 | def clearcaches(self): |
|
814 | 815 | self._revisioncache = None |
|
815 | 816 | self._chainbasecache.clear() |
|
816 | 817 | self._chunkcache = (0, b'') |
|
817 | 818 | self._pcache = {} |
|
818 | 819 | self._nodemap_docket = None |
|
819 | 820 | self.index.clearcaches() |
|
820 | 821 | # The python code is the one responsible for validating the docket, we |
|
821 | 822 | # end up having to refresh it here. |
|
822 | 823 | use_nodemap = ( |
|
823 | 824 | not self._inline |
|
824 | 825 | and self.nodemap_file is not None |
|
825 | 826 | and util.safehasattr(self.index, 'update_nodemap_data') |
|
826 | 827 | ) |
|
827 | 828 | if use_nodemap: |
|
828 | 829 | nodemap_data = nodemaputil.persisted_data(self) |
|
829 | 830 | if nodemap_data is not None: |
|
830 | 831 | self._nodemap_docket = nodemap_data[0] |
|
831 | 832 | self.index.update_nodemap_data(*nodemap_data) |
|
832 | 833 | |
|
833 | 834 | def rev(self, node): |
|
834 | 835 | try: |
|
835 | 836 | return self.index.rev(node) |
|
836 | 837 | except TypeError: |
|
837 | 838 | raise |
|
838 | 839 | except error.RevlogError: |
|
839 | 840 | # parsers.c radix tree lookup failed |
|
840 | 841 | if node == wdirid or node in wdirfilenodeids: |
|
841 | 842 | raise error.WdirUnsupported |
|
842 | 843 | raise error.LookupError(node, self.indexfile, _(b'no node')) |
|
843 | 844 | |
|
844 | 845 | # Accessors for index entries. |
|
845 | 846 | |
|
846 | 847 | # First tuple entry is 8 bytes. First 6 bytes are offset. Last 2 bytes |
|
847 | 848 | # are flags. |
|
848 | 849 | def start(self, rev): |
|
849 | 850 | return int(self.index[rev][0] >> 16) |
|
850 | 851 | |
|
851 | 852 | def flags(self, rev): |
|
852 | 853 | return self.index[rev][0] & 0xFFFF |
|
853 | 854 | |
|
854 | 855 | def length(self, rev): |
|
855 | 856 | return self.index[rev][1] |
|
856 | 857 | |
|
857 | 858 | def sidedata_length(self, rev): |
|
858 | 859 | if self.version & 0xFFFF != REVLOGV2: |
|
859 | 860 | return 0 |
|
860 | 861 | return self.index[rev][9] |
|
861 | 862 | |
|
862 | 863 | def rawsize(self, rev): |
|
863 | 864 | """return the length of the uncompressed text for a given revision""" |
|
864 | 865 | l = self.index[rev][2] |
|
865 | 866 | if l >= 0: |
|
866 | 867 | return l |
|
867 | 868 | |
|
868 | 869 | t = self.rawdata(rev) |
|
869 | 870 | return len(t) |
|
870 | 871 | |
|
871 | 872 | def size(self, rev): |
|
872 | 873 | """length of non-raw text (processed by a "read" flag processor)""" |
|
873 | 874 | # fast path: if no "read" flag processor could change the content, |
|
874 | 875 | # size is rawsize. note: ELLIPSIS is known to not change the content. |
|
875 | 876 | flags = self.flags(rev) |
|
876 | 877 | if flags & (flagutil.REVIDX_KNOWN_FLAGS ^ REVIDX_ELLIPSIS) == 0: |
|
877 | 878 | return self.rawsize(rev) |
|
878 | 879 | |
|
879 | 880 | return len(self.revision(rev, raw=False)) |
|
880 | 881 | |
|
881 | 882 | def chainbase(self, rev): |
|
882 | 883 | base = self._chainbasecache.get(rev) |
|
883 | 884 | if base is not None: |
|
884 | 885 | return base |
|
885 | 886 | |
|
886 | 887 | index = self.index |
|
887 | 888 | iterrev = rev |
|
888 | 889 | base = index[iterrev][3] |
|
889 | 890 | while base != iterrev: |
|
890 | 891 | iterrev = base |
|
891 | 892 | base = index[iterrev][3] |
|
892 | 893 | |
|
893 | 894 | self._chainbasecache[rev] = base |
|
894 | 895 | return base |
|
895 | 896 | |
|
896 | 897 | def linkrev(self, rev): |
|
897 | 898 | return self.index[rev][4] |
|
898 | 899 | |
|
899 | 900 | def parentrevs(self, rev): |
|
900 | 901 | try: |
|
901 | 902 | entry = self.index[rev] |
|
902 | 903 | except IndexError: |
|
903 | 904 | if rev == wdirrev: |
|
904 | 905 | raise error.WdirUnsupported |
|
905 | 906 | raise |
|
906 | 907 | |
|
907 | 908 | return entry[5], entry[6] |
|
908 | 909 | |
|
909 | 910 | # fast parentrevs(rev) where rev isn't filtered |
|
910 | 911 | _uncheckedparentrevs = parentrevs |
|
911 | 912 | |
|
912 | 913 | def node(self, rev): |
|
913 | 914 | try: |
|
914 | 915 | return self.index[rev][7] |
|
915 | 916 | except IndexError: |
|
916 | 917 | if rev == wdirrev: |
|
917 | 918 | raise error.WdirUnsupported |
|
918 | 919 | raise |
|
919 | 920 | |
|
920 | 921 | # Derived from index values. |
|
921 | 922 | |
|
922 | 923 | def end(self, rev): |
|
923 | 924 | return self.start(rev) + self.length(rev) |
|
924 | 925 | |
|
925 | 926 | def parents(self, node): |
|
926 | 927 | i = self.index |
|
927 | 928 | d = i[self.rev(node)] |
|
928 | 929 | return i[d[5]][7], i[d[6]][7] # map revisions to nodes inline |
|
929 | 930 | |
|
930 | 931 | def chainlen(self, rev): |
|
931 | 932 | return self._chaininfo(rev)[0] |
|
932 | 933 | |
|
933 | 934 | def _chaininfo(self, rev): |
|
934 | 935 | chaininfocache = self._chaininfocache |
|
935 | 936 | if rev in chaininfocache: |
|
936 | 937 | return chaininfocache[rev] |
|
937 | 938 | index = self.index |
|
938 | 939 | generaldelta = self._generaldelta |
|
939 | 940 | iterrev = rev |
|
940 | 941 | e = index[iterrev] |
|
941 | 942 | clen = 0 |
|
942 | 943 | compresseddeltalen = 0 |
|
943 | 944 | while iterrev != e[3]: |
|
944 | 945 | clen += 1 |
|
945 | 946 | compresseddeltalen += e[1] |
|
946 | 947 | if generaldelta: |
|
947 | 948 | iterrev = e[3] |
|
948 | 949 | else: |
|
949 | 950 | iterrev -= 1 |
|
950 | 951 | if iterrev in chaininfocache: |
|
951 | 952 | t = chaininfocache[iterrev] |
|
952 | 953 | clen += t[0] |
|
953 | 954 | compresseddeltalen += t[1] |
|
954 | 955 | break |
|
955 | 956 | e = index[iterrev] |
|
956 | 957 | else: |
|
957 | 958 | # Add text length of base since decompressing that also takes |
|
958 | 959 | # work. For cache hits the length is already included. |
|
959 | 960 | compresseddeltalen += e[1] |
|
960 | 961 | r = (clen, compresseddeltalen) |
|
961 | 962 | chaininfocache[rev] = r |
|
962 | 963 | return r |
|
963 | 964 | |
|
964 | 965 | def _deltachain(self, rev, stoprev=None): |
|
965 | 966 | """Obtain the delta chain for a revision. |
|
966 | 967 | |
|
967 | 968 | ``stoprev`` specifies a revision to stop at. If not specified, we |
|
968 | 969 | stop at the base of the chain. |
|
969 | 970 | |
|
970 | 971 | Returns a 2-tuple of (chain, stopped) where ``chain`` is a list of |
|
971 | 972 | revs in ascending order and ``stopped`` is a bool indicating whether |
|
972 | 973 | ``stoprev`` was hit. |
|
973 | 974 | """ |
|
974 | 975 | # Try C implementation. |
|
975 | 976 | try: |
|
976 | 977 | return self.index.deltachain(rev, stoprev, self._generaldelta) |
|
977 | 978 | except AttributeError: |
|
978 | 979 | pass |
|
979 | 980 | |
|
980 | 981 | chain = [] |
|
981 | 982 | |
|
982 | 983 | # Alias to prevent attribute lookup in tight loop. |
|
983 | 984 | index = self.index |
|
984 | 985 | generaldelta = self._generaldelta |
|
985 | 986 | |
|
986 | 987 | iterrev = rev |
|
987 | 988 | e = index[iterrev] |
|
988 | 989 | while iterrev != e[3] and iterrev != stoprev: |
|
989 | 990 | chain.append(iterrev) |
|
990 | 991 | if generaldelta: |
|
991 | 992 | iterrev = e[3] |
|
992 | 993 | else: |
|
993 | 994 | iterrev -= 1 |
|
994 | 995 | e = index[iterrev] |
|
995 | 996 | |
|
996 | 997 | if iterrev == stoprev: |
|
997 | 998 | stopped = True |
|
998 | 999 | else: |
|
999 | 1000 | chain.append(iterrev) |
|
1000 | 1001 | stopped = False |
|
1001 | 1002 | |
|
1002 | 1003 | chain.reverse() |
|
1003 | 1004 | return chain, stopped |
|
1004 | 1005 | |
|
1005 | 1006 | def ancestors(self, revs, stoprev=0, inclusive=False): |
|
1006 | 1007 | """Generate the ancestors of 'revs' in reverse revision order. |
|
1007 | 1008 | Does not generate revs lower than stoprev. |
|
1008 | 1009 | |
|
1009 | 1010 | See the documentation for ancestor.lazyancestors for more details.""" |
|
1010 | 1011 | |
|
1011 | 1012 | # first, make sure start revisions aren't filtered |
|
1012 | 1013 | revs = list(revs) |
|
1013 | 1014 | checkrev = self.node |
|
1014 | 1015 | for r in revs: |
|
1015 | 1016 | checkrev(r) |
|
1016 | 1017 | # and we're sure ancestors aren't filtered as well |
|
1017 | 1018 | |
|
1018 | 1019 | if rustancestor is not None: |
|
1019 | 1020 | lazyancestors = rustancestor.LazyAncestors |
|
1020 | 1021 | arg = self.index |
|
1021 | 1022 | else: |
|
1022 | 1023 | lazyancestors = ancestor.lazyancestors |
|
1023 | 1024 | arg = self._uncheckedparentrevs |
|
1024 | 1025 | return lazyancestors(arg, revs, stoprev=stoprev, inclusive=inclusive) |
|
1025 | 1026 | |
|
1026 | 1027 | def descendants(self, revs): |
|
1027 | 1028 | return dagop.descendantrevs(revs, self.revs, self.parentrevs) |
|
1028 | 1029 | |
|
1029 | 1030 | def findcommonmissing(self, common=None, heads=None): |
|
1030 | 1031 | """Return a tuple of the ancestors of common and the ancestors of heads |
|
1031 | 1032 | that are not ancestors of common. In revset terminology, we return the |
|
1032 | 1033 | tuple: |
|
1033 | 1034 | |
|
1034 | 1035 | ::common, (::heads) - (::common) |
|
1035 | 1036 | |
|
1036 | 1037 | The list is sorted by revision number, meaning it is |
|
1037 | 1038 | topologically sorted. |
|
1038 | 1039 | |
|
1039 | 1040 | 'heads' and 'common' are both lists of node IDs. If heads is |
|
1040 | 1041 | not supplied, uses all of the revlog's heads. If common is not |
|
1041 | 1042 | supplied, uses nullid.""" |
|
1042 | 1043 | if common is None: |
|
1043 | 1044 | common = [nullid] |
|
1044 | 1045 | if heads is None: |
|
1045 | 1046 | heads = self.heads() |
|
1046 | 1047 | |
|
1047 | 1048 | common = [self.rev(n) for n in common] |
|
1048 | 1049 | heads = [self.rev(n) for n in heads] |
|
1049 | 1050 | |
|
1050 | 1051 | # we want the ancestors, but inclusive |
|
1051 | 1052 | class lazyset(object): |
|
1052 | 1053 | def __init__(self, lazyvalues): |
|
1053 | 1054 | self.addedvalues = set() |
|
1054 | 1055 | self.lazyvalues = lazyvalues |
|
1055 | 1056 | |
|
1056 | 1057 | def __contains__(self, value): |
|
1057 | 1058 | return value in self.addedvalues or value in self.lazyvalues |
|
1058 | 1059 | |
|
1059 | 1060 | def __iter__(self): |
|
1060 | 1061 | added = self.addedvalues |
|
1061 | 1062 | for r in added: |
|
1062 | 1063 | yield r |
|
1063 | 1064 | for r in self.lazyvalues: |
|
1064 | 1065 | if not r in added: |
|
1065 | 1066 | yield r |
|
1066 | 1067 | |
|
1067 | 1068 | def add(self, value): |
|
1068 | 1069 | self.addedvalues.add(value) |
|
1069 | 1070 | |
|
1070 | 1071 | def update(self, values): |
|
1071 | 1072 | self.addedvalues.update(values) |
|
1072 | 1073 | |
|
1073 | 1074 | has = lazyset(self.ancestors(common)) |
|
1074 | 1075 | has.add(nullrev) |
|
1075 | 1076 | has.update(common) |
|
1076 | 1077 | |
|
1077 | 1078 | # take all ancestors from heads that aren't in has |
|
1078 | 1079 | missing = set() |
|
1079 | 1080 | visit = collections.deque(r for r in heads if r not in has) |
|
1080 | 1081 | while visit: |
|
1081 | 1082 | r = visit.popleft() |
|
1082 | 1083 | if r in missing: |
|
1083 | 1084 | continue |
|
1084 | 1085 | else: |
|
1085 | 1086 | missing.add(r) |
|
1086 | 1087 | for p in self.parentrevs(r): |
|
1087 | 1088 | if p not in has: |
|
1088 | 1089 | visit.append(p) |
|
1089 | 1090 | missing = list(missing) |
|
1090 | 1091 | missing.sort() |
|
1091 | 1092 | return has, [self.node(miss) for miss in missing] |
|
1092 | 1093 | |
|
1093 | 1094 | def incrementalmissingrevs(self, common=None): |
|
1094 | 1095 | """Return an object that can be used to incrementally compute the |
|
1095 | 1096 | revision numbers of the ancestors of arbitrary sets that are not |
|
1096 | 1097 | ancestors of common. This is an ancestor.incrementalmissingancestors |
|
1097 | 1098 | object. |
|
1098 | 1099 | |
|
1099 | 1100 | 'common' is a list of revision numbers. If common is not supplied, uses |
|
1100 | 1101 | nullrev. |
|
1101 | 1102 | """ |
|
1102 | 1103 | if common is None: |
|
1103 | 1104 | common = [nullrev] |
|
1104 | 1105 | |
|
1105 | 1106 | if rustancestor is not None: |
|
1106 | 1107 | return rustancestor.MissingAncestors(self.index, common) |
|
1107 | 1108 | return ancestor.incrementalmissingancestors(self.parentrevs, common) |
|
1108 | 1109 | |
|
1109 | 1110 | def findmissingrevs(self, common=None, heads=None): |
|
1110 | 1111 | """Return the revision numbers of the ancestors of heads that |
|
1111 | 1112 | are not ancestors of common. |
|
1112 | 1113 | |
|
1113 | 1114 | More specifically, return a list of revision numbers corresponding to |
|
1114 | 1115 | nodes N such that every N satisfies the following constraints: |
|
1115 | 1116 | |
|
1116 | 1117 | 1. N is an ancestor of some node in 'heads' |
|
1117 | 1118 | 2. N is not an ancestor of any node in 'common' |
|
1118 | 1119 | |
|
1119 | 1120 | The list is sorted by revision number, meaning it is |
|
1120 | 1121 | topologically sorted. |
|
1121 | 1122 | |
|
1122 | 1123 | 'heads' and 'common' are both lists of revision numbers. If heads is |
|
1123 | 1124 | not supplied, uses all of the revlog's heads. If common is not |
|
1124 | 1125 | supplied, uses nullid.""" |
|
1125 | 1126 | if common is None: |
|
1126 | 1127 | common = [nullrev] |
|
1127 | 1128 | if heads is None: |
|
1128 | 1129 | heads = self.headrevs() |
|
1129 | 1130 | |
|
1130 | 1131 | inc = self.incrementalmissingrevs(common=common) |
|
1131 | 1132 | return inc.missingancestors(heads) |
|
1132 | 1133 | |
|
1133 | 1134 | def findmissing(self, common=None, heads=None): |
|
1134 | 1135 | """Return the ancestors of heads that are not ancestors of common. |
|
1135 | 1136 | |
|
1136 | 1137 | More specifically, return a list of nodes N such that every N |
|
1137 | 1138 | satisfies the following constraints: |
|
1138 | 1139 | |
|
1139 | 1140 | 1. N is an ancestor of some node in 'heads' |
|
1140 | 1141 | 2. N is not an ancestor of any node in 'common' |
|
1141 | 1142 | |
|
1142 | 1143 | The list is sorted by revision number, meaning it is |
|
1143 | 1144 | topologically sorted. |
|
1144 | 1145 | |
|
1145 | 1146 | 'heads' and 'common' are both lists of node IDs. If heads is |
|
1146 | 1147 | not supplied, uses all of the revlog's heads. If common is not |
|
1147 | 1148 | supplied, uses nullid.""" |
|
1148 | 1149 | if common is None: |
|
1149 | 1150 | common = [nullid] |
|
1150 | 1151 | if heads is None: |
|
1151 | 1152 | heads = self.heads() |
|
1152 | 1153 | |
|
1153 | 1154 | common = [self.rev(n) for n in common] |
|
1154 | 1155 | heads = [self.rev(n) for n in heads] |
|
1155 | 1156 | |
|
1156 | 1157 | inc = self.incrementalmissingrevs(common=common) |
|
1157 | 1158 | return [self.node(r) for r in inc.missingancestors(heads)] |
|
1158 | 1159 | |
|
1159 | 1160 | def nodesbetween(self, roots=None, heads=None): |
|
1160 | 1161 | """Return a topological path from 'roots' to 'heads'. |
|
1161 | 1162 | |
|
1162 | 1163 | Return a tuple (nodes, outroots, outheads) where 'nodes' is a |
|
1163 | 1164 | topologically sorted list of all nodes N that satisfy both of |
|
1164 | 1165 | these constraints: |
|
1165 | 1166 | |
|
1166 | 1167 | 1. N is a descendant of some node in 'roots' |
|
1167 | 1168 | 2. N is an ancestor of some node in 'heads' |
|
1168 | 1169 | |
|
1169 | 1170 | Every node is considered to be both a descendant and an ancestor |
|
1170 | 1171 | of itself, so every reachable node in 'roots' and 'heads' will be |
|
1171 | 1172 | included in 'nodes'. |
|
1172 | 1173 | |
|
1173 | 1174 | 'outroots' is the list of reachable nodes in 'roots', i.e., the |
|
1174 | 1175 | subset of 'roots' that is returned in 'nodes'. Likewise, |
|
1175 | 1176 | 'outheads' is the subset of 'heads' that is also in 'nodes'. |
|
1176 | 1177 | |
|
1177 | 1178 | 'roots' and 'heads' are both lists of node IDs. If 'roots' is |
|
1178 | 1179 | unspecified, uses nullid as the only root. If 'heads' is |
|
1179 | 1180 | unspecified, uses list of all of the revlog's heads.""" |
|
1180 | 1181 | nonodes = ([], [], []) |
|
1181 | 1182 | if roots is not None: |
|
1182 | 1183 | roots = list(roots) |
|
1183 | 1184 | if not roots: |
|
1184 | 1185 | return nonodes |
|
1185 | 1186 | lowestrev = min([self.rev(n) for n in roots]) |
|
1186 | 1187 | else: |
|
1187 | 1188 | roots = [nullid] # Everybody's a descendant of nullid |
|
1188 | 1189 | lowestrev = nullrev |
|
1189 | 1190 | if (lowestrev == nullrev) and (heads is None): |
|
1190 | 1191 | # We want _all_ the nodes! |
|
1191 | 1192 | return ([self.node(r) for r in self], [nullid], list(self.heads())) |
|
1192 | 1193 | if heads is None: |
|
1193 | 1194 | # All nodes are ancestors, so the latest ancestor is the last |
|
1194 | 1195 | # node. |
|
1195 | 1196 | highestrev = len(self) - 1 |
|
1196 | 1197 | # Set ancestors to None to signal that every node is an ancestor. |
|
1197 | 1198 | ancestors = None |
|
1198 | 1199 | # Set heads to an empty dictionary for later discovery of heads |
|
1199 | 1200 | heads = {} |
|
1200 | 1201 | else: |
|
1201 | 1202 | heads = list(heads) |
|
1202 | 1203 | if not heads: |
|
1203 | 1204 | return nonodes |
|
1204 | 1205 | ancestors = set() |
|
1205 | 1206 | # Turn heads into a dictionary so we can remove 'fake' heads. |
|
1206 | 1207 | # Also, later we will be using it to filter out the heads we can't |
|
1207 | 1208 | # find from roots. |
|
1208 | 1209 | heads = dict.fromkeys(heads, False) |
|
1209 | 1210 | # Start at the top and keep marking parents until we're done. |
|
1210 | 1211 | nodestotag = set(heads) |
|
1211 | 1212 | # Remember where the top was so we can use it as a limit later. |
|
1212 | 1213 | highestrev = max([self.rev(n) for n in nodestotag]) |
|
1213 | 1214 | while nodestotag: |
|
1214 | 1215 | # grab a node to tag |
|
1215 | 1216 | n = nodestotag.pop() |
|
1216 | 1217 | # Never tag nullid |
|
1217 | 1218 | if n == nullid: |
|
1218 | 1219 | continue |
|
1219 | 1220 | # A node's revision number represents its place in a |
|
1220 | 1221 | # topologically sorted list of nodes. |
|
1221 | 1222 | r = self.rev(n) |
|
1222 | 1223 | if r >= lowestrev: |
|
1223 | 1224 | if n not in ancestors: |
|
1224 | 1225 | # If we are possibly a descendant of one of the roots |
|
1225 | 1226 | # and we haven't already been marked as an ancestor |
|
1226 | 1227 | ancestors.add(n) # Mark as ancestor |
|
1227 | 1228 | # Add non-nullid parents to list of nodes to tag. |
|
1228 | 1229 | nodestotag.update( |
|
1229 | 1230 | [p for p in self.parents(n) if p != nullid] |
|
1230 | 1231 | ) |
|
1231 | 1232 | elif n in heads: # We've seen it before, is it a fake head? |
|
1232 | 1233 | # So it is, real heads should not be the ancestors of |
|
1233 | 1234 | # any other heads. |
|
1234 | 1235 | heads.pop(n) |
|
1235 | 1236 | if not ancestors: |
|
1236 | 1237 | return nonodes |
|
1237 | 1238 | # Now that we have our set of ancestors, we want to remove any |
|
1238 | 1239 | # roots that are not ancestors. |
|
1239 | 1240 | |
|
1240 | 1241 | # If one of the roots was nullid, everything is included anyway. |
|
1241 | 1242 | if lowestrev > nullrev: |
|
1242 | 1243 | # But, since we weren't, let's recompute the lowest rev to not |
|
1243 | 1244 | # include roots that aren't ancestors. |
|
1244 | 1245 | |
|
1245 | 1246 | # Filter out roots that aren't ancestors of heads |
|
1246 | 1247 | roots = [root for root in roots if root in ancestors] |
|
1247 | 1248 | # Recompute the lowest revision |
|
1248 | 1249 | if roots: |
|
1249 | 1250 | lowestrev = min([self.rev(root) for root in roots]) |
|
1250 | 1251 | else: |
|
1251 | 1252 | # No more roots? Return empty list |
|
1252 | 1253 | return nonodes |
|
1253 | 1254 | else: |
|
1254 | 1255 | # We are descending from nullid, and don't need to care about |
|
1255 | 1256 | # any other roots. |
|
1256 | 1257 | lowestrev = nullrev |
|
1257 | 1258 | roots = [nullid] |
|
1258 | 1259 | # Transform our roots list into a set. |
|
1259 | 1260 | descendants = set(roots) |
|
1260 | 1261 | # Also, keep the original roots so we can filter out roots that aren't |
|
1261 | 1262 | # 'real' roots (i.e. are descended from other roots). |
|
1262 | 1263 | roots = descendants.copy() |
|
1263 | 1264 | # Our topologically sorted list of output nodes. |
|
1264 | 1265 | orderedout = [] |
|
1265 | 1266 | # Don't start at nullid since we don't want nullid in our output list, |
|
1266 | 1267 | # and if nullid shows up in descendants, empty parents will look like |
|
1267 | 1268 | # they're descendants. |
|
1268 | 1269 | for r in self.revs(start=max(lowestrev, 0), stop=highestrev + 1): |
|
1269 | 1270 | n = self.node(r) |
|
1270 | 1271 | isdescendant = False |
|
1271 | 1272 | if lowestrev == nullrev: # Everybody is a descendant of nullid |
|
1272 | 1273 | isdescendant = True |
|
1273 | 1274 | elif n in descendants: |
|
1274 | 1275 | # n is already a descendant |
|
1275 | 1276 | isdescendant = True |
|
1276 | 1277 | # This check only needs to be done here because all the roots |
|
1277 | 1278 | # will start being marked is descendants before the loop. |
|
1278 | 1279 | if n in roots: |
|
1279 | 1280 | # If n was a root, check if it's a 'real' root. |
|
1280 | 1281 | p = tuple(self.parents(n)) |
|
1281 | 1282 | # If any of its parents are descendants, it's not a root. |
|
1282 | 1283 | if (p[0] in descendants) or (p[1] in descendants): |
|
1283 | 1284 | roots.remove(n) |
|
1284 | 1285 | else: |
|
1285 | 1286 | p = tuple(self.parents(n)) |
|
1286 | 1287 | # A node is a descendant if either of its parents are |
|
1287 | 1288 | # descendants. (We seeded the dependents list with the roots |
|
1288 | 1289 | # up there, remember?) |
|
1289 | 1290 | if (p[0] in descendants) or (p[1] in descendants): |
|
1290 | 1291 | descendants.add(n) |
|
1291 | 1292 | isdescendant = True |
|
1292 | 1293 | if isdescendant and ((ancestors is None) or (n in ancestors)): |
|
1293 | 1294 | # Only include nodes that are both descendants and ancestors. |
|
1294 | 1295 | orderedout.append(n) |
|
1295 | 1296 | if (ancestors is not None) and (n in heads): |
|
1296 | 1297 | # We're trying to figure out which heads are reachable |
|
1297 | 1298 | # from roots. |
|
1298 | 1299 | # Mark this head as having been reached |
|
1299 | 1300 | heads[n] = True |
|
1300 | 1301 | elif ancestors is None: |
|
1301 | 1302 | # Otherwise, we're trying to discover the heads. |
|
1302 | 1303 | # Assume this is a head because if it isn't, the next step |
|
1303 | 1304 | # will eventually remove it. |
|
1304 | 1305 | heads[n] = True |
|
1305 | 1306 | # But, obviously its parents aren't. |
|
1306 | 1307 | for p in self.parents(n): |
|
1307 | 1308 | heads.pop(p, None) |
|
1308 | 1309 | heads = [head for head, flag in pycompat.iteritems(heads) if flag] |
|
1309 | 1310 | roots = list(roots) |
|
1310 | 1311 | assert orderedout |
|
1311 | 1312 | assert roots |
|
1312 | 1313 | assert heads |
|
1313 | 1314 | return (orderedout, roots, heads) |
|
1314 | 1315 | |
|
1315 | 1316 | def headrevs(self, revs=None): |
|
1316 | 1317 | if revs is None: |
|
1317 | 1318 | try: |
|
1318 | 1319 | return self.index.headrevs() |
|
1319 | 1320 | except AttributeError: |
|
1320 | 1321 | return self._headrevs() |
|
1321 | 1322 | if rustdagop is not None: |
|
1322 | 1323 | return rustdagop.headrevs(self.index, revs) |
|
1323 | 1324 | return dagop.headrevs(revs, self._uncheckedparentrevs) |
|
1324 | 1325 | |
|
1325 | 1326 | def computephases(self, roots): |
|
1326 | 1327 | return self.index.computephasesmapsets(roots) |
|
1327 | 1328 | |
|
1328 | 1329 | def _headrevs(self): |
|
1329 | 1330 | count = len(self) |
|
1330 | 1331 | if not count: |
|
1331 | 1332 | return [nullrev] |
|
1332 | 1333 | # we won't iter over filtered rev so nobody is a head at start |
|
1333 | 1334 | ishead = [0] * (count + 1) |
|
1334 | 1335 | index = self.index |
|
1335 | 1336 | for r in self: |
|
1336 | 1337 | ishead[r] = 1 # I may be an head |
|
1337 | 1338 | e = index[r] |
|
1338 | 1339 | ishead[e[5]] = ishead[e[6]] = 0 # my parent are not |
|
1339 | 1340 | return [r for r, val in enumerate(ishead) if val] |
|
1340 | 1341 | |
|
1341 | 1342 | def heads(self, start=None, stop=None): |
|
1342 | 1343 | """return the list of all nodes that have no children |
|
1343 | 1344 | |
|
1344 | 1345 | if start is specified, only heads that are descendants of |
|
1345 | 1346 | start will be returned |
|
1346 | 1347 | if stop is specified, it will consider all the revs from stop |
|
1347 | 1348 | as if they had no children |
|
1348 | 1349 | """ |
|
1349 | 1350 | if start is None and stop is None: |
|
1350 | 1351 | if not len(self): |
|
1351 | 1352 | return [nullid] |
|
1352 | 1353 | return [self.node(r) for r in self.headrevs()] |
|
1353 | 1354 | |
|
1354 | 1355 | if start is None: |
|
1355 | 1356 | start = nullrev |
|
1356 | 1357 | else: |
|
1357 | 1358 | start = self.rev(start) |
|
1358 | 1359 | |
|
1359 | 1360 | stoprevs = {self.rev(n) for n in stop or []} |
|
1360 | 1361 | |
|
1361 | 1362 | revs = dagop.headrevssubset( |
|
1362 | 1363 | self.revs, self.parentrevs, startrev=start, stoprevs=stoprevs |
|
1363 | 1364 | ) |
|
1364 | 1365 | |
|
1365 | 1366 | return [self.node(rev) for rev in revs] |
|
1366 | 1367 | |
|
1367 | 1368 | def children(self, node): |
|
1368 | 1369 | """find the children of a given node""" |
|
1369 | 1370 | c = [] |
|
1370 | 1371 | p = self.rev(node) |
|
1371 | 1372 | for r in self.revs(start=p + 1): |
|
1372 | 1373 | prevs = [pr for pr in self.parentrevs(r) if pr != nullrev] |
|
1373 | 1374 | if prevs: |
|
1374 | 1375 | for pr in prevs: |
|
1375 | 1376 | if pr == p: |
|
1376 | 1377 | c.append(self.node(r)) |
|
1377 | 1378 | elif p == nullrev: |
|
1378 | 1379 | c.append(self.node(r)) |
|
1379 | 1380 | return c |
|
1380 | 1381 | |
|
1381 | 1382 | def commonancestorsheads(self, a, b): |
|
1382 | 1383 | """calculate all the heads of the common ancestors of nodes a and b""" |
|
1383 | 1384 | a, b = self.rev(a), self.rev(b) |
|
1384 | 1385 | ancs = self._commonancestorsheads(a, b) |
|
1385 | 1386 | return pycompat.maplist(self.node, ancs) |
|
1386 | 1387 | |
|
1387 | 1388 | def _commonancestorsheads(self, *revs): |
|
1388 | 1389 | """calculate all the heads of the common ancestors of revs""" |
|
1389 | 1390 | try: |
|
1390 | 1391 | ancs = self.index.commonancestorsheads(*revs) |
|
1391 | 1392 | except (AttributeError, OverflowError): # C implementation failed |
|
1392 | 1393 | ancs = ancestor.commonancestorsheads(self.parentrevs, *revs) |
|
1393 | 1394 | return ancs |
|
1394 | 1395 | |
|
1395 | 1396 | def isancestor(self, a, b): |
|
1396 | 1397 | """return True if node a is an ancestor of node b |
|
1397 | 1398 | |
|
1398 | 1399 | A revision is considered an ancestor of itself.""" |
|
1399 | 1400 | a, b = self.rev(a), self.rev(b) |
|
1400 | 1401 | return self.isancestorrev(a, b) |
|
1401 | 1402 | |
|
1402 | 1403 | def isancestorrev(self, a, b): |
|
1403 | 1404 | """return True if revision a is an ancestor of revision b |
|
1404 | 1405 | |
|
1405 | 1406 | A revision is considered an ancestor of itself. |
|
1406 | 1407 | |
|
1407 | 1408 | The implementation of this is trivial but the use of |
|
1408 | 1409 | reachableroots is not.""" |
|
1409 | 1410 | if a == nullrev: |
|
1410 | 1411 | return True |
|
1411 | 1412 | elif a == b: |
|
1412 | 1413 | return True |
|
1413 | 1414 | elif a > b: |
|
1414 | 1415 | return False |
|
1415 | 1416 | return bool(self.reachableroots(a, [b], [a], includepath=False)) |
|
1416 | 1417 | |
|
1417 | 1418 | def reachableroots(self, minroot, heads, roots, includepath=False): |
|
1418 | 1419 | """return (heads(::(<roots> and <roots>::<heads>))) |
|
1419 | 1420 | |
|
1420 | 1421 | If includepath is True, return (<roots>::<heads>).""" |
|
1421 | 1422 | try: |
|
1422 | 1423 | return self.index.reachableroots2( |
|
1423 | 1424 | minroot, heads, roots, includepath |
|
1424 | 1425 | ) |
|
1425 | 1426 | except AttributeError: |
|
1426 | 1427 | return dagop._reachablerootspure( |
|
1427 | 1428 | self.parentrevs, minroot, roots, heads, includepath |
|
1428 | 1429 | ) |
|
1429 | 1430 | |
|
1430 | 1431 | def ancestor(self, a, b): |
|
1431 | 1432 | """calculate the "best" common ancestor of nodes a and b""" |
|
1432 | 1433 | |
|
1433 | 1434 | a, b = self.rev(a), self.rev(b) |
|
1434 | 1435 | try: |
|
1435 | 1436 | ancs = self.index.ancestors(a, b) |
|
1436 | 1437 | except (AttributeError, OverflowError): |
|
1437 | 1438 | ancs = ancestor.ancestors(self.parentrevs, a, b) |
|
1438 | 1439 | if ancs: |
|
1439 | 1440 | # choose a consistent winner when there's a tie |
|
1440 | 1441 | return min(map(self.node, ancs)) |
|
1441 | 1442 | return nullid |
|
1442 | 1443 | |
|
1443 | 1444 | def _match(self, id): |
|
1444 | 1445 | if isinstance(id, int): |
|
1445 | 1446 | # rev |
|
1446 | 1447 | return self.node(id) |
|
1447 | 1448 | if len(id) == 20: |
|
1448 | 1449 | # possibly a binary node |
|
1449 | 1450 | # odds of a binary node being all hex in ASCII are 1 in 10**25 |
|
1450 | 1451 | try: |
|
1451 | 1452 | node = id |
|
1452 | 1453 | self.rev(node) # quick search the index |
|
1453 | 1454 | return node |
|
1454 | 1455 | except error.LookupError: |
|
1455 | 1456 | pass # may be partial hex id |
|
1456 | 1457 | try: |
|
1457 | 1458 | # str(rev) |
|
1458 | 1459 | rev = int(id) |
|
1459 | 1460 | if b"%d" % rev != id: |
|
1460 | 1461 | raise ValueError |
|
1461 | 1462 | if rev < 0: |
|
1462 | 1463 | rev = len(self) + rev |
|
1463 | 1464 | if rev < 0 or rev >= len(self): |
|
1464 | 1465 | raise ValueError |
|
1465 | 1466 | return self.node(rev) |
|
1466 | 1467 | except (ValueError, OverflowError): |
|
1467 | 1468 | pass |
|
1468 | 1469 | if len(id) == 40: |
|
1469 | 1470 | try: |
|
1470 | 1471 | # a full hex nodeid? |
|
1471 | 1472 | node = bin(id) |
|
1472 | 1473 | self.rev(node) |
|
1473 | 1474 | return node |
|
1474 | 1475 | except (TypeError, error.LookupError): |
|
1475 | 1476 | pass |
|
1476 | 1477 | |
|
1477 | 1478 | def _partialmatch(self, id): |
|
1478 | 1479 | # we don't care wdirfilenodeids as they should be always full hash |
|
1479 | 1480 | maybewdir = wdirhex.startswith(id) |
|
1480 | 1481 | try: |
|
1481 | 1482 | partial = self.index.partialmatch(id) |
|
1482 | 1483 | if partial and self.hasnode(partial): |
|
1483 | 1484 | if maybewdir: |
|
1484 | 1485 | # single 'ff...' match in radix tree, ambiguous with wdir |
|
1485 | 1486 | raise error.RevlogError |
|
1486 | 1487 | return partial |
|
1487 | 1488 | if maybewdir: |
|
1488 | 1489 | # no 'ff...' match in radix tree, wdir identified |
|
1489 | 1490 | raise error.WdirUnsupported |
|
1490 | 1491 | return None |
|
1491 | 1492 | except error.RevlogError: |
|
1492 | 1493 | # parsers.c radix tree lookup gave multiple matches |
|
1493 | 1494 | # fast path: for unfiltered changelog, radix tree is accurate |
|
1494 | 1495 | if not getattr(self, 'filteredrevs', None): |
|
1495 | 1496 | raise error.AmbiguousPrefixLookupError( |
|
1496 | 1497 | id, self.indexfile, _(b'ambiguous identifier') |
|
1497 | 1498 | ) |
|
1498 | 1499 | # fall through to slow path that filters hidden revisions |
|
1499 | 1500 | except (AttributeError, ValueError): |
|
1500 | 1501 | # we are pure python, or key was too short to search radix tree |
|
1501 | 1502 | pass |
|
1502 | 1503 | |
|
1503 | 1504 | if id in self._pcache: |
|
1504 | 1505 | return self._pcache[id] |
|
1505 | 1506 | |
|
1506 | 1507 | if len(id) <= 40: |
|
1507 | 1508 | try: |
|
1508 | 1509 | # hex(node)[:...] |
|
1509 | 1510 | l = len(id) // 2 # grab an even number of digits |
|
1510 | 1511 | prefix = bin(id[: l * 2]) |
|
1511 | 1512 | nl = [e[7] for e in self.index if e[7].startswith(prefix)] |
|
1512 | 1513 | nl = [ |
|
1513 | 1514 | n for n in nl if hex(n).startswith(id) and self.hasnode(n) |
|
1514 | 1515 | ] |
|
1515 | 1516 | if nullhex.startswith(id): |
|
1516 | 1517 | nl.append(nullid) |
|
1517 | 1518 | if len(nl) > 0: |
|
1518 | 1519 | if len(nl) == 1 and not maybewdir: |
|
1519 | 1520 | self._pcache[id] = nl[0] |
|
1520 | 1521 | return nl[0] |
|
1521 | 1522 | raise error.AmbiguousPrefixLookupError( |
|
1522 | 1523 | id, self.indexfile, _(b'ambiguous identifier') |
|
1523 | 1524 | ) |
|
1524 | 1525 | if maybewdir: |
|
1525 | 1526 | raise error.WdirUnsupported |
|
1526 | 1527 | return None |
|
1527 | 1528 | except TypeError: |
|
1528 | 1529 | pass |
|
1529 | 1530 | |
|
1530 | 1531 | def lookup(self, id): |
|
1531 | 1532 | """locate a node based on: |
|
1532 | 1533 | - revision number or str(revision number) |
|
1533 | 1534 | - nodeid or subset of hex nodeid |
|
1534 | 1535 | """ |
|
1535 | 1536 | n = self._match(id) |
|
1536 | 1537 | if n is not None: |
|
1537 | 1538 | return n |
|
1538 | 1539 | n = self._partialmatch(id) |
|
1539 | 1540 | if n: |
|
1540 | 1541 | return n |
|
1541 | 1542 | |
|
1542 | 1543 | raise error.LookupError(id, self.indexfile, _(b'no match found')) |
|
1543 | 1544 | |
|
1544 | 1545 | def shortest(self, node, minlength=1): |
|
1545 | 1546 | """Find the shortest unambiguous prefix that matches node.""" |
|
1546 | 1547 | |
|
1547 | 1548 | def isvalid(prefix): |
|
1548 | 1549 | try: |
|
1549 | 1550 | matchednode = self._partialmatch(prefix) |
|
1550 | 1551 | except error.AmbiguousPrefixLookupError: |
|
1551 | 1552 | return False |
|
1552 | 1553 | except error.WdirUnsupported: |
|
1553 | 1554 | # single 'ff...' match |
|
1554 | 1555 | return True |
|
1555 | 1556 | if matchednode is None: |
|
1556 | 1557 | raise error.LookupError(node, self.indexfile, _(b'no node')) |
|
1557 | 1558 | return True |
|
1558 | 1559 | |
|
1559 | 1560 | def maybewdir(prefix): |
|
1560 | 1561 | return all(c == b'f' for c in pycompat.iterbytestr(prefix)) |
|
1561 | 1562 | |
|
1562 | 1563 | hexnode = hex(node) |
|
1563 | 1564 | |
|
1564 | 1565 | def disambiguate(hexnode, minlength): |
|
1565 | 1566 | """Disambiguate against wdirid.""" |
|
1566 | 1567 | for length in range(minlength, len(hexnode) + 1): |
|
1567 | 1568 | prefix = hexnode[:length] |
|
1568 | 1569 | if not maybewdir(prefix): |
|
1569 | 1570 | return prefix |
|
1570 | 1571 | |
|
1571 | 1572 | if not getattr(self, 'filteredrevs', None): |
|
1572 | 1573 | try: |
|
1573 | 1574 | length = max(self.index.shortest(node), minlength) |
|
1574 | 1575 | return disambiguate(hexnode, length) |
|
1575 | 1576 | except error.RevlogError: |
|
1576 | 1577 | if node != wdirid: |
|
1577 | 1578 | raise error.LookupError(node, self.indexfile, _(b'no node')) |
|
1578 | 1579 | except AttributeError: |
|
1579 | 1580 | # Fall through to pure code |
|
1580 | 1581 | pass |
|
1581 | 1582 | |
|
1582 | 1583 | if node == wdirid: |
|
1583 | 1584 | for length in range(minlength, len(hexnode) + 1): |
|
1584 | 1585 | prefix = hexnode[:length] |
|
1585 | 1586 | if isvalid(prefix): |
|
1586 | 1587 | return prefix |
|
1587 | 1588 | |
|
1588 | 1589 | for length in range(minlength, len(hexnode) + 1): |
|
1589 | 1590 | prefix = hexnode[:length] |
|
1590 | 1591 | if isvalid(prefix): |
|
1591 | 1592 | return disambiguate(hexnode, length) |
|
1592 | 1593 | |
|
1593 | 1594 | def cmp(self, node, text): |
|
1594 | 1595 | """compare text with a given file revision |
|
1595 | 1596 | |
|
1596 | 1597 | returns True if text is different than what is stored. |
|
1597 | 1598 | """ |
|
1598 | 1599 | p1, p2 = self.parents(node) |
|
1599 | 1600 | return storageutil.hashrevisionsha1(text, p1, p2) != node |
|
1600 | 1601 | |
|
1601 | 1602 | def _cachesegment(self, offset, data): |
|
1602 | 1603 | """Add a segment to the revlog cache. |
|
1603 | 1604 | |
|
1604 | 1605 | Accepts an absolute offset and the data that is at that location. |
|
1605 | 1606 | """ |
|
1606 | 1607 | o, d = self._chunkcache |
|
1607 | 1608 | # try to add to existing cache |
|
1608 | 1609 | if o + len(d) == offset and len(d) + len(data) < _chunksize: |
|
1609 | 1610 | self._chunkcache = o, d + data |
|
1610 | 1611 | else: |
|
1611 | 1612 | self._chunkcache = offset, data |
|
1612 | 1613 | |
|
1613 | 1614 | def _readsegment(self, offset, length, df=None): |
|
1614 | 1615 | """Load a segment of raw data from the revlog. |
|
1615 | 1616 | |
|
1616 | 1617 | Accepts an absolute offset, length to read, and an optional existing |
|
1617 | 1618 | file handle to read from. |
|
1618 | 1619 | |
|
1619 | 1620 | If an existing file handle is passed, it will be seeked and the |
|
1620 | 1621 | original seek position will NOT be restored. |
|
1621 | 1622 | |
|
1622 | 1623 | Returns a str or buffer of raw byte data. |
|
1623 | 1624 | |
|
1624 | 1625 | Raises if the requested number of bytes could not be read. |
|
1625 | 1626 | """ |
|
1626 | 1627 | # Cache data both forward and backward around the requested |
|
1627 | 1628 | # data, in a fixed size window. This helps speed up operations |
|
1628 | 1629 | # involving reading the revlog backwards. |
|
1629 | 1630 | cachesize = self._chunkcachesize |
|
1630 | 1631 | realoffset = offset & ~(cachesize - 1) |
|
1631 | 1632 | reallength = ( |
|
1632 | 1633 | (offset + length + cachesize) & ~(cachesize - 1) |
|
1633 | 1634 | ) - realoffset |
|
1634 | 1635 | with self._datareadfp(df) as df: |
|
1635 | 1636 | df.seek(realoffset) |
|
1636 | 1637 | d = df.read(reallength) |
|
1637 | 1638 | |
|
1638 | 1639 | self._cachesegment(realoffset, d) |
|
1639 | 1640 | if offset != realoffset or reallength != length: |
|
1640 | 1641 | startoffset = offset - realoffset |
|
1641 | 1642 | if len(d) - startoffset < length: |
|
1642 | 1643 | raise error.RevlogError( |
|
1643 | 1644 | _( |
|
1644 | 1645 | b'partial read of revlog %s; expected %d bytes from ' |
|
1645 | 1646 | b'offset %d, got %d' |
|
1646 | 1647 | ) |
|
1647 | 1648 | % ( |
|
1648 | 1649 | self.indexfile if self._inline else self.datafile, |
|
1649 | 1650 | length, |
|
1650 | 1651 | realoffset, |
|
1651 | 1652 | len(d) - startoffset, |
|
1652 | 1653 | ) |
|
1653 | 1654 | ) |
|
1654 | 1655 | |
|
1655 | 1656 | return util.buffer(d, startoffset, length) |
|
1656 | 1657 | |
|
1657 | 1658 | if len(d) < length: |
|
1658 | 1659 | raise error.RevlogError( |
|
1659 | 1660 | _( |
|
1660 | 1661 | b'partial read of revlog %s; expected %d bytes from offset ' |
|
1661 | 1662 | b'%d, got %d' |
|
1662 | 1663 | ) |
|
1663 | 1664 | % ( |
|
1664 | 1665 | self.indexfile if self._inline else self.datafile, |
|
1665 | 1666 | length, |
|
1666 | 1667 | offset, |
|
1667 | 1668 | len(d), |
|
1668 | 1669 | ) |
|
1669 | 1670 | ) |
|
1670 | 1671 | |
|
1671 | 1672 | return d |
|
1672 | 1673 | |
|
1673 | 1674 | def _getsegment(self, offset, length, df=None): |
|
1674 | 1675 | """Obtain a segment of raw data from the revlog. |
|
1675 | 1676 | |
|
1676 | 1677 | Accepts an absolute offset, length of bytes to obtain, and an |
|
1677 | 1678 | optional file handle to the already-opened revlog. If the file |
|
1678 | 1679 | handle is used, it's original seek position will not be preserved. |
|
1679 | 1680 | |
|
1680 | 1681 | Requests for data may be returned from a cache. |
|
1681 | 1682 | |
|
1682 | 1683 | Returns a str or a buffer instance of raw byte data. |
|
1683 | 1684 | """ |
|
1684 | 1685 | o, d = self._chunkcache |
|
1685 | 1686 | l = len(d) |
|
1686 | 1687 | |
|
1687 | 1688 | # is it in the cache? |
|
1688 | 1689 | cachestart = offset - o |
|
1689 | 1690 | cacheend = cachestart + length |
|
1690 | 1691 | if cachestart >= 0 and cacheend <= l: |
|
1691 | 1692 | if cachestart == 0 and cacheend == l: |
|
1692 | 1693 | return d # avoid a copy |
|
1693 | 1694 | return util.buffer(d, cachestart, cacheend - cachestart) |
|
1694 | 1695 | |
|
1695 | 1696 | return self._readsegment(offset, length, df=df) |
|
1696 | 1697 | |
|
1697 | 1698 | def _getsegmentforrevs(self, startrev, endrev, df=None): |
|
1698 | 1699 | """Obtain a segment of raw data corresponding to a range of revisions. |
|
1699 | 1700 | |
|
1700 | 1701 | Accepts the start and end revisions and an optional already-open |
|
1701 | 1702 | file handle to be used for reading. If the file handle is read, its |
|
1702 | 1703 | seek position will not be preserved. |
|
1703 | 1704 | |
|
1704 | 1705 | Requests for data may be satisfied by a cache. |
|
1705 | 1706 | |
|
1706 | 1707 | Returns a 2-tuple of (offset, data) for the requested range of |
|
1707 | 1708 | revisions. Offset is the integer offset from the beginning of the |
|
1708 | 1709 | revlog and data is a str or buffer of the raw byte data. |
|
1709 | 1710 | |
|
1710 | 1711 | Callers will need to call ``self.start(rev)`` and ``self.length(rev)`` |
|
1711 | 1712 | to determine where each revision's data begins and ends. |
|
1712 | 1713 | """ |
|
1713 | 1714 | # Inlined self.start(startrev) & self.end(endrev) for perf reasons |
|
1714 | 1715 | # (functions are expensive). |
|
1715 | 1716 | index = self.index |
|
1716 | 1717 | istart = index[startrev] |
|
1717 | 1718 | start = int(istart[0] >> 16) |
|
1718 | 1719 | if startrev == endrev: |
|
1719 | 1720 | end = start + istart[1] |
|
1720 | 1721 | else: |
|
1721 | 1722 | iend = index[endrev] |
|
1722 | 1723 | end = int(iend[0] >> 16) + iend[1] |
|
1723 | 1724 | |
|
1724 | 1725 | if self._inline: |
|
1725 | 1726 | start += (startrev + 1) * self._io.size |
|
1726 | 1727 | end += (endrev + 1) * self._io.size |
|
1727 | 1728 | length = end - start |
|
1728 | 1729 | |
|
1729 | 1730 | return start, self._getsegment(start, length, df=df) |
|
1730 | 1731 | |
|
1731 | 1732 | def _chunk(self, rev, df=None): |
|
1732 | 1733 | """Obtain a single decompressed chunk for a revision. |
|
1733 | 1734 | |
|
1734 | 1735 | Accepts an integer revision and an optional already-open file handle |
|
1735 | 1736 | to be used for reading. If used, the seek position of the file will not |
|
1736 | 1737 | be preserved. |
|
1737 | 1738 | |
|
1738 | 1739 | Returns a str holding uncompressed data for the requested revision. |
|
1739 | 1740 | """ |
|
1740 | 1741 | return self.decompress(self._getsegmentforrevs(rev, rev, df=df)[1]) |
|
1741 | 1742 | |
|
1742 | 1743 | def _chunks(self, revs, df=None, targetsize=None): |
|
1743 | 1744 | """Obtain decompressed chunks for the specified revisions. |
|
1744 | 1745 | |
|
1745 | 1746 | Accepts an iterable of numeric revisions that are assumed to be in |
|
1746 | 1747 | ascending order. Also accepts an optional already-open file handle |
|
1747 | 1748 | to be used for reading. If used, the seek position of the file will |
|
1748 | 1749 | not be preserved. |
|
1749 | 1750 | |
|
1750 | 1751 | This function is similar to calling ``self._chunk()`` multiple times, |
|
1751 | 1752 | but is faster. |
|
1752 | 1753 | |
|
1753 | 1754 | Returns a list with decompressed data for each requested revision. |
|
1754 | 1755 | """ |
|
1755 | 1756 | if not revs: |
|
1756 | 1757 | return [] |
|
1757 | 1758 | start = self.start |
|
1758 | 1759 | length = self.length |
|
1759 | 1760 | inline = self._inline |
|
1760 | 1761 | iosize = self._io.size |
|
1761 | 1762 | buffer = util.buffer |
|
1762 | 1763 | |
|
1763 | 1764 | l = [] |
|
1764 | 1765 | ladd = l.append |
|
1765 | 1766 | |
|
1766 | 1767 | if not self._withsparseread: |
|
1767 | 1768 | slicedchunks = (revs,) |
|
1768 | 1769 | else: |
|
1769 | 1770 | slicedchunks = deltautil.slicechunk( |
|
1770 | 1771 | self, revs, targetsize=targetsize |
|
1771 | 1772 | ) |
|
1772 | 1773 | |
|
1773 | 1774 | for revschunk in slicedchunks: |
|
1774 | 1775 | firstrev = revschunk[0] |
|
1775 | 1776 | # Skip trailing revisions with empty diff |
|
1776 | 1777 | for lastrev in revschunk[::-1]: |
|
1777 | 1778 | if length(lastrev) != 0: |
|
1778 | 1779 | break |
|
1779 | 1780 | |
|
1780 | 1781 | try: |
|
1781 | 1782 | offset, data = self._getsegmentforrevs(firstrev, lastrev, df=df) |
|
1782 | 1783 | except OverflowError: |
|
1783 | 1784 | # issue4215 - we can't cache a run of chunks greater than |
|
1784 | 1785 | # 2G on Windows |
|
1785 | 1786 | return [self._chunk(rev, df=df) for rev in revschunk] |
|
1786 | 1787 | |
|
1787 | 1788 | decomp = self.decompress |
|
1788 | 1789 | for rev in revschunk: |
|
1789 | 1790 | chunkstart = start(rev) |
|
1790 | 1791 | if inline: |
|
1791 | 1792 | chunkstart += (rev + 1) * iosize |
|
1792 | 1793 | chunklength = length(rev) |
|
1793 | 1794 | ladd(decomp(buffer(data, chunkstart - offset, chunklength))) |
|
1794 | 1795 | |
|
1795 | 1796 | return l |
|
1796 | 1797 | |
|
1797 | 1798 | def _chunkclear(self): |
|
1798 | 1799 | """Clear the raw chunk cache.""" |
|
1799 | 1800 | self._chunkcache = (0, b'') |
|
1800 | 1801 | |
|
1801 | 1802 | def deltaparent(self, rev): |
|
1802 | 1803 | """return deltaparent of the given revision""" |
|
1803 | 1804 | base = self.index[rev][3] |
|
1804 | 1805 | if base == rev: |
|
1805 | 1806 | return nullrev |
|
1806 | 1807 | elif self._generaldelta: |
|
1807 | 1808 | return base |
|
1808 | 1809 | else: |
|
1809 | 1810 | return rev - 1 |
|
1810 | 1811 | |
|
1811 | 1812 | def issnapshot(self, rev): |
|
1812 | 1813 | """tells whether rev is a snapshot""" |
|
1813 | 1814 | if not self._sparserevlog: |
|
1814 | 1815 | return self.deltaparent(rev) == nullrev |
|
1815 | 1816 | elif util.safehasattr(self.index, b'issnapshot'): |
|
1816 | 1817 | # directly assign the method to cache the testing and access |
|
1817 | 1818 | self.issnapshot = self.index.issnapshot |
|
1818 | 1819 | return self.issnapshot(rev) |
|
1819 | 1820 | if rev == nullrev: |
|
1820 | 1821 | return True |
|
1821 | 1822 | entry = self.index[rev] |
|
1822 | 1823 | base = entry[3] |
|
1823 | 1824 | if base == rev: |
|
1824 | 1825 | return True |
|
1825 | 1826 | if base == nullrev: |
|
1826 | 1827 | return True |
|
1827 | 1828 | p1 = entry[5] |
|
1828 | 1829 | p2 = entry[6] |
|
1829 | 1830 | if base == p1 or base == p2: |
|
1830 | 1831 | return False |
|
1831 | 1832 | return self.issnapshot(base) |
|
1832 | 1833 | |
|
1833 | 1834 | def snapshotdepth(self, rev): |
|
1834 | 1835 | """number of snapshot in the chain before this one""" |
|
1835 | 1836 | if not self.issnapshot(rev): |
|
1836 | 1837 | raise error.ProgrammingError(b'revision %d not a snapshot') |
|
1837 | 1838 | return len(self._deltachain(rev)[0]) - 1 |
|
1838 | 1839 | |
|
1839 | 1840 | def revdiff(self, rev1, rev2): |
|
1840 | 1841 | """return or calculate a delta between two revisions |
|
1841 | 1842 | |
|
1842 | 1843 | The delta calculated is in binary form and is intended to be written to |
|
1843 | 1844 | revlog data directly. So this function needs raw revision data. |
|
1844 | 1845 | """ |
|
1845 | 1846 | if rev1 != nullrev and self.deltaparent(rev2) == rev1: |
|
1846 | 1847 | return bytes(self._chunk(rev2)) |
|
1847 | 1848 | |
|
1848 | 1849 | return mdiff.textdiff(self.rawdata(rev1), self.rawdata(rev2)) |
|
1849 | 1850 | |
|
1850 | 1851 | def _processflags(self, text, flags, operation, raw=False): |
|
1851 | 1852 | """deprecated entry point to access flag processors""" |
|
1852 | 1853 | msg = b'_processflag(...) use the specialized variant' |
|
1853 | 1854 | util.nouideprecwarn(msg, b'5.2', stacklevel=2) |
|
1854 | 1855 | if raw: |
|
1855 | 1856 | return text, flagutil.processflagsraw(self, text, flags) |
|
1856 | 1857 | elif operation == b'read': |
|
1857 | 1858 | return flagutil.processflagsread(self, text, flags) |
|
1858 | 1859 | else: # write operation |
|
1859 | 1860 | return flagutil.processflagswrite(self, text, flags) |
|
1860 | 1861 | |
|
1861 | 1862 | def revision(self, nodeorrev, _df=None, raw=False): |
|
1862 | 1863 | """return an uncompressed revision of a given node or revision |
|
1863 | 1864 | number. |
|
1864 | 1865 | |
|
1865 | 1866 | _df - an existing file handle to read from. (internal-only) |
|
1866 | 1867 | raw - an optional argument specifying if the revision data is to be |
|
1867 | 1868 | treated as raw data when applying flag transforms. 'raw' should be set |
|
1868 | 1869 | to True when generating changegroups or in debug commands. |
|
1869 | 1870 | """ |
|
1870 | 1871 | if raw: |
|
1871 | 1872 | msg = ( |
|
1872 | 1873 | b'revlog.revision(..., raw=True) is deprecated, ' |
|
1873 | 1874 | b'use revlog.rawdata(...)' |
|
1874 | 1875 | ) |
|
1875 | 1876 | util.nouideprecwarn(msg, b'5.2', stacklevel=2) |
|
1876 | 1877 | return self._revisiondata(nodeorrev, _df, raw=raw)[0] |
|
1877 | 1878 | |
|
1878 | 1879 | def sidedata(self, nodeorrev, _df=None): |
|
1879 | 1880 | """a map of extra data related to the changeset but not part of the hash |
|
1880 | 1881 | |
|
1881 | 1882 | This function currently return a dictionary. However, more advanced |
|
1882 | 1883 | mapping object will likely be used in the future for a more |
|
1883 | 1884 | efficient/lazy code. |
|
1884 | 1885 | """ |
|
1885 | 1886 | return self._revisiondata(nodeorrev, _df)[1] |
|
1886 | 1887 | |
|
1887 | 1888 | def _revisiondata(self, nodeorrev, _df=None, raw=False): |
|
1888 | 1889 | # deal with <nodeorrev> argument type |
|
1889 | 1890 | if isinstance(nodeorrev, int): |
|
1890 | 1891 | rev = nodeorrev |
|
1891 | 1892 | node = self.node(rev) |
|
1892 | 1893 | else: |
|
1893 | 1894 | node = nodeorrev |
|
1894 | 1895 | rev = None |
|
1895 | 1896 | |
|
1896 | 1897 | # fast path the special `nullid` rev |
|
1897 | 1898 | if node == nullid: |
|
1898 | 1899 | return b"", {} |
|
1899 | 1900 | |
|
1900 | 1901 | # ``rawtext`` is the text as stored inside the revlog. Might be the |
|
1901 | 1902 | # revision or might need to be processed to retrieve the revision. |
|
1902 | 1903 | rev, rawtext, validated = self._rawtext(node, rev, _df=_df) |
|
1903 | 1904 | |
|
1904 | 1905 | if self.version & 0xFFFF == REVLOGV2: |
|
1905 | 1906 | if rev is None: |
|
1906 | 1907 | rev = self.rev(node) |
|
1907 | 1908 | sidedata = self._sidedata(rev) |
|
1908 | 1909 | else: |
|
1909 | 1910 | sidedata = {} |
|
1910 | 1911 | |
|
1911 | 1912 | if raw and validated: |
|
1912 | 1913 | # if we don't want to process the raw text and that raw |
|
1913 | 1914 | # text is cached, we can exit early. |
|
1914 | 1915 | return rawtext, sidedata |
|
1915 | 1916 | if rev is None: |
|
1916 | 1917 | rev = self.rev(node) |
|
1917 | 1918 | # the revlog's flag for this revision |
|
1918 | 1919 | # (usually alter its state or content) |
|
1919 | 1920 | flags = self.flags(rev) |
|
1920 | 1921 | |
|
1921 | 1922 | if validated and flags == REVIDX_DEFAULT_FLAGS: |
|
1922 | 1923 | # no extra flags set, no flag processor runs, text = rawtext |
|
1923 | 1924 | return rawtext, sidedata |
|
1924 | 1925 | |
|
1925 | 1926 | if raw: |
|
1926 | 1927 | validatehash = flagutil.processflagsraw(self, rawtext, flags) |
|
1927 | 1928 | text = rawtext |
|
1928 | 1929 | else: |
|
1929 | 1930 | r = flagutil.processflagsread(self, rawtext, flags) |
|
1930 | 1931 | text, validatehash = r |
|
1931 | 1932 | if validatehash: |
|
1932 | 1933 | self.checkhash(text, node, rev=rev) |
|
1933 | 1934 | if not validated: |
|
1934 | 1935 | self._revisioncache = (node, rev, rawtext) |
|
1935 | 1936 | |
|
1936 | 1937 | return text, sidedata |
|
1937 | 1938 | |
|
1938 | 1939 | def _rawtext(self, node, rev, _df=None): |
|
1939 | 1940 | """return the possibly unvalidated rawtext for a revision |
|
1940 | 1941 | |
|
1941 | 1942 | returns (rev, rawtext, validated) |
|
1942 | 1943 | """ |
|
1943 | 1944 | |
|
1944 | 1945 | # revision in the cache (could be useful to apply delta) |
|
1945 | 1946 | cachedrev = None |
|
1946 | 1947 | # An intermediate text to apply deltas to |
|
1947 | 1948 | basetext = None |
|
1948 | 1949 | |
|
1949 | 1950 | # Check if we have the entry in cache |
|
1950 | 1951 | # The cache entry looks like (node, rev, rawtext) |
|
1951 | 1952 | if self._revisioncache: |
|
1952 | 1953 | if self._revisioncache[0] == node: |
|
1953 | 1954 | return (rev, self._revisioncache[2], True) |
|
1954 | 1955 | cachedrev = self._revisioncache[1] |
|
1955 | 1956 | |
|
1956 | 1957 | if rev is None: |
|
1957 | 1958 | rev = self.rev(node) |
|
1958 | 1959 | |
|
1959 | 1960 | chain, stopped = self._deltachain(rev, stoprev=cachedrev) |
|
1960 | 1961 | if stopped: |
|
1961 | 1962 | basetext = self._revisioncache[2] |
|
1962 | 1963 | |
|
1963 | 1964 | # drop cache to save memory, the caller is expected to |
|
1964 | 1965 | # update self._revisioncache after validating the text |
|
1965 | 1966 | self._revisioncache = None |
|
1966 | 1967 | |
|
1967 | 1968 | targetsize = None |
|
1968 | 1969 | rawsize = self.index[rev][2] |
|
1969 | 1970 | if 0 <= rawsize: |
|
1970 | 1971 | targetsize = 4 * rawsize |
|
1971 | 1972 | |
|
1972 | 1973 | bins = self._chunks(chain, df=_df, targetsize=targetsize) |
|
1973 | 1974 | if basetext is None: |
|
1974 | 1975 | basetext = bytes(bins[0]) |
|
1975 | 1976 | bins = bins[1:] |
|
1976 | 1977 | |
|
1977 | 1978 | rawtext = mdiff.patches(basetext, bins) |
|
1978 | 1979 | del basetext # let us have a chance to free memory early |
|
1979 | 1980 | return (rev, rawtext, False) |
|
1980 | 1981 | |
|
1981 | 1982 | def _sidedata(self, rev): |
|
1982 | 1983 | """Return the sidedata for a given revision number.""" |
|
1983 | 1984 | index_entry = self.index[rev] |
|
1984 | 1985 | sidedata_offset = index_entry[8] |
|
1985 | 1986 | sidedata_size = index_entry[9] |
|
1986 | 1987 | |
|
1987 | 1988 | if self._inline: |
|
1988 | 1989 | sidedata_offset += self._io.size * (1 + rev) |
|
1989 | 1990 | if sidedata_size == 0: |
|
1990 | 1991 | return {} |
|
1991 | 1992 | |
|
1992 | 1993 | segment = self._getsegment(sidedata_offset, sidedata_size) |
|
1993 | 1994 | sidedata = sidedatautil.deserialize_sidedata(segment) |
|
1994 | 1995 | return sidedata |
|
1995 | 1996 | |
|
1996 | 1997 | def rawdata(self, nodeorrev, _df=None): |
|
1997 | 1998 | """return an uncompressed raw data of a given node or revision number. |
|
1998 | 1999 | |
|
1999 | 2000 | _df - an existing file handle to read from. (internal-only) |
|
2000 | 2001 | """ |
|
2001 | 2002 | return self._revisiondata(nodeorrev, _df, raw=True)[0] |
|
2002 | 2003 | |
|
2003 | 2004 | def hash(self, text, p1, p2): |
|
2004 | 2005 | """Compute a node hash. |
|
2005 | 2006 | |
|
2006 | 2007 | Available as a function so that subclasses can replace the hash |
|
2007 | 2008 | as needed. |
|
2008 | 2009 | """ |
|
2009 | 2010 | return storageutil.hashrevisionsha1(text, p1, p2) |
|
2010 | 2011 | |
|
2011 | 2012 | def checkhash(self, text, node, p1=None, p2=None, rev=None): |
|
2012 | 2013 | """Check node hash integrity. |
|
2013 | 2014 | |
|
2014 | 2015 | Available as a function so that subclasses can extend hash mismatch |
|
2015 | 2016 | behaviors as needed. |
|
2016 | 2017 | """ |
|
2017 | 2018 | try: |
|
2018 | 2019 | if p1 is None and p2 is None: |
|
2019 | 2020 | p1, p2 = self.parents(node) |
|
2020 | 2021 | if node != self.hash(text, p1, p2): |
|
2021 | 2022 | # Clear the revision cache on hash failure. The revision cache |
|
2022 | 2023 | # only stores the raw revision and clearing the cache does have |
|
2023 | 2024 | # the side-effect that we won't have a cache hit when the raw |
|
2024 | 2025 | # revision data is accessed. But this case should be rare and |
|
2025 | 2026 | # it is extra work to teach the cache about the hash |
|
2026 | 2027 | # verification state. |
|
2027 | 2028 | if self._revisioncache and self._revisioncache[0] == node: |
|
2028 | 2029 | self._revisioncache = None |
|
2029 | 2030 | |
|
2030 | 2031 | revornode = rev |
|
2031 | 2032 | if revornode is None: |
|
2032 | 2033 | revornode = templatefilters.short(hex(node)) |
|
2033 | 2034 | raise error.RevlogError( |
|
2034 | 2035 | _(b"integrity check failed on %s:%s") |
|
2035 | 2036 | % (self.indexfile, pycompat.bytestr(revornode)) |
|
2036 | 2037 | ) |
|
2037 | 2038 | except error.RevlogError: |
|
2038 | 2039 | if self._censorable and storageutil.iscensoredtext(text): |
|
2039 | 2040 | raise error.CensoredNodeError(self.indexfile, node, text) |
|
2040 | 2041 | raise |
|
2041 | 2042 | |
|
2042 | 2043 | def _enforceinlinesize(self, tr, fp=None): |
|
2043 | 2044 | """Check if the revlog is too big for inline and convert if so. |
|
2044 | 2045 | |
|
2045 | 2046 | This should be called after revisions are added to the revlog. If the |
|
2046 | 2047 | revlog has grown too large to be an inline revlog, it will convert it |
|
2047 | 2048 | to use multiple index and data files. |
|
2048 | 2049 | """ |
|
2049 | 2050 | tiprev = len(self) - 1 |
|
2050 | 2051 | if ( |
|
2051 | 2052 | not self._inline |
|
2052 | 2053 | or (self.start(tiprev) + self.length(tiprev)) < _maxinline |
|
2053 | 2054 | ): |
|
2054 | 2055 | return |
|
2055 | 2056 | |
|
2056 | 2057 | troffset = tr.findoffset(self.indexfile) |
|
2057 | 2058 | if troffset is None: |
|
2058 | 2059 | raise error.RevlogError( |
|
2059 | 2060 | _(b"%s not found in the transaction") % self.indexfile |
|
2060 | 2061 | ) |
|
2061 | 2062 | trindex = 0 |
|
2062 | 2063 | tr.add(self.datafile, 0) |
|
2063 | 2064 | |
|
2064 | 2065 | if fp: |
|
2065 | 2066 | fp.flush() |
|
2066 | 2067 | fp.close() |
|
2067 | 2068 | # We can't use the cached file handle after close(). So prevent |
|
2068 | 2069 | # its usage. |
|
2069 | 2070 | self._writinghandles = None |
|
2070 | 2071 | |
|
2071 | 2072 | with self._indexfp(b'r') as ifh, self._datafp(b'w') as dfh: |
|
2072 | 2073 | for r in self: |
|
2073 | 2074 | dfh.write(self._getsegmentforrevs(r, r, df=ifh)[1]) |
|
2074 | 2075 | if troffset <= self.start(r): |
|
2075 | 2076 | trindex = r |
|
2076 | 2077 | |
|
2077 | 2078 | with self._indexfp(b'w') as fp: |
|
2078 | 2079 | self.version &= ~FLAG_INLINE_DATA |
|
2079 | 2080 | self._inline = False |
|
2080 | 2081 | io = self._io |
|
2081 | 2082 | for i in self: |
|
2082 | 2083 | e = io.packentry(self.index[i], self.node, self.version, i) |
|
2083 | 2084 | fp.write(e) |
|
2084 | 2085 | |
|
2085 | 2086 | # the temp file replace the real index when we exit the context |
|
2086 | 2087 | # manager |
|
2087 | 2088 | |
|
2088 | 2089 | tr.replace(self.indexfile, trindex * self._io.size) |
|
2089 | 2090 | nodemaputil.setup_persistent_nodemap(tr, self) |
|
2090 | 2091 | self._chunkclear() |
|
2091 | 2092 | |
|
2092 | 2093 | def _nodeduplicatecallback(self, transaction, node): |
|
2093 | 2094 | """called when trying to add a node already stored.""" |
|
2094 | 2095 | |
|
2095 | 2096 | def addrevision( |
|
2096 | 2097 | self, |
|
2097 | 2098 | text, |
|
2098 | 2099 | transaction, |
|
2099 | 2100 | link, |
|
2100 | 2101 | p1, |
|
2101 | 2102 | p2, |
|
2102 | 2103 | cachedelta=None, |
|
2103 | 2104 | node=None, |
|
2104 | 2105 | flags=REVIDX_DEFAULT_FLAGS, |
|
2105 | 2106 | deltacomputer=None, |
|
2106 | 2107 | sidedata=None, |
|
2107 | 2108 | ): |
|
2108 | 2109 | """add a revision to the log |
|
2109 | 2110 | |
|
2110 | 2111 | text - the revision data to add |
|
2111 | 2112 | transaction - the transaction object used for rollback |
|
2112 | 2113 | link - the linkrev data to add |
|
2113 | 2114 | p1, p2 - the parent nodeids of the revision |
|
2114 | 2115 | cachedelta - an optional precomputed delta |
|
2115 | 2116 | node - nodeid of revision; typically node is not specified, and it is |
|
2116 | 2117 | computed by default as hash(text, p1, p2), however subclasses might |
|
2117 | 2118 | use different hashing method (and override checkhash() in such case) |
|
2118 | 2119 | flags - the known flags to set on the revision |
|
2119 | 2120 | deltacomputer - an optional deltacomputer instance shared between |
|
2120 | 2121 | multiple calls |
|
2121 | 2122 | """ |
|
2122 | 2123 | if link == nullrev: |
|
2123 | 2124 | raise error.RevlogError( |
|
2124 | 2125 | _(b"attempted to add linkrev -1 to %s") % self.indexfile |
|
2125 | 2126 | ) |
|
2126 | 2127 | |
|
2127 | 2128 | if sidedata is None: |
|
2128 | 2129 | sidedata = {} |
|
2129 | 2130 | elif not self.hassidedata: |
|
2130 | 2131 | raise error.ProgrammingError( |
|
2131 | 2132 | _(b"trying to add sidedata to a revlog who don't support them") |
|
2132 | 2133 | ) |
|
2133 | 2134 | |
|
2134 | 2135 | if flags: |
|
2135 | 2136 | node = node or self.hash(text, p1, p2) |
|
2136 | 2137 | |
|
2137 | 2138 | rawtext, validatehash = flagutil.processflagswrite(self, text, flags) |
|
2138 | 2139 | |
|
2139 | 2140 | # If the flag processor modifies the revision data, ignore any provided |
|
2140 | 2141 | # cachedelta. |
|
2141 | 2142 | if rawtext != text: |
|
2142 | 2143 | cachedelta = None |
|
2143 | 2144 | |
|
2144 | 2145 | if len(rawtext) > _maxentrysize: |
|
2145 | 2146 | raise error.RevlogError( |
|
2146 | 2147 | _( |
|
2147 | 2148 | b"%s: size of %d bytes exceeds maximum revlog storage of 2GiB" |
|
2148 | 2149 | ) |
|
2149 | 2150 | % (self.indexfile, len(rawtext)) |
|
2150 | 2151 | ) |
|
2151 | 2152 | |
|
2152 | 2153 | node = node or self.hash(rawtext, p1, p2) |
|
2153 | 2154 | rev = self.index.get_rev(node) |
|
2154 | 2155 | if rev is not None: |
|
2155 | 2156 | return rev |
|
2156 | 2157 | |
|
2157 | 2158 | if validatehash: |
|
2158 | 2159 | self.checkhash(rawtext, node, p1=p1, p2=p2) |
|
2159 | 2160 | |
|
2160 | 2161 | return self.addrawrevision( |
|
2161 | 2162 | rawtext, |
|
2162 | 2163 | transaction, |
|
2163 | 2164 | link, |
|
2164 | 2165 | p1, |
|
2165 | 2166 | p2, |
|
2166 | 2167 | node, |
|
2167 | 2168 | flags, |
|
2168 | 2169 | cachedelta=cachedelta, |
|
2169 | 2170 | deltacomputer=deltacomputer, |
|
2170 | 2171 | sidedata=sidedata, |
|
2171 | 2172 | ) |
|
2172 | 2173 | |
|
2173 | 2174 | def addrawrevision( |
|
2174 | 2175 | self, |
|
2175 | 2176 | rawtext, |
|
2176 | 2177 | transaction, |
|
2177 | 2178 | link, |
|
2178 | 2179 | p1, |
|
2179 | 2180 | p2, |
|
2180 | 2181 | node, |
|
2181 | 2182 | flags, |
|
2182 | 2183 | cachedelta=None, |
|
2183 | 2184 | deltacomputer=None, |
|
2184 | 2185 | sidedata=None, |
|
2185 | 2186 | ): |
|
2186 | 2187 | """add a raw revision with known flags, node and parents |
|
2187 | 2188 | useful when reusing a revision not stored in this revlog (ex: received |
|
2188 | 2189 | over wire, or read from an external bundle). |
|
2189 | 2190 | """ |
|
2190 | 2191 | dfh = None |
|
2191 | 2192 | if not self._inline: |
|
2192 | 2193 | dfh = self._datafp(b"a+") |
|
2193 | 2194 | ifh = self._indexfp(b"a+") |
|
2194 | 2195 | try: |
|
2195 | 2196 | return self._addrevision( |
|
2196 | 2197 | node, |
|
2197 | 2198 | rawtext, |
|
2198 | 2199 | transaction, |
|
2199 | 2200 | link, |
|
2200 | 2201 | p1, |
|
2201 | 2202 | p2, |
|
2202 | 2203 | flags, |
|
2203 | 2204 | cachedelta, |
|
2204 | 2205 | ifh, |
|
2205 | 2206 | dfh, |
|
2206 | 2207 | deltacomputer=deltacomputer, |
|
2207 | 2208 | sidedata=sidedata, |
|
2208 | 2209 | ) |
|
2209 | 2210 | finally: |
|
2210 | 2211 | if dfh: |
|
2211 | 2212 | dfh.close() |
|
2212 | 2213 | ifh.close() |
|
2213 | 2214 | |
|
2214 | 2215 | def compress(self, data): |
|
2215 | 2216 | """Generate a possibly-compressed representation of data.""" |
|
2216 | 2217 | if not data: |
|
2217 | 2218 | return b'', data |
|
2218 | 2219 | |
|
2219 | 2220 | compressed = self._compressor.compress(data) |
|
2220 | 2221 | |
|
2221 | 2222 | if compressed: |
|
2222 | 2223 | # The revlog compressor added the header in the returned data. |
|
2223 | 2224 | return b'', compressed |
|
2224 | 2225 | |
|
2225 | 2226 | if data[0:1] == b'\0': |
|
2226 | 2227 | return b'', data |
|
2227 | 2228 | return b'u', data |
|
2228 | 2229 | |
|
2229 | 2230 | def decompress(self, data): |
|
2230 | 2231 | """Decompress a revlog chunk. |
|
2231 | 2232 | |
|
2232 | 2233 | The chunk is expected to begin with a header identifying the |
|
2233 | 2234 | format type so it can be routed to an appropriate decompressor. |
|
2234 | 2235 | """ |
|
2235 | 2236 | if not data: |
|
2236 | 2237 | return data |
|
2237 | 2238 | |
|
2238 | 2239 | # Revlogs are read much more frequently than they are written and many |
|
2239 | 2240 | # chunks only take microseconds to decompress, so performance is |
|
2240 | 2241 | # important here. |
|
2241 | 2242 | # |
|
2242 | 2243 | # We can make a few assumptions about revlogs: |
|
2243 | 2244 | # |
|
2244 | 2245 | # 1) the majority of chunks will be compressed (as opposed to inline |
|
2245 | 2246 | # raw data). |
|
2246 | 2247 | # 2) decompressing *any* data will likely by at least 10x slower than |
|
2247 | 2248 | # returning raw inline data. |
|
2248 | 2249 | # 3) we want to prioritize common and officially supported compression |
|
2249 | 2250 | # engines |
|
2250 | 2251 | # |
|
2251 | 2252 | # It follows that we want to optimize for "decompress compressed data |
|
2252 | 2253 | # when encoded with common and officially supported compression engines" |
|
2253 | 2254 | # case over "raw data" and "data encoded by less common or non-official |
|
2254 | 2255 | # compression engines." That is why we have the inline lookup first |
|
2255 | 2256 | # followed by the compengines lookup. |
|
2256 | 2257 | # |
|
2257 | 2258 | # According to `hg perfrevlogchunks`, this is ~0.5% faster for zlib |
|
2258 | 2259 | # compressed chunks. And this matters for changelog and manifest reads. |
|
2259 | 2260 | t = data[0:1] |
|
2260 | 2261 | |
|
2261 | 2262 | if t == b'x': |
|
2262 | 2263 | try: |
|
2263 | 2264 | return _zlibdecompress(data) |
|
2264 | 2265 | except zlib.error as e: |
|
2265 | 2266 | raise error.RevlogError( |
|
2266 | 2267 | _(b'revlog decompress error: %s') |
|
2267 | 2268 | % stringutil.forcebytestr(e) |
|
2268 | 2269 | ) |
|
2269 | 2270 | # '\0' is more common than 'u' so it goes first. |
|
2270 | 2271 | elif t == b'\0': |
|
2271 | 2272 | return data |
|
2272 | 2273 | elif t == b'u': |
|
2273 | 2274 | return util.buffer(data, 1) |
|
2274 | 2275 | |
|
2275 | 2276 | try: |
|
2276 | 2277 | compressor = self._decompressors[t] |
|
2277 | 2278 | except KeyError: |
|
2278 | 2279 | try: |
|
2279 | 2280 | engine = util.compengines.forrevlogheader(t) |
|
2280 | 2281 | compressor = engine.revlogcompressor(self._compengineopts) |
|
2281 | 2282 | self._decompressors[t] = compressor |
|
2282 | 2283 | except KeyError: |
|
2283 | 2284 | raise error.RevlogError(_(b'unknown compression type %r') % t) |
|
2284 | 2285 | |
|
2285 | 2286 | return compressor.decompress(data) |
|
2286 | 2287 | |
|
2287 | 2288 | def _addrevision( |
|
2288 | 2289 | self, |
|
2289 | 2290 | node, |
|
2290 | 2291 | rawtext, |
|
2291 | 2292 | transaction, |
|
2292 | 2293 | link, |
|
2293 | 2294 | p1, |
|
2294 | 2295 | p2, |
|
2295 | 2296 | flags, |
|
2296 | 2297 | cachedelta, |
|
2297 | 2298 | ifh, |
|
2298 | 2299 | dfh, |
|
2299 | 2300 | alwayscache=False, |
|
2300 | 2301 | deltacomputer=None, |
|
2301 | 2302 | sidedata=None, |
|
2302 | 2303 | ): |
|
2303 | 2304 | """internal function to add revisions to the log |
|
2304 | 2305 | |
|
2305 | 2306 | see addrevision for argument descriptions. |
|
2306 | 2307 | |
|
2307 | 2308 | note: "addrevision" takes non-raw text, "_addrevision" takes raw text. |
|
2308 | 2309 | |
|
2309 | 2310 | if "deltacomputer" is not provided or None, a defaultdeltacomputer will |
|
2310 | 2311 | be used. |
|
2311 | 2312 | |
|
2312 | 2313 | invariants: |
|
2313 | 2314 | - rawtext is optional (can be None); if not set, cachedelta must be set. |
|
2314 | 2315 | if both are set, they must correspond to each other. |
|
2315 | 2316 | """ |
|
2316 | 2317 | if node == nullid: |
|
2317 | 2318 | raise error.RevlogError( |
|
2318 | 2319 | _(b"%s: attempt to add null revision") % self.indexfile |
|
2319 | 2320 | ) |
|
2320 | 2321 | if node == wdirid or node in wdirfilenodeids: |
|
2321 | 2322 | raise error.RevlogError( |
|
2322 | 2323 | _(b"%s: attempt to add wdir revision") % self.indexfile |
|
2323 | 2324 | ) |
|
2324 | 2325 | |
|
2325 | 2326 | if self._inline: |
|
2326 | 2327 | fh = ifh |
|
2327 | 2328 | else: |
|
2328 | 2329 | fh = dfh |
|
2329 | 2330 | |
|
2330 | 2331 | btext = [rawtext] |
|
2331 | 2332 | |
|
2332 | 2333 | curr = len(self) |
|
2333 | 2334 | prev = curr - 1 |
|
2334 | 2335 | |
|
2335 | 2336 | offset = self._get_data_offset(prev) |
|
2336 | 2337 | |
|
2337 | 2338 | if self._concurrencychecker: |
|
2338 | 2339 | if self._inline: |
|
2339 | 2340 | # offset is "as if" it were in the .d file, so we need to add on |
|
2340 | 2341 | # the size of the entry metadata. |
|
2341 | 2342 | self._concurrencychecker( |
|
2342 | 2343 | ifh, self.indexfile, offset + curr * self._io.size |
|
2343 | 2344 | ) |
|
2344 | 2345 | else: |
|
2345 | 2346 | # Entries in the .i are a consistent size. |
|
2346 | 2347 | self._concurrencychecker( |
|
2347 | 2348 | ifh, self.indexfile, curr * self._io.size |
|
2348 | 2349 | ) |
|
2349 | 2350 | self._concurrencychecker(dfh, self.datafile, offset) |
|
2350 | 2351 | |
|
2351 | 2352 | p1r, p2r = self.rev(p1), self.rev(p2) |
|
2352 | 2353 | |
|
2353 | 2354 | # full versions are inserted when the needed deltas |
|
2354 | 2355 | # become comparable to the uncompressed text |
|
2355 | 2356 | if rawtext is None: |
|
2356 | 2357 | # need rawtext size, before changed by flag processors, which is |
|
2357 | 2358 | # the non-raw size. use revlog explicitly to avoid filelog's extra |
|
2358 | 2359 | # logic that might remove metadata size. |
|
2359 | 2360 | textlen = mdiff.patchedsize( |
|
2360 | 2361 | revlog.size(self, cachedelta[0]), cachedelta[1] |
|
2361 | 2362 | ) |
|
2362 | 2363 | else: |
|
2363 | 2364 | textlen = len(rawtext) |
|
2364 | 2365 | |
|
2365 | 2366 | if deltacomputer is None: |
|
2366 | 2367 | deltacomputer = deltautil.deltacomputer(self) |
|
2367 | 2368 | |
|
2368 | 2369 | revinfo = _revisioninfo(node, p1, p2, btext, textlen, cachedelta, flags) |
|
2369 | 2370 | |
|
2370 | 2371 | deltainfo = deltacomputer.finddeltainfo(revinfo, fh) |
|
2371 | 2372 | |
|
2372 | 2373 | if sidedata: |
|
2373 | 2374 | serialized_sidedata = sidedatautil.serialize_sidedata(sidedata) |
|
2374 | 2375 | sidedata_offset = offset + deltainfo.deltalen |
|
2375 | 2376 | else: |
|
2376 | 2377 | serialized_sidedata = b"" |
|
2377 | 2378 | # Don't store the offset if the sidedata is empty, that way |
|
2378 | 2379 | # we can easily detect empty sidedata and they will be no different |
|
2379 | 2380 | # than ones we manually add. |
|
2380 | 2381 | sidedata_offset = 0 |
|
2381 | 2382 | |
|
2382 | 2383 | e = ( |
|
2383 | 2384 | offset_type(offset, flags), |
|
2384 | 2385 | deltainfo.deltalen, |
|
2385 | 2386 | textlen, |
|
2386 | 2387 | deltainfo.base, |
|
2387 | 2388 | link, |
|
2388 | 2389 | p1r, |
|
2389 | 2390 | p2r, |
|
2390 | 2391 | node, |
|
2391 | 2392 | sidedata_offset, |
|
2392 | 2393 | len(serialized_sidedata), |
|
2393 | 2394 | ) |
|
2394 | 2395 | |
|
2395 | 2396 | if self.version & 0xFFFF != REVLOGV2: |
|
2396 | 2397 | e = e[:8] |
|
2397 | 2398 | |
|
2398 | 2399 | self.index.append(e) |
|
2399 | 2400 | entry = self._io.packentry(e, self.node, self.version, curr) |
|
2400 | 2401 | self._writeentry( |
|
2401 | 2402 | transaction, |
|
2402 | 2403 | ifh, |
|
2403 | 2404 | dfh, |
|
2404 | 2405 | entry, |
|
2405 | 2406 | deltainfo.data, |
|
2406 | 2407 | link, |
|
2407 | 2408 | offset, |
|
2408 | 2409 | serialized_sidedata, |
|
2409 | 2410 | ) |
|
2410 | 2411 | |
|
2411 | 2412 | rawtext = btext[0] |
|
2412 | 2413 | |
|
2413 | 2414 | if alwayscache and rawtext is None: |
|
2414 | 2415 | rawtext = deltacomputer.buildtext(revinfo, fh) |
|
2415 | 2416 | |
|
2416 | 2417 | if type(rawtext) == bytes: # only accept immutable objects |
|
2417 | 2418 | self._revisioncache = (node, curr, rawtext) |
|
2418 | 2419 | self._chainbasecache[curr] = deltainfo.chainbase |
|
2419 | 2420 | return curr |
|
2420 | 2421 | |
|
2421 | 2422 | def _get_data_offset(self, prev): |
|
2422 | 2423 | """Returns the current offset in the (in-transaction) data file. |
|
2423 | 2424 | Versions < 2 of the revlog can get this 0(1), revlog v2 needs a docket |
|
2424 | 2425 | file to store that information: since sidedata can be rewritten to the |
|
2425 | 2426 | end of the data file within a transaction, you can have cases where, for |
|
2426 | 2427 | example, rev `n` does not have sidedata while rev `n - 1` does, leading |
|
2427 | 2428 | to `n - 1`'s sidedata being written after `n`'s data. |
|
2428 | 2429 | |
|
2429 | 2430 | TODO cache this in a docket file before getting out of experimental.""" |
|
2430 | 2431 | if self.version & 0xFFFF != REVLOGV2: |
|
2431 | 2432 | return self.end(prev) |
|
2432 | 2433 | |
|
2433 | 2434 | offset = 0 |
|
2434 | 2435 | for rev, entry in enumerate(self.index): |
|
2435 | 2436 | sidedata_end = entry[8] + entry[9] |
|
2436 | 2437 | # Sidedata for a previous rev has potentially been written after |
|
2437 | 2438 | # this rev's end, so take the max. |
|
2438 | 2439 | offset = max(self.end(rev), offset, sidedata_end) |
|
2439 | 2440 | return offset |
|
2440 | 2441 | |
|
2441 | 2442 | def _writeentry( |
|
2442 | 2443 | self, transaction, ifh, dfh, entry, data, link, offset, sidedata |
|
2443 | 2444 | ): |
|
2444 | 2445 | # Files opened in a+ mode have inconsistent behavior on various |
|
2445 | 2446 | # platforms. Windows requires that a file positioning call be made |
|
2446 | 2447 | # when the file handle transitions between reads and writes. See |
|
2447 | 2448 | # 3686fa2b8eee and the mixedfilemodewrapper in windows.py. On other |
|
2448 | 2449 | # platforms, Python or the platform itself can be buggy. Some versions |
|
2449 | 2450 | # of Solaris have been observed to not append at the end of the file |
|
2450 | 2451 | # if the file was seeked to before the end. See issue4943 for more. |
|
2451 | 2452 | # |
|
2452 | 2453 | # We work around this issue by inserting a seek() before writing. |
|
2453 | 2454 | # Note: This is likely not necessary on Python 3. However, because |
|
2454 | 2455 | # the file handle is reused for reads and may be seeked there, we need |
|
2455 | 2456 | # to be careful before changing this. |
|
2456 | 2457 | ifh.seek(0, os.SEEK_END) |
|
2457 | 2458 | if dfh: |
|
2458 | 2459 | dfh.seek(0, os.SEEK_END) |
|
2459 | 2460 | |
|
2460 | 2461 | curr = len(self) - 1 |
|
2461 | 2462 | if not self._inline: |
|
2462 | 2463 | transaction.add(self.datafile, offset) |
|
2463 | 2464 | transaction.add(self.indexfile, curr * len(entry)) |
|
2464 | 2465 | if data[0]: |
|
2465 | 2466 | dfh.write(data[0]) |
|
2466 | 2467 | dfh.write(data[1]) |
|
2467 | 2468 | if sidedata: |
|
2468 | 2469 | dfh.write(sidedata) |
|
2469 | 2470 | ifh.write(entry) |
|
2470 | 2471 | else: |
|
2471 | 2472 | offset += curr * self._io.size |
|
2472 | 2473 | transaction.add(self.indexfile, offset) |
|
2473 | 2474 | ifh.write(entry) |
|
2474 | 2475 | ifh.write(data[0]) |
|
2475 | 2476 | ifh.write(data[1]) |
|
2476 | 2477 | if sidedata: |
|
2477 | 2478 | ifh.write(sidedata) |
|
2478 | 2479 | self._enforceinlinesize(transaction, ifh) |
|
2479 | 2480 | nodemaputil.setup_persistent_nodemap(transaction, self) |
|
2480 | 2481 | |
|
2481 | 2482 | def addgroup( |
|
2482 | 2483 | self, |
|
2483 | 2484 | deltas, |
|
2484 | 2485 | linkmapper, |
|
2485 | 2486 | transaction, |
|
2486 | 2487 | alwayscache=False, |
|
2487 | 2488 | addrevisioncb=None, |
|
2488 | 2489 | duplicaterevisioncb=None, |
|
2489 | 2490 | ): |
|
2490 | 2491 | """ |
|
2491 | 2492 | add a delta group |
|
2492 | 2493 | |
|
2493 | 2494 | given a set of deltas, add them to the revision log. the |
|
2494 | 2495 | first delta is against its parent, which should be in our |
|
2495 | 2496 | log, the rest are against the previous delta. |
|
2496 | 2497 | |
|
2497 | 2498 | If ``addrevisioncb`` is defined, it will be called with arguments of |
|
2498 | 2499 | this revlog and the node that was added. |
|
2499 | 2500 | """ |
|
2500 | 2501 | |
|
2501 | 2502 | if self._writinghandles: |
|
2502 | 2503 | raise error.ProgrammingError(b'cannot nest addgroup() calls') |
|
2503 | 2504 | |
|
2504 | 2505 | r = len(self) |
|
2505 | 2506 | end = 0 |
|
2506 | 2507 | if r: |
|
2507 | 2508 | end = self.end(r - 1) |
|
2508 | 2509 | ifh = self._indexfp(b"a+") |
|
2509 | 2510 | isize = r * self._io.size |
|
2510 | 2511 | if self._inline: |
|
2511 | 2512 | transaction.add(self.indexfile, end + isize) |
|
2512 | 2513 | dfh = None |
|
2513 | 2514 | else: |
|
2514 | 2515 | transaction.add(self.indexfile, isize) |
|
2515 | 2516 | transaction.add(self.datafile, end) |
|
2516 | 2517 | dfh = self._datafp(b"a+") |
|
2517 | 2518 | |
|
2518 | 2519 | def flush(): |
|
2519 | 2520 | if dfh: |
|
2520 | 2521 | dfh.flush() |
|
2521 | 2522 | ifh.flush() |
|
2522 | 2523 | |
|
2523 | 2524 | self._writinghandles = (ifh, dfh) |
|
2524 | 2525 | empty = True |
|
2525 | 2526 | |
|
2526 | 2527 | try: |
|
2527 | 2528 | deltacomputer = deltautil.deltacomputer(self) |
|
2528 | 2529 | # loop through our set of deltas |
|
2529 | 2530 | for data in deltas: |
|
2530 | 2531 | node, p1, p2, linknode, deltabase, delta, flags, sidedata = data |
|
2531 | 2532 | link = linkmapper(linknode) |
|
2532 | 2533 | flags = flags or REVIDX_DEFAULT_FLAGS |
|
2533 | 2534 | |
|
2534 | 2535 | rev = self.index.get_rev(node) |
|
2535 | 2536 | if rev is not None: |
|
2536 | 2537 | # this can happen if two branches make the same change |
|
2537 | 2538 | self._nodeduplicatecallback(transaction, rev) |
|
2538 | 2539 | if duplicaterevisioncb: |
|
2539 | 2540 | duplicaterevisioncb(self, rev) |
|
2540 | 2541 | empty = False |
|
2541 | 2542 | continue |
|
2542 | 2543 | |
|
2543 | 2544 | for p in (p1, p2): |
|
2544 | 2545 | if not self.index.has_node(p): |
|
2545 | 2546 | raise error.LookupError( |
|
2546 | 2547 | p, self.indexfile, _(b'unknown parent') |
|
2547 | 2548 | ) |
|
2548 | 2549 | |
|
2549 | 2550 | if not self.index.has_node(deltabase): |
|
2550 | 2551 | raise error.LookupError( |
|
2551 | 2552 | deltabase, self.indexfile, _(b'unknown delta base') |
|
2552 | 2553 | ) |
|
2553 | 2554 | |
|
2554 | 2555 | baserev = self.rev(deltabase) |
|
2555 | 2556 | |
|
2556 | 2557 | if baserev != nullrev and self.iscensored(baserev): |
|
2557 | 2558 | # if base is censored, delta must be full replacement in a |
|
2558 | 2559 | # single patch operation |
|
2559 | 2560 | hlen = struct.calcsize(b">lll") |
|
2560 | 2561 | oldlen = self.rawsize(baserev) |
|
2561 | 2562 | newlen = len(delta) - hlen |
|
2562 | 2563 | if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen): |
|
2563 | 2564 | raise error.CensoredBaseError( |
|
2564 | 2565 | self.indexfile, self.node(baserev) |
|
2565 | 2566 | ) |
|
2566 | 2567 | |
|
2567 | 2568 | if not flags and self._peek_iscensored(baserev, delta, flush): |
|
2568 | 2569 | flags |= REVIDX_ISCENSORED |
|
2569 | 2570 | |
|
2570 | 2571 | # We assume consumers of addrevisioncb will want to retrieve |
|
2571 | 2572 | # the added revision, which will require a call to |
|
2572 | 2573 | # revision(). revision() will fast path if there is a cache |
|
2573 | 2574 | # hit. So, we tell _addrevision() to always cache in this case. |
|
2574 | 2575 | # We're only using addgroup() in the context of changegroup |
|
2575 | 2576 | # generation so the revision data can always be handled as raw |
|
2576 | 2577 | # by the flagprocessor. |
|
2577 | 2578 | rev = self._addrevision( |
|
2578 | 2579 | node, |
|
2579 | 2580 | None, |
|
2580 | 2581 | transaction, |
|
2581 | 2582 | link, |
|
2582 | 2583 | p1, |
|
2583 | 2584 | p2, |
|
2584 | 2585 | flags, |
|
2585 | 2586 | (baserev, delta), |
|
2586 | 2587 | ifh, |
|
2587 | 2588 | dfh, |
|
2588 | 2589 | alwayscache=alwayscache, |
|
2589 | 2590 | deltacomputer=deltacomputer, |
|
2591 | sidedata=sidedata, | |
|
2590 | 2592 | ) |
|
2591 | 2593 | |
|
2592 | 2594 | if addrevisioncb: |
|
2593 | 2595 | addrevisioncb(self, rev) |
|
2594 | 2596 | empty = False |
|
2595 | 2597 | |
|
2596 | 2598 | if not dfh and not self._inline: |
|
2597 | 2599 | # addrevision switched from inline to conventional |
|
2598 | 2600 | # reopen the index |
|
2599 | 2601 | ifh.close() |
|
2600 | 2602 | dfh = self._datafp(b"a+") |
|
2601 | 2603 | ifh = self._indexfp(b"a+") |
|
2602 | 2604 | self._writinghandles = (ifh, dfh) |
|
2603 | 2605 | finally: |
|
2604 | 2606 | self._writinghandles = None |
|
2605 | 2607 | |
|
2606 | 2608 | if dfh: |
|
2607 | 2609 | dfh.close() |
|
2608 | 2610 | ifh.close() |
|
2609 | 2611 | return not empty |
|
2610 | 2612 | |
|
2611 | 2613 | def iscensored(self, rev): |
|
2612 | 2614 | """Check if a file revision is censored.""" |
|
2613 | 2615 | if not self._censorable: |
|
2614 | 2616 | return False |
|
2615 | 2617 | |
|
2616 | 2618 | return self.flags(rev) & REVIDX_ISCENSORED |
|
2617 | 2619 | |
|
2618 | 2620 | def _peek_iscensored(self, baserev, delta, flush): |
|
2619 | 2621 | """Quickly check if a delta produces a censored revision.""" |
|
2620 | 2622 | if not self._censorable: |
|
2621 | 2623 | return False |
|
2622 | 2624 | |
|
2623 | 2625 | return storageutil.deltaiscensored(delta, baserev, self.rawsize) |
|
2624 | 2626 | |
|
2625 | 2627 | def getstrippoint(self, minlink): |
|
2626 | 2628 | """find the minimum rev that must be stripped to strip the linkrev |
|
2627 | 2629 | |
|
2628 | 2630 | Returns a tuple containing the minimum rev and a set of all revs that |
|
2629 | 2631 | have linkrevs that will be broken by this strip. |
|
2630 | 2632 | """ |
|
2631 | 2633 | return storageutil.resolvestripinfo( |
|
2632 | 2634 | minlink, |
|
2633 | 2635 | len(self) - 1, |
|
2634 | 2636 | self.headrevs(), |
|
2635 | 2637 | self.linkrev, |
|
2636 | 2638 | self.parentrevs, |
|
2637 | 2639 | ) |
|
2638 | 2640 | |
|
2639 | 2641 | def strip(self, minlink, transaction): |
|
2640 | 2642 | """truncate the revlog on the first revision with a linkrev >= minlink |
|
2641 | 2643 | |
|
2642 | 2644 | This function is called when we're stripping revision minlink and |
|
2643 | 2645 | its descendants from the repository. |
|
2644 | 2646 | |
|
2645 | 2647 | We have to remove all revisions with linkrev >= minlink, because |
|
2646 | 2648 | the equivalent changelog revisions will be renumbered after the |
|
2647 | 2649 | strip. |
|
2648 | 2650 | |
|
2649 | 2651 | So we truncate the revlog on the first of these revisions, and |
|
2650 | 2652 | trust that the caller has saved the revisions that shouldn't be |
|
2651 | 2653 | removed and that it'll re-add them after this truncation. |
|
2652 | 2654 | """ |
|
2653 | 2655 | if len(self) == 0: |
|
2654 | 2656 | return |
|
2655 | 2657 | |
|
2656 | 2658 | rev, _ = self.getstrippoint(minlink) |
|
2657 | 2659 | if rev == len(self): |
|
2658 | 2660 | return |
|
2659 | 2661 | |
|
2660 | 2662 | # first truncate the files on disk |
|
2661 | 2663 | end = self.start(rev) |
|
2662 | 2664 | if not self._inline: |
|
2663 | 2665 | transaction.add(self.datafile, end) |
|
2664 | 2666 | end = rev * self._io.size |
|
2665 | 2667 | else: |
|
2666 | 2668 | end += rev * self._io.size |
|
2667 | 2669 | |
|
2668 | 2670 | transaction.add(self.indexfile, end) |
|
2669 | 2671 | |
|
2670 | 2672 | # then reset internal state in memory to forget those revisions |
|
2671 | 2673 | self._revisioncache = None |
|
2672 | 2674 | self._chaininfocache = util.lrucachedict(500) |
|
2673 | 2675 | self._chunkclear() |
|
2674 | 2676 | |
|
2675 | 2677 | del self.index[rev:-1] |
|
2676 | 2678 | |
|
2677 | 2679 | def checksize(self): |
|
2678 | 2680 | """Check size of index and data files |
|
2679 | 2681 | |
|
2680 | 2682 | return a (dd, di) tuple. |
|
2681 | 2683 | - dd: extra bytes for the "data" file |
|
2682 | 2684 | - di: extra bytes for the "index" file |
|
2683 | 2685 | |
|
2684 | 2686 | A healthy revlog will return (0, 0). |
|
2685 | 2687 | """ |
|
2686 | 2688 | expected = 0 |
|
2687 | 2689 | if len(self): |
|
2688 | 2690 | expected = max(0, self.end(len(self) - 1)) |
|
2689 | 2691 | |
|
2690 | 2692 | try: |
|
2691 | 2693 | with self._datafp() as f: |
|
2692 | 2694 | f.seek(0, io.SEEK_END) |
|
2693 | 2695 | actual = f.tell() |
|
2694 | 2696 | dd = actual - expected |
|
2695 | 2697 | except IOError as inst: |
|
2696 | 2698 | if inst.errno != errno.ENOENT: |
|
2697 | 2699 | raise |
|
2698 | 2700 | dd = 0 |
|
2699 | 2701 | |
|
2700 | 2702 | try: |
|
2701 | 2703 | f = self.opener(self.indexfile) |
|
2702 | 2704 | f.seek(0, io.SEEK_END) |
|
2703 | 2705 | actual = f.tell() |
|
2704 | 2706 | f.close() |
|
2705 | 2707 | s = self._io.size |
|
2706 | 2708 | i = max(0, actual // s) |
|
2707 | 2709 | di = actual - (i * s) |
|
2708 | 2710 | if self._inline: |
|
2709 | 2711 | databytes = 0 |
|
2710 | 2712 | for r in self: |
|
2711 | 2713 | databytes += max(0, self.length(r)) |
|
2712 | 2714 | dd = 0 |
|
2713 | 2715 | di = actual - len(self) * s - databytes |
|
2714 | 2716 | except IOError as inst: |
|
2715 | 2717 | if inst.errno != errno.ENOENT: |
|
2716 | 2718 | raise |
|
2717 | 2719 | di = 0 |
|
2718 | 2720 | |
|
2719 | 2721 | return (dd, di) |
|
2720 | 2722 | |
|
2721 | 2723 | def files(self): |
|
2722 | 2724 | res = [self.indexfile] |
|
2723 | 2725 | if not self._inline: |
|
2724 | 2726 | res.append(self.datafile) |
|
2725 | 2727 | return res |
|
2726 | 2728 | |
|
2727 | 2729 | def emitrevisions( |
|
2728 | 2730 | self, |
|
2729 | 2731 | nodes, |
|
2730 | 2732 | nodesorder=None, |
|
2731 | 2733 | revisiondata=False, |
|
2732 | 2734 | assumehaveparentrevisions=False, |
|
2733 | 2735 | deltamode=repository.CG_DELTAMODE_STD, |
|
2734 | 2736 | ): |
|
2735 | 2737 | if nodesorder not in (b'nodes', b'storage', b'linear', None): |
|
2736 | 2738 | raise error.ProgrammingError( |
|
2737 | 2739 | b'unhandled value for nodesorder: %s' % nodesorder |
|
2738 | 2740 | ) |
|
2739 | 2741 | |
|
2740 | 2742 | if nodesorder is None and not self._generaldelta: |
|
2741 | 2743 | nodesorder = b'storage' |
|
2742 | 2744 | |
|
2743 | 2745 | if ( |
|
2744 | 2746 | not self._storedeltachains |
|
2745 | 2747 | and deltamode != repository.CG_DELTAMODE_PREV |
|
2746 | 2748 | ): |
|
2747 | 2749 | deltamode = repository.CG_DELTAMODE_FULL |
|
2748 | 2750 | |
|
2749 | 2751 | return storageutil.emitrevisions( |
|
2750 | 2752 | self, |
|
2751 | 2753 | nodes, |
|
2752 | 2754 | nodesorder, |
|
2753 | 2755 | revlogrevisiondelta, |
|
2754 | 2756 | deltaparentfn=self.deltaparent, |
|
2755 | 2757 | candeltafn=self.candelta, |
|
2756 | 2758 | rawsizefn=self.rawsize, |
|
2757 | 2759 | revdifffn=self.revdiff, |
|
2758 | 2760 | flagsfn=self.flags, |
|
2759 | 2761 | deltamode=deltamode, |
|
2760 | 2762 | revisiondata=revisiondata, |
|
2761 | 2763 | assumehaveparentrevisions=assumehaveparentrevisions, |
|
2762 | 2764 | ) |
|
2763 | 2765 | |
|
2764 | 2766 | DELTAREUSEALWAYS = b'always' |
|
2765 | 2767 | DELTAREUSESAMEREVS = b'samerevs' |
|
2766 | 2768 | DELTAREUSENEVER = b'never' |
|
2767 | 2769 | |
|
2768 | 2770 | DELTAREUSEFULLADD = b'fulladd' |
|
2769 | 2771 | |
|
2770 | 2772 | DELTAREUSEALL = {b'always', b'samerevs', b'never', b'fulladd'} |
|
2771 | 2773 | |
|
2772 | 2774 | def clone( |
|
2773 | 2775 | self, |
|
2774 | 2776 | tr, |
|
2775 | 2777 | destrevlog, |
|
2776 | 2778 | addrevisioncb=None, |
|
2777 | 2779 | deltareuse=DELTAREUSESAMEREVS, |
|
2778 | 2780 | forcedeltabothparents=None, |
|
2779 | 2781 | sidedatacompanion=None, |
|
2780 | 2782 | ): |
|
2781 | 2783 | """Copy this revlog to another, possibly with format changes. |
|
2782 | 2784 | |
|
2783 | 2785 | The destination revlog will contain the same revisions and nodes. |
|
2784 | 2786 | However, it may not be bit-for-bit identical due to e.g. delta encoding |
|
2785 | 2787 | differences. |
|
2786 | 2788 | |
|
2787 | 2789 | The ``deltareuse`` argument control how deltas from the existing revlog |
|
2788 | 2790 | are preserved in the destination revlog. The argument can have the |
|
2789 | 2791 | following values: |
|
2790 | 2792 | |
|
2791 | 2793 | DELTAREUSEALWAYS |
|
2792 | 2794 | Deltas will always be reused (if possible), even if the destination |
|
2793 | 2795 | revlog would not select the same revisions for the delta. This is the |
|
2794 | 2796 | fastest mode of operation. |
|
2795 | 2797 | DELTAREUSESAMEREVS |
|
2796 | 2798 | Deltas will be reused if the destination revlog would pick the same |
|
2797 | 2799 | revisions for the delta. This mode strikes a balance between speed |
|
2798 | 2800 | and optimization. |
|
2799 | 2801 | DELTAREUSENEVER |
|
2800 | 2802 | Deltas will never be reused. This is the slowest mode of execution. |
|
2801 | 2803 | This mode can be used to recompute deltas (e.g. if the diff/delta |
|
2802 | 2804 | algorithm changes). |
|
2803 | 2805 | DELTAREUSEFULLADD |
|
2804 | 2806 | Revision will be re-added as if their were new content. This is |
|
2805 | 2807 | slower than DELTAREUSEALWAYS but allow more mechanism to kicks in. |
|
2806 | 2808 | eg: large file detection and handling. |
|
2807 | 2809 | |
|
2808 | 2810 | Delta computation can be slow, so the choice of delta reuse policy can |
|
2809 | 2811 | significantly affect run time. |
|
2810 | 2812 | |
|
2811 | 2813 | The default policy (``DELTAREUSESAMEREVS``) strikes a balance between |
|
2812 | 2814 | two extremes. Deltas will be reused if they are appropriate. But if the |
|
2813 | 2815 | delta could choose a better revision, it will do so. This means if you |
|
2814 | 2816 | are converting a non-generaldelta revlog to a generaldelta revlog, |
|
2815 | 2817 | deltas will be recomputed if the delta's parent isn't a parent of the |
|
2816 | 2818 | revision. |
|
2817 | 2819 | |
|
2818 | 2820 | In addition to the delta policy, the ``forcedeltabothparents`` |
|
2819 | 2821 | argument controls whether to force compute deltas against both parents |
|
2820 | 2822 | for merges. By default, the current default is used. |
|
2821 | 2823 | |
|
2822 | 2824 | If not None, the `sidedatacompanion` is callable that accept two |
|
2823 | 2825 | arguments: |
|
2824 | 2826 | |
|
2825 | 2827 | (srcrevlog, rev) |
|
2826 | 2828 | |
|
2827 | 2829 | and return a quintet that control changes to sidedata content from the |
|
2828 | 2830 | old revision to the new clone result: |
|
2829 | 2831 | |
|
2830 | 2832 | (dropall, filterout, update, new_flags, dropped_flags) |
|
2831 | 2833 | |
|
2832 | 2834 | * if `dropall` is True, all sidedata should be dropped |
|
2833 | 2835 | * `filterout` is a set of sidedata keys that should be dropped |
|
2834 | 2836 | * `update` is a mapping of additionnal/new key -> value |
|
2835 | 2837 | * new_flags is a bitfields of new flags that the revision should get |
|
2836 | 2838 | * dropped_flags is a bitfields of new flags that the revision shoudl not longer have |
|
2837 | 2839 | """ |
|
2838 | 2840 | if deltareuse not in self.DELTAREUSEALL: |
|
2839 | 2841 | raise ValueError( |
|
2840 | 2842 | _(b'value for deltareuse invalid: %s') % deltareuse |
|
2841 | 2843 | ) |
|
2842 | 2844 | |
|
2843 | 2845 | if len(destrevlog): |
|
2844 | 2846 | raise ValueError(_(b'destination revlog is not empty')) |
|
2845 | 2847 | |
|
2846 | 2848 | if getattr(self, 'filteredrevs', None): |
|
2847 | 2849 | raise ValueError(_(b'source revlog has filtered revisions')) |
|
2848 | 2850 | if getattr(destrevlog, 'filteredrevs', None): |
|
2849 | 2851 | raise ValueError(_(b'destination revlog has filtered revisions')) |
|
2850 | 2852 | |
|
2851 | 2853 | # lazydelta and lazydeltabase controls whether to reuse a cached delta, |
|
2852 | 2854 | # if possible. |
|
2853 | 2855 | oldlazydelta = destrevlog._lazydelta |
|
2854 | 2856 | oldlazydeltabase = destrevlog._lazydeltabase |
|
2855 | 2857 | oldamd = destrevlog._deltabothparents |
|
2856 | 2858 | |
|
2857 | 2859 | try: |
|
2858 | 2860 | if deltareuse == self.DELTAREUSEALWAYS: |
|
2859 | 2861 | destrevlog._lazydeltabase = True |
|
2860 | 2862 | destrevlog._lazydelta = True |
|
2861 | 2863 | elif deltareuse == self.DELTAREUSESAMEREVS: |
|
2862 | 2864 | destrevlog._lazydeltabase = False |
|
2863 | 2865 | destrevlog._lazydelta = True |
|
2864 | 2866 | elif deltareuse == self.DELTAREUSENEVER: |
|
2865 | 2867 | destrevlog._lazydeltabase = False |
|
2866 | 2868 | destrevlog._lazydelta = False |
|
2867 | 2869 | |
|
2868 | 2870 | destrevlog._deltabothparents = forcedeltabothparents or oldamd |
|
2869 | 2871 | |
|
2870 | 2872 | self._clone( |
|
2871 | 2873 | tr, |
|
2872 | 2874 | destrevlog, |
|
2873 | 2875 | addrevisioncb, |
|
2874 | 2876 | deltareuse, |
|
2875 | 2877 | forcedeltabothparents, |
|
2876 | 2878 | sidedatacompanion, |
|
2877 | 2879 | ) |
|
2878 | 2880 | |
|
2879 | 2881 | finally: |
|
2880 | 2882 | destrevlog._lazydelta = oldlazydelta |
|
2881 | 2883 | destrevlog._lazydeltabase = oldlazydeltabase |
|
2882 | 2884 | destrevlog._deltabothparents = oldamd |
|
2883 | 2885 | |
|
2884 | 2886 | def _clone( |
|
2885 | 2887 | self, |
|
2886 | 2888 | tr, |
|
2887 | 2889 | destrevlog, |
|
2888 | 2890 | addrevisioncb, |
|
2889 | 2891 | deltareuse, |
|
2890 | 2892 | forcedeltabothparents, |
|
2891 | 2893 | sidedatacompanion, |
|
2892 | 2894 | ): |
|
2893 | 2895 | """perform the core duty of `revlog.clone` after parameter processing""" |
|
2894 | 2896 | deltacomputer = deltautil.deltacomputer(destrevlog) |
|
2895 | 2897 | index = self.index |
|
2896 | 2898 | for rev in self: |
|
2897 | 2899 | entry = index[rev] |
|
2898 | 2900 | |
|
2899 | 2901 | # Some classes override linkrev to take filtered revs into |
|
2900 | 2902 | # account. Use raw entry from index. |
|
2901 | 2903 | flags = entry[0] & 0xFFFF |
|
2902 | 2904 | linkrev = entry[4] |
|
2903 | 2905 | p1 = index[entry[5]][7] |
|
2904 | 2906 | p2 = index[entry[6]][7] |
|
2905 | 2907 | node = entry[7] |
|
2906 | 2908 | |
|
2907 | 2909 | sidedataactions = (False, [], {}, 0, 0) |
|
2908 | 2910 | if sidedatacompanion is not None: |
|
2909 | 2911 | sidedataactions = sidedatacompanion(self, rev) |
|
2910 | 2912 | |
|
2911 | 2913 | # (Possibly) reuse the delta from the revlog if allowed and |
|
2912 | 2914 | # the revlog chunk is a delta. |
|
2913 | 2915 | cachedelta = None |
|
2914 | 2916 | rawtext = None |
|
2915 | 2917 | if any(sidedataactions) or deltareuse == self.DELTAREUSEFULLADD: |
|
2916 | 2918 | dropall = sidedataactions[0] |
|
2917 | 2919 | filterout = sidedataactions[1] |
|
2918 | 2920 | update = sidedataactions[2] |
|
2919 | 2921 | new_flags = sidedataactions[3] |
|
2920 | 2922 | dropped_flags = sidedataactions[4] |
|
2921 | 2923 | text, sidedata = self._revisiondata(rev) |
|
2922 | 2924 | if dropall: |
|
2923 | 2925 | sidedata = {} |
|
2924 | 2926 | for key in filterout: |
|
2925 | 2927 | sidedata.pop(key, None) |
|
2926 | 2928 | sidedata.update(update) |
|
2927 | 2929 | if not sidedata: |
|
2928 | 2930 | sidedata = None |
|
2929 | 2931 | |
|
2930 | 2932 | flags |= new_flags |
|
2931 | 2933 | flags &= ~dropped_flags |
|
2932 | 2934 | |
|
2933 | 2935 | destrevlog.addrevision( |
|
2934 | 2936 | text, |
|
2935 | 2937 | tr, |
|
2936 | 2938 | linkrev, |
|
2937 | 2939 | p1, |
|
2938 | 2940 | p2, |
|
2939 | 2941 | cachedelta=cachedelta, |
|
2940 | 2942 | node=node, |
|
2941 | 2943 | flags=flags, |
|
2942 | 2944 | deltacomputer=deltacomputer, |
|
2943 | 2945 | sidedata=sidedata, |
|
2944 | 2946 | ) |
|
2945 | 2947 | else: |
|
2946 | 2948 | if destrevlog._lazydelta: |
|
2947 | 2949 | dp = self.deltaparent(rev) |
|
2948 | 2950 | if dp != nullrev: |
|
2949 | 2951 | cachedelta = (dp, bytes(self._chunk(rev))) |
|
2950 | 2952 | |
|
2951 | 2953 | if not cachedelta: |
|
2952 | 2954 | rawtext = self.rawdata(rev) |
|
2953 | 2955 | |
|
2954 | 2956 | ifh = destrevlog.opener( |
|
2955 | 2957 | destrevlog.indexfile, b'a+', checkambig=False |
|
2956 | 2958 | ) |
|
2957 | 2959 | dfh = None |
|
2958 | 2960 | if not destrevlog._inline: |
|
2959 | 2961 | dfh = destrevlog.opener(destrevlog.datafile, b'a+') |
|
2960 | 2962 | try: |
|
2961 | 2963 | destrevlog._addrevision( |
|
2962 | 2964 | node, |
|
2963 | 2965 | rawtext, |
|
2964 | 2966 | tr, |
|
2965 | 2967 | linkrev, |
|
2966 | 2968 | p1, |
|
2967 | 2969 | p2, |
|
2968 | 2970 | flags, |
|
2969 | 2971 | cachedelta, |
|
2970 | 2972 | ifh, |
|
2971 | 2973 | dfh, |
|
2972 | 2974 | deltacomputer=deltacomputer, |
|
2973 | 2975 | ) |
|
2974 | 2976 | finally: |
|
2975 | 2977 | if dfh: |
|
2976 | 2978 | dfh.close() |
|
2977 | 2979 | ifh.close() |
|
2978 | 2980 | |
|
2979 | 2981 | if addrevisioncb: |
|
2980 | 2982 | addrevisioncb(self, rev, node) |
|
2981 | 2983 | |
|
2982 | 2984 | def censorrevision(self, tr, censornode, tombstone=b''): |
|
2983 | 2985 | if (self.version & 0xFFFF) == REVLOGV0: |
|
2984 | 2986 | raise error.RevlogError( |
|
2985 | 2987 | _(b'cannot censor with version %d revlogs') % self.version |
|
2986 | 2988 | ) |
|
2987 | 2989 | |
|
2988 | 2990 | censorrev = self.rev(censornode) |
|
2989 | 2991 | tombstone = storageutil.packmeta({b'censored': tombstone}, b'') |
|
2990 | 2992 | |
|
2991 | 2993 | if len(tombstone) > self.rawsize(censorrev): |
|
2992 | 2994 | raise error.Abort( |
|
2993 | 2995 | _(b'censor tombstone must be no longer than censored data') |
|
2994 | 2996 | ) |
|
2995 | 2997 | |
|
2996 | 2998 | # Rewriting the revlog in place is hard. Our strategy for censoring is |
|
2997 | 2999 | # to create a new revlog, copy all revisions to it, then replace the |
|
2998 | 3000 | # revlogs on transaction close. |
|
2999 | 3001 | |
|
3000 | 3002 | newindexfile = self.indexfile + b'.tmpcensored' |
|
3001 | 3003 | newdatafile = self.datafile + b'.tmpcensored' |
|
3002 | 3004 | |
|
3003 | 3005 | # This is a bit dangerous. We could easily have a mismatch of state. |
|
3004 | 3006 | newrl = revlog(self.opener, newindexfile, newdatafile, censorable=True) |
|
3005 | 3007 | newrl.version = self.version |
|
3006 | 3008 | newrl._generaldelta = self._generaldelta |
|
3007 | 3009 | newrl._io = self._io |
|
3008 | 3010 | |
|
3009 | 3011 | for rev in self.revs(): |
|
3010 | 3012 | node = self.node(rev) |
|
3011 | 3013 | p1, p2 = self.parents(node) |
|
3012 | 3014 | |
|
3013 | 3015 | if rev == censorrev: |
|
3014 | 3016 | newrl.addrawrevision( |
|
3015 | 3017 | tombstone, |
|
3016 | 3018 | tr, |
|
3017 | 3019 | self.linkrev(censorrev), |
|
3018 | 3020 | p1, |
|
3019 | 3021 | p2, |
|
3020 | 3022 | censornode, |
|
3021 | 3023 | REVIDX_ISCENSORED, |
|
3022 | 3024 | ) |
|
3023 | 3025 | |
|
3024 | 3026 | if newrl.deltaparent(rev) != nullrev: |
|
3025 | 3027 | raise error.Abort( |
|
3026 | 3028 | _( |
|
3027 | 3029 | b'censored revision stored as delta; ' |
|
3028 | 3030 | b'cannot censor' |
|
3029 | 3031 | ), |
|
3030 | 3032 | hint=_( |
|
3031 | 3033 | b'censoring of revlogs is not ' |
|
3032 | 3034 | b'fully implemented; please report ' |
|
3033 | 3035 | b'this bug' |
|
3034 | 3036 | ), |
|
3035 | 3037 | ) |
|
3036 | 3038 | continue |
|
3037 | 3039 | |
|
3038 | 3040 | if self.iscensored(rev): |
|
3039 | 3041 | if self.deltaparent(rev) != nullrev: |
|
3040 | 3042 | raise error.Abort( |
|
3041 | 3043 | _( |
|
3042 | 3044 | b'cannot censor due to censored ' |
|
3043 | 3045 | b'revision having delta stored' |
|
3044 | 3046 | ) |
|
3045 | 3047 | ) |
|
3046 | 3048 | rawtext = self._chunk(rev) |
|
3047 | 3049 | else: |
|
3048 | 3050 | rawtext = self.rawdata(rev) |
|
3049 | 3051 | |
|
3050 | 3052 | newrl.addrawrevision( |
|
3051 | 3053 | rawtext, tr, self.linkrev(rev), p1, p2, node, self.flags(rev) |
|
3052 | 3054 | ) |
|
3053 | 3055 | |
|
3054 | 3056 | tr.addbackup(self.indexfile, location=b'store') |
|
3055 | 3057 | if not self._inline: |
|
3056 | 3058 | tr.addbackup(self.datafile, location=b'store') |
|
3057 | 3059 | |
|
3058 | 3060 | self.opener.rename(newrl.indexfile, self.indexfile) |
|
3059 | 3061 | if not self._inline: |
|
3060 | 3062 | self.opener.rename(newrl.datafile, self.datafile) |
|
3061 | 3063 | |
|
3062 | 3064 | self.clearcaches() |
|
3063 | 3065 | self._loadindex() |
|
3064 | 3066 | |
|
3065 | 3067 | def verifyintegrity(self, state): |
|
3066 | 3068 | """Verifies the integrity of the revlog. |
|
3067 | 3069 | |
|
3068 | 3070 | Yields ``revlogproblem`` instances describing problems that are |
|
3069 | 3071 | found. |
|
3070 | 3072 | """ |
|
3071 | 3073 | dd, di = self.checksize() |
|
3072 | 3074 | if dd: |
|
3073 | 3075 | yield revlogproblem(error=_(b'data length off by %d bytes') % dd) |
|
3074 | 3076 | if di: |
|
3075 | 3077 | yield revlogproblem(error=_(b'index contains %d extra bytes') % di) |
|
3076 | 3078 | |
|
3077 | 3079 | version = self.version & 0xFFFF |
|
3078 | 3080 | |
|
3079 | 3081 | # The verifier tells us what version revlog we should be. |
|
3080 | 3082 | if version != state[b'expectedversion']: |
|
3081 | 3083 | yield revlogproblem( |
|
3082 | 3084 | warning=_(b"warning: '%s' uses revlog format %d; expected %d") |
|
3083 | 3085 | % (self.indexfile, version, state[b'expectedversion']) |
|
3084 | 3086 | ) |
|
3085 | 3087 | |
|
3086 | 3088 | state[b'skipread'] = set() |
|
3087 | 3089 | state[b'safe_renamed'] = set() |
|
3088 | 3090 | |
|
3089 | 3091 | for rev in self: |
|
3090 | 3092 | node = self.node(rev) |
|
3091 | 3093 | |
|
3092 | 3094 | # Verify contents. 4 cases to care about: |
|
3093 | 3095 | # |
|
3094 | 3096 | # common: the most common case |
|
3095 | 3097 | # rename: with a rename |
|
3096 | 3098 | # meta: file content starts with b'\1\n', the metadata |
|
3097 | 3099 | # header defined in filelog.py, but without a rename |
|
3098 | 3100 | # ext: content stored externally |
|
3099 | 3101 | # |
|
3100 | 3102 | # More formally, their differences are shown below: |
|
3101 | 3103 | # |
|
3102 | 3104 | # | common | rename | meta | ext |
|
3103 | 3105 | # ------------------------------------------------------- |
|
3104 | 3106 | # flags() | 0 | 0 | 0 | not 0 |
|
3105 | 3107 | # renamed() | False | True | False | ? |
|
3106 | 3108 | # rawtext[0:2]=='\1\n'| False | True | True | ? |
|
3107 | 3109 | # |
|
3108 | 3110 | # "rawtext" means the raw text stored in revlog data, which |
|
3109 | 3111 | # could be retrieved by "rawdata(rev)". "text" |
|
3110 | 3112 | # mentioned below is "revision(rev)". |
|
3111 | 3113 | # |
|
3112 | 3114 | # There are 3 different lengths stored physically: |
|
3113 | 3115 | # 1. L1: rawsize, stored in revlog index |
|
3114 | 3116 | # 2. L2: len(rawtext), stored in revlog data |
|
3115 | 3117 | # 3. L3: len(text), stored in revlog data if flags==0, or |
|
3116 | 3118 | # possibly somewhere else if flags!=0 |
|
3117 | 3119 | # |
|
3118 | 3120 | # L1 should be equal to L2. L3 could be different from them. |
|
3119 | 3121 | # "text" may or may not affect commit hash depending on flag |
|
3120 | 3122 | # processors (see flagutil.addflagprocessor). |
|
3121 | 3123 | # |
|
3122 | 3124 | # | common | rename | meta | ext |
|
3123 | 3125 | # ------------------------------------------------- |
|
3124 | 3126 | # rawsize() | L1 | L1 | L1 | L1 |
|
3125 | 3127 | # size() | L1 | L2-LM | L1(*) | L1 (?) |
|
3126 | 3128 | # len(rawtext) | L2 | L2 | L2 | L2 |
|
3127 | 3129 | # len(text) | L2 | L2 | L2 | L3 |
|
3128 | 3130 | # len(read()) | L2 | L2-LM | L2-LM | L3 (?) |
|
3129 | 3131 | # |
|
3130 | 3132 | # LM: length of metadata, depending on rawtext |
|
3131 | 3133 | # (*): not ideal, see comment in filelog.size |
|
3132 | 3134 | # (?): could be "- len(meta)" if the resolved content has |
|
3133 | 3135 | # rename metadata |
|
3134 | 3136 | # |
|
3135 | 3137 | # Checks needed to be done: |
|
3136 | 3138 | # 1. length check: L1 == L2, in all cases. |
|
3137 | 3139 | # 2. hash check: depending on flag processor, we may need to |
|
3138 | 3140 | # use either "text" (external), or "rawtext" (in revlog). |
|
3139 | 3141 | |
|
3140 | 3142 | try: |
|
3141 | 3143 | skipflags = state.get(b'skipflags', 0) |
|
3142 | 3144 | if skipflags: |
|
3143 | 3145 | skipflags &= self.flags(rev) |
|
3144 | 3146 | |
|
3145 | 3147 | _verify_revision(self, skipflags, state, node) |
|
3146 | 3148 | |
|
3147 | 3149 | l1 = self.rawsize(rev) |
|
3148 | 3150 | l2 = len(self.rawdata(node)) |
|
3149 | 3151 | |
|
3150 | 3152 | if l1 != l2: |
|
3151 | 3153 | yield revlogproblem( |
|
3152 | 3154 | error=_(b'unpacked size is %d, %d expected') % (l2, l1), |
|
3153 | 3155 | node=node, |
|
3154 | 3156 | ) |
|
3155 | 3157 | |
|
3156 | 3158 | except error.CensoredNodeError: |
|
3157 | 3159 | if state[b'erroroncensored']: |
|
3158 | 3160 | yield revlogproblem( |
|
3159 | 3161 | error=_(b'censored file data'), node=node |
|
3160 | 3162 | ) |
|
3161 | 3163 | state[b'skipread'].add(node) |
|
3162 | 3164 | except Exception as e: |
|
3163 | 3165 | yield revlogproblem( |
|
3164 | 3166 | error=_(b'unpacking %s: %s') |
|
3165 | 3167 | % (short(node), stringutil.forcebytestr(e)), |
|
3166 | 3168 | node=node, |
|
3167 | 3169 | ) |
|
3168 | 3170 | state[b'skipread'].add(node) |
|
3169 | 3171 | |
|
3170 | 3172 | def storageinfo( |
|
3171 | 3173 | self, |
|
3172 | 3174 | exclusivefiles=False, |
|
3173 | 3175 | sharedfiles=False, |
|
3174 | 3176 | revisionscount=False, |
|
3175 | 3177 | trackedsize=False, |
|
3176 | 3178 | storedsize=False, |
|
3177 | 3179 | ): |
|
3178 | 3180 | d = {} |
|
3179 | 3181 | |
|
3180 | 3182 | if exclusivefiles: |
|
3181 | 3183 | d[b'exclusivefiles'] = [(self.opener, self.indexfile)] |
|
3182 | 3184 | if not self._inline: |
|
3183 | 3185 | d[b'exclusivefiles'].append((self.opener, self.datafile)) |
|
3184 | 3186 | |
|
3185 | 3187 | if sharedfiles: |
|
3186 | 3188 | d[b'sharedfiles'] = [] |
|
3187 | 3189 | |
|
3188 | 3190 | if revisionscount: |
|
3189 | 3191 | d[b'revisionscount'] = len(self) |
|
3190 | 3192 | |
|
3191 | 3193 | if trackedsize: |
|
3192 | 3194 | d[b'trackedsize'] = sum(map(self.rawsize, iter(self))) |
|
3193 | 3195 | |
|
3194 | 3196 | if storedsize: |
|
3195 | 3197 | d[b'storedsize'] = sum( |
|
3196 | 3198 | self.opener.stat(path).st_size for path in self.files() |
|
3197 | 3199 | ) |
|
3198 | 3200 | |
|
3199 | 3201 | return d |
@@ -1,1450 +1,1454 b'' | |||
|
1 | 1 | # storage.py - Testing of storage primitives. |
|
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 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import unittest |
|
11 | 11 | |
|
12 | 12 | from ..node import ( |
|
13 | 13 | hex, |
|
14 | 14 | nullid, |
|
15 | 15 | nullrev, |
|
16 | 16 | ) |
|
17 | 17 | from ..pycompat import getattr |
|
18 | 18 | from .. import ( |
|
19 | 19 | error, |
|
20 | 20 | mdiff, |
|
21 | 21 | ) |
|
22 | 22 | from ..interfaces import repository |
|
23 | 23 | from ..utils import storageutil |
|
24 | 24 | |
|
25 | 25 | |
|
26 | 26 | class basetestcase(unittest.TestCase): |
|
27 | 27 | if not getattr(unittest.TestCase, 'assertRaisesRegex', False): |
|
28 | 28 | assertRaisesRegex = ( # camelcase-required |
|
29 | 29 | unittest.TestCase.assertRaisesRegexp |
|
30 | 30 | ) |
|
31 | 31 | |
|
32 | 32 | |
|
33 | 33 | class ifileindextests(basetestcase): |
|
34 | 34 | """Generic tests for the ifileindex interface. |
|
35 | 35 | |
|
36 | 36 | All file storage backends for index data should conform to the tests in this |
|
37 | 37 | class. |
|
38 | 38 | |
|
39 | 39 | Use ``makeifileindextests()`` to create an instance of this type. |
|
40 | 40 | """ |
|
41 | 41 | |
|
42 | 42 | def testempty(self): |
|
43 | 43 | f = self._makefilefn() |
|
44 | 44 | self.assertEqual(len(f), 0, b'new file store has 0 length by default') |
|
45 | 45 | self.assertEqual(list(f), [], b'iter yields nothing by default') |
|
46 | 46 | |
|
47 | 47 | gen = iter(f) |
|
48 | 48 | with self.assertRaises(StopIteration): |
|
49 | 49 | next(gen) |
|
50 | 50 | |
|
51 | 51 | self.assertFalse(f.hasnode(None)) |
|
52 | 52 | self.assertFalse(f.hasnode(0)) |
|
53 | 53 | self.assertFalse(f.hasnode(nullrev)) |
|
54 | 54 | self.assertFalse(f.hasnode(nullid)) |
|
55 | 55 | self.assertFalse(f.hasnode(b'0')) |
|
56 | 56 | self.assertFalse(f.hasnode(b'a' * 20)) |
|
57 | 57 | |
|
58 | 58 | # revs() should evaluate to an empty list. |
|
59 | 59 | self.assertEqual(list(f.revs()), []) |
|
60 | 60 | |
|
61 | 61 | revs = iter(f.revs()) |
|
62 | 62 | with self.assertRaises(StopIteration): |
|
63 | 63 | next(revs) |
|
64 | 64 | |
|
65 | 65 | self.assertEqual(list(f.revs(start=20)), []) |
|
66 | 66 | |
|
67 | 67 | # parents() and parentrevs() work with nullid/nullrev. |
|
68 | 68 | self.assertEqual(f.parents(nullid), (nullid, nullid)) |
|
69 | 69 | self.assertEqual(f.parentrevs(nullrev), (nullrev, nullrev)) |
|
70 | 70 | |
|
71 | 71 | with self.assertRaises(error.LookupError): |
|
72 | 72 | f.parents(b'\x01' * 20) |
|
73 | 73 | |
|
74 | 74 | for i in range(-5, 5): |
|
75 | 75 | if i == nullrev: |
|
76 | 76 | continue |
|
77 | 77 | |
|
78 | 78 | with self.assertRaises(IndexError): |
|
79 | 79 | f.parentrevs(i) |
|
80 | 80 | |
|
81 | 81 | # nullid/nullrev lookup always works. |
|
82 | 82 | self.assertEqual(f.rev(nullid), nullrev) |
|
83 | 83 | self.assertEqual(f.node(nullrev), nullid) |
|
84 | 84 | |
|
85 | 85 | with self.assertRaises(error.LookupError): |
|
86 | 86 | f.rev(b'\x01' * 20) |
|
87 | 87 | |
|
88 | 88 | for i in range(-5, 5): |
|
89 | 89 | if i == nullrev: |
|
90 | 90 | continue |
|
91 | 91 | |
|
92 | 92 | with self.assertRaises(IndexError): |
|
93 | 93 | f.node(i) |
|
94 | 94 | |
|
95 | 95 | self.assertEqual(f.lookup(nullid), nullid) |
|
96 | 96 | self.assertEqual(f.lookup(nullrev), nullid) |
|
97 | 97 | self.assertEqual(f.lookup(hex(nullid)), nullid) |
|
98 | 98 | self.assertEqual(f.lookup(b'%d' % nullrev), nullid) |
|
99 | 99 | |
|
100 | 100 | with self.assertRaises(error.LookupError): |
|
101 | 101 | f.lookup(b'badvalue') |
|
102 | 102 | |
|
103 | 103 | with self.assertRaises(error.LookupError): |
|
104 | 104 | f.lookup(hex(nullid)[0:12]) |
|
105 | 105 | |
|
106 | 106 | with self.assertRaises(error.LookupError): |
|
107 | 107 | f.lookup(b'-2') |
|
108 | 108 | |
|
109 | 109 | with self.assertRaises(error.LookupError): |
|
110 | 110 | f.lookup(b'0') |
|
111 | 111 | |
|
112 | 112 | with self.assertRaises(error.LookupError): |
|
113 | 113 | f.lookup(b'1') |
|
114 | 114 | |
|
115 | 115 | with self.assertRaises(error.LookupError): |
|
116 | 116 | f.lookup(b'11111111111111111111111111111111111111') |
|
117 | 117 | |
|
118 | 118 | for i in range(-5, 5): |
|
119 | 119 | if i == nullrev: |
|
120 | 120 | continue |
|
121 | 121 | |
|
122 | 122 | with self.assertRaises(LookupError): |
|
123 | 123 | f.lookup(i) |
|
124 | 124 | |
|
125 | 125 | self.assertEqual(f.linkrev(nullrev), nullrev) |
|
126 | 126 | |
|
127 | 127 | for i in range(-5, 5): |
|
128 | 128 | if i == nullrev: |
|
129 | 129 | continue |
|
130 | 130 | |
|
131 | 131 | with self.assertRaises(IndexError): |
|
132 | 132 | f.linkrev(i) |
|
133 | 133 | |
|
134 | 134 | self.assertFalse(f.iscensored(nullrev)) |
|
135 | 135 | |
|
136 | 136 | for i in range(-5, 5): |
|
137 | 137 | if i == nullrev: |
|
138 | 138 | continue |
|
139 | 139 | |
|
140 | 140 | with self.assertRaises(IndexError): |
|
141 | 141 | f.iscensored(i) |
|
142 | 142 | |
|
143 | 143 | self.assertEqual(list(f.commonancestorsheads(nullid, nullid)), []) |
|
144 | 144 | |
|
145 | 145 | with self.assertRaises(ValueError): |
|
146 | 146 | self.assertEqual(list(f.descendants([])), []) |
|
147 | 147 | |
|
148 | 148 | self.assertEqual(list(f.descendants([nullrev])), []) |
|
149 | 149 | |
|
150 | 150 | self.assertEqual(f.heads(), [nullid]) |
|
151 | 151 | self.assertEqual(f.heads(nullid), [nullid]) |
|
152 | 152 | self.assertEqual(f.heads(None, [nullid]), [nullid]) |
|
153 | 153 | self.assertEqual(f.heads(nullid, [nullid]), [nullid]) |
|
154 | 154 | |
|
155 | 155 | self.assertEqual(f.children(nullid), []) |
|
156 | 156 | |
|
157 | 157 | with self.assertRaises(error.LookupError): |
|
158 | 158 | f.children(b'\x01' * 20) |
|
159 | 159 | |
|
160 | 160 | def testsinglerevision(self): |
|
161 | 161 | f = self._makefilefn() |
|
162 | 162 | with self._maketransactionfn() as tr: |
|
163 | 163 | node = f.add(b'initial', None, tr, 0, nullid, nullid) |
|
164 | 164 | |
|
165 | 165 | self.assertEqual(len(f), 1) |
|
166 | 166 | self.assertEqual(list(f), [0]) |
|
167 | 167 | |
|
168 | 168 | gen = iter(f) |
|
169 | 169 | self.assertEqual(next(gen), 0) |
|
170 | 170 | |
|
171 | 171 | with self.assertRaises(StopIteration): |
|
172 | 172 | next(gen) |
|
173 | 173 | |
|
174 | 174 | self.assertTrue(f.hasnode(node)) |
|
175 | 175 | self.assertFalse(f.hasnode(hex(node))) |
|
176 | 176 | self.assertFalse(f.hasnode(nullrev)) |
|
177 | 177 | self.assertFalse(f.hasnode(nullid)) |
|
178 | 178 | self.assertFalse(f.hasnode(node[0:12])) |
|
179 | 179 | self.assertFalse(f.hasnode(hex(node)[0:20])) |
|
180 | 180 | |
|
181 | 181 | self.assertEqual(list(f.revs()), [0]) |
|
182 | 182 | self.assertEqual(list(f.revs(start=1)), []) |
|
183 | 183 | self.assertEqual(list(f.revs(start=0)), [0]) |
|
184 | 184 | self.assertEqual(list(f.revs(stop=0)), [0]) |
|
185 | 185 | self.assertEqual(list(f.revs(stop=1)), [0]) |
|
186 | 186 | self.assertEqual(list(f.revs(1, 1)), []) |
|
187 | 187 | # TODO buggy |
|
188 | 188 | self.assertEqual(list(f.revs(1, 0)), [1, 0]) |
|
189 | 189 | self.assertEqual(list(f.revs(2, 0)), [2, 1, 0]) |
|
190 | 190 | |
|
191 | 191 | self.assertEqual(f.parents(node), (nullid, nullid)) |
|
192 | 192 | self.assertEqual(f.parentrevs(0), (nullrev, nullrev)) |
|
193 | 193 | |
|
194 | 194 | with self.assertRaises(error.LookupError): |
|
195 | 195 | f.parents(b'\x01' * 20) |
|
196 | 196 | |
|
197 | 197 | with self.assertRaises(IndexError): |
|
198 | 198 | f.parentrevs(1) |
|
199 | 199 | |
|
200 | 200 | self.assertEqual(f.rev(node), 0) |
|
201 | 201 | |
|
202 | 202 | with self.assertRaises(error.LookupError): |
|
203 | 203 | f.rev(b'\x01' * 20) |
|
204 | 204 | |
|
205 | 205 | self.assertEqual(f.node(0), node) |
|
206 | 206 | |
|
207 | 207 | with self.assertRaises(IndexError): |
|
208 | 208 | f.node(1) |
|
209 | 209 | |
|
210 | 210 | self.assertEqual(f.lookup(node), node) |
|
211 | 211 | self.assertEqual(f.lookup(0), node) |
|
212 | 212 | self.assertEqual(f.lookup(-1), nullid) |
|
213 | 213 | self.assertEqual(f.lookup(b'0'), node) |
|
214 | 214 | self.assertEqual(f.lookup(hex(node)), node) |
|
215 | 215 | |
|
216 | 216 | with self.assertRaises(error.LookupError): |
|
217 | 217 | f.lookup(hex(node)[0:12]) |
|
218 | 218 | |
|
219 | 219 | with self.assertRaises(error.LookupError): |
|
220 | 220 | f.lookup(-2) |
|
221 | 221 | |
|
222 | 222 | with self.assertRaises(error.LookupError): |
|
223 | 223 | f.lookup(b'-2') |
|
224 | 224 | |
|
225 | 225 | with self.assertRaises(error.LookupError): |
|
226 | 226 | f.lookup(1) |
|
227 | 227 | |
|
228 | 228 | with self.assertRaises(error.LookupError): |
|
229 | 229 | f.lookup(b'1') |
|
230 | 230 | |
|
231 | 231 | self.assertEqual(f.linkrev(0), 0) |
|
232 | 232 | |
|
233 | 233 | with self.assertRaises(IndexError): |
|
234 | 234 | f.linkrev(1) |
|
235 | 235 | |
|
236 | 236 | self.assertFalse(f.iscensored(0)) |
|
237 | 237 | |
|
238 | 238 | with self.assertRaises(IndexError): |
|
239 | 239 | f.iscensored(1) |
|
240 | 240 | |
|
241 | 241 | self.assertEqual(list(f.descendants([0])), []) |
|
242 | 242 | |
|
243 | 243 | self.assertEqual(f.heads(), [node]) |
|
244 | 244 | self.assertEqual(f.heads(node), [node]) |
|
245 | 245 | self.assertEqual(f.heads(stop=[node]), [node]) |
|
246 | 246 | |
|
247 | 247 | with self.assertRaises(error.LookupError): |
|
248 | 248 | f.heads(stop=[b'\x01' * 20]) |
|
249 | 249 | |
|
250 | 250 | self.assertEqual(f.children(node), []) |
|
251 | 251 | |
|
252 | 252 | def testmultiplerevisions(self): |
|
253 | 253 | fulltext0 = b'x' * 1024 |
|
254 | 254 | fulltext1 = fulltext0 + b'y' |
|
255 | 255 | fulltext2 = b'y' + fulltext0 + b'z' |
|
256 | 256 | |
|
257 | 257 | f = self._makefilefn() |
|
258 | 258 | with self._maketransactionfn() as tr: |
|
259 | 259 | node0 = f.add(fulltext0, None, tr, 0, nullid, nullid) |
|
260 | 260 | node1 = f.add(fulltext1, None, tr, 1, node0, nullid) |
|
261 | 261 | node2 = f.add(fulltext2, None, tr, 3, node1, nullid) |
|
262 | 262 | |
|
263 | 263 | self.assertEqual(len(f), 3) |
|
264 | 264 | self.assertEqual(list(f), [0, 1, 2]) |
|
265 | 265 | |
|
266 | 266 | gen = iter(f) |
|
267 | 267 | self.assertEqual(next(gen), 0) |
|
268 | 268 | self.assertEqual(next(gen), 1) |
|
269 | 269 | self.assertEqual(next(gen), 2) |
|
270 | 270 | |
|
271 | 271 | with self.assertRaises(StopIteration): |
|
272 | 272 | next(gen) |
|
273 | 273 | |
|
274 | 274 | self.assertEqual(list(f.revs()), [0, 1, 2]) |
|
275 | 275 | self.assertEqual(list(f.revs(0)), [0, 1, 2]) |
|
276 | 276 | self.assertEqual(list(f.revs(1)), [1, 2]) |
|
277 | 277 | self.assertEqual(list(f.revs(2)), [2]) |
|
278 | 278 | self.assertEqual(list(f.revs(3)), []) |
|
279 | 279 | self.assertEqual(list(f.revs(stop=1)), [0, 1]) |
|
280 | 280 | self.assertEqual(list(f.revs(stop=2)), [0, 1, 2]) |
|
281 | 281 | self.assertEqual(list(f.revs(stop=3)), [0, 1, 2]) |
|
282 | 282 | self.assertEqual(list(f.revs(2, 0)), [2, 1, 0]) |
|
283 | 283 | self.assertEqual(list(f.revs(2, 1)), [2, 1]) |
|
284 | 284 | # TODO this is wrong |
|
285 | 285 | self.assertEqual(list(f.revs(3, 2)), [3, 2]) |
|
286 | 286 | |
|
287 | 287 | self.assertEqual(f.parents(node0), (nullid, nullid)) |
|
288 | 288 | self.assertEqual(f.parents(node1), (node0, nullid)) |
|
289 | 289 | self.assertEqual(f.parents(node2), (node1, nullid)) |
|
290 | 290 | |
|
291 | 291 | self.assertEqual(f.parentrevs(0), (nullrev, nullrev)) |
|
292 | 292 | self.assertEqual(f.parentrevs(1), (0, nullrev)) |
|
293 | 293 | self.assertEqual(f.parentrevs(2), (1, nullrev)) |
|
294 | 294 | |
|
295 | 295 | self.assertEqual(f.rev(node0), 0) |
|
296 | 296 | self.assertEqual(f.rev(node1), 1) |
|
297 | 297 | self.assertEqual(f.rev(node2), 2) |
|
298 | 298 | |
|
299 | 299 | with self.assertRaises(error.LookupError): |
|
300 | 300 | f.rev(b'\x01' * 20) |
|
301 | 301 | |
|
302 | 302 | self.assertEqual(f.node(0), node0) |
|
303 | 303 | self.assertEqual(f.node(1), node1) |
|
304 | 304 | self.assertEqual(f.node(2), node2) |
|
305 | 305 | |
|
306 | 306 | with self.assertRaises(IndexError): |
|
307 | 307 | f.node(3) |
|
308 | 308 | |
|
309 | 309 | self.assertEqual(f.lookup(node0), node0) |
|
310 | 310 | self.assertEqual(f.lookup(0), node0) |
|
311 | 311 | self.assertEqual(f.lookup(b'0'), node0) |
|
312 | 312 | self.assertEqual(f.lookup(hex(node0)), node0) |
|
313 | 313 | |
|
314 | 314 | self.assertEqual(f.lookup(node1), node1) |
|
315 | 315 | self.assertEqual(f.lookup(1), node1) |
|
316 | 316 | self.assertEqual(f.lookup(b'1'), node1) |
|
317 | 317 | self.assertEqual(f.lookup(hex(node1)), node1) |
|
318 | 318 | |
|
319 | 319 | self.assertEqual(f.linkrev(0), 0) |
|
320 | 320 | self.assertEqual(f.linkrev(1), 1) |
|
321 | 321 | self.assertEqual(f.linkrev(2), 3) |
|
322 | 322 | |
|
323 | 323 | with self.assertRaises(IndexError): |
|
324 | 324 | f.linkrev(3) |
|
325 | 325 | |
|
326 | 326 | self.assertFalse(f.iscensored(0)) |
|
327 | 327 | self.assertFalse(f.iscensored(1)) |
|
328 | 328 | self.assertFalse(f.iscensored(2)) |
|
329 | 329 | |
|
330 | 330 | with self.assertRaises(IndexError): |
|
331 | 331 | f.iscensored(3) |
|
332 | 332 | |
|
333 | 333 | self.assertEqual(f.commonancestorsheads(node1, nullid), []) |
|
334 | 334 | self.assertEqual(f.commonancestorsheads(node1, node0), [node0]) |
|
335 | 335 | self.assertEqual(f.commonancestorsheads(node1, node1), [node1]) |
|
336 | 336 | self.assertEqual(f.commonancestorsheads(node0, node1), [node0]) |
|
337 | 337 | self.assertEqual(f.commonancestorsheads(node1, node2), [node1]) |
|
338 | 338 | self.assertEqual(f.commonancestorsheads(node2, node1), [node1]) |
|
339 | 339 | |
|
340 | 340 | self.assertEqual(list(f.descendants([0])), [1, 2]) |
|
341 | 341 | self.assertEqual(list(f.descendants([1])), [2]) |
|
342 | 342 | self.assertEqual(list(f.descendants([0, 1])), [1, 2]) |
|
343 | 343 | |
|
344 | 344 | self.assertEqual(f.heads(), [node2]) |
|
345 | 345 | self.assertEqual(f.heads(node0), [node2]) |
|
346 | 346 | self.assertEqual(f.heads(node1), [node2]) |
|
347 | 347 | self.assertEqual(f.heads(node2), [node2]) |
|
348 | 348 | |
|
349 | 349 | # TODO this behavior seems wonky. Is it correct? If so, the |
|
350 | 350 | # docstring for heads() should be updated to reflect desired |
|
351 | 351 | # behavior. |
|
352 | 352 | self.assertEqual(f.heads(stop=[node1]), [node1, node2]) |
|
353 | 353 | self.assertEqual(f.heads(stop=[node0]), [node0, node2]) |
|
354 | 354 | self.assertEqual(f.heads(stop=[node1, node2]), [node1, node2]) |
|
355 | 355 | |
|
356 | 356 | with self.assertRaises(error.LookupError): |
|
357 | 357 | f.heads(stop=[b'\x01' * 20]) |
|
358 | 358 | |
|
359 | 359 | self.assertEqual(f.children(node0), [node1]) |
|
360 | 360 | self.assertEqual(f.children(node1), [node2]) |
|
361 | 361 | self.assertEqual(f.children(node2), []) |
|
362 | 362 | |
|
363 | 363 | def testmultipleheads(self): |
|
364 | 364 | f = self._makefilefn() |
|
365 | 365 | |
|
366 | 366 | with self._maketransactionfn() as tr: |
|
367 | 367 | node0 = f.add(b'0', None, tr, 0, nullid, nullid) |
|
368 | 368 | node1 = f.add(b'1', None, tr, 1, node0, nullid) |
|
369 | 369 | node2 = f.add(b'2', None, tr, 2, node1, nullid) |
|
370 | 370 | node3 = f.add(b'3', None, tr, 3, node0, nullid) |
|
371 | 371 | node4 = f.add(b'4', None, tr, 4, node3, nullid) |
|
372 | 372 | node5 = f.add(b'5', None, tr, 5, node0, nullid) |
|
373 | 373 | |
|
374 | 374 | self.assertEqual(len(f), 6) |
|
375 | 375 | |
|
376 | 376 | self.assertEqual(list(f.descendants([0])), [1, 2, 3, 4, 5]) |
|
377 | 377 | self.assertEqual(list(f.descendants([1])), [2]) |
|
378 | 378 | self.assertEqual(list(f.descendants([2])), []) |
|
379 | 379 | self.assertEqual(list(f.descendants([3])), [4]) |
|
380 | 380 | self.assertEqual(list(f.descendants([0, 1])), [1, 2, 3, 4, 5]) |
|
381 | 381 | self.assertEqual(list(f.descendants([1, 3])), [2, 4]) |
|
382 | 382 | |
|
383 | 383 | self.assertEqual(f.heads(), [node2, node4, node5]) |
|
384 | 384 | self.assertEqual(f.heads(node0), [node2, node4, node5]) |
|
385 | 385 | self.assertEqual(f.heads(node1), [node2]) |
|
386 | 386 | self.assertEqual(f.heads(node2), [node2]) |
|
387 | 387 | self.assertEqual(f.heads(node3), [node4]) |
|
388 | 388 | self.assertEqual(f.heads(node4), [node4]) |
|
389 | 389 | self.assertEqual(f.heads(node5), [node5]) |
|
390 | 390 | |
|
391 | 391 | # TODO this seems wrong. |
|
392 | 392 | self.assertEqual(f.heads(stop=[node0]), [node0, node2, node4, node5]) |
|
393 | 393 | self.assertEqual(f.heads(stop=[node1]), [node1, node2, node4, node5]) |
|
394 | 394 | |
|
395 | 395 | self.assertEqual(f.children(node0), [node1, node3, node5]) |
|
396 | 396 | self.assertEqual(f.children(node1), [node2]) |
|
397 | 397 | self.assertEqual(f.children(node2), []) |
|
398 | 398 | self.assertEqual(f.children(node3), [node4]) |
|
399 | 399 | self.assertEqual(f.children(node4), []) |
|
400 | 400 | self.assertEqual(f.children(node5), []) |
|
401 | 401 | |
|
402 | 402 | |
|
403 | 403 | class ifiledatatests(basetestcase): |
|
404 | 404 | """Generic tests for the ifiledata interface. |
|
405 | 405 | |
|
406 | 406 | All file storage backends for data should conform to the tests in this |
|
407 | 407 | class. |
|
408 | 408 | |
|
409 | 409 | Use ``makeifiledatatests()`` to create an instance of this type. |
|
410 | 410 | """ |
|
411 | 411 | |
|
412 | 412 | def testempty(self): |
|
413 | 413 | f = self._makefilefn() |
|
414 | 414 | |
|
415 | 415 | self.assertEqual(f.storageinfo(), {}) |
|
416 | 416 | self.assertEqual( |
|
417 | 417 | f.storageinfo(revisionscount=True, trackedsize=True), |
|
418 | 418 | {b'revisionscount': 0, b'trackedsize': 0}, |
|
419 | 419 | ) |
|
420 | 420 | |
|
421 | 421 | self.assertEqual(f.size(nullrev), 0) |
|
422 | 422 | |
|
423 | 423 | for i in range(-5, 5): |
|
424 | 424 | if i == nullrev: |
|
425 | 425 | continue |
|
426 | 426 | |
|
427 | 427 | with self.assertRaises(IndexError): |
|
428 | 428 | f.size(i) |
|
429 | 429 | |
|
430 | 430 | self.assertEqual(f.revision(nullid), b'') |
|
431 | 431 | self.assertEqual(f.rawdata(nullid), b'') |
|
432 | 432 | |
|
433 | 433 | with self.assertRaises(error.LookupError): |
|
434 | 434 | f.revision(b'\x01' * 20) |
|
435 | 435 | |
|
436 | 436 | self.assertEqual(f.read(nullid), b'') |
|
437 | 437 | |
|
438 | 438 | with self.assertRaises(error.LookupError): |
|
439 | 439 | f.read(b'\x01' * 20) |
|
440 | 440 | |
|
441 | 441 | self.assertFalse(f.renamed(nullid)) |
|
442 | 442 | |
|
443 | 443 | with self.assertRaises(error.LookupError): |
|
444 | 444 | f.read(b'\x01' * 20) |
|
445 | 445 | |
|
446 | 446 | self.assertTrue(f.cmp(nullid, b'')) |
|
447 | 447 | self.assertTrue(f.cmp(nullid, b'foo')) |
|
448 | 448 | |
|
449 | 449 | with self.assertRaises(error.LookupError): |
|
450 | 450 | f.cmp(b'\x01' * 20, b'irrelevant') |
|
451 | 451 | |
|
452 | 452 | # Emitting empty list is an empty generator. |
|
453 | 453 | gen = f.emitrevisions([]) |
|
454 | 454 | with self.assertRaises(StopIteration): |
|
455 | 455 | next(gen) |
|
456 | 456 | |
|
457 | 457 | # Emitting null node yields nothing. |
|
458 | 458 | gen = f.emitrevisions([nullid]) |
|
459 | 459 | with self.assertRaises(StopIteration): |
|
460 | 460 | next(gen) |
|
461 | 461 | |
|
462 | 462 | # Requesting unknown node fails. |
|
463 | 463 | with self.assertRaises(error.LookupError): |
|
464 | 464 | list(f.emitrevisions([b'\x01' * 20])) |
|
465 | 465 | |
|
466 | 466 | def testsinglerevision(self): |
|
467 | 467 | fulltext = b'initial' |
|
468 | 468 | |
|
469 | 469 | f = self._makefilefn() |
|
470 | 470 | with self._maketransactionfn() as tr: |
|
471 | 471 | node = f.add(fulltext, None, tr, 0, nullid, nullid) |
|
472 | 472 | |
|
473 | 473 | self.assertEqual(f.storageinfo(), {}) |
|
474 | 474 | self.assertEqual( |
|
475 | 475 | f.storageinfo(revisionscount=True, trackedsize=True), |
|
476 | 476 | {b'revisionscount': 1, b'trackedsize': len(fulltext)}, |
|
477 | 477 | ) |
|
478 | 478 | |
|
479 | 479 | self.assertEqual(f.size(0), len(fulltext)) |
|
480 | 480 | |
|
481 | 481 | with self.assertRaises(IndexError): |
|
482 | 482 | f.size(1) |
|
483 | 483 | |
|
484 | 484 | self.assertEqual(f.revision(node), fulltext) |
|
485 | 485 | self.assertEqual(f.rawdata(node), fulltext) |
|
486 | 486 | |
|
487 | 487 | self.assertEqual(f.read(node), fulltext) |
|
488 | 488 | |
|
489 | 489 | self.assertFalse(f.renamed(node)) |
|
490 | 490 | |
|
491 | 491 | self.assertFalse(f.cmp(node, fulltext)) |
|
492 | 492 | self.assertTrue(f.cmp(node, fulltext + b'extra')) |
|
493 | 493 | |
|
494 | 494 | # Emitting a single revision works. |
|
495 | 495 | gen = f.emitrevisions([node]) |
|
496 | 496 | rev = next(gen) |
|
497 | 497 | |
|
498 | 498 | self.assertEqual(rev.node, node) |
|
499 | 499 | self.assertEqual(rev.p1node, nullid) |
|
500 | 500 | self.assertEqual(rev.p2node, nullid) |
|
501 | 501 | self.assertIsNone(rev.linknode) |
|
502 | 502 | self.assertEqual(rev.basenode, nullid) |
|
503 | 503 | self.assertIsNone(rev.baserevisionsize) |
|
504 | 504 | self.assertIsNone(rev.revision) |
|
505 | 505 | self.assertIsNone(rev.delta) |
|
506 | 506 | |
|
507 | 507 | with self.assertRaises(StopIteration): |
|
508 | 508 | next(gen) |
|
509 | 509 | |
|
510 | 510 | # Requesting revision data works. |
|
511 | 511 | gen = f.emitrevisions([node], revisiondata=True) |
|
512 | 512 | rev = next(gen) |
|
513 | 513 | |
|
514 | 514 | self.assertEqual(rev.node, node) |
|
515 | 515 | self.assertEqual(rev.p1node, nullid) |
|
516 | 516 | self.assertEqual(rev.p2node, nullid) |
|
517 | 517 | self.assertIsNone(rev.linknode) |
|
518 | 518 | self.assertEqual(rev.basenode, nullid) |
|
519 | 519 | self.assertIsNone(rev.baserevisionsize) |
|
520 | 520 | self.assertEqual(rev.revision, fulltext) |
|
521 | 521 | self.assertIsNone(rev.delta) |
|
522 | 522 | |
|
523 | 523 | with self.assertRaises(StopIteration): |
|
524 | 524 | next(gen) |
|
525 | 525 | |
|
526 | 526 | # Emitting an unknown node after a known revision results in error. |
|
527 | 527 | with self.assertRaises(error.LookupError): |
|
528 | 528 | list(f.emitrevisions([node, b'\x01' * 20])) |
|
529 | 529 | |
|
530 | 530 | def testmultiplerevisions(self): |
|
531 | 531 | fulltext0 = b'x' * 1024 |
|
532 | 532 | fulltext1 = fulltext0 + b'y' |
|
533 | 533 | fulltext2 = b'y' + fulltext0 + b'z' |
|
534 | 534 | |
|
535 | 535 | f = self._makefilefn() |
|
536 | 536 | with self._maketransactionfn() as tr: |
|
537 | 537 | node0 = f.add(fulltext0, None, tr, 0, nullid, nullid) |
|
538 | 538 | node1 = f.add(fulltext1, None, tr, 1, node0, nullid) |
|
539 | 539 | node2 = f.add(fulltext2, None, tr, 3, node1, nullid) |
|
540 | 540 | |
|
541 | 541 | self.assertEqual(f.storageinfo(), {}) |
|
542 | 542 | self.assertEqual( |
|
543 | 543 | f.storageinfo(revisionscount=True, trackedsize=True), |
|
544 | 544 | { |
|
545 | 545 | b'revisionscount': 3, |
|
546 | 546 | b'trackedsize': len(fulltext0) |
|
547 | 547 | + len(fulltext1) |
|
548 | 548 | + len(fulltext2), |
|
549 | 549 | }, |
|
550 | 550 | ) |
|
551 | 551 | |
|
552 | 552 | self.assertEqual(f.size(0), len(fulltext0)) |
|
553 | 553 | self.assertEqual(f.size(1), len(fulltext1)) |
|
554 | 554 | self.assertEqual(f.size(2), len(fulltext2)) |
|
555 | 555 | |
|
556 | 556 | with self.assertRaises(IndexError): |
|
557 | 557 | f.size(3) |
|
558 | 558 | |
|
559 | 559 | self.assertEqual(f.revision(node0), fulltext0) |
|
560 | 560 | self.assertEqual(f.rawdata(node0), fulltext0) |
|
561 | 561 | self.assertEqual(f.revision(node1), fulltext1) |
|
562 | 562 | self.assertEqual(f.rawdata(node1), fulltext1) |
|
563 | 563 | self.assertEqual(f.revision(node2), fulltext2) |
|
564 | 564 | self.assertEqual(f.rawdata(node2), fulltext2) |
|
565 | 565 | |
|
566 | 566 | with self.assertRaises(error.LookupError): |
|
567 | 567 | f.revision(b'\x01' * 20) |
|
568 | 568 | |
|
569 | 569 | self.assertEqual(f.read(node0), fulltext0) |
|
570 | 570 | self.assertEqual(f.read(node1), fulltext1) |
|
571 | 571 | self.assertEqual(f.read(node2), fulltext2) |
|
572 | 572 | |
|
573 | 573 | with self.assertRaises(error.LookupError): |
|
574 | 574 | f.read(b'\x01' * 20) |
|
575 | 575 | |
|
576 | 576 | self.assertFalse(f.renamed(node0)) |
|
577 | 577 | self.assertFalse(f.renamed(node1)) |
|
578 | 578 | self.assertFalse(f.renamed(node2)) |
|
579 | 579 | |
|
580 | 580 | with self.assertRaises(error.LookupError): |
|
581 | 581 | f.renamed(b'\x01' * 20) |
|
582 | 582 | |
|
583 | 583 | self.assertFalse(f.cmp(node0, fulltext0)) |
|
584 | 584 | self.assertFalse(f.cmp(node1, fulltext1)) |
|
585 | 585 | self.assertFalse(f.cmp(node2, fulltext2)) |
|
586 | 586 | |
|
587 | 587 | self.assertTrue(f.cmp(node1, fulltext0)) |
|
588 | 588 | self.assertTrue(f.cmp(node2, fulltext1)) |
|
589 | 589 | |
|
590 | 590 | with self.assertRaises(error.LookupError): |
|
591 | 591 | f.cmp(b'\x01' * 20, b'irrelevant') |
|
592 | 592 | |
|
593 | 593 | # Nodes should be emitted in order. |
|
594 | 594 | gen = f.emitrevisions([node0, node1, node2], revisiondata=True) |
|
595 | 595 | |
|
596 | 596 | rev = next(gen) |
|
597 | 597 | |
|
598 | 598 | self.assertEqual(rev.node, node0) |
|
599 | 599 | self.assertEqual(rev.p1node, nullid) |
|
600 | 600 | self.assertEqual(rev.p2node, nullid) |
|
601 | 601 | self.assertIsNone(rev.linknode) |
|
602 | 602 | self.assertEqual(rev.basenode, nullid) |
|
603 | 603 | self.assertIsNone(rev.baserevisionsize) |
|
604 | 604 | self.assertEqual(rev.revision, fulltext0) |
|
605 | 605 | self.assertIsNone(rev.delta) |
|
606 | 606 | |
|
607 | 607 | rev = next(gen) |
|
608 | 608 | |
|
609 | 609 | self.assertEqual(rev.node, node1) |
|
610 | 610 | self.assertEqual(rev.p1node, node0) |
|
611 | 611 | self.assertEqual(rev.p2node, nullid) |
|
612 | 612 | self.assertIsNone(rev.linknode) |
|
613 | 613 | self.assertEqual(rev.basenode, node0) |
|
614 | 614 | self.assertIsNone(rev.baserevisionsize) |
|
615 | 615 | self.assertIsNone(rev.revision) |
|
616 | 616 | self.assertEqual( |
|
617 | 617 | rev.delta, |
|
618 | 618 | b'\x00\x00\x00\x00\x00\x00\x04\x00\x00\x00\x04\x01' + fulltext1, |
|
619 | 619 | ) |
|
620 | 620 | |
|
621 | 621 | rev = next(gen) |
|
622 | 622 | |
|
623 | 623 | self.assertEqual(rev.node, node2) |
|
624 | 624 | self.assertEqual(rev.p1node, node1) |
|
625 | 625 | self.assertEqual(rev.p2node, nullid) |
|
626 | 626 | self.assertIsNone(rev.linknode) |
|
627 | 627 | self.assertEqual(rev.basenode, node1) |
|
628 | 628 | self.assertIsNone(rev.baserevisionsize) |
|
629 | 629 | self.assertIsNone(rev.revision) |
|
630 | 630 | self.assertEqual( |
|
631 | 631 | rev.delta, |
|
632 | 632 | b'\x00\x00\x00\x00\x00\x00\x04\x01\x00\x00\x04\x02' + fulltext2, |
|
633 | 633 | ) |
|
634 | 634 | |
|
635 | 635 | with self.assertRaises(StopIteration): |
|
636 | 636 | next(gen) |
|
637 | 637 | |
|
638 | 638 | # Request not in DAG order is reordered to be in DAG order. |
|
639 | 639 | gen = f.emitrevisions([node2, node1, node0], revisiondata=True) |
|
640 | 640 | |
|
641 | 641 | rev = next(gen) |
|
642 | 642 | |
|
643 | 643 | self.assertEqual(rev.node, node0) |
|
644 | 644 | self.assertEqual(rev.p1node, nullid) |
|
645 | 645 | self.assertEqual(rev.p2node, nullid) |
|
646 | 646 | self.assertIsNone(rev.linknode) |
|
647 | 647 | self.assertEqual(rev.basenode, nullid) |
|
648 | 648 | self.assertIsNone(rev.baserevisionsize) |
|
649 | 649 | self.assertEqual(rev.revision, fulltext0) |
|
650 | 650 | self.assertIsNone(rev.delta) |
|
651 | 651 | |
|
652 | 652 | rev = next(gen) |
|
653 | 653 | |
|
654 | 654 | self.assertEqual(rev.node, node1) |
|
655 | 655 | self.assertEqual(rev.p1node, node0) |
|
656 | 656 | self.assertEqual(rev.p2node, nullid) |
|
657 | 657 | self.assertIsNone(rev.linknode) |
|
658 | 658 | self.assertEqual(rev.basenode, node0) |
|
659 | 659 | self.assertIsNone(rev.baserevisionsize) |
|
660 | 660 | self.assertIsNone(rev.revision) |
|
661 | 661 | self.assertEqual( |
|
662 | 662 | rev.delta, |
|
663 | 663 | b'\x00\x00\x00\x00\x00\x00\x04\x00\x00\x00\x04\x01' + fulltext1, |
|
664 | 664 | ) |
|
665 | 665 | |
|
666 | 666 | rev = next(gen) |
|
667 | 667 | |
|
668 | 668 | self.assertEqual(rev.node, node2) |
|
669 | 669 | self.assertEqual(rev.p1node, node1) |
|
670 | 670 | self.assertEqual(rev.p2node, nullid) |
|
671 | 671 | self.assertIsNone(rev.linknode) |
|
672 | 672 | self.assertEqual(rev.basenode, node1) |
|
673 | 673 | self.assertIsNone(rev.baserevisionsize) |
|
674 | 674 | self.assertIsNone(rev.revision) |
|
675 | 675 | self.assertEqual( |
|
676 | 676 | rev.delta, |
|
677 | 677 | b'\x00\x00\x00\x00\x00\x00\x04\x01\x00\x00\x04\x02' + fulltext2, |
|
678 | 678 | ) |
|
679 | 679 | |
|
680 | 680 | with self.assertRaises(StopIteration): |
|
681 | 681 | next(gen) |
|
682 | 682 | |
|
683 | 683 | # Unrecognized nodesorder value raises ProgrammingError. |
|
684 | 684 | with self.assertRaises(error.ProgrammingError): |
|
685 | 685 | list(f.emitrevisions([], nodesorder=b'bad')) |
|
686 | 686 | |
|
687 | 687 | # nodesorder=storage is recognized. But we can't test it thoroughly |
|
688 | 688 | # because behavior is storage-dependent. |
|
689 | 689 | res = list( |
|
690 | 690 | f.emitrevisions([node2, node1, node0], nodesorder=b'storage') |
|
691 | 691 | ) |
|
692 | 692 | self.assertEqual(len(res), 3) |
|
693 | 693 | self.assertEqual({o.node for o in res}, {node0, node1, node2}) |
|
694 | 694 | |
|
695 | 695 | # nodesorder=nodes forces the order. |
|
696 | 696 | gen = f.emitrevisions( |
|
697 | 697 | [node2, node0], nodesorder=b'nodes', revisiondata=True |
|
698 | 698 | ) |
|
699 | 699 | |
|
700 | 700 | rev = next(gen) |
|
701 | 701 | self.assertEqual(rev.node, node2) |
|
702 | 702 | self.assertEqual(rev.p1node, node1) |
|
703 | 703 | self.assertEqual(rev.p2node, nullid) |
|
704 | 704 | self.assertEqual(rev.basenode, nullid) |
|
705 | 705 | self.assertIsNone(rev.baserevisionsize) |
|
706 | 706 | self.assertEqual(rev.revision, fulltext2) |
|
707 | 707 | self.assertIsNone(rev.delta) |
|
708 | 708 | |
|
709 | 709 | rev = next(gen) |
|
710 | 710 | self.assertEqual(rev.node, node0) |
|
711 | 711 | self.assertEqual(rev.p1node, nullid) |
|
712 | 712 | self.assertEqual(rev.p2node, nullid) |
|
713 | 713 | # Delta behavior is storage dependent, so we can't easily test it. |
|
714 | 714 | |
|
715 | 715 | with self.assertRaises(StopIteration): |
|
716 | 716 | next(gen) |
|
717 | 717 | |
|
718 | 718 | # assumehaveparentrevisions=False (the default) won't send a delta for |
|
719 | 719 | # the first revision. |
|
720 | 720 | gen = f.emitrevisions({node2, node1}, revisiondata=True) |
|
721 | 721 | |
|
722 | 722 | rev = next(gen) |
|
723 | 723 | self.assertEqual(rev.node, node1) |
|
724 | 724 | self.assertEqual(rev.p1node, node0) |
|
725 | 725 | self.assertEqual(rev.p2node, nullid) |
|
726 | 726 | self.assertEqual(rev.basenode, nullid) |
|
727 | 727 | self.assertIsNone(rev.baserevisionsize) |
|
728 | 728 | self.assertEqual(rev.revision, fulltext1) |
|
729 | 729 | self.assertIsNone(rev.delta) |
|
730 | 730 | |
|
731 | 731 | rev = next(gen) |
|
732 | 732 | self.assertEqual(rev.node, node2) |
|
733 | 733 | self.assertEqual(rev.p1node, node1) |
|
734 | 734 | self.assertEqual(rev.p2node, nullid) |
|
735 | 735 | self.assertEqual(rev.basenode, node1) |
|
736 | 736 | self.assertIsNone(rev.baserevisionsize) |
|
737 | 737 | self.assertIsNone(rev.revision) |
|
738 | 738 | self.assertEqual( |
|
739 | 739 | rev.delta, |
|
740 | 740 | b'\x00\x00\x00\x00\x00\x00\x04\x01\x00\x00\x04\x02' + fulltext2, |
|
741 | 741 | ) |
|
742 | 742 | |
|
743 | 743 | with self.assertRaises(StopIteration): |
|
744 | 744 | next(gen) |
|
745 | 745 | |
|
746 | 746 | # assumehaveparentrevisions=True allows delta against initial revision. |
|
747 | 747 | gen = f.emitrevisions( |
|
748 | 748 | [node2, node1], revisiondata=True, assumehaveparentrevisions=True |
|
749 | 749 | ) |
|
750 | 750 | |
|
751 | 751 | rev = next(gen) |
|
752 | 752 | self.assertEqual(rev.node, node1) |
|
753 | 753 | self.assertEqual(rev.p1node, node0) |
|
754 | 754 | self.assertEqual(rev.p2node, nullid) |
|
755 | 755 | self.assertEqual(rev.basenode, node0) |
|
756 | 756 | self.assertIsNone(rev.baserevisionsize) |
|
757 | 757 | self.assertIsNone(rev.revision) |
|
758 | 758 | self.assertEqual( |
|
759 | 759 | rev.delta, |
|
760 | 760 | b'\x00\x00\x00\x00\x00\x00\x04\x00\x00\x00\x04\x01' + fulltext1, |
|
761 | 761 | ) |
|
762 | 762 | |
|
763 | 763 | # forceprevious=True forces a delta against the previous revision. |
|
764 | 764 | # Special case for initial revision. |
|
765 | 765 | gen = f.emitrevisions( |
|
766 | 766 | [node0], revisiondata=True, deltamode=repository.CG_DELTAMODE_PREV |
|
767 | 767 | ) |
|
768 | 768 | |
|
769 | 769 | rev = next(gen) |
|
770 | 770 | self.assertEqual(rev.node, node0) |
|
771 | 771 | self.assertEqual(rev.p1node, nullid) |
|
772 | 772 | self.assertEqual(rev.p2node, nullid) |
|
773 | 773 | self.assertEqual(rev.basenode, nullid) |
|
774 | 774 | self.assertIsNone(rev.baserevisionsize) |
|
775 | 775 | self.assertIsNone(rev.revision) |
|
776 | 776 | self.assertEqual( |
|
777 | 777 | rev.delta, |
|
778 | 778 | b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x04\x00' + fulltext0, |
|
779 | 779 | ) |
|
780 | 780 | |
|
781 | 781 | with self.assertRaises(StopIteration): |
|
782 | 782 | next(gen) |
|
783 | 783 | |
|
784 | 784 | gen = f.emitrevisions( |
|
785 | 785 | [node0, node2], |
|
786 | 786 | revisiondata=True, |
|
787 | 787 | deltamode=repository.CG_DELTAMODE_PREV, |
|
788 | 788 | ) |
|
789 | 789 | |
|
790 | 790 | rev = next(gen) |
|
791 | 791 | self.assertEqual(rev.node, node0) |
|
792 | 792 | self.assertEqual(rev.p1node, nullid) |
|
793 | 793 | self.assertEqual(rev.p2node, nullid) |
|
794 | 794 | self.assertEqual(rev.basenode, nullid) |
|
795 | 795 | self.assertIsNone(rev.baserevisionsize) |
|
796 | 796 | self.assertIsNone(rev.revision) |
|
797 | 797 | self.assertEqual( |
|
798 | 798 | rev.delta, |
|
799 | 799 | b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x04\x00' + fulltext0, |
|
800 | 800 | ) |
|
801 | 801 | |
|
802 | 802 | rev = next(gen) |
|
803 | 803 | self.assertEqual(rev.node, node2) |
|
804 | 804 | self.assertEqual(rev.p1node, node1) |
|
805 | 805 | self.assertEqual(rev.p2node, nullid) |
|
806 | 806 | self.assertEqual(rev.basenode, node0) |
|
807 | 807 | |
|
808 | 808 | with self.assertRaises(StopIteration): |
|
809 | 809 | next(gen) |
|
810 | 810 | |
|
811 | 811 | def testrenamed(self): |
|
812 | 812 | fulltext0 = b'foo' |
|
813 | 813 | fulltext1 = b'bar' |
|
814 | 814 | fulltext2 = b'baz' |
|
815 | 815 | |
|
816 | 816 | meta1 = { |
|
817 | 817 | b'copy': b'source0', |
|
818 | 818 | b'copyrev': b'a' * 40, |
|
819 | 819 | } |
|
820 | 820 | |
|
821 | 821 | meta2 = { |
|
822 | 822 | b'copy': b'source1', |
|
823 | 823 | b'copyrev': b'b' * 40, |
|
824 | 824 | } |
|
825 | 825 | |
|
826 | 826 | stored1 = b''.join( |
|
827 | 827 | [ |
|
828 | 828 | b'\x01\ncopy: source0\n', |
|
829 | 829 | b'copyrev: aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\n\x01\n', |
|
830 | 830 | fulltext1, |
|
831 | 831 | ] |
|
832 | 832 | ) |
|
833 | 833 | |
|
834 | 834 | stored2 = b''.join( |
|
835 | 835 | [ |
|
836 | 836 | b'\x01\ncopy: source1\n', |
|
837 | 837 | b'copyrev: bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\n\x01\n', |
|
838 | 838 | fulltext2, |
|
839 | 839 | ] |
|
840 | 840 | ) |
|
841 | 841 | |
|
842 | 842 | f = self._makefilefn() |
|
843 | 843 | with self._maketransactionfn() as tr: |
|
844 | 844 | node0 = f.add(fulltext0, None, tr, 0, nullid, nullid) |
|
845 | 845 | node1 = f.add(fulltext1, meta1, tr, 1, node0, nullid) |
|
846 | 846 | node2 = f.add(fulltext2, meta2, tr, 2, nullid, nullid) |
|
847 | 847 | |
|
848 | 848 | # Metadata header isn't recognized when parent isn't nullid. |
|
849 | 849 | self.assertEqual(f.size(1), len(stored1)) |
|
850 | 850 | self.assertEqual(f.size(2), len(fulltext2)) |
|
851 | 851 | |
|
852 | 852 | self.assertEqual(f.revision(node1), stored1) |
|
853 | 853 | self.assertEqual(f.rawdata(node1), stored1) |
|
854 | 854 | self.assertEqual(f.revision(node2), stored2) |
|
855 | 855 | self.assertEqual(f.rawdata(node2), stored2) |
|
856 | 856 | |
|
857 | 857 | self.assertEqual(f.read(node1), fulltext1) |
|
858 | 858 | self.assertEqual(f.read(node2), fulltext2) |
|
859 | 859 | |
|
860 | 860 | # Returns False when first parent is set. |
|
861 | 861 | self.assertFalse(f.renamed(node1)) |
|
862 | 862 | self.assertEqual(f.renamed(node2), (b'source1', b'\xbb' * 20)) |
|
863 | 863 | |
|
864 | 864 | self.assertTrue(f.cmp(node1, fulltext1)) |
|
865 | 865 | self.assertTrue(f.cmp(node1, stored1)) |
|
866 | 866 | self.assertFalse(f.cmp(node2, fulltext2)) |
|
867 | 867 | self.assertTrue(f.cmp(node2, stored2)) |
|
868 | 868 | |
|
869 | 869 | def testmetadataprefix(self): |
|
870 | 870 | # Content with metadata prefix has extra prefix inserted in storage. |
|
871 | 871 | fulltext0 = b'\x01\nfoo' |
|
872 | 872 | stored0 = b'\x01\n\x01\n\x01\nfoo' |
|
873 | 873 | |
|
874 | 874 | fulltext1 = b'\x01\nbar' |
|
875 | 875 | meta1 = { |
|
876 | 876 | b'copy': b'source0', |
|
877 | 877 | b'copyrev': b'b' * 40, |
|
878 | 878 | } |
|
879 | 879 | stored1 = b''.join( |
|
880 | 880 | [ |
|
881 | 881 | b'\x01\ncopy: source0\n', |
|
882 | 882 | b'copyrev: %s\n' % (b'b' * 40), |
|
883 | 883 | b'\x01\n\x01\nbar', |
|
884 | 884 | ] |
|
885 | 885 | ) |
|
886 | 886 | |
|
887 | 887 | f = self._makefilefn() |
|
888 | 888 | with self._maketransactionfn() as tr: |
|
889 | 889 | node0 = f.add(fulltext0, {}, tr, 0, nullid, nullid) |
|
890 | 890 | node1 = f.add(fulltext1, meta1, tr, 1, nullid, nullid) |
|
891 | 891 | |
|
892 | 892 | # TODO this is buggy. |
|
893 | 893 | self.assertEqual(f.size(0), len(fulltext0) + 4) |
|
894 | 894 | |
|
895 | 895 | self.assertEqual(f.size(1), len(fulltext1)) |
|
896 | 896 | |
|
897 | 897 | self.assertEqual(f.revision(node0), stored0) |
|
898 | 898 | self.assertEqual(f.rawdata(node0), stored0) |
|
899 | 899 | |
|
900 | 900 | self.assertEqual(f.revision(node1), stored1) |
|
901 | 901 | self.assertEqual(f.rawdata(node1), stored1) |
|
902 | 902 | |
|
903 | 903 | self.assertEqual(f.read(node0), fulltext0) |
|
904 | 904 | self.assertEqual(f.read(node1), fulltext1) |
|
905 | 905 | |
|
906 | 906 | self.assertFalse(f.cmp(node0, fulltext0)) |
|
907 | 907 | self.assertTrue(f.cmp(node0, stored0)) |
|
908 | 908 | |
|
909 | 909 | self.assertFalse(f.cmp(node1, fulltext1)) |
|
910 | 910 | self.assertTrue(f.cmp(node1, stored0)) |
|
911 | 911 | |
|
912 | 912 | def testbadnoderead(self): |
|
913 | 913 | f = self._makefilefn() |
|
914 | 914 | |
|
915 | 915 | fulltext0 = b'foo\n' * 30 |
|
916 | 916 | fulltext1 = fulltext0 + b'bar\n' |
|
917 | 917 | |
|
918 | 918 | with self._maketransactionfn() as tr: |
|
919 | 919 | node0 = f.add(fulltext0, None, tr, 0, nullid, nullid) |
|
920 | 920 | node1 = b'\xaa' * 20 |
|
921 | 921 | |
|
922 | 922 | self._addrawrevisionfn( |
|
923 | 923 | f, tr, node1, node0, nullid, 1, rawtext=fulltext1 |
|
924 | 924 | ) |
|
925 | 925 | |
|
926 | 926 | self.assertEqual(len(f), 2) |
|
927 | 927 | self.assertEqual(f.parents(node1), (node0, nullid)) |
|
928 | 928 | |
|
929 | 929 | # revision() raises since it performs hash verification. |
|
930 | 930 | with self.assertRaises(error.StorageError): |
|
931 | 931 | f.revision(node1) |
|
932 | 932 | |
|
933 | 933 | # rawdata() still verifies because there are no special storage |
|
934 | 934 | # settings. |
|
935 | 935 | with self.assertRaises(error.StorageError): |
|
936 | 936 | f.rawdata(node1) |
|
937 | 937 | |
|
938 | 938 | # read() behaves like revision(). |
|
939 | 939 | with self.assertRaises(error.StorageError): |
|
940 | 940 | f.read(node1) |
|
941 | 941 | |
|
942 | 942 | # We can't test renamed() here because some backends may not require |
|
943 | 943 | # reading/validating the fulltext to return rename metadata. |
|
944 | 944 | |
|
945 | 945 | def testbadnoderevisionraw(self): |
|
946 | 946 | # Like above except we test rawdata() first to isolate |
|
947 | 947 | # revision caching behavior. |
|
948 | 948 | f = self._makefilefn() |
|
949 | 949 | |
|
950 | 950 | fulltext0 = b'foo\n' * 30 |
|
951 | 951 | fulltext1 = fulltext0 + b'bar\n' |
|
952 | 952 | |
|
953 | 953 | with self._maketransactionfn() as tr: |
|
954 | 954 | node0 = f.add(fulltext0, None, tr, 0, nullid, nullid) |
|
955 | 955 | node1 = b'\xaa' * 20 |
|
956 | 956 | |
|
957 | 957 | self._addrawrevisionfn( |
|
958 | 958 | f, tr, node1, node0, nullid, 1, rawtext=fulltext1 |
|
959 | 959 | ) |
|
960 | 960 | |
|
961 | 961 | with self.assertRaises(error.StorageError): |
|
962 | 962 | f.rawdata(node1) |
|
963 | 963 | |
|
964 | 964 | with self.assertRaises(error.StorageError): |
|
965 | 965 | f.rawdata(node1) |
|
966 | 966 | |
|
967 | 967 | def testbadnoderevision(self): |
|
968 | 968 | # Like above except we test read() first to isolate revision caching |
|
969 | 969 | # behavior. |
|
970 | 970 | f = self._makefilefn() |
|
971 | 971 | |
|
972 | 972 | fulltext0 = b'foo\n' * 30 |
|
973 | 973 | fulltext1 = fulltext0 + b'bar\n' |
|
974 | 974 | |
|
975 | 975 | with self._maketransactionfn() as tr: |
|
976 | 976 | node0 = f.add(fulltext0, None, tr, 0, nullid, nullid) |
|
977 | 977 | node1 = b'\xaa' * 20 |
|
978 | 978 | |
|
979 | 979 | self._addrawrevisionfn( |
|
980 | 980 | f, tr, node1, node0, nullid, 1, rawtext=fulltext1 |
|
981 | 981 | ) |
|
982 | 982 | |
|
983 | 983 | with self.assertRaises(error.StorageError): |
|
984 | 984 | f.read(node1) |
|
985 | 985 | |
|
986 | 986 | with self.assertRaises(error.StorageError): |
|
987 | 987 | f.read(node1) |
|
988 | 988 | |
|
989 | 989 | def testbadnodedelta(self): |
|
990 | 990 | f = self._makefilefn() |
|
991 | 991 | |
|
992 | 992 | fulltext0 = b'foo\n' * 31 |
|
993 | 993 | fulltext1 = fulltext0 + b'bar\n' |
|
994 | 994 | fulltext2 = fulltext1 + b'baz\n' |
|
995 | 995 | |
|
996 | 996 | with self._maketransactionfn() as tr: |
|
997 | 997 | node0 = f.add(fulltext0, None, tr, 0, nullid, nullid) |
|
998 | 998 | node1 = b'\xaa' * 20 |
|
999 | 999 | |
|
1000 | 1000 | self._addrawrevisionfn( |
|
1001 | 1001 | f, tr, node1, node0, nullid, 1, rawtext=fulltext1 |
|
1002 | 1002 | ) |
|
1003 | 1003 | |
|
1004 | 1004 | with self.assertRaises(error.StorageError): |
|
1005 | 1005 | f.read(node1) |
|
1006 | 1006 | |
|
1007 | 1007 | node2 = storageutil.hashrevisionsha1(fulltext2, node1, nullid) |
|
1008 | 1008 | |
|
1009 | 1009 | with self._maketransactionfn() as tr: |
|
1010 | 1010 | delta = mdiff.textdiff(fulltext1, fulltext2) |
|
1011 | 1011 | self._addrawrevisionfn( |
|
1012 | 1012 | f, tr, node2, node1, nullid, 2, delta=(1, delta) |
|
1013 | 1013 | ) |
|
1014 | 1014 | |
|
1015 | 1015 | self.assertEqual(len(f), 3) |
|
1016 | 1016 | |
|
1017 | 1017 | # Assuming a delta is stored, we shouldn't need to validate node1 in |
|
1018 | 1018 | # order to retrieve node2. |
|
1019 | 1019 | self.assertEqual(f.read(node2), fulltext2) |
|
1020 | 1020 | |
|
1021 | 1021 | def testcensored(self): |
|
1022 | 1022 | f = self._makefilefn() |
|
1023 | 1023 | |
|
1024 | 1024 | stored1 = storageutil.packmeta( |
|
1025 | 1025 | { |
|
1026 | 1026 | b'censored': b'tombstone', |
|
1027 | 1027 | }, |
|
1028 | 1028 | b'', |
|
1029 | 1029 | ) |
|
1030 | 1030 | |
|
1031 | 1031 | with self._maketransactionfn() as tr: |
|
1032 | 1032 | node0 = f.add(b'foo', None, tr, 0, nullid, nullid) |
|
1033 | 1033 | |
|
1034 | 1034 | # The node value doesn't matter since we can't verify it. |
|
1035 | 1035 | node1 = b'\xbb' * 20 |
|
1036 | 1036 | |
|
1037 | 1037 | self._addrawrevisionfn( |
|
1038 | 1038 | f, tr, node1, node0, nullid, 1, stored1, censored=True |
|
1039 | 1039 | ) |
|
1040 | 1040 | |
|
1041 | 1041 | self.assertTrue(f.iscensored(1)) |
|
1042 | 1042 | |
|
1043 | 1043 | with self.assertRaises(error.CensoredNodeError): |
|
1044 | 1044 | f.revision(1) |
|
1045 | 1045 | |
|
1046 | 1046 | with self.assertRaises(error.CensoredNodeError): |
|
1047 | 1047 | f.rawdata(1) |
|
1048 | 1048 | |
|
1049 | 1049 | with self.assertRaises(error.CensoredNodeError): |
|
1050 | 1050 | f.read(1) |
|
1051 | 1051 | |
|
1052 | 1052 | def testcensoredrawrevision(self): |
|
1053 | 1053 | # Like above, except we do the rawdata() request first to |
|
1054 | 1054 | # isolate revision caching behavior. |
|
1055 | 1055 | |
|
1056 | 1056 | f = self._makefilefn() |
|
1057 | 1057 | |
|
1058 | 1058 | stored1 = storageutil.packmeta( |
|
1059 | 1059 | { |
|
1060 | 1060 | b'censored': b'tombstone', |
|
1061 | 1061 | }, |
|
1062 | 1062 | b'', |
|
1063 | 1063 | ) |
|
1064 | 1064 | |
|
1065 | 1065 | with self._maketransactionfn() as tr: |
|
1066 | 1066 | node0 = f.add(b'foo', None, tr, 0, nullid, nullid) |
|
1067 | 1067 | |
|
1068 | 1068 | # The node value doesn't matter since we can't verify it. |
|
1069 | 1069 | node1 = b'\xbb' * 20 |
|
1070 | 1070 | |
|
1071 | 1071 | self._addrawrevisionfn( |
|
1072 | 1072 | f, tr, node1, node0, nullid, 1, stored1, censored=True |
|
1073 | 1073 | ) |
|
1074 | 1074 | |
|
1075 | 1075 | with self.assertRaises(error.CensoredNodeError): |
|
1076 | 1076 | f.rawdata(1) |
|
1077 | 1077 | |
|
1078 | 1078 | |
|
1079 | 1079 | class ifilemutationtests(basetestcase): |
|
1080 | 1080 | """Generic tests for the ifilemutation interface. |
|
1081 | 1081 | |
|
1082 | 1082 | All file storage backends that support writing should conform to this |
|
1083 | 1083 | interface. |
|
1084 | 1084 | |
|
1085 | 1085 | Use ``makeifilemutationtests()`` to create an instance of this type. |
|
1086 | 1086 | """ |
|
1087 | 1087 | |
|
1088 | 1088 | def testaddnoop(self): |
|
1089 | 1089 | f = self._makefilefn() |
|
1090 | 1090 | with self._maketransactionfn() as tr: |
|
1091 | 1091 | node0 = f.add(b'foo', None, tr, 0, nullid, nullid) |
|
1092 | 1092 | node1 = f.add(b'foo', None, tr, 0, nullid, nullid) |
|
1093 | 1093 | # Varying by linkrev shouldn't impact hash. |
|
1094 | 1094 | node2 = f.add(b'foo', None, tr, 1, nullid, nullid) |
|
1095 | 1095 | |
|
1096 | 1096 | self.assertEqual(node1, node0) |
|
1097 | 1097 | self.assertEqual(node2, node0) |
|
1098 | 1098 | self.assertEqual(len(f), 1) |
|
1099 | 1099 | |
|
1100 | 1100 | def testaddrevisionbadnode(self): |
|
1101 | 1101 | f = self._makefilefn() |
|
1102 | 1102 | with self._maketransactionfn() as tr: |
|
1103 | 1103 | # Adding a revision with bad node value fails. |
|
1104 | 1104 | with self.assertRaises(error.StorageError): |
|
1105 | 1105 | f.addrevision(b'foo', tr, 0, nullid, nullid, node=b'\x01' * 20) |
|
1106 | 1106 | |
|
1107 | 1107 | def testaddrevisionunknownflag(self): |
|
1108 | 1108 | f = self._makefilefn() |
|
1109 | 1109 | with self._maketransactionfn() as tr: |
|
1110 | 1110 | for i in range(15, 0, -1): |
|
1111 | 1111 | if (1 << i) & ~repository.REVISION_FLAGS_KNOWN: |
|
1112 | 1112 | flags = 1 << i |
|
1113 | 1113 | break |
|
1114 | 1114 | |
|
1115 | 1115 | with self.assertRaises(error.StorageError): |
|
1116 | 1116 | f.addrevision(b'foo', tr, 0, nullid, nullid, flags=flags) |
|
1117 | 1117 | |
|
1118 | 1118 | def testaddgroupsimple(self): |
|
1119 | 1119 | f = self._makefilefn() |
|
1120 | 1120 | |
|
1121 | 1121 | callbackargs = [] |
|
1122 | 1122 | |
|
1123 | 1123 | def cb(*args, **kwargs): |
|
1124 | 1124 | callbackargs.append((args, kwargs)) |
|
1125 | 1125 | |
|
1126 | 1126 | def linkmapper(node): |
|
1127 | 1127 | return 0 |
|
1128 | 1128 | |
|
1129 | 1129 | with self._maketransactionfn() as tr: |
|
1130 | 1130 | nodes = [] |
|
1131 | 1131 | |
|
1132 | 1132 | def onchangeset(cl, rev): |
|
1133 | 1133 | node = cl.node(rev) |
|
1134 | 1134 | nodes.append(node) |
|
1135 | 1135 | cb(cl, node) |
|
1136 | 1136 | |
|
1137 | 1137 | def ondupchangeset(cl, rev): |
|
1138 | 1138 | nodes.append(cl.node(rev)) |
|
1139 | 1139 | |
|
1140 | 1140 | f.addgroup( |
|
1141 | 1141 | [], |
|
1142 | 1142 | None, |
|
1143 | 1143 | tr, |
|
1144 | 1144 | addrevisioncb=onchangeset, |
|
1145 | 1145 | duplicaterevisioncb=ondupchangeset, |
|
1146 | 1146 | ) |
|
1147 | 1147 | |
|
1148 | 1148 | self.assertEqual(nodes, []) |
|
1149 | 1149 | self.assertEqual(callbackargs, []) |
|
1150 | 1150 | self.assertEqual(len(f), 0) |
|
1151 | 1151 | |
|
1152 | 1152 | fulltext0 = b'foo' |
|
1153 | 1153 | delta0 = mdiff.trivialdiffheader(len(fulltext0)) + fulltext0 |
|
1154 | 1154 | |
|
1155 | 1155 | with self._maketransactionfn() as tr: |
|
1156 | 1156 | node0 = f.add(fulltext0, None, tr, 0, nullid, nullid) |
|
1157 | 1157 | |
|
1158 | 1158 | f = self._makefilefn() |
|
1159 | 1159 | |
|
1160 | 1160 | deltas = [ |
|
1161 | (node0, nullid, nullid, nullid, nullid, delta0, 0), | |
|
1161 | (node0, nullid, nullid, nullid, nullid, delta0, 0, {}), | |
|
1162 | 1162 | ] |
|
1163 | 1163 | |
|
1164 | 1164 | with self._maketransactionfn() as tr: |
|
1165 | 1165 | nodes = [] |
|
1166 | 1166 | |
|
1167 | 1167 | def onchangeset(cl, rev): |
|
1168 | 1168 | node = cl.node(rev) |
|
1169 | 1169 | nodes.append(node) |
|
1170 | 1170 | cb(cl, node) |
|
1171 | 1171 | |
|
1172 | 1172 | def ondupchangeset(cl, rev): |
|
1173 | 1173 | nodes.append(cl.node(rev)) |
|
1174 | 1174 | |
|
1175 | 1175 | f.addgroup( |
|
1176 | 1176 | deltas, |
|
1177 | 1177 | linkmapper, |
|
1178 | 1178 | tr, |
|
1179 | 1179 | addrevisioncb=onchangeset, |
|
1180 | 1180 | duplicaterevisioncb=ondupchangeset, |
|
1181 | 1181 | ) |
|
1182 | 1182 | |
|
1183 | 1183 | self.assertEqual( |
|
1184 | 1184 | nodes, |
|
1185 | 1185 | [ |
|
1186 | 1186 | b'\x49\xd8\xcb\xb1\x5c\xe2\x57\x92\x04\x47' |
|
1187 | 1187 | b'\x00\x6b\x46\x97\x8b\x7a\xf9\x80\xa9\x79' |
|
1188 | 1188 | ], |
|
1189 | 1189 | ) |
|
1190 | 1190 | |
|
1191 | 1191 | self.assertEqual(len(callbackargs), 1) |
|
1192 | 1192 | self.assertEqual(callbackargs[0][0][1], nodes[0]) |
|
1193 | 1193 | |
|
1194 | 1194 | self.assertEqual(list(f.revs()), [0]) |
|
1195 | 1195 | self.assertEqual(f.rev(nodes[0]), 0) |
|
1196 | 1196 | self.assertEqual(f.node(0), nodes[0]) |
|
1197 | 1197 | |
|
1198 | 1198 | def testaddgroupmultiple(self): |
|
1199 | 1199 | f = self._makefilefn() |
|
1200 | 1200 | |
|
1201 | 1201 | fulltexts = [ |
|
1202 | 1202 | b'foo', |
|
1203 | 1203 | b'bar', |
|
1204 | 1204 | b'x' * 1024, |
|
1205 | 1205 | ] |
|
1206 | 1206 | |
|
1207 | 1207 | nodes = [] |
|
1208 | 1208 | with self._maketransactionfn() as tr: |
|
1209 | 1209 | for fulltext in fulltexts: |
|
1210 | 1210 | nodes.append(f.add(fulltext, None, tr, 0, nullid, nullid)) |
|
1211 | 1211 | |
|
1212 | 1212 | f = self._makefilefn() |
|
1213 | 1213 | deltas = [] |
|
1214 | 1214 | for i, fulltext in enumerate(fulltexts): |
|
1215 | 1215 | delta = mdiff.trivialdiffheader(len(fulltext)) + fulltext |
|
1216 | 1216 | |
|
1217 | deltas.append((nodes[i], nullid, nullid, nullid, nullid, delta, 0)) | |
|
1217 | deltas.append( | |
|
1218 | (nodes[i], nullid, nullid, nullid, nullid, delta, 0, {}) | |
|
1219 | ) | |
|
1218 | 1220 | |
|
1219 | 1221 | with self._maketransactionfn() as tr: |
|
1220 | 1222 | newnodes = [] |
|
1221 | 1223 | |
|
1222 | 1224 | def onchangeset(cl, rev): |
|
1223 | 1225 | newnodes.append(cl.node(rev)) |
|
1224 | 1226 | |
|
1225 | 1227 | f.addgroup( |
|
1226 | 1228 | deltas, |
|
1227 | 1229 | lambda x: 0, |
|
1228 | 1230 | tr, |
|
1229 | 1231 | addrevisioncb=onchangeset, |
|
1230 | 1232 | duplicaterevisioncb=onchangeset, |
|
1231 | 1233 | ) |
|
1232 | 1234 | self.assertEqual(newnodes, nodes) |
|
1233 | 1235 | |
|
1234 | 1236 | self.assertEqual(len(f), len(deltas)) |
|
1235 | 1237 | self.assertEqual(list(f.revs()), [0, 1, 2]) |
|
1236 | 1238 | self.assertEqual(f.rev(nodes[0]), 0) |
|
1237 | 1239 | self.assertEqual(f.rev(nodes[1]), 1) |
|
1238 | 1240 | self.assertEqual(f.rev(nodes[2]), 2) |
|
1239 | 1241 | self.assertEqual(f.node(0), nodes[0]) |
|
1240 | 1242 | self.assertEqual(f.node(1), nodes[1]) |
|
1241 | 1243 | self.assertEqual(f.node(2), nodes[2]) |
|
1242 | 1244 | |
|
1243 | 1245 | def testdeltaagainstcensored(self): |
|
1244 | 1246 | # Attempt to apply a delta made against a censored revision. |
|
1245 | 1247 | f = self._makefilefn() |
|
1246 | 1248 | |
|
1247 | 1249 | stored1 = storageutil.packmeta( |
|
1248 | 1250 | { |
|
1249 | 1251 | b'censored': b'tombstone', |
|
1250 | 1252 | }, |
|
1251 | 1253 | b'', |
|
1252 | 1254 | ) |
|
1253 | 1255 | |
|
1254 | 1256 | with self._maketransactionfn() as tr: |
|
1255 | 1257 | node0 = f.add(b'foo\n' * 30, None, tr, 0, nullid, nullid) |
|
1256 | 1258 | |
|
1257 | 1259 | # The node value doesn't matter since we can't verify it. |
|
1258 | 1260 | node1 = b'\xbb' * 20 |
|
1259 | 1261 | |
|
1260 | 1262 | self._addrawrevisionfn( |
|
1261 | 1263 | f, tr, node1, node0, nullid, 1, stored1, censored=True |
|
1262 | 1264 | ) |
|
1263 | 1265 | |
|
1264 | 1266 | delta = mdiff.textdiff(b'bar\n' * 30, (b'bar\n' * 30) + b'baz\n') |
|
1265 | deltas = [(b'\xcc' * 20, node1, nullid, b'\x01' * 20, node1, delta, 0)] | |
|
1267 | deltas = [ | |
|
1268 | (b'\xcc' * 20, node1, nullid, b'\x01' * 20, node1, delta, 0, {}) | |
|
1269 | ] | |
|
1266 | 1270 | |
|
1267 | 1271 | with self._maketransactionfn() as tr: |
|
1268 | 1272 | with self.assertRaises(error.CensoredBaseError): |
|
1269 | 1273 | f.addgroup(deltas, lambda x: 0, tr) |
|
1270 | 1274 | |
|
1271 | 1275 | def testcensorrevisionbasic(self): |
|
1272 | 1276 | f = self._makefilefn() |
|
1273 | 1277 | |
|
1274 | 1278 | with self._maketransactionfn() as tr: |
|
1275 | 1279 | node0 = f.add(b'foo\n' * 30, None, tr, 0, nullid, nullid) |
|
1276 | 1280 | node1 = f.add(b'foo\n' * 31, None, tr, 1, node0, nullid) |
|
1277 | 1281 | node2 = f.add(b'foo\n' * 32, None, tr, 2, node1, nullid) |
|
1278 | 1282 | |
|
1279 | 1283 | with self._maketransactionfn() as tr: |
|
1280 | 1284 | f.censorrevision(tr, node1) |
|
1281 | 1285 | |
|
1282 | 1286 | self.assertEqual(len(f), 3) |
|
1283 | 1287 | self.assertEqual(list(f.revs()), [0, 1, 2]) |
|
1284 | 1288 | |
|
1285 | 1289 | self.assertEqual(f.read(node0), b'foo\n' * 30) |
|
1286 | 1290 | self.assertEqual(f.read(node2), b'foo\n' * 32) |
|
1287 | 1291 | |
|
1288 | 1292 | with self.assertRaises(error.CensoredNodeError): |
|
1289 | 1293 | f.read(node1) |
|
1290 | 1294 | |
|
1291 | 1295 | def testgetstrippointnoparents(self): |
|
1292 | 1296 | # N revisions where none have parents. |
|
1293 | 1297 | f = self._makefilefn() |
|
1294 | 1298 | |
|
1295 | 1299 | with self._maketransactionfn() as tr: |
|
1296 | 1300 | for rev in range(10): |
|
1297 | 1301 | f.add(b'%d' % rev, None, tr, rev, nullid, nullid) |
|
1298 | 1302 | |
|
1299 | 1303 | for rev in range(10): |
|
1300 | 1304 | self.assertEqual(f.getstrippoint(rev), (rev, set())) |
|
1301 | 1305 | |
|
1302 | 1306 | def testgetstrippointlinear(self): |
|
1303 | 1307 | # N revisions in a linear chain. |
|
1304 | 1308 | f = self._makefilefn() |
|
1305 | 1309 | |
|
1306 | 1310 | with self._maketransactionfn() as tr: |
|
1307 | 1311 | p1 = nullid |
|
1308 | 1312 | |
|
1309 | 1313 | for rev in range(10): |
|
1310 | 1314 | f.add(b'%d' % rev, None, tr, rev, p1, nullid) |
|
1311 | 1315 | |
|
1312 | 1316 | for rev in range(10): |
|
1313 | 1317 | self.assertEqual(f.getstrippoint(rev), (rev, set())) |
|
1314 | 1318 | |
|
1315 | 1319 | def testgetstrippointmultipleheads(self): |
|
1316 | 1320 | f = self._makefilefn() |
|
1317 | 1321 | |
|
1318 | 1322 | with self._maketransactionfn() as tr: |
|
1319 | 1323 | node0 = f.add(b'0', None, tr, 0, nullid, nullid) |
|
1320 | 1324 | node1 = f.add(b'1', None, tr, 1, node0, nullid) |
|
1321 | 1325 | f.add(b'2', None, tr, 2, node1, nullid) |
|
1322 | 1326 | f.add(b'3', None, tr, 3, node0, nullid) |
|
1323 | 1327 | f.add(b'4', None, tr, 4, node0, nullid) |
|
1324 | 1328 | |
|
1325 | 1329 | for rev in range(5): |
|
1326 | 1330 | self.assertEqual(f.getstrippoint(rev), (rev, set())) |
|
1327 | 1331 | |
|
1328 | 1332 | def testgetstrippointearlierlinkrevs(self): |
|
1329 | 1333 | f = self._makefilefn() |
|
1330 | 1334 | |
|
1331 | 1335 | with self._maketransactionfn() as tr: |
|
1332 | 1336 | node0 = f.add(b'0', None, tr, 0, nullid, nullid) |
|
1333 | 1337 | f.add(b'1', None, tr, 10, node0, nullid) |
|
1334 | 1338 | f.add(b'2', None, tr, 5, node0, nullid) |
|
1335 | 1339 | |
|
1336 | 1340 | self.assertEqual(f.getstrippoint(0), (0, set())) |
|
1337 | 1341 | self.assertEqual(f.getstrippoint(1), (1, set())) |
|
1338 | 1342 | self.assertEqual(f.getstrippoint(2), (1, set())) |
|
1339 | 1343 | self.assertEqual(f.getstrippoint(3), (1, set())) |
|
1340 | 1344 | self.assertEqual(f.getstrippoint(4), (1, set())) |
|
1341 | 1345 | self.assertEqual(f.getstrippoint(5), (1, set())) |
|
1342 | 1346 | self.assertEqual(f.getstrippoint(6), (1, {2})) |
|
1343 | 1347 | self.assertEqual(f.getstrippoint(7), (1, {2})) |
|
1344 | 1348 | self.assertEqual(f.getstrippoint(8), (1, {2})) |
|
1345 | 1349 | self.assertEqual(f.getstrippoint(9), (1, {2})) |
|
1346 | 1350 | self.assertEqual(f.getstrippoint(10), (1, {2})) |
|
1347 | 1351 | self.assertEqual(f.getstrippoint(11), (3, set())) |
|
1348 | 1352 | |
|
1349 | 1353 | def teststripempty(self): |
|
1350 | 1354 | f = self._makefilefn() |
|
1351 | 1355 | |
|
1352 | 1356 | with self._maketransactionfn() as tr: |
|
1353 | 1357 | f.strip(0, tr) |
|
1354 | 1358 | |
|
1355 | 1359 | self.assertEqual(len(f), 0) |
|
1356 | 1360 | |
|
1357 | 1361 | def teststripall(self): |
|
1358 | 1362 | f = self._makefilefn() |
|
1359 | 1363 | |
|
1360 | 1364 | with self._maketransactionfn() as tr: |
|
1361 | 1365 | p1 = nullid |
|
1362 | 1366 | for rev in range(10): |
|
1363 | 1367 | p1 = f.add(b'%d' % rev, None, tr, rev, p1, nullid) |
|
1364 | 1368 | |
|
1365 | 1369 | self.assertEqual(len(f), 10) |
|
1366 | 1370 | |
|
1367 | 1371 | with self._maketransactionfn() as tr: |
|
1368 | 1372 | f.strip(0, tr) |
|
1369 | 1373 | |
|
1370 | 1374 | self.assertEqual(len(f), 0) |
|
1371 | 1375 | |
|
1372 | 1376 | def teststrippartial(self): |
|
1373 | 1377 | f = self._makefilefn() |
|
1374 | 1378 | |
|
1375 | 1379 | with self._maketransactionfn() as tr: |
|
1376 | 1380 | f.add(b'0', None, tr, 0, nullid, nullid) |
|
1377 | 1381 | node1 = f.add(b'1', None, tr, 5, nullid, nullid) |
|
1378 | 1382 | node2 = f.add(b'2', None, tr, 10, nullid, nullid) |
|
1379 | 1383 | |
|
1380 | 1384 | self.assertEqual(len(f), 3) |
|
1381 | 1385 | |
|
1382 | 1386 | with self._maketransactionfn() as tr: |
|
1383 | 1387 | f.strip(11, tr) |
|
1384 | 1388 | |
|
1385 | 1389 | self.assertEqual(len(f), 3) |
|
1386 | 1390 | |
|
1387 | 1391 | with self._maketransactionfn() as tr: |
|
1388 | 1392 | f.strip(10, tr) |
|
1389 | 1393 | |
|
1390 | 1394 | self.assertEqual(len(f), 2) |
|
1391 | 1395 | |
|
1392 | 1396 | with self.assertRaises(error.LookupError): |
|
1393 | 1397 | f.rev(node2) |
|
1394 | 1398 | |
|
1395 | 1399 | with self._maketransactionfn() as tr: |
|
1396 | 1400 | f.strip(6, tr) |
|
1397 | 1401 | |
|
1398 | 1402 | self.assertEqual(len(f), 2) |
|
1399 | 1403 | |
|
1400 | 1404 | with self._maketransactionfn() as tr: |
|
1401 | 1405 | f.strip(3, tr) |
|
1402 | 1406 | |
|
1403 | 1407 | self.assertEqual(len(f), 1) |
|
1404 | 1408 | |
|
1405 | 1409 | with self.assertRaises(error.LookupError): |
|
1406 | 1410 | f.rev(node1) |
|
1407 | 1411 | |
|
1408 | 1412 | |
|
1409 | 1413 | def makeifileindextests(makefilefn, maketransactionfn, addrawrevisionfn): |
|
1410 | 1414 | """Create a unittest.TestCase class suitable for testing file storage. |
|
1411 | 1415 | |
|
1412 | 1416 | ``makefilefn`` is a callable which receives the test case as an |
|
1413 | 1417 | argument and returns an object implementing the ``ifilestorage`` interface. |
|
1414 | 1418 | |
|
1415 | 1419 | ``maketransactionfn`` is a callable which receives the test case as an |
|
1416 | 1420 | argument and returns a transaction object. |
|
1417 | 1421 | |
|
1418 | 1422 | ``addrawrevisionfn`` is a callable which receives arguments describing a |
|
1419 | 1423 | low-level revision to add. This callable allows the insertion of |
|
1420 | 1424 | potentially bad data into the store in order to facilitate testing. |
|
1421 | 1425 | |
|
1422 | 1426 | Returns a type that is a ``unittest.TestCase`` that can be used for |
|
1423 | 1427 | testing the object implementing the file storage interface. Simply |
|
1424 | 1428 | assign the returned value to a module-level attribute and a test loader |
|
1425 | 1429 | should find and run it automatically. |
|
1426 | 1430 | """ |
|
1427 | 1431 | d = { |
|
1428 | 1432 | '_makefilefn': makefilefn, |
|
1429 | 1433 | '_maketransactionfn': maketransactionfn, |
|
1430 | 1434 | '_addrawrevisionfn': addrawrevisionfn, |
|
1431 | 1435 | } |
|
1432 | 1436 | return type('ifileindextests', (ifileindextests,), d) |
|
1433 | 1437 | |
|
1434 | 1438 | |
|
1435 | 1439 | def makeifiledatatests(makefilefn, maketransactionfn, addrawrevisionfn): |
|
1436 | 1440 | d = { |
|
1437 | 1441 | '_makefilefn': makefilefn, |
|
1438 | 1442 | '_maketransactionfn': maketransactionfn, |
|
1439 | 1443 | '_addrawrevisionfn': addrawrevisionfn, |
|
1440 | 1444 | } |
|
1441 | 1445 | return type('ifiledatatests', (ifiledatatests,), d) |
|
1442 | 1446 | |
|
1443 | 1447 | |
|
1444 | 1448 | def makeifilemutationtests(makefilefn, maketransactionfn, addrawrevisionfn): |
|
1445 | 1449 | d = { |
|
1446 | 1450 | '_makefilefn': makefilefn, |
|
1447 | 1451 | '_maketransactionfn': maketransactionfn, |
|
1448 | 1452 | '_addrawrevisionfn': addrawrevisionfn, |
|
1449 | 1453 | } |
|
1450 | 1454 | return type('ifilemutationtests', (ifilemutationtests,), d) |
@@ -1,512 +1,513 b'' | |||
|
1 | 1 | # storageutil.py - Storage functionality agnostic of backend implementation. |
|
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 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import re |
|
11 | 11 | import struct |
|
12 | 12 | |
|
13 | 13 | from ..i18n import _ |
|
14 | 14 | from ..node import ( |
|
15 | 15 | bin, |
|
16 | 16 | nullid, |
|
17 | 17 | nullrev, |
|
18 | 18 | ) |
|
19 | 19 | from .. import ( |
|
20 | 20 | dagop, |
|
21 | 21 | error, |
|
22 | 22 | mdiff, |
|
23 | 23 | pycompat, |
|
24 | 24 | ) |
|
25 | 25 | from ..interfaces import repository |
|
26 | 26 | from ..utils import hashutil |
|
27 | 27 | |
|
28 | 28 | _nullhash = hashutil.sha1(nullid) |
|
29 | 29 | |
|
30 | 30 | |
|
31 | 31 | def hashrevisionsha1(text, p1, p2): |
|
32 | 32 | """Compute the SHA-1 for revision data and its parents. |
|
33 | 33 | |
|
34 | 34 | This hash combines both the current file contents and its history |
|
35 | 35 | in a manner that makes it easy to distinguish nodes with the same |
|
36 | 36 | content in the revision graph. |
|
37 | 37 | """ |
|
38 | 38 | # As of now, if one of the parent node is null, p2 is null |
|
39 | 39 | if p2 == nullid: |
|
40 | 40 | # deep copy of a hash is faster than creating one |
|
41 | 41 | s = _nullhash.copy() |
|
42 | 42 | s.update(p1) |
|
43 | 43 | else: |
|
44 | 44 | # none of the parent nodes are nullid |
|
45 | 45 | if p1 < p2: |
|
46 | 46 | a = p1 |
|
47 | 47 | b = p2 |
|
48 | 48 | else: |
|
49 | 49 | a = p2 |
|
50 | 50 | b = p1 |
|
51 | 51 | s = hashutil.sha1(a) |
|
52 | 52 | s.update(b) |
|
53 | 53 | s.update(text) |
|
54 | 54 | return s.digest() |
|
55 | 55 | |
|
56 | 56 | |
|
57 | 57 | METADATA_RE = re.compile(b'\x01\n') |
|
58 | 58 | |
|
59 | 59 | |
|
60 | 60 | def parsemeta(text): |
|
61 | 61 | """Parse metadata header from revision data. |
|
62 | 62 | |
|
63 | 63 | Returns a 2-tuple of (metadata, offset), where both can be None if there |
|
64 | 64 | is no metadata. |
|
65 | 65 | """ |
|
66 | 66 | # text can be buffer, so we can't use .startswith or .index |
|
67 | 67 | if text[:2] != b'\x01\n': |
|
68 | 68 | return None, None |
|
69 | 69 | s = METADATA_RE.search(text, 2).start() |
|
70 | 70 | mtext = text[2:s] |
|
71 | 71 | meta = {} |
|
72 | 72 | for l in mtext.splitlines(): |
|
73 | 73 | k, v = l.split(b': ', 1) |
|
74 | 74 | meta[k] = v |
|
75 | 75 | return meta, s + 2 |
|
76 | 76 | |
|
77 | 77 | |
|
78 | 78 | def packmeta(meta, text): |
|
79 | 79 | """Add metadata to fulltext to produce revision text.""" |
|
80 | 80 | keys = sorted(meta) |
|
81 | 81 | metatext = b''.join(b'%s: %s\n' % (k, meta[k]) for k in keys) |
|
82 | 82 | return b'\x01\n%s\x01\n%s' % (metatext, text) |
|
83 | 83 | |
|
84 | 84 | |
|
85 | 85 | def iscensoredtext(text): |
|
86 | 86 | meta = parsemeta(text)[0] |
|
87 | 87 | return meta and b'censored' in meta |
|
88 | 88 | |
|
89 | 89 | |
|
90 | 90 | def filtermetadata(text): |
|
91 | 91 | """Extract just the revision data from source text. |
|
92 | 92 | |
|
93 | 93 | Returns ``text`` unless it has a metadata header, in which case we return |
|
94 | 94 | a new buffer without hte metadata. |
|
95 | 95 | """ |
|
96 | 96 | if not text.startswith(b'\x01\n'): |
|
97 | 97 | return text |
|
98 | 98 | |
|
99 | 99 | offset = text.index(b'\x01\n', 2) |
|
100 | 100 | return text[offset + 2 :] |
|
101 | 101 | |
|
102 | 102 | |
|
103 | 103 | def filerevisioncopied(store, node): |
|
104 | 104 | """Resolve file revision copy metadata. |
|
105 | 105 | |
|
106 | 106 | Returns ``False`` if the file has no copy metadata. Otherwise a |
|
107 | 107 | 2-tuple of the source filename and node. |
|
108 | 108 | """ |
|
109 | 109 | if store.parents(node)[0] != nullid: |
|
110 | 110 | return False |
|
111 | 111 | |
|
112 | 112 | meta = parsemeta(store.revision(node))[0] |
|
113 | 113 | |
|
114 | 114 | # copy and copyrev occur in pairs. In rare cases due to old bugs, |
|
115 | 115 | # one can occur without the other. So ensure both are present to flag |
|
116 | 116 | # as a copy. |
|
117 | 117 | if meta and b'copy' in meta and b'copyrev' in meta: |
|
118 | 118 | return meta[b'copy'], bin(meta[b'copyrev']) |
|
119 | 119 | |
|
120 | 120 | return False |
|
121 | 121 | |
|
122 | 122 | |
|
123 | 123 | def filedataequivalent(store, node, filedata): |
|
124 | 124 | """Determines whether file data is equivalent to a stored node. |
|
125 | 125 | |
|
126 | 126 | Returns True if the passed file data would hash to the same value |
|
127 | 127 | as a stored revision and False otherwise. |
|
128 | 128 | |
|
129 | 129 | When a stored revision is censored, filedata must be empty to have |
|
130 | 130 | equivalence. |
|
131 | 131 | |
|
132 | 132 | When a stored revision has copy metadata, it is ignored as part |
|
133 | 133 | of the compare. |
|
134 | 134 | """ |
|
135 | 135 | |
|
136 | 136 | if filedata.startswith(b'\x01\n'): |
|
137 | 137 | revisiontext = b'\x01\n\x01\n' + filedata |
|
138 | 138 | else: |
|
139 | 139 | revisiontext = filedata |
|
140 | 140 | |
|
141 | 141 | p1, p2 = store.parents(node) |
|
142 | 142 | |
|
143 | 143 | computednode = hashrevisionsha1(revisiontext, p1, p2) |
|
144 | 144 | |
|
145 | 145 | if computednode == node: |
|
146 | 146 | return True |
|
147 | 147 | |
|
148 | 148 | # Censored files compare against the empty file. |
|
149 | 149 | if store.iscensored(store.rev(node)): |
|
150 | 150 | return filedata == b'' |
|
151 | 151 | |
|
152 | 152 | # Renaming a file produces a different hash, even if the data |
|
153 | 153 | # remains unchanged. Check if that's the case. |
|
154 | 154 | if store.renamed(node): |
|
155 | 155 | return store.read(node) == filedata |
|
156 | 156 | |
|
157 | 157 | return False |
|
158 | 158 | |
|
159 | 159 | |
|
160 | 160 | def iterrevs(storelen, start=0, stop=None): |
|
161 | 161 | """Iterate over revision numbers in a store.""" |
|
162 | 162 | step = 1 |
|
163 | 163 | |
|
164 | 164 | if stop is not None: |
|
165 | 165 | if start > stop: |
|
166 | 166 | step = -1 |
|
167 | 167 | stop += step |
|
168 | 168 | if stop > storelen: |
|
169 | 169 | stop = storelen |
|
170 | 170 | else: |
|
171 | 171 | stop = storelen |
|
172 | 172 | |
|
173 | 173 | return pycompat.xrange(start, stop, step) |
|
174 | 174 | |
|
175 | 175 | |
|
176 | 176 | def fileidlookup(store, fileid, identifier): |
|
177 | 177 | """Resolve the file node for a value. |
|
178 | 178 | |
|
179 | 179 | ``store`` is an object implementing the ``ifileindex`` interface. |
|
180 | 180 | |
|
181 | 181 | ``fileid`` can be: |
|
182 | 182 | |
|
183 | 183 | * A 20 or 32 byte binary node. |
|
184 | 184 | * An integer revision number |
|
185 | 185 | * A 40 or 64 byte hex node. |
|
186 | 186 | * A bytes that can be parsed as an integer representing a revision number. |
|
187 | 187 | |
|
188 | 188 | ``identifier`` is used to populate ``error.LookupError`` with an identifier |
|
189 | 189 | for the store. |
|
190 | 190 | |
|
191 | 191 | Raises ``error.LookupError`` on failure. |
|
192 | 192 | """ |
|
193 | 193 | if isinstance(fileid, int): |
|
194 | 194 | try: |
|
195 | 195 | return store.node(fileid) |
|
196 | 196 | except IndexError: |
|
197 | 197 | raise error.LookupError( |
|
198 | 198 | b'%d' % fileid, identifier, _(b'no match found') |
|
199 | 199 | ) |
|
200 | 200 | |
|
201 | 201 | if len(fileid) in (20, 32): |
|
202 | 202 | try: |
|
203 | 203 | store.rev(fileid) |
|
204 | 204 | return fileid |
|
205 | 205 | except error.LookupError: |
|
206 | 206 | pass |
|
207 | 207 | |
|
208 | 208 | if len(fileid) in (40, 64): |
|
209 | 209 | try: |
|
210 | 210 | rawnode = bin(fileid) |
|
211 | 211 | store.rev(rawnode) |
|
212 | 212 | return rawnode |
|
213 | 213 | except TypeError: |
|
214 | 214 | pass |
|
215 | 215 | |
|
216 | 216 | try: |
|
217 | 217 | rev = int(fileid) |
|
218 | 218 | |
|
219 | 219 | if b'%d' % rev != fileid: |
|
220 | 220 | raise ValueError |
|
221 | 221 | |
|
222 | 222 | try: |
|
223 | 223 | return store.node(rev) |
|
224 | 224 | except (IndexError, TypeError): |
|
225 | 225 | pass |
|
226 | 226 | except (ValueError, OverflowError): |
|
227 | 227 | pass |
|
228 | 228 | |
|
229 | 229 | raise error.LookupError(fileid, identifier, _(b'no match found')) |
|
230 | 230 | |
|
231 | 231 | |
|
232 | 232 | def resolvestripinfo(minlinkrev, tiprev, headrevs, linkrevfn, parentrevsfn): |
|
233 | 233 | """Resolve information needed to strip revisions. |
|
234 | 234 | |
|
235 | 235 | Finds the minimum revision number that must be stripped in order to |
|
236 | 236 | strip ``minlinkrev``. |
|
237 | 237 | |
|
238 | 238 | Returns a 2-tuple of the minimum revision number to do that and a set |
|
239 | 239 | of all revision numbers that have linkrevs that would be broken |
|
240 | 240 | by that strip. |
|
241 | 241 | |
|
242 | 242 | ``tiprev`` is the current tip-most revision. It is ``len(store) - 1``. |
|
243 | 243 | ``headrevs`` is an iterable of head revisions. |
|
244 | 244 | ``linkrevfn`` is a callable that receives a revision and returns a linked |
|
245 | 245 | revision. |
|
246 | 246 | ``parentrevsfn`` is a callable that receives a revision number and returns |
|
247 | 247 | an iterable of its parent revision numbers. |
|
248 | 248 | """ |
|
249 | 249 | brokenrevs = set() |
|
250 | 250 | strippoint = tiprev + 1 |
|
251 | 251 | |
|
252 | 252 | heads = {} |
|
253 | 253 | futurelargelinkrevs = set() |
|
254 | 254 | for head in headrevs: |
|
255 | 255 | headlinkrev = linkrevfn(head) |
|
256 | 256 | heads[head] = headlinkrev |
|
257 | 257 | if headlinkrev >= minlinkrev: |
|
258 | 258 | futurelargelinkrevs.add(headlinkrev) |
|
259 | 259 | |
|
260 | 260 | # This algorithm involves walking down the rev graph, starting at the |
|
261 | 261 | # heads. Since the revs are topologically sorted according to linkrev, |
|
262 | 262 | # once all head linkrevs are below the minlink, we know there are |
|
263 | 263 | # no more revs that could have a linkrev greater than minlink. |
|
264 | 264 | # So we can stop walking. |
|
265 | 265 | while futurelargelinkrevs: |
|
266 | 266 | strippoint -= 1 |
|
267 | 267 | linkrev = heads.pop(strippoint) |
|
268 | 268 | |
|
269 | 269 | if linkrev < minlinkrev: |
|
270 | 270 | brokenrevs.add(strippoint) |
|
271 | 271 | else: |
|
272 | 272 | futurelargelinkrevs.remove(linkrev) |
|
273 | 273 | |
|
274 | 274 | for p in parentrevsfn(strippoint): |
|
275 | 275 | if p != nullrev: |
|
276 | 276 | plinkrev = linkrevfn(p) |
|
277 | 277 | heads[p] = plinkrev |
|
278 | 278 | if plinkrev >= minlinkrev: |
|
279 | 279 | futurelargelinkrevs.add(plinkrev) |
|
280 | 280 | |
|
281 | 281 | return strippoint, brokenrevs |
|
282 | 282 | |
|
283 | 283 | |
|
284 | 284 | def emitrevisions( |
|
285 | 285 | store, |
|
286 | 286 | nodes, |
|
287 | 287 | nodesorder, |
|
288 | 288 | resultcls, |
|
289 | 289 | deltaparentfn=None, |
|
290 | 290 | candeltafn=None, |
|
291 | 291 | rawsizefn=None, |
|
292 | 292 | revdifffn=None, |
|
293 | 293 | flagsfn=None, |
|
294 | 294 | deltamode=repository.CG_DELTAMODE_STD, |
|
295 | 295 | revisiondata=False, |
|
296 | 296 | assumehaveparentrevisions=False, |
|
297 | 297 | ): |
|
298 | 298 | """Generic implementation of ifiledata.emitrevisions(). |
|
299 | 299 | |
|
300 | 300 | Emitting revision data is subtly complex. This function attempts to |
|
301 | 301 | encapsulate all the logic for doing so in a backend-agnostic way. |
|
302 | 302 | |
|
303 | 303 | ``store`` |
|
304 | 304 | Object conforming to ``ifilestorage`` interface. |
|
305 | 305 | |
|
306 | 306 | ``nodes`` |
|
307 | 307 | List of revision nodes whose data to emit. |
|
308 | 308 | |
|
309 | 309 | ``resultcls`` |
|
310 | 310 | A type implementing the ``irevisiondelta`` interface that will be |
|
311 | 311 | constructed and returned. |
|
312 | 312 | |
|
313 | 313 | ``deltaparentfn`` (optional) |
|
314 | 314 | Callable receiving a revision number and returning the revision number |
|
315 | 315 | of a revision that the internal delta is stored against. This delta |
|
316 | 316 | will be preferred over computing a new arbitrary delta. |
|
317 | 317 | |
|
318 | 318 | If not defined, a delta will always be computed from raw revision |
|
319 | 319 | data. |
|
320 | 320 | |
|
321 | 321 | ``candeltafn`` (optional) |
|
322 | 322 | Callable receiving a pair of revision numbers that returns a bool |
|
323 | 323 | indicating whether a delta between them can be produced. |
|
324 | 324 | |
|
325 | 325 | If not defined, it is assumed that any two revisions can delta with |
|
326 | 326 | each other. |
|
327 | 327 | |
|
328 | 328 | ``rawsizefn`` (optional) |
|
329 | 329 | Callable receiving a revision number and returning the length of the |
|
330 | 330 | ``store.rawdata(rev)``. |
|
331 | 331 | |
|
332 | 332 | If not defined, ``len(store.rawdata(rev))`` will be called. |
|
333 | 333 | |
|
334 | 334 | ``revdifffn`` (optional) |
|
335 | 335 | Callable receiving a pair of revision numbers that returns a delta |
|
336 | 336 | between them. |
|
337 | 337 | |
|
338 | 338 | If not defined, a delta will be computed by invoking mdiff code |
|
339 | 339 | on ``store.revision()`` results. |
|
340 | 340 | |
|
341 | 341 | Defining this function allows a precomputed or stored delta to be |
|
342 | 342 | used without having to compute on. |
|
343 | 343 | |
|
344 | 344 | ``flagsfn`` (optional) |
|
345 | 345 | Callable receiving a revision number and returns the integer flags |
|
346 | 346 | value for it. If not defined, flags value will be 0. |
|
347 | 347 | |
|
348 | 348 | ``deltamode`` |
|
349 | 349 | constaint on delta to be sent: |
|
350 | 350 | * CG_DELTAMODE_STD - normal mode, try to reuse storage deltas, |
|
351 | 351 | * CG_DELTAMODE_PREV - only delta against "prev", |
|
352 | 352 | * CG_DELTAMODE_FULL - only issue full snapshot. |
|
353 | 353 | |
|
354 | 354 | Whether to send fulltext revisions instead of deltas, if allowed. |
|
355 | 355 | |
|
356 | 356 | ``nodesorder`` |
|
357 | 357 | ``revisiondata`` |
|
358 | 358 | ``assumehaveparentrevisions`` |
|
359 | 359 | """ |
|
360 | 360 | |
|
361 | 361 | fnode = store.node |
|
362 | 362 | frev = store.rev |
|
363 | 363 | |
|
364 | 364 | if nodesorder == b'nodes': |
|
365 | 365 | revs = [frev(n) for n in nodes] |
|
366 | 366 | elif nodesorder == b'linear': |
|
367 | 367 | revs = {frev(n) for n in nodes} |
|
368 | 368 | revs = dagop.linearize(revs, store.parentrevs) |
|
369 | 369 | else: # storage and default |
|
370 | 370 | revs = sorted(frev(n) for n in nodes) |
|
371 | 371 | |
|
372 | 372 | prevrev = None |
|
373 | 373 | |
|
374 | 374 | if deltamode == repository.CG_DELTAMODE_PREV or assumehaveparentrevisions: |
|
375 | 375 | prevrev = store.parentrevs(revs[0])[0] |
|
376 | 376 | |
|
377 | 377 | # Set of revs available to delta against. |
|
378 | 378 | available = set() |
|
379 | 379 | |
|
380 | 380 | for rev in revs: |
|
381 | 381 | if rev == nullrev: |
|
382 | 382 | continue |
|
383 | 383 | |
|
384 | 384 | node = fnode(rev) |
|
385 | 385 | p1rev, p2rev = store.parentrevs(rev) |
|
386 | 386 | |
|
387 | 387 | if deltaparentfn: |
|
388 | 388 | deltaparentrev = deltaparentfn(rev) |
|
389 | 389 | else: |
|
390 | 390 | deltaparentrev = nullrev |
|
391 | 391 | |
|
392 | 392 | # Forced delta against previous mode. |
|
393 | 393 | if deltamode == repository.CG_DELTAMODE_PREV: |
|
394 | 394 | baserev = prevrev |
|
395 | 395 | |
|
396 | 396 | # We're instructed to send fulltext. Honor that. |
|
397 | 397 | elif deltamode == repository.CG_DELTAMODE_FULL: |
|
398 | 398 | baserev = nullrev |
|
399 | 399 | # We're instructed to use p1. Honor that |
|
400 | 400 | elif deltamode == repository.CG_DELTAMODE_P1: |
|
401 | 401 | baserev = p1rev |
|
402 | 402 | |
|
403 | 403 | # There is a delta in storage. We try to use that because it |
|
404 | 404 | # amounts to effectively copying data from storage and is |
|
405 | 405 | # therefore the fastest. |
|
406 | 406 | elif deltaparentrev != nullrev: |
|
407 | 407 | # Base revision was already emitted in this group. We can |
|
408 | 408 | # always safely use the delta. |
|
409 | 409 | if deltaparentrev in available: |
|
410 | 410 | baserev = deltaparentrev |
|
411 | 411 | |
|
412 | 412 | # Base revision is a parent that hasn't been emitted already. |
|
413 | 413 | # Use it if we can assume the receiver has the parent revision. |
|
414 | 414 | elif assumehaveparentrevisions and deltaparentrev in (p1rev, p2rev): |
|
415 | 415 | baserev = deltaparentrev |
|
416 | 416 | |
|
417 | 417 | # No guarantee the receiver has the delta parent. Send delta |
|
418 | 418 | # against last revision (if possible), which in the common case |
|
419 | 419 | # should be similar enough to this revision that the delta is |
|
420 | 420 | # reasonable. |
|
421 | 421 | elif prevrev is not None: |
|
422 | 422 | baserev = prevrev |
|
423 | 423 | else: |
|
424 | 424 | baserev = nullrev |
|
425 | 425 | |
|
426 | 426 | # Storage has a fulltext revision. |
|
427 | 427 | |
|
428 | 428 | # Let's use the previous revision, which is as good a guess as any. |
|
429 | 429 | # There is definitely room to improve this logic. |
|
430 | 430 | elif prevrev is not None: |
|
431 | 431 | baserev = prevrev |
|
432 | 432 | else: |
|
433 | 433 | baserev = nullrev |
|
434 | 434 | |
|
435 | 435 | # But we can't actually use our chosen delta base for whatever |
|
436 | 436 | # reason. Reset to fulltext. |
|
437 | 437 | if baserev != nullrev and (candeltafn and not candeltafn(baserev, rev)): |
|
438 | 438 | baserev = nullrev |
|
439 | 439 | |
|
440 | 440 | revision = None |
|
441 | 441 | delta = None |
|
442 | 442 | baserevisionsize = None |
|
443 | 443 | |
|
444 | 444 | if revisiondata: |
|
445 | 445 | if store.iscensored(baserev) or store.iscensored(rev): |
|
446 | 446 | try: |
|
447 | 447 | revision = store.rawdata(node) |
|
448 | 448 | except error.CensoredNodeError as e: |
|
449 | 449 | revision = e.tombstone |
|
450 | 450 | |
|
451 | 451 | if baserev != nullrev: |
|
452 | 452 | if rawsizefn: |
|
453 | 453 | baserevisionsize = rawsizefn(baserev) |
|
454 | 454 | else: |
|
455 | 455 | baserevisionsize = len(store.rawdata(baserev)) |
|
456 | 456 | |
|
457 | 457 | elif ( |
|
458 | 458 | baserev == nullrev and deltamode != repository.CG_DELTAMODE_PREV |
|
459 | 459 | ): |
|
460 | 460 | revision = store.rawdata(node) |
|
461 | 461 | available.add(rev) |
|
462 | 462 | else: |
|
463 | 463 | if revdifffn: |
|
464 | 464 | delta = revdifffn(baserev, rev) |
|
465 | 465 | else: |
|
466 | 466 | delta = mdiff.textdiff( |
|
467 | 467 | store.rawdata(baserev), store.rawdata(rev) |
|
468 | 468 | ) |
|
469 | 469 | |
|
470 | 470 | available.add(rev) |
|
471 | 471 | |
|
472 | 472 | yield resultcls( |
|
473 | 473 | node=node, |
|
474 | 474 | p1node=fnode(p1rev), |
|
475 | 475 | p2node=fnode(p2rev), |
|
476 | 476 | basenode=fnode(baserev), |
|
477 | 477 | flags=flagsfn(rev) if flagsfn else 0, |
|
478 | 478 | baserevisionsize=baserevisionsize, |
|
479 | 479 | revision=revision, |
|
480 | 480 | delta=delta, |
|
481 | sidedata=sidedata, | |
|
481 | 482 | ) |
|
482 | 483 | |
|
483 | 484 | prevrev = rev |
|
484 | 485 | |
|
485 | 486 | |
|
486 | 487 | def deltaiscensored(delta, baserev, baselenfn): |
|
487 | 488 | """Determine if a delta represents censored revision data. |
|
488 | 489 | |
|
489 | 490 | ``baserev`` is the base revision this delta is encoded against. |
|
490 | 491 | ``baselenfn`` is a callable receiving a revision number that resolves the |
|
491 | 492 | length of the revision fulltext. |
|
492 | 493 | |
|
493 | 494 | Returns a bool indicating if the result of the delta represents a censored |
|
494 | 495 | revision. |
|
495 | 496 | """ |
|
496 | 497 | # Fragile heuristic: unless new file meta keys are added alphabetically |
|
497 | 498 | # preceding "censored", all censored revisions are prefixed by |
|
498 | 499 | # "\1\ncensored:". A delta producing such a censored revision must be a |
|
499 | 500 | # full-replacement delta, so we inspect the first and only patch in the |
|
500 | 501 | # delta for this prefix. |
|
501 | 502 | hlen = struct.calcsize(b">lll") |
|
502 | 503 | if len(delta) <= hlen: |
|
503 | 504 | return False |
|
504 | 505 | |
|
505 | 506 | oldlen = baselenfn(baserev) |
|
506 | 507 | newlen = len(delta) - hlen |
|
507 | 508 | if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen): |
|
508 | 509 | return False |
|
509 | 510 | |
|
510 | 511 | add = b"\1\ncensored:" |
|
511 | 512 | addlen = len(add) |
|
512 | 513 | return newlen >= addlen and delta[hlen : hlen + addlen] == add |
General Comments 0
You need to be logged in to leave comments.
Login now