Show More
@@ -1,750 +1,749 b'' | |||
|
1 | 1 | # store.py - repository store handling for Mercurial |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2008 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 errno |
|
11 | 11 | import functools |
|
12 | 12 | import os |
|
13 | 13 | import stat |
|
14 | 14 | |
|
15 | 15 | from .i18n import _ |
|
16 | 16 | from .pycompat import getattr |
|
17 | 17 | from .node import hex |
|
18 | 18 | from . import ( |
|
19 | 19 | changelog, |
|
20 | 20 | error, |
|
21 | 21 | manifest, |
|
22 | 22 | policy, |
|
23 | 23 | pycompat, |
|
24 | 24 | util, |
|
25 | 25 | vfs as vfsmod, |
|
26 | 26 | ) |
|
27 | 27 | from .utils import hashutil |
|
28 | 28 | |
|
29 | 29 | parsers = policy.importmod('parsers') |
|
30 | 30 | # how much bytes should be read from fncache in one read |
|
31 | 31 | # It is done to prevent loading large fncache files into memory |
|
32 | 32 | fncache_chunksize = 10 ** 6 |
|
33 | 33 | |
|
34 | 34 | |
|
35 | 35 | def _matchtrackedpath(path, matcher): |
|
36 | 36 | """parses a fncache entry and returns whether the entry is tracking a path |
|
37 | 37 | matched by matcher or not. |
|
38 | 38 | |
|
39 | 39 | If matcher is None, returns True""" |
|
40 | 40 | |
|
41 | 41 | if matcher is None: |
|
42 | 42 | return True |
|
43 | 43 | path = decodedir(path) |
|
44 | 44 | if path.startswith(b'data/'): |
|
45 | 45 | return matcher(path[len(b'data/') : -len(b'.i')]) |
|
46 | 46 | elif path.startswith(b'meta/'): |
|
47 | 47 | return matcher.visitdir(path[len(b'meta/') : -len(b'/00manifest.i')]) |
|
48 | 48 | |
|
49 | 49 | raise error.ProgrammingError(b"cannot decode path %s" % path) |
|
50 | 50 | |
|
51 | 51 | |
|
52 | 52 | # This avoids a collision between a file named foo and a dir named |
|
53 | 53 | # foo.i or foo.d |
|
54 | 54 | def _encodedir(path): |
|
55 | 55 | """ |
|
56 | 56 | >>> _encodedir(b'data/foo.i') |
|
57 | 57 | 'data/foo.i' |
|
58 | 58 | >>> _encodedir(b'data/foo.i/bla.i') |
|
59 | 59 | 'data/foo.i.hg/bla.i' |
|
60 | 60 | >>> _encodedir(b'data/foo.i.hg/bla.i') |
|
61 | 61 | 'data/foo.i.hg.hg/bla.i' |
|
62 | 62 | >>> _encodedir(b'data/foo.i\\ndata/foo.i/bla.i\\ndata/foo.i.hg/bla.i\\n') |
|
63 | 63 | 'data/foo.i\\ndata/foo.i.hg/bla.i\\ndata/foo.i.hg.hg/bla.i\\n' |
|
64 | 64 | """ |
|
65 | 65 | return ( |
|
66 | 66 | path.replace(b".hg/", b".hg.hg/") |
|
67 | 67 | .replace(b".i/", b".i.hg/") |
|
68 | 68 | .replace(b".d/", b".d.hg/") |
|
69 | 69 | ) |
|
70 | 70 | |
|
71 | 71 | |
|
72 | 72 | encodedir = getattr(parsers, 'encodedir', _encodedir) |
|
73 | 73 | |
|
74 | 74 | |
|
75 | 75 | def decodedir(path): |
|
76 | 76 | """ |
|
77 | 77 | >>> decodedir(b'data/foo.i') |
|
78 | 78 | 'data/foo.i' |
|
79 | 79 | >>> decodedir(b'data/foo.i.hg/bla.i') |
|
80 | 80 | 'data/foo.i/bla.i' |
|
81 | 81 | >>> decodedir(b'data/foo.i.hg.hg/bla.i') |
|
82 | 82 | 'data/foo.i.hg/bla.i' |
|
83 | 83 | """ |
|
84 | 84 | if b".hg/" not in path: |
|
85 | 85 | return path |
|
86 | 86 | return ( |
|
87 | 87 | path.replace(b".d.hg/", b".d/") |
|
88 | 88 | .replace(b".i.hg/", b".i/") |
|
89 | 89 | .replace(b".hg.hg/", b".hg/") |
|
90 | 90 | ) |
|
91 | 91 | |
|
92 | 92 | |
|
93 | 93 | def _reserved(): |
|
94 | 94 | """characters that are problematic for filesystems |
|
95 | 95 | |
|
96 | 96 | * ascii escapes (0..31) |
|
97 | 97 | * ascii hi (126..255) |
|
98 | 98 | * windows specials |
|
99 | 99 | |
|
100 | 100 | these characters will be escaped by encodefunctions |
|
101 | 101 | """ |
|
102 | 102 | winreserved = [ord(x) for x in u'\\:*?"<>|'] |
|
103 | 103 | for x in range(32): |
|
104 | 104 | yield x |
|
105 | 105 | for x in range(126, 256): |
|
106 | 106 | yield x |
|
107 | 107 | for x in winreserved: |
|
108 | 108 | yield x |
|
109 | 109 | |
|
110 | 110 | |
|
111 | 111 | def _buildencodefun(): |
|
112 | 112 | """ |
|
113 | 113 | >>> enc, dec = _buildencodefun() |
|
114 | 114 | |
|
115 | 115 | >>> enc(b'nothing/special.txt') |
|
116 | 116 | 'nothing/special.txt' |
|
117 | 117 | >>> dec(b'nothing/special.txt') |
|
118 | 118 | 'nothing/special.txt' |
|
119 | 119 | |
|
120 | 120 | >>> enc(b'HELLO') |
|
121 | 121 | '_h_e_l_l_o' |
|
122 | 122 | >>> dec(b'_h_e_l_l_o') |
|
123 | 123 | 'HELLO' |
|
124 | 124 | |
|
125 | 125 | >>> enc(b'hello:world?') |
|
126 | 126 | 'hello~3aworld~3f' |
|
127 | 127 | >>> dec(b'hello~3aworld~3f') |
|
128 | 128 | 'hello:world?' |
|
129 | 129 | |
|
130 | 130 | >>> enc(b'the\\x07quick\\xADshot') |
|
131 | 131 | 'the~07quick~adshot' |
|
132 | 132 | >>> dec(b'the~07quick~adshot') |
|
133 | 133 | 'the\\x07quick\\xadshot' |
|
134 | 134 | """ |
|
135 | 135 | e = b'_' |
|
136 | 136 | xchr = pycompat.bytechr |
|
137 | 137 | asciistr = list(map(xchr, range(127))) |
|
138 | 138 | capitals = list(range(ord(b"A"), ord(b"Z") + 1)) |
|
139 | 139 | |
|
140 | 140 | cmap = {x: x for x in asciistr} |
|
141 | 141 | for x in _reserved(): |
|
142 | 142 | cmap[xchr(x)] = b"~%02x" % x |
|
143 | 143 | for x in capitals + [ord(e)]: |
|
144 | 144 | cmap[xchr(x)] = e + xchr(x).lower() |
|
145 | 145 | |
|
146 | 146 | dmap = {} |
|
147 | 147 | for k, v in pycompat.iteritems(cmap): |
|
148 | 148 | dmap[v] = k |
|
149 | 149 | |
|
150 | 150 | def decode(s): |
|
151 | 151 | i = 0 |
|
152 | 152 | while i < len(s): |
|
153 | 153 | for l in pycompat.xrange(1, 4): |
|
154 | 154 | try: |
|
155 | 155 | yield dmap[s[i : i + l]] |
|
156 | 156 | i += l |
|
157 | 157 | break |
|
158 | 158 | except KeyError: |
|
159 | 159 | pass |
|
160 | 160 | else: |
|
161 | 161 | raise KeyError |
|
162 | 162 | |
|
163 | 163 | return ( |
|
164 | 164 | lambda s: b''.join( |
|
165 | 165 | [cmap[s[c : c + 1]] for c in pycompat.xrange(len(s))] |
|
166 | 166 | ), |
|
167 | 167 | lambda s: b''.join(list(decode(s))), |
|
168 | 168 | ) |
|
169 | 169 | |
|
170 | 170 | |
|
171 | 171 | _encodefname, _decodefname = _buildencodefun() |
|
172 | 172 | |
|
173 | 173 | |
|
174 | 174 | def encodefilename(s): |
|
175 | 175 | """ |
|
176 | 176 | >>> encodefilename(b'foo.i/bar.d/bla.hg/hi:world?/HELLO') |
|
177 | 177 | 'foo.i.hg/bar.d.hg/bla.hg.hg/hi~3aworld~3f/_h_e_l_l_o' |
|
178 | 178 | """ |
|
179 | 179 | return _encodefname(encodedir(s)) |
|
180 | 180 | |
|
181 | 181 | |
|
182 | 182 | def decodefilename(s): |
|
183 | 183 | """ |
|
184 | 184 | >>> decodefilename(b'foo.i.hg/bar.d.hg/bla.hg.hg/hi~3aworld~3f/_h_e_l_l_o') |
|
185 | 185 | 'foo.i/bar.d/bla.hg/hi:world?/HELLO' |
|
186 | 186 | """ |
|
187 | 187 | return decodedir(_decodefname(s)) |
|
188 | 188 | |
|
189 | 189 | |
|
190 | 190 | def _buildlowerencodefun(): |
|
191 | 191 | """ |
|
192 | 192 | >>> f = _buildlowerencodefun() |
|
193 | 193 | >>> f(b'nothing/special.txt') |
|
194 | 194 | 'nothing/special.txt' |
|
195 | 195 | >>> f(b'HELLO') |
|
196 | 196 | 'hello' |
|
197 | 197 | >>> f(b'hello:world?') |
|
198 | 198 | 'hello~3aworld~3f' |
|
199 | 199 | >>> f(b'the\\x07quick\\xADshot') |
|
200 | 200 | 'the~07quick~adshot' |
|
201 | 201 | """ |
|
202 | 202 | xchr = pycompat.bytechr |
|
203 | 203 | cmap = {xchr(x): xchr(x) for x in pycompat.xrange(127)} |
|
204 | 204 | for x in _reserved(): |
|
205 | 205 | cmap[xchr(x)] = b"~%02x" % x |
|
206 | 206 | for x in range(ord(b"A"), ord(b"Z") + 1): |
|
207 | 207 | cmap[xchr(x)] = xchr(x).lower() |
|
208 | 208 | |
|
209 | 209 | def lowerencode(s): |
|
210 | 210 | return b"".join([cmap[c] for c in pycompat.iterbytestr(s)]) |
|
211 | 211 | |
|
212 | 212 | return lowerencode |
|
213 | 213 | |
|
214 | 214 | |
|
215 | 215 | lowerencode = getattr(parsers, 'lowerencode', None) or _buildlowerencodefun() |
|
216 | 216 | |
|
217 | 217 | # Windows reserved names: con, prn, aux, nul, com1..com9, lpt1..lpt9 |
|
218 | 218 | _winres3 = (b'aux', b'con', b'prn', b'nul') # length 3 |
|
219 | 219 | _winres4 = (b'com', b'lpt') # length 4 (with trailing 1..9) |
|
220 | 220 | |
|
221 | 221 | |
|
222 | 222 | def _auxencode(path, dotencode): |
|
223 | 223 | """ |
|
224 | 224 | Encodes filenames containing names reserved by Windows or which end in |
|
225 | 225 | period or space. Does not touch other single reserved characters c. |
|
226 | 226 | Specifically, c in '\\:*?"<>|' or ord(c) <= 31 are *not* encoded here. |
|
227 | 227 | Additionally encodes space or period at the beginning, if dotencode is |
|
228 | 228 | True. Parameter path is assumed to be all lowercase. |
|
229 | 229 | A segment only needs encoding if a reserved name appears as a |
|
230 | 230 | basename (e.g. "aux", "aux.foo"). A directory or file named "foo.aux" |
|
231 | 231 | doesn't need encoding. |
|
232 | 232 | |
|
233 | 233 | >>> s = b'.foo/aux.txt/txt.aux/con/prn/nul/foo.' |
|
234 | 234 | >>> _auxencode(s.split(b'/'), True) |
|
235 | 235 | ['~2efoo', 'au~78.txt', 'txt.aux', 'co~6e', 'pr~6e', 'nu~6c', 'foo~2e'] |
|
236 | 236 | >>> s = b'.com1com2/lpt9.lpt4.lpt1/conprn/com0/lpt0/foo.' |
|
237 | 237 | >>> _auxencode(s.split(b'/'), False) |
|
238 | 238 | ['.com1com2', 'lp~749.lpt4.lpt1', 'conprn', 'com0', 'lpt0', 'foo~2e'] |
|
239 | 239 | >>> _auxencode([b'foo. '], True) |
|
240 | 240 | ['foo.~20'] |
|
241 | 241 | >>> _auxencode([b' .foo'], True) |
|
242 | 242 | ['~20.foo'] |
|
243 | 243 | """ |
|
244 | 244 | for i, n in enumerate(path): |
|
245 | 245 | if not n: |
|
246 | 246 | continue |
|
247 | 247 | if dotencode and n[0] in b'. ': |
|
248 | 248 | n = b"~%02x" % ord(n[0:1]) + n[1:] |
|
249 | 249 | path[i] = n |
|
250 | 250 | else: |
|
251 | 251 | l = n.find(b'.') |
|
252 | 252 | if l == -1: |
|
253 | 253 | l = len(n) |
|
254 | 254 | if (l == 3 and n[:3] in _winres3) or ( |
|
255 | 255 | l == 4 |
|
256 | 256 | and n[3:4] <= b'9' |
|
257 | 257 | and n[3:4] >= b'1' |
|
258 | 258 | and n[:3] in _winres4 |
|
259 | 259 | ): |
|
260 | 260 | # encode third letter ('aux' -> 'au~78') |
|
261 | 261 | ec = b"~%02x" % ord(n[2:3]) |
|
262 | 262 | n = n[0:2] + ec + n[3:] |
|
263 | 263 | path[i] = n |
|
264 | 264 | if n[-1] in b'. ': |
|
265 | 265 | # encode last period or space ('foo...' -> 'foo..~2e') |
|
266 | 266 | path[i] = n[:-1] + b"~%02x" % ord(n[-1:]) |
|
267 | 267 | return path |
|
268 | 268 | |
|
269 | 269 | |
|
270 | 270 | _maxstorepathlen = 120 |
|
271 | 271 | _dirprefixlen = 8 |
|
272 | 272 | _maxshortdirslen = 8 * (_dirprefixlen + 1) - 4 |
|
273 | 273 | |
|
274 | 274 | |
|
275 | 275 | def _hashencode(path, dotencode): |
|
276 | 276 | digest = hex(hashutil.sha1(path).digest()) |
|
277 | 277 | le = lowerencode(path[5:]).split(b'/') # skips prefix 'data/' or 'meta/' |
|
278 | 278 | parts = _auxencode(le, dotencode) |
|
279 | 279 | basename = parts[-1] |
|
280 | 280 | _root, ext = os.path.splitext(basename) |
|
281 | 281 | sdirs = [] |
|
282 | 282 | sdirslen = 0 |
|
283 | 283 | for p in parts[:-1]: |
|
284 | 284 | d = p[:_dirprefixlen] |
|
285 | 285 | if d[-1] in b'. ': |
|
286 | 286 | # Windows can't access dirs ending in period or space |
|
287 | 287 | d = d[:-1] + b'_' |
|
288 | 288 | if sdirslen == 0: |
|
289 | 289 | t = len(d) |
|
290 | 290 | else: |
|
291 | 291 | t = sdirslen + 1 + len(d) |
|
292 | 292 | if t > _maxshortdirslen: |
|
293 | 293 | break |
|
294 | 294 | sdirs.append(d) |
|
295 | 295 | sdirslen = t |
|
296 | 296 | dirs = b'/'.join(sdirs) |
|
297 | 297 | if len(dirs) > 0: |
|
298 | 298 | dirs += b'/' |
|
299 | 299 | res = b'dh/' + dirs + digest + ext |
|
300 | 300 | spaceleft = _maxstorepathlen - len(res) |
|
301 | 301 | if spaceleft > 0: |
|
302 | 302 | filler = basename[:spaceleft] |
|
303 | 303 | res = b'dh/' + dirs + filler + digest + ext |
|
304 | 304 | return res |
|
305 | 305 | |
|
306 | 306 | |
|
307 | 307 | def _hybridencode(path, dotencode): |
|
308 | 308 | """encodes path with a length limit |
|
309 | 309 | |
|
310 | 310 | Encodes all paths that begin with 'data/', according to the following. |
|
311 | 311 | |
|
312 | 312 | Default encoding (reversible): |
|
313 | 313 | |
|
314 | 314 | Encodes all uppercase letters 'X' as '_x'. All reserved or illegal |
|
315 | 315 | characters are encoded as '~xx', where xx is the two digit hex code |
|
316 | 316 | of the character (see encodefilename). |
|
317 | 317 | Relevant path components consisting of Windows reserved filenames are |
|
318 | 318 | masked by encoding the third character ('aux' -> 'au~78', see _auxencode). |
|
319 | 319 | |
|
320 | 320 | Hashed encoding (not reversible): |
|
321 | 321 | |
|
322 | 322 | If the default-encoded path is longer than _maxstorepathlen, a |
|
323 | 323 | non-reversible hybrid hashing of the path is done instead. |
|
324 | 324 | This encoding uses up to _dirprefixlen characters of all directory |
|
325 | 325 | levels of the lowerencoded path, but not more levels than can fit into |
|
326 | 326 | _maxshortdirslen. |
|
327 | 327 | Then follows the filler followed by the sha digest of the full path. |
|
328 | 328 | The filler is the beginning of the basename of the lowerencoded path |
|
329 | 329 | (the basename is everything after the last path separator). The filler |
|
330 | 330 | is as long as possible, filling in characters from the basename until |
|
331 | 331 | the encoded path has _maxstorepathlen characters (or all chars of the |
|
332 | 332 | basename have been taken). |
|
333 | 333 | The extension (e.g. '.i' or '.d') is preserved. |
|
334 | 334 | |
|
335 | 335 | The string 'data/' at the beginning is replaced with 'dh/', if the hashed |
|
336 | 336 | encoding was used. |
|
337 | 337 | """ |
|
338 | 338 | path = encodedir(path) |
|
339 | 339 | ef = _encodefname(path).split(b'/') |
|
340 | 340 | res = b'/'.join(_auxencode(ef, dotencode)) |
|
341 | 341 | if len(res) > _maxstorepathlen: |
|
342 | 342 | res = _hashencode(path, dotencode) |
|
343 | 343 | return res |
|
344 | 344 | |
|
345 | 345 | |
|
346 | 346 | def _pathencode(path): |
|
347 | 347 | de = encodedir(path) |
|
348 | 348 | if len(path) > _maxstorepathlen: |
|
349 | 349 | return _hashencode(de, True) |
|
350 | 350 | ef = _encodefname(de).split(b'/') |
|
351 | 351 | res = b'/'.join(_auxencode(ef, True)) |
|
352 | 352 | if len(res) > _maxstorepathlen: |
|
353 | 353 | return _hashencode(de, True) |
|
354 | 354 | return res |
|
355 | 355 | |
|
356 | 356 | |
|
357 | 357 | _pathencode = getattr(parsers, 'pathencode', _pathencode) |
|
358 | 358 | |
|
359 | 359 | |
|
360 | 360 | def _plainhybridencode(f): |
|
361 | 361 | return _hybridencode(f, False) |
|
362 | 362 | |
|
363 | 363 | |
|
364 | 364 | def _calcmode(vfs): |
|
365 | 365 | try: |
|
366 | 366 | # files in .hg/ will be created using this mode |
|
367 | 367 | mode = vfs.stat().st_mode |
|
368 | 368 | # avoid some useless chmods |
|
369 | 369 | if (0o777 & ~util.umask) == (0o777 & mode): |
|
370 | 370 | mode = None |
|
371 | 371 | except OSError: |
|
372 | 372 | mode = None |
|
373 | 373 | return mode |
|
374 | 374 | |
|
375 | 375 | |
|
376 | 376 | _data = [ |
|
377 | 377 | b'bookmarks', |
|
378 | 378 | b'narrowspec', |
|
379 | 379 | b'data', |
|
380 | 380 | b'meta', |
|
381 | 381 | b'00manifest.d', |
|
382 | 382 | b'00manifest.i', |
|
383 | 383 | b'00changelog.d', |
|
384 | 384 | b'00changelog.i', |
|
385 | 385 | b'phaseroots', |
|
386 | 386 | b'obsstore', |
|
387 | 387 | b'requires', |
|
388 | 388 | ] |
|
389 | 389 | |
|
390 | REVLOG_FILES_EXT = (b'.i', b'.d', b'.n', b'.nd') | |
|
390 | 391 | |
|
391 | 392 | def isrevlog(f, kind, st): |
|
392 | 393 | if kind != stat.S_IFREG: |
|
393 | 394 | return False |
|
394 | if f[-2:] in (b'.i', b'.d', b'.n'): | |
|
395 | return True | |
|
396 | return f[-3:] == b'.nd' | |
|
395 | return f.endswith(REVLOG_FILES_EXT) | |
|
397 | 396 | |
|
398 | 397 | |
|
399 | 398 | class basicstore(object): |
|
400 | 399 | '''base class for local repository stores''' |
|
401 | 400 | |
|
402 | 401 | def __init__(self, path, vfstype): |
|
403 | 402 | vfs = vfstype(path) |
|
404 | 403 | self.path = vfs.base |
|
405 | 404 | self.createmode = _calcmode(vfs) |
|
406 | 405 | vfs.createmode = self.createmode |
|
407 | 406 | self.rawvfs = vfs |
|
408 | 407 | self.vfs = vfsmod.filtervfs(vfs, encodedir) |
|
409 | 408 | self.opener = self.vfs |
|
410 | 409 | |
|
411 | 410 | def join(self, f): |
|
412 | 411 | return self.path + b'/' + encodedir(f) |
|
413 | 412 | |
|
414 | 413 | def _walk(self, relpath, recurse, filefilter=isrevlog): |
|
415 | 414 | '''yields (unencoded, encoded, size)''' |
|
416 | 415 | path = self.path |
|
417 | 416 | if relpath: |
|
418 | 417 | path += b'/' + relpath |
|
419 | 418 | striplen = len(self.path) + 1 |
|
420 | 419 | l = [] |
|
421 | 420 | if self.rawvfs.isdir(path): |
|
422 | 421 | visit = [path] |
|
423 | 422 | readdir = self.rawvfs.readdir |
|
424 | 423 | while visit: |
|
425 | 424 | p = visit.pop() |
|
426 | 425 | for f, kind, st in readdir(p, stat=True): |
|
427 | 426 | fp = p + b'/' + f |
|
428 | 427 | if filefilter(f, kind, st): |
|
429 | 428 | n = util.pconvert(fp[striplen:]) |
|
430 | 429 | l.append((decodedir(n), n, st.st_size)) |
|
431 | 430 | elif kind == stat.S_IFDIR and recurse: |
|
432 | 431 | visit.append(fp) |
|
433 | 432 | l.sort() |
|
434 | 433 | return l |
|
435 | 434 | |
|
436 | 435 | def changelog(self, trypending): |
|
437 | 436 | return changelog.changelog(self.vfs, trypending=trypending) |
|
438 | 437 | |
|
439 | 438 | def manifestlog(self, repo, storenarrowmatch): |
|
440 | 439 | rootstore = manifest.manifestrevlog(self.vfs) |
|
441 | 440 | return manifest.manifestlog(self.vfs, repo, rootstore, storenarrowmatch) |
|
442 | 441 | |
|
443 | 442 | def datafiles(self, matcher=None): |
|
444 | 443 | return self._walk(b'data', True) + self._walk(b'meta', True) |
|
445 | 444 | |
|
446 | 445 | def topfiles(self): |
|
447 | 446 | # yield manifest before changelog |
|
448 | 447 | return reversed(self._walk(b'', False)) |
|
449 | 448 | |
|
450 | 449 | def walk(self, matcher=None): |
|
451 | 450 | """yields (unencoded, encoded, size) |
|
452 | 451 | |
|
453 | 452 | if a matcher is passed, storage files of only those tracked paths |
|
454 | 453 | are passed with matches the matcher |
|
455 | 454 | """ |
|
456 | 455 | # yield data files first |
|
457 | 456 | for x in self.datafiles(matcher): |
|
458 | 457 | yield x |
|
459 | 458 | for x in self.topfiles(): |
|
460 | 459 | yield x |
|
461 | 460 | |
|
462 | 461 | def copylist(self): |
|
463 | 462 | return _data |
|
464 | 463 | |
|
465 | 464 | def write(self, tr): |
|
466 | 465 | pass |
|
467 | 466 | |
|
468 | 467 | def invalidatecaches(self): |
|
469 | 468 | pass |
|
470 | 469 | |
|
471 | 470 | def markremoved(self, fn): |
|
472 | 471 | pass |
|
473 | 472 | |
|
474 | 473 | def __contains__(self, path): |
|
475 | 474 | '''Checks if the store contains path''' |
|
476 | 475 | path = b"/".join((b"data", path)) |
|
477 | 476 | # file? |
|
478 | 477 | if self.vfs.exists(path + b".i"): |
|
479 | 478 | return True |
|
480 | 479 | # dir? |
|
481 | 480 | if not path.endswith(b"/"): |
|
482 | 481 | path = path + b"/" |
|
483 | 482 | return self.vfs.exists(path) |
|
484 | 483 | |
|
485 | 484 | |
|
486 | 485 | class encodedstore(basicstore): |
|
487 | 486 | def __init__(self, path, vfstype): |
|
488 | 487 | vfs = vfstype(path + b'/store') |
|
489 | 488 | self.path = vfs.base |
|
490 | 489 | self.createmode = _calcmode(vfs) |
|
491 | 490 | vfs.createmode = self.createmode |
|
492 | 491 | self.rawvfs = vfs |
|
493 | 492 | self.vfs = vfsmod.filtervfs(vfs, encodefilename) |
|
494 | 493 | self.opener = self.vfs |
|
495 | 494 | |
|
496 | 495 | def datafiles(self, matcher=None): |
|
497 | 496 | for a, b, size in super(encodedstore, self).datafiles(): |
|
498 | 497 | try: |
|
499 | 498 | a = decodefilename(a) |
|
500 | 499 | except KeyError: |
|
501 | 500 | a = None |
|
502 | 501 | if a is not None and not _matchtrackedpath(a, matcher): |
|
503 | 502 | continue |
|
504 | 503 | yield a, b, size |
|
505 | 504 | |
|
506 | 505 | def join(self, f): |
|
507 | 506 | return self.path + b'/' + encodefilename(f) |
|
508 | 507 | |
|
509 | 508 | def copylist(self): |
|
510 | 509 | return [b'requires', b'00changelog.i'] + [b'store/' + f for f in _data] |
|
511 | 510 | |
|
512 | 511 | |
|
513 | 512 | class fncache(object): |
|
514 | 513 | # the filename used to be partially encoded |
|
515 | 514 | # hence the encodedir/decodedir dance |
|
516 | 515 | def __init__(self, vfs): |
|
517 | 516 | self.vfs = vfs |
|
518 | 517 | self.entries = None |
|
519 | 518 | self._dirty = False |
|
520 | 519 | # set of new additions to fncache |
|
521 | 520 | self.addls = set() |
|
522 | 521 | |
|
523 | 522 | def ensureloaded(self, warn=None): |
|
524 | 523 | """read the fncache file if not already read. |
|
525 | 524 | |
|
526 | 525 | If the file on disk is corrupted, raise. If warn is provided, |
|
527 | 526 | warn and keep going instead.""" |
|
528 | 527 | if self.entries is None: |
|
529 | 528 | self._load(warn) |
|
530 | 529 | |
|
531 | 530 | def _load(self, warn=None): |
|
532 | 531 | '''fill the entries from the fncache file''' |
|
533 | 532 | self._dirty = False |
|
534 | 533 | try: |
|
535 | 534 | fp = self.vfs(b'fncache', mode=b'rb') |
|
536 | 535 | except IOError: |
|
537 | 536 | # skip nonexistent file |
|
538 | 537 | self.entries = set() |
|
539 | 538 | return |
|
540 | 539 | |
|
541 | 540 | self.entries = set() |
|
542 | 541 | chunk = b'' |
|
543 | 542 | for c in iter(functools.partial(fp.read, fncache_chunksize), b''): |
|
544 | 543 | chunk += c |
|
545 | 544 | try: |
|
546 | 545 | p = chunk.rindex(b'\n') |
|
547 | 546 | self.entries.update(decodedir(chunk[: p + 1]).splitlines()) |
|
548 | 547 | chunk = chunk[p + 1 :] |
|
549 | 548 | except ValueError: |
|
550 | 549 | # substring '\n' not found, maybe the entry is bigger than the |
|
551 | 550 | # chunksize, so let's keep iterating |
|
552 | 551 | pass |
|
553 | 552 | |
|
554 | 553 | if chunk: |
|
555 | 554 | msg = _(b"fncache does not ends with a newline") |
|
556 | 555 | if warn: |
|
557 | 556 | warn(msg + b'\n') |
|
558 | 557 | else: |
|
559 | 558 | raise error.Abort( |
|
560 | 559 | msg, |
|
561 | 560 | hint=_( |
|
562 | 561 | b"use 'hg debugrebuildfncache' to " |
|
563 | 562 | b"rebuild the fncache" |
|
564 | 563 | ), |
|
565 | 564 | ) |
|
566 | 565 | self._checkentries(fp, warn) |
|
567 | 566 | fp.close() |
|
568 | 567 | |
|
569 | 568 | def _checkentries(self, fp, warn): |
|
570 | 569 | """ make sure there is no empty string in entries """ |
|
571 | 570 | if b'' in self.entries: |
|
572 | 571 | fp.seek(0) |
|
573 | 572 | for n, line in enumerate(util.iterfile(fp)): |
|
574 | 573 | if not line.rstrip(b'\n'): |
|
575 | 574 | t = _(b'invalid entry in fncache, line %d') % (n + 1) |
|
576 | 575 | if warn: |
|
577 | 576 | warn(t + b'\n') |
|
578 | 577 | else: |
|
579 | 578 | raise error.Abort(t) |
|
580 | 579 | |
|
581 | 580 | def write(self, tr): |
|
582 | 581 | if self._dirty: |
|
583 | 582 | assert self.entries is not None |
|
584 | 583 | self.entries = self.entries | self.addls |
|
585 | 584 | self.addls = set() |
|
586 | 585 | tr.addbackup(b'fncache') |
|
587 | 586 | fp = self.vfs(b'fncache', mode=b'wb', atomictemp=True) |
|
588 | 587 | if self.entries: |
|
589 | 588 | fp.write(encodedir(b'\n'.join(self.entries) + b'\n')) |
|
590 | 589 | fp.close() |
|
591 | 590 | self._dirty = False |
|
592 | 591 | if self.addls: |
|
593 | 592 | # if we have just new entries, let's append them to the fncache |
|
594 | 593 | tr.addbackup(b'fncache') |
|
595 | 594 | fp = self.vfs(b'fncache', mode=b'ab', atomictemp=True) |
|
596 | 595 | if self.addls: |
|
597 | 596 | fp.write(encodedir(b'\n'.join(self.addls) + b'\n')) |
|
598 | 597 | fp.close() |
|
599 | 598 | self.entries = None |
|
600 | 599 | self.addls = set() |
|
601 | 600 | |
|
602 | 601 | def add(self, fn): |
|
603 | 602 | if self.entries is None: |
|
604 | 603 | self._load() |
|
605 | 604 | if fn not in self.entries: |
|
606 | 605 | self.addls.add(fn) |
|
607 | 606 | |
|
608 | 607 | def remove(self, fn): |
|
609 | 608 | if self.entries is None: |
|
610 | 609 | self._load() |
|
611 | 610 | if fn in self.addls: |
|
612 | 611 | self.addls.remove(fn) |
|
613 | 612 | return |
|
614 | 613 | try: |
|
615 | 614 | self.entries.remove(fn) |
|
616 | 615 | self._dirty = True |
|
617 | 616 | except KeyError: |
|
618 | 617 | pass |
|
619 | 618 | |
|
620 | 619 | def __contains__(self, fn): |
|
621 | 620 | if fn in self.addls: |
|
622 | 621 | return True |
|
623 | 622 | if self.entries is None: |
|
624 | 623 | self._load() |
|
625 | 624 | return fn in self.entries |
|
626 | 625 | |
|
627 | 626 | def __iter__(self): |
|
628 | 627 | if self.entries is None: |
|
629 | 628 | self._load() |
|
630 | 629 | return iter(self.entries | self.addls) |
|
631 | 630 | |
|
632 | 631 | |
|
633 | 632 | class _fncachevfs(vfsmod.proxyvfs): |
|
634 | 633 | def __init__(self, vfs, fnc, encode): |
|
635 | 634 | vfsmod.proxyvfs.__init__(self, vfs) |
|
636 | 635 | self.fncache = fnc |
|
637 | 636 | self.encode = encode |
|
638 | 637 | |
|
639 | 638 | def __call__(self, path, mode=b'r', *args, **kw): |
|
640 | 639 | encoded = self.encode(path) |
|
641 | 640 | if mode not in (b'r', b'rb') and ( |
|
642 | 641 | path.startswith(b'data/') or path.startswith(b'meta/') |
|
643 | 642 | ): |
|
644 | 643 | # do not trigger a fncache load when adding a file that already is |
|
645 | 644 | # known to exist. |
|
646 | 645 | notload = self.fncache.entries is None and self.vfs.exists(encoded) |
|
647 | 646 | if notload and b'a' in mode and not self.vfs.stat(encoded).st_size: |
|
648 | 647 | # when appending to an existing file, if the file has size zero, |
|
649 | 648 | # it should be considered as missing. Such zero-size files are |
|
650 | 649 | # the result of truncation when a transaction is aborted. |
|
651 | 650 | notload = False |
|
652 | 651 | if not notload: |
|
653 | 652 | self.fncache.add(path) |
|
654 | 653 | return self.vfs(encoded, mode, *args, **kw) |
|
655 | 654 | |
|
656 | 655 | def join(self, path): |
|
657 | 656 | if path: |
|
658 | 657 | return self.vfs.join(self.encode(path)) |
|
659 | 658 | else: |
|
660 | 659 | return self.vfs.join(path) |
|
661 | 660 | |
|
662 | 661 | |
|
663 | 662 | class fncachestore(basicstore): |
|
664 | 663 | def __init__(self, path, vfstype, dotencode): |
|
665 | 664 | if dotencode: |
|
666 | 665 | encode = _pathencode |
|
667 | 666 | else: |
|
668 | 667 | encode = _plainhybridencode |
|
669 | 668 | self.encode = encode |
|
670 | 669 | vfs = vfstype(path + b'/store') |
|
671 | 670 | self.path = vfs.base |
|
672 | 671 | self.pathsep = self.path + b'/' |
|
673 | 672 | self.createmode = _calcmode(vfs) |
|
674 | 673 | vfs.createmode = self.createmode |
|
675 | 674 | self.rawvfs = vfs |
|
676 | 675 | fnc = fncache(vfs) |
|
677 | 676 | self.fncache = fnc |
|
678 | 677 | self.vfs = _fncachevfs(vfs, fnc, encode) |
|
679 | 678 | self.opener = self.vfs |
|
680 | 679 | |
|
681 | 680 | def join(self, f): |
|
682 | 681 | return self.pathsep + self.encode(f) |
|
683 | 682 | |
|
684 | 683 | def getsize(self, path): |
|
685 | 684 | return self.rawvfs.stat(path).st_size |
|
686 | 685 | |
|
687 | 686 | def datafiles(self, matcher=None): |
|
688 | 687 | for f in sorted(self.fncache): |
|
689 | 688 | if not _matchtrackedpath(f, matcher): |
|
690 | 689 | continue |
|
691 | 690 | ef = self.encode(f) |
|
692 | 691 | try: |
|
693 | 692 | yield f, ef, self.getsize(ef) |
|
694 | 693 | except OSError as err: |
|
695 | 694 | if err.errno != errno.ENOENT: |
|
696 | 695 | raise |
|
697 | 696 | |
|
698 | 697 | def copylist(self): |
|
699 | 698 | d = ( |
|
700 | 699 | b'bookmarks', |
|
701 | 700 | b'narrowspec', |
|
702 | 701 | b'data', |
|
703 | 702 | b'meta', |
|
704 | 703 | b'dh', |
|
705 | 704 | b'fncache', |
|
706 | 705 | b'phaseroots', |
|
707 | 706 | b'obsstore', |
|
708 | 707 | b'00manifest.d', |
|
709 | 708 | b'00manifest.i', |
|
710 | 709 | b'00changelog.d', |
|
711 | 710 | b'00changelog.i', |
|
712 | 711 | b'requires', |
|
713 | 712 | ) |
|
714 | 713 | return [b'requires', b'00changelog.i'] + [b'store/' + f for f in d] |
|
715 | 714 | |
|
716 | 715 | def write(self, tr): |
|
717 | 716 | self.fncache.write(tr) |
|
718 | 717 | |
|
719 | 718 | def invalidatecaches(self): |
|
720 | 719 | self.fncache.entries = None |
|
721 | 720 | self.fncache.addls = set() |
|
722 | 721 | |
|
723 | 722 | def markremoved(self, fn): |
|
724 | 723 | self.fncache.remove(fn) |
|
725 | 724 | |
|
726 | 725 | def _exists(self, f): |
|
727 | 726 | ef = self.encode(f) |
|
728 | 727 | try: |
|
729 | 728 | self.getsize(ef) |
|
730 | 729 | return True |
|
731 | 730 | except OSError as err: |
|
732 | 731 | if err.errno != errno.ENOENT: |
|
733 | 732 | raise |
|
734 | 733 | # nonexistent entry |
|
735 | 734 | return False |
|
736 | 735 | |
|
737 | 736 | def __contains__(self, path): |
|
738 | 737 | '''Checks if the store contains path''' |
|
739 | 738 | path = b"/".join((b"data", path)) |
|
740 | 739 | # check for files (exact match) |
|
741 | 740 | e = path + b'.i' |
|
742 | 741 | if e in self.fncache and self._exists(e): |
|
743 | 742 | return True |
|
744 | 743 | # now check for directories (prefix match) |
|
745 | 744 | if not path.endswith(b'/'): |
|
746 | 745 | path += b'/' |
|
747 | 746 | for e in self.fncache: |
|
748 | 747 | if e.startswith(path) and self._exists(e): |
|
749 | 748 | return True |
|
750 | 749 | return False |
General Comments 0
You need to be logged in to leave comments.
Login now