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