##// END OF EJS Templates
store: rename "op" variables to "vfs"
FUJIWARA Katsunori -
r17652:2c6f7231 default
parent child Browse files
Show More
@@ -1,494 +1,494 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 i18n import _
9 9 import osutil, scmutil, util, parsers
10 10 import os, stat, errno
11 11
12 12 _sha = util.sha1
13 13
14 14 # This avoids a collision between a file named foo and a dir named
15 15 # foo.i or foo.d
16 16 def _encodedir(path):
17 17 '''
18 18 >>> _encodedir('data/foo.i')
19 19 'data/foo.i'
20 20 >>> _encodedir('data/foo.i/bla.i')
21 21 'data/foo.i.hg/bla.i'
22 22 >>> _encodedir('data/foo.i.hg/bla.i')
23 23 'data/foo.i.hg.hg/bla.i'
24 24 >>> _encodedir('data/foo.i\\ndata/foo.i/bla.i\\ndata/foo.i.hg/bla.i\\n')
25 25 'data/foo.i\\ndata/foo.i.hg/bla.i\\ndata/foo.i.hg.hg/bla.i\\n'
26 26 '''
27 27 return (path
28 28 .replace(".hg/", ".hg.hg/")
29 29 .replace(".i/", ".i.hg/")
30 30 .replace(".d/", ".d.hg/"))
31 31
32 32 encodedir = getattr(parsers, 'encodedir', _encodedir)
33 33
34 34 def decodedir(path):
35 35 '''
36 36 >>> decodedir('data/foo.i')
37 37 'data/foo.i'
38 38 >>> decodedir('data/foo.i.hg/bla.i')
39 39 'data/foo.i/bla.i'
40 40 >>> decodedir('data/foo.i.hg.hg/bla.i')
41 41 'data/foo.i.hg/bla.i'
42 42 '''
43 43 if ".hg/" not in path:
44 44 return path
45 45 return (path
46 46 .replace(".d.hg/", ".d/")
47 47 .replace(".i.hg/", ".i/")
48 48 .replace(".hg.hg/", ".hg/"))
49 49
50 50 def _buildencodefun():
51 51 '''
52 52 >>> enc, dec = _buildencodefun()
53 53
54 54 >>> enc('nothing/special.txt')
55 55 'nothing/special.txt'
56 56 >>> dec('nothing/special.txt')
57 57 'nothing/special.txt'
58 58
59 59 >>> enc('HELLO')
60 60 '_h_e_l_l_o'
61 61 >>> dec('_h_e_l_l_o')
62 62 'HELLO'
63 63
64 64 >>> enc('hello:world?')
65 65 'hello~3aworld~3f'
66 66 >>> dec('hello~3aworld~3f')
67 67 'hello:world?'
68 68
69 69 >>> enc('the\x07quick\xADshot')
70 70 'the~07quick~adshot'
71 71 >>> dec('the~07quick~adshot')
72 72 'the\\x07quick\\xadshot'
73 73 '''
74 74 e = '_'
75 75 winreserved = [ord(x) for x in '\\:*?"<>|']
76 76 cmap = dict([(chr(x), chr(x)) for x in xrange(127)])
77 77 for x in (range(32) + range(126, 256) + winreserved):
78 78 cmap[chr(x)] = "~%02x" % x
79 79 for x in range(ord("A"), ord("Z")+1) + [ord(e)]:
80 80 cmap[chr(x)] = e + chr(x).lower()
81 81 dmap = {}
82 82 for k, v in cmap.iteritems():
83 83 dmap[v] = k
84 84 def decode(s):
85 85 i = 0
86 86 while i < len(s):
87 87 for l in xrange(1, 4):
88 88 try:
89 89 yield dmap[s[i:i + l]]
90 90 i += l
91 91 break
92 92 except KeyError:
93 93 pass
94 94 else:
95 95 raise KeyError
96 96 return (lambda s: ''.join([cmap[c] for c in s]),
97 97 lambda s: ''.join(list(decode(s))))
98 98
99 99 _encodefname, _decodefname = _buildencodefun()
100 100
101 101 def encodefilename(s):
102 102 '''
103 103 >>> encodefilename('foo.i/bar.d/bla.hg/hi:world?/HELLO')
104 104 'foo.i.hg/bar.d.hg/bla.hg.hg/hi~3aworld~3f/_h_e_l_l_o'
105 105 '''
106 106 return _encodefname(encodedir(s))
107 107
108 108 def decodefilename(s):
109 109 '''
110 110 >>> decodefilename('foo.i.hg/bar.d.hg/bla.hg.hg/hi~3aworld~3f/_h_e_l_l_o')
111 111 'foo.i/bar.d/bla.hg/hi:world?/HELLO'
112 112 '''
113 113 return decodedir(_decodefname(s))
114 114
115 115 def _buildlowerencodefun():
116 116 '''
117 117 >>> f = _buildlowerencodefun()
118 118 >>> f('nothing/special.txt')
119 119 'nothing/special.txt'
120 120 >>> f('HELLO')
121 121 'hello'
122 122 >>> f('hello:world?')
123 123 'hello~3aworld~3f'
124 124 >>> f('the\x07quick\xADshot')
125 125 'the~07quick~adshot'
126 126 '''
127 127 winreserved = [ord(x) for x in '\\:*?"<>|']
128 128 cmap = dict([(chr(x), chr(x)) for x in xrange(127)])
129 129 for x in (range(32) + range(126, 256) + winreserved):
130 130 cmap[chr(x)] = "~%02x" % x
131 131 for x in range(ord("A"), ord("Z")+1):
132 132 cmap[chr(x)] = chr(x).lower()
133 133 return lambda s: "".join([cmap[c] for c in s])
134 134
135 135 lowerencode = _buildlowerencodefun()
136 136
137 137 # Windows reserved names: con, prn, aux, nul, com1..com9, lpt1..lpt9
138 138 _winres3 = ('aux', 'con', 'prn', 'nul') # length 3
139 139 _winres4 = ('com', 'lpt') # length 4 (with trailing 1..9)
140 140 def _auxencode(path, dotencode):
141 141 '''
142 142 Encodes filenames containing names reserved by Windows or which end in
143 143 period or space. Does not touch other single reserved characters c.
144 144 Specifically, c in '\\:*?"<>|' or ord(c) <= 31 are *not* encoded here.
145 145 Additionally encodes space or period at the beginning, if dotencode is
146 146 True. Parameter path is assumed to be all lowercase.
147 147 A segment only needs encoding if a reserved name appears as a
148 148 basename (e.g. "aux", "aux.foo"). A directory or file named "foo.aux"
149 149 doesn't need encoding.
150 150
151 151 >>> s = '.foo/aux.txt/txt.aux/con/prn/nul/foo.'
152 152 >>> _auxencode(s.split('/'), True)
153 153 ['~2efoo', 'au~78.txt', 'txt.aux', 'co~6e', 'pr~6e', 'nu~6c', 'foo~2e']
154 154 >>> s = '.com1com2/lpt9.lpt4.lpt1/conprn/com0/lpt0/foo.'
155 155 >>> _auxencode(s.split('/'), False)
156 156 ['.com1com2', 'lp~749.lpt4.lpt1', 'conprn', 'com0', 'lpt0', 'foo~2e']
157 157 >>> _auxencode(['foo. '], True)
158 158 ['foo.~20']
159 159 >>> _auxencode([' .foo'], True)
160 160 ['~20.foo']
161 161 '''
162 162 for i, n in enumerate(path):
163 163 if not n:
164 164 continue
165 165 if dotencode and n[0] in '. ':
166 166 n = "~%02x" % ord(n[0]) + n[1:]
167 167 path[i] = n
168 168 else:
169 169 l = n.find('.')
170 170 if l == -1:
171 171 l = len(n)
172 172 if ((l == 3 and n[:3] in _winres3) or
173 173 (l == 4 and n[3] <= '9' and n[3] >= '1'
174 174 and n[:3] in _winres4)):
175 175 # encode third letter ('aux' -> 'au~78')
176 176 ec = "~%02x" % ord(n[2])
177 177 n = n[0:2] + ec + n[3:]
178 178 path[i] = n
179 179 if n[-1] in '. ':
180 180 # encode last period or space ('foo...' -> 'foo..~2e')
181 181 path[i] = n[:-1] + "~%02x" % ord(n[-1])
182 182 return path
183 183
184 184 _maxstorepathlen = 120
185 185 _dirprefixlen = 8
186 186 _maxshortdirslen = 8 * (_dirprefixlen + 1) - 4
187 187
188 188 def _hashencode(path, dotencode):
189 189 digest = _sha(path).hexdigest()
190 190 le = lowerencode(path).split('/')[1:]
191 191 parts = _auxencode(le, dotencode)
192 192 basename = parts[-1]
193 193 _root, ext = os.path.splitext(basename)
194 194 sdirs = []
195 195 sdirslen = 0
196 196 for p in parts[:-1]:
197 197 d = p[:_dirprefixlen]
198 198 if d[-1] in '. ':
199 199 # Windows can't access dirs ending in period or space
200 200 d = d[:-1] + '_'
201 201 if sdirslen == 0:
202 202 t = len(d)
203 203 else:
204 204 t = sdirslen + 1 + len(d)
205 205 if t > _maxshortdirslen:
206 206 break
207 207 sdirs.append(d)
208 208 sdirslen = t
209 209 dirs = '/'.join(sdirs)
210 210 if len(dirs) > 0:
211 211 dirs += '/'
212 212 res = 'dh/' + dirs + digest + ext
213 213 spaceleft = _maxstorepathlen - len(res)
214 214 if spaceleft > 0:
215 215 filler = basename[:spaceleft]
216 216 res = 'dh/' + dirs + filler + digest + ext
217 217 return res
218 218
219 219 def _hybridencode(path, dotencode):
220 220 '''encodes path with a length limit
221 221
222 222 Encodes all paths that begin with 'data/', according to the following.
223 223
224 224 Default encoding (reversible):
225 225
226 226 Encodes all uppercase letters 'X' as '_x'. All reserved or illegal
227 227 characters are encoded as '~xx', where xx is the two digit hex code
228 228 of the character (see encodefilename).
229 229 Relevant path components consisting of Windows reserved filenames are
230 230 masked by encoding the third character ('aux' -> 'au~78', see auxencode).
231 231
232 232 Hashed encoding (not reversible):
233 233
234 234 If the default-encoded path is longer than _maxstorepathlen, a
235 235 non-reversible hybrid hashing of the path is done instead.
236 236 This encoding uses up to _dirprefixlen characters of all directory
237 237 levels of the lowerencoded path, but not more levels than can fit into
238 238 _maxshortdirslen.
239 239 Then follows the filler followed by the sha digest of the full path.
240 240 The filler is the beginning of the basename of the lowerencoded path
241 241 (the basename is everything after the last path separator). The filler
242 242 is as long as possible, filling in characters from the basename until
243 243 the encoded path has _maxstorepathlen characters (or all chars of the
244 244 basename have been taken).
245 245 The extension (e.g. '.i' or '.d') is preserved.
246 246
247 247 The string 'data/' at the beginning is replaced with 'dh/', if the hashed
248 248 encoding was used.
249 249 '''
250 250 path = encodedir(path)
251 251 ef = _encodefname(path).split('/')
252 252 res = '/'.join(_auxencode(ef, dotencode))
253 253 if len(res) > _maxstorepathlen:
254 254 res = _hashencode(path, dotencode)
255 255 return res
256 256
257 257 def _pathencode(path):
258 258 ef = _encodefname(encodedir(path)).split('/')
259 259 res = '/'.join(_auxencode(ef, True))
260 260 if len(res) > _maxstorepathlen:
261 261 return None
262 262 return res
263 263
264 264 _pathencode = getattr(parsers, 'pathencode', _pathencode)
265 265
266 266 def _dothybridencode(f):
267 267 ef = _pathencode(f)
268 268 if ef is None:
269 269 return _hashencode(encodedir(f), True)
270 270 return ef
271 271
272 272 def _plainhybridencode(f):
273 273 return _hybridencode(f, False)
274 274
275 275 def _calcmode(path):
276 276 try:
277 277 # files in .hg/ will be created using this mode
278 278 mode = os.stat(path).st_mode
279 279 # avoid some useless chmods
280 280 if (0777 & ~util.umask) == (0777 & mode):
281 281 mode = None
282 282 except OSError:
283 283 mode = None
284 284 return mode
285 285
286 286 _data = ('data 00manifest.d 00manifest.i 00changelog.d 00changelog.i'
287 287 ' phaseroots obsstore')
288 288
289 289 class basicstore(object):
290 290 '''base class for local repository stores'''
291 291 def __init__(self, path, vfstype):
292 292 self.path = path
293 293 self.createmode = _calcmode(path)
294 op = vfstype(self.path)
295 op.createmode = self.createmode
296 self.opener = scmutil.filteropener(op, encodedir)
294 vfs = vfstype(self.path)
295 vfs.createmode = self.createmode
296 self.opener = scmutil.filteropener(vfs, encodedir)
297 297
298 298 def join(self, f):
299 299 return self.path + '/' + encodedir(f)
300 300
301 301 def _walk(self, relpath, recurse):
302 302 '''yields (unencoded, encoded, size)'''
303 303 path = self.path
304 304 if relpath:
305 305 path += '/' + relpath
306 306 striplen = len(self.path) + 1
307 307 l = []
308 308 if os.path.isdir(path):
309 309 visit = [path]
310 310 while visit:
311 311 p = visit.pop()
312 312 for f, kind, st in osutil.listdir(p, stat=True):
313 313 fp = p + '/' + f
314 314 if kind == stat.S_IFREG and f[-2:] in ('.d', '.i'):
315 315 n = util.pconvert(fp[striplen:])
316 316 l.append((decodedir(n), n, st.st_size))
317 317 elif kind == stat.S_IFDIR and recurse:
318 318 visit.append(fp)
319 319 l.sort()
320 320 return l
321 321
322 322 def datafiles(self):
323 323 return self._walk('data', True)
324 324
325 325 def walk(self):
326 326 '''yields (unencoded, encoded, size)'''
327 327 # yield data files first
328 328 for x in self.datafiles():
329 329 yield x
330 330 # yield manifest before changelog
331 331 for x in reversed(self._walk('', False)):
332 332 yield x
333 333
334 334 def copylist(self):
335 335 return ['requires'] + _data.split()
336 336
337 337 def write(self):
338 338 pass
339 339
340 340 class encodedstore(basicstore):
341 341 def __init__(self, path, vfstype):
342 342 self.path = path + '/store'
343 343 self.createmode = _calcmode(self.path)
344 op = vfstype(self.path)
345 op.createmode = self.createmode
346 self.opener = scmutil.filteropener(op, encodefilename)
344 vfs = vfstype(self.path)
345 vfs.createmode = self.createmode
346 self.opener = scmutil.filteropener(vfs, encodefilename)
347 347
348 348 def datafiles(self):
349 349 for a, b, size in self._walk('data', True):
350 350 try:
351 351 a = decodefilename(a)
352 352 except KeyError:
353 353 a = None
354 354 yield a, b, size
355 355
356 356 def join(self, f):
357 357 return self.path + '/' + encodefilename(f)
358 358
359 359 def copylist(self):
360 360 return (['requires', '00changelog.i'] +
361 361 ['store/' + f for f in _data.split()])
362 362
363 363 class fncache(object):
364 364 # the filename used to be partially encoded
365 365 # hence the encodedir/decodedir dance
366 366 def __init__(self, opener):
367 367 self.opener = opener
368 368 self.entries = None
369 369 self._dirty = False
370 370
371 371 def _load(self):
372 372 '''fill the entries from the fncache file'''
373 373 self._dirty = False
374 374 try:
375 375 fp = self.opener('fncache', mode='rb')
376 376 except IOError:
377 377 # skip nonexistent file
378 378 self.entries = set()
379 379 return
380 380 self.entries = set(decodedir(fp.read()).splitlines())
381 381 if '' in self.entries:
382 382 fp.seek(0)
383 383 for n, line in enumerate(fp):
384 384 if not line.rstrip('\n'):
385 385 t = _('invalid entry in fncache, line %s') % (n + 1)
386 386 raise util.Abort(t)
387 387 fp.close()
388 388
389 389 def _write(self, files, atomictemp):
390 390 fp = self.opener('fncache', mode='wb', atomictemp=atomictemp)
391 391 if files:
392 392 fp.write(encodedir('\n'.join(files) + '\n'))
393 393 fp.close()
394 394 self._dirty = False
395 395
396 396 def rewrite(self, files):
397 397 self._write(files, False)
398 398 self.entries = set(files)
399 399
400 400 def write(self):
401 401 if self._dirty:
402 402 self._write(self.entries, True)
403 403
404 404 def add(self, fn):
405 405 if self.entries is None:
406 406 self._load()
407 407 if fn not in self.entries:
408 408 self._dirty = True
409 409 self.entries.add(fn)
410 410
411 411 def __contains__(self, fn):
412 412 if self.entries is None:
413 413 self._load()
414 414 return fn in self.entries
415 415
416 416 def __iter__(self):
417 417 if self.entries is None:
418 418 self._load()
419 419 return iter(self.entries)
420 420
421 421 class _fncachevfs(scmutil.abstractvfs):
422 422 def __init__(self, op, fnc, encode):
423 423 self.opener = op
424 424 self.fncache = fnc
425 425 self.encode = encode
426 426
427 427 def _getmustaudit(self):
428 428 return self.opener.mustaudit
429 429
430 430 def _setmustaudit(self, onoff):
431 431 self.opener.mustaudit = onoff
432 432
433 433 mustaudit = property(_getmustaudit, _setmustaudit)
434 434
435 435 def __call__(self, path, mode='r', *args, **kw):
436 436 if mode not in ('r', 'rb') and path.startswith('data/'):
437 437 self.fncache.add(path)
438 438 return self.opener(self.encode(path), mode, *args, **kw)
439 439
440 440 class fncachestore(basicstore):
441 441 def __init__(self, path, vfstype, dotencode):
442 442 if dotencode:
443 443 encode = _dothybridencode
444 444 else:
445 445 encode = _plainhybridencode
446 446 self.encode = encode
447 447 self.path = path + '/store'
448 448 self.pathsep = self.path + '/'
449 449 self.createmode = _calcmode(self.path)
450 op = vfstype(self.path)
451 op.createmode = self.createmode
452 fnc = fncache(op)
450 vfs = vfstype(self.path)
451 vfs.createmode = self.createmode
452 fnc = fncache(vfs)
453 453 self.fncache = fnc
454 self.opener = _fncachevfs(op, fnc, encode)
454 self.opener = _fncachevfs(vfs, fnc, encode)
455 455
456 456 def join(self, f):
457 457 return self.pathsep + self.encode(f)
458 458
459 459 def getsize(self, path):
460 460 return os.stat(self.pathsep + path).st_size
461 461
462 462 def datafiles(self):
463 463 rewrite = False
464 464 existing = []
465 465 for f in sorted(self.fncache):
466 466 ef = self.encode(f)
467 467 try:
468 468 yield f, ef, self.getsize(ef)
469 469 existing.append(f)
470 470 except OSError, err:
471 471 if err.errno != errno.ENOENT:
472 472 raise
473 473 # nonexistent entry
474 474 rewrite = True
475 475 if rewrite:
476 476 # rewrite fncache to remove nonexistent entries
477 477 # (may be caused by rollback / strip)
478 478 self.fncache.rewrite(existing)
479 479
480 480 def copylist(self):
481 481 d = ('data dh fncache phaseroots obsstore'
482 482 ' 00manifest.d 00manifest.i 00changelog.d 00changelog.i')
483 483 return (['requires', '00changelog.i'] +
484 484 ['store/' + f for f in d.split()])
485 485
486 486 def write(self):
487 487 self.fncache.write()
488 488
489 489 def store(requirements, path, vfstype):
490 490 if 'store' in requirements:
491 491 if 'fncache' in requirements:
492 492 return fncachestore(path, vfstype, 'dotencode' in requirements)
493 493 return encodedstore(path, vfstype)
494 494 return basicstore(path, vfstype)
General Comments 0
You need to be logged in to leave comments. Login now