Show More
@@ -1,1560 +1,1565 b'' | |||
|
1 | 1 | # manifest.py - manifest revision class for mercurial |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2005-2007 Matt Mackall <mpm@selenic.com> |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import array |
|
11 | 11 | import heapq |
|
12 | 12 | import os |
|
13 | 13 | import struct |
|
14 | 14 | |
|
15 | 15 | from .i18n import _ |
|
16 | 16 | from . import ( |
|
17 | 17 | error, |
|
18 | 18 | mdiff, |
|
19 | 19 | parsers, |
|
20 | 20 | revlog, |
|
21 | 21 | util, |
|
22 | 22 | ) |
|
23 | 23 | |
|
24 | 24 | propertycache = util.propertycache |
|
25 | 25 | |
|
26 | 26 | def _parsev1(data): |
|
27 | 27 | # This method does a little bit of excessive-looking |
|
28 | 28 | # precondition checking. This is so that the behavior of this |
|
29 | 29 | # class exactly matches its C counterpart to try and help |
|
30 | 30 | # prevent surprise breakage for anyone that develops against |
|
31 | 31 | # the pure version. |
|
32 | 32 | if data and data[-1] != '\n': |
|
33 | 33 | raise ValueError('Manifest did not end in a newline.') |
|
34 | 34 | prev = None |
|
35 | 35 | for l in data.splitlines(): |
|
36 | 36 | if prev is not None and prev > l: |
|
37 | 37 | raise ValueError('Manifest lines not in sorted order.') |
|
38 | 38 | prev = l |
|
39 | 39 | f, n = l.split('\0') |
|
40 | 40 | if len(n) > 40: |
|
41 | 41 | yield f, revlog.bin(n[:40]), n[40:] |
|
42 | 42 | else: |
|
43 | 43 | yield f, revlog.bin(n), '' |
|
44 | 44 | |
|
45 | 45 | def _parsev2(data): |
|
46 | 46 | metadataend = data.find('\n') |
|
47 | 47 | # Just ignore metadata for now |
|
48 | 48 | pos = metadataend + 1 |
|
49 | 49 | prevf = '' |
|
50 | 50 | while pos < len(data): |
|
51 | 51 | end = data.find('\n', pos + 1) # +1 to skip stem length byte |
|
52 | 52 | if end == -1: |
|
53 | 53 | raise ValueError('Manifest ended with incomplete file entry.') |
|
54 | 54 | stemlen = ord(data[pos]) |
|
55 | 55 | items = data[pos + 1:end].split('\0') |
|
56 | 56 | f = prevf[:stemlen] + items[0] |
|
57 | 57 | if prevf > f: |
|
58 | 58 | raise ValueError('Manifest entries not in sorted order.') |
|
59 | 59 | fl = items[1] |
|
60 | 60 | # Just ignore metadata (items[2:] for now) |
|
61 | 61 | n = data[end + 1:end + 21] |
|
62 | 62 | yield f, n, fl |
|
63 | 63 | pos = end + 22 |
|
64 | 64 | prevf = f |
|
65 | 65 | |
|
66 | 66 | def _parse(data): |
|
67 | 67 | """Generates (path, node, flags) tuples from a manifest text""" |
|
68 | 68 | if data.startswith('\0'): |
|
69 | 69 | return iter(_parsev2(data)) |
|
70 | 70 | else: |
|
71 | 71 | return iter(_parsev1(data)) |
|
72 | 72 | |
|
73 | 73 | def _text(it, usemanifestv2): |
|
74 | 74 | """Given an iterator over (path, node, flags) tuples, returns a manifest |
|
75 | 75 | text""" |
|
76 | 76 | if usemanifestv2: |
|
77 | 77 | return _textv2(it) |
|
78 | 78 | else: |
|
79 | 79 | return _textv1(it) |
|
80 | 80 | |
|
81 | 81 | def _textv1(it): |
|
82 | 82 | files = [] |
|
83 | 83 | lines = [] |
|
84 | 84 | _hex = revlog.hex |
|
85 | 85 | for f, n, fl in it: |
|
86 | 86 | files.append(f) |
|
87 | 87 | # if this is changed to support newlines in filenames, |
|
88 | 88 | # be sure to check the templates/ dir again (especially *-raw.tmpl) |
|
89 | 89 | lines.append("%s\0%s%s\n" % (f, _hex(n), fl)) |
|
90 | 90 | |
|
91 | 91 | _checkforbidden(files) |
|
92 | 92 | return ''.join(lines) |
|
93 | 93 | |
|
94 | 94 | def _textv2(it): |
|
95 | 95 | files = [] |
|
96 | 96 | lines = ['\0\n'] |
|
97 | 97 | prevf = '' |
|
98 | 98 | for f, n, fl in it: |
|
99 | 99 | files.append(f) |
|
100 | 100 | stem = os.path.commonprefix([prevf, f]) |
|
101 | 101 | stemlen = min(len(stem), 255) |
|
102 | 102 | lines.append("%c%s\0%s\n%s\n" % (stemlen, f[stemlen:], fl, n)) |
|
103 | 103 | prevf = f |
|
104 | 104 | _checkforbidden(files) |
|
105 | 105 | return ''.join(lines) |
|
106 | 106 | |
|
107 | 107 | class lazymanifestiter(object): |
|
108 | 108 | def __init__(self, lm): |
|
109 | 109 | self.pos = 0 |
|
110 | 110 | self.lm = lm |
|
111 | 111 | |
|
112 | 112 | def __iter__(self): |
|
113 | 113 | return self |
|
114 | 114 | |
|
115 | 115 | def next(self): |
|
116 | 116 | try: |
|
117 | 117 | data, pos = self.lm._get(self.pos) |
|
118 | 118 | except IndexError: |
|
119 | 119 | raise StopIteration |
|
120 | 120 | if pos == -1: |
|
121 | 121 | self.pos += 1 |
|
122 | 122 | return data[0] |
|
123 | 123 | self.pos += 1 |
|
124 | 124 | zeropos = data.find('\x00', pos) |
|
125 | 125 | return data[pos:zeropos] |
|
126 | 126 | |
|
127 | 127 | class lazymanifestiterentries(object): |
|
128 | 128 | def __init__(self, lm): |
|
129 | 129 | self.lm = lm |
|
130 | 130 | self.pos = 0 |
|
131 | 131 | |
|
132 | 132 | def __iter__(self): |
|
133 | 133 | return self |
|
134 | 134 | |
|
135 | 135 | def next(self): |
|
136 | 136 | try: |
|
137 | 137 | data, pos = self.lm._get(self.pos) |
|
138 | 138 | except IndexError: |
|
139 | 139 | raise StopIteration |
|
140 | 140 | if pos == -1: |
|
141 | 141 | self.pos += 1 |
|
142 | 142 | return data |
|
143 | 143 | zeropos = data.find('\x00', pos) |
|
144 | 144 | hashval = unhexlify(data, self.lm.extrainfo[self.pos], |
|
145 | 145 | zeropos + 1, 40) |
|
146 | 146 | flags = self.lm._getflags(data, self.pos, zeropos) |
|
147 | 147 | self.pos += 1 |
|
148 | 148 | return (data[pos:zeropos], hashval, flags) |
|
149 | 149 | |
|
150 | 150 | def unhexlify(data, extra, pos, length): |
|
151 | 151 | s = data[pos:pos + length].decode('hex') |
|
152 | 152 | if extra: |
|
153 | 153 | s += chr(extra & 0xff) |
|
154 | 154 | return s |
|
155 | 155 | |
|
156 | 156 | def _cmp(a, b): |
|
157 | 157 | return (a > b) - (a < b) |
|
158 | 158 | |
|
159 | 159 | class _lazymanifest(object): |
|
160 | 160 | def __init__(self, data, positions=None, extrainfo=None, extradata=None): |
|
161 | 161 | if positions is None: |
|
162 | 162 | self.positions = self.findlines(data) |
|
163 | 163 | self.extrainfo = [0] * len(self.positions) |
|
164 | 164 | self.data = data |
|
165 | 165 | self.extradata = [] |
|
166 | 166 | else: |
|
167 | 167 | self.positions = positions[:] |
|
168 | 168 | self.extrainfo = extrainfo[:] |
|
169 | 169 | self.extradata = extradata[:] |
|
170 | 170 | self.data = data |
|
171 | 171 | |
|
172 | 172 | def findlines(self, data): |
|
173 | 173 | if not data: |
|
174 | 174 | return [] |
|
175 | 175 | pos = data.find("\n") |
|
176 | 176 | if pos == -1 or data[-1] != '\n': |
|
177 | 177 | raise ValueError("Manifest did not end in a newline.") |
|
178 | 178 | positions = [0] |
|
179 | 179 | prev = data[:data.find('\x00')] |
|
180 | 180 | while pos < len(data) - 1 and pos != -1: |
|
181 | 181 | positions.append(pos + 1) |
|
182 | 182 | nexts = data[pos + 1:data.find('\x00', pos + 1)] |
|
183 | 183 | if nexts < prev: |
|
184 | 184 | raise ValueError("Manifest lines not in sorted order.") |
|
185 | 185 | prev = nexts |
|
186 | 186 | pos = data.find("\n", pos + 1) |
|
187 | 187 | return positions |
|
188 | 188 | |
|
189 | 189 | def _get(self, index): |
|
190 | 190 | # get the position encoded in pos: |
|
191 | 191 | # positive number is an index in 'data' |
|
192 | 192 | # negative number is in extrapieces |
|
193 | 193 | pos = self.positions[index] |
|
194 | 194 | if pos >= 0: |
|
195 | 195 | return self.data, pos |
|
196 | 196 | return self.extradata[-pos - 1], -1 |
|
197 | 197 | |
|
198 | 198 | def _getkey(self, pos): |
|
199 | 199 | if pos >= 0: |
|
200 | 200 | return self.data[pos:self.data.find('\x00', pos + 1)] |
|
201 | 201 | return self.extradata[-pos - 1][0] |
|
202 | 202 | |
|
203 | 203 | def bsearch(self, key): |
|
204 | 204 | first = 0 |
|
205 | 205 | last = len(self.positions) - 1 |
|
206 | 206 | |
|
207 | 207 | while first <= last: |
|
208 | 208 | midpoint = (first + last)//2 |
|
209 | 209 | nextpos = self.positions[midpoint] |
|
210 | 210 | candidate = self._getkey(nextpos) |
|
211 | 211 | r = _cmp(key, candidate) |
|
212 | 212 | if r == 0: |
|
213 | 213 | return midpoint |
|
214 | 214 | else: |
|
215 | 215 | if r < 0: |
|
216 | 216 | last = midpoint - 1 |
|
217 | 217 | else: |
|
218 | 218 | first = midpoint + 1 |
|
219 | 219 | return -1 |
|
220 | 220 | |
|
221 | 221 | def bsearch2(self, key): |
|
222 | 222 | # same as the above, but will always return the position |
|
223 | 223 | # done for performance reasons |
|
224 | 224 | first = 0 |
|
225 | 225 | last = len(self.positions) - 1 |
|
226 | 226 | |
|
227 | 227 | while first <= last: |
|
228 | 228 | midpoint = (first + last)//2 |
|
229 | 229 | nextpos = self.positions[midpoint] |
|
230 | 230 | candidate = self._getkey(nextpos) |
|
231 | 231 | r = _cmp(key, candidate) |
|
232 | 232 | if r == 0: |
|
233 | 233 | return (midpoint, True) |
|
234 | 234 | else: |
|
235 | 235 | if r < 0: |
|
236 | 236 | last = midpoint - 1 |
|
237 | 237 | else: |
|
238 | 238 | first = midpoint + 1 |
|
239 | 239 | return (first, False) |
|
240 | 240 | |
|
241 | 241 | def __contains__(self, key): |
|
242 | 242 | return self.bsearch(key) != -1 |
|
243 | 243 | |
|
244 | 244 | def _getflags(self, data, needle, pos): |
|
245 | 245 | start = pos + 41 |
|
246 | 246 | end = data.find("\n", start) |
|
247 | 247 | if end == -1: |
|
248 | 248 | end = len(data) - 1 |
|
249 | 249 | if start == end: |
|
250 | 250 | return '' |
|
251 | 251 | return self.data[start:end] |
|
252 | 252 | |
|
253 | 253 | def __getitem__(self, key): |
|
254 | 254 | if not isinstance(key, str): |
|
255 | 255 | raise TypeError("getitem: manifest keys must be a string.") |
|
256 | 256 | needle = self.bsearch(key) |
|
257 | 257 | if needle == -1: |
|
258 | 258 | raise KeyError |
|
259 | 259 | data, pos = self._get(needle) |
|
260 | 260 | if pos == -1: |
|
261 | 261 | return (data[1], data[2]) |
|
262 | 262 | zeropos = data.find('\x00', pos) |
|
263 | 263 | assert 0 <= needle <= len(self.positions) |
|
264 | 264 | assert len(self.extrainfo) == len(self.positions) |
|
265 | 265 | hashval = unhexlify(data, self.extrainfo[needle], zeropos + 1, 40) |
|
266 | 266 | flags = self._getflags(data, needle, zeropos) |
|
267 | 267 | return (hashval, flags) |
|
268 | 268 | |
|
269 | 269 | def __delitem__(self, key): |
|
270 | 270 | needle, found = self.bsearch2(key) |
|
271 | 271 | if not found: |
|
272 | 272 | raise KeyError |
|
273 | 273 | cur = self.positions[needle] |
|
274 | 274 | self.positions = self.positions[:needle] + self.positions[needle + 1:] |
|
275 | 275 | self.extrainfo = self.extrainfo[:needle] + self.extrainfo[needle + 1:] |
|
276 | 276 | if cur >= 0: |
|
277 | 277 | self.data = self.data[:cur] + '\x00' + self.data[cur + 1:] |
|
278 | 278 | |
|
279 | 279 | def __setitem__(self, key, value): |
|
280 | 280 | if not isinstance(key, str): |
|
281 | 281 | raise TypeError("setitem: manifest keys must be a string.") |
|
282 | 282 | if not isinstance(value, tuple) or len(value) != 2: |
|
283 | 283 | raise TypeError("Manifest values must be a tuple of (node, flags).") |
|
284 | 284 | hashval = value[0] |
|
285 | 285 | if not isinstance(hashval, str) or not 20 <= len(hashval) <= 22: |
|
286 | 286 | raise TypeError("node must be a 20-byte string") |
|
287 | 287 | flags = value[1] |
|
288 | 288 | if len(hashval) == 22: |
|
289 | 289 | hashval = hashval[:-1] |
|
290 | 290 | if not isinstance(flags, str) or len(flags) > 1: |
|
291 | 291 | raise TypeError("flags must a 0 or 1 byte string, got %r", flags) |
|
292 | 292 | needle, found = self.bsearch2(key) |
|
293 | 293 | if found: |
|
294 | 294 | # put the item |
|
295 | 295 | pos = self.positions[needle] |
|
296 | 296 | if pos < 0: |
|
297 | 297 | self.extradata[-pos - 1] = (key, hashval, value[1]) |
|
298 | 298 | else: |
|
299 | 299 | # just don't bother |
|
300 | 300 | self.extradata.append((key, hashval, value[1])) |
|
301 | 301 | self.positions[needle] = -len(self.extradata) |
|
302 | 302 | else: |
|
303 | 303 | # not found, put it in with extra positions |
|
304 | 304 | self.extradata.append((key, hashval, value[1])) |
|
305 | 305 | self.positions = (self.positions[:needle] + [-len(self.extradata)] |
|
306 | 306 | + self.positions[needle:]) |
|
307 | 307 | self.extrainfo = (self.extrainfo[:needle] + [0] + |
|
308 | 308 | self.extrainfo[needle:]) |
|
309 | 309 | |
|
310 | 310 | def copy(self): |
|
311 | 311 | # XXX call _compact like in C? |
|
312 | 312 | return _lazymanifest(self.data, self.positions, self.extrainfo, |
|
313 | 313 | self.extradata) |
|
314 | 314 | |
|
315 | 315 | def _compact(self): |
|
316 | 316 | # hopefully not called TOO often |
|
317 | 317 | if len(self.extradata) == 0: |
|
318 | 318 | return |
|
319 | 319 | l = [] |
|
320 | 320 | last_cut = 0 |
|
321 | 321 | i = 0 |
|
322 | 322 | offset = 0 |
|
323 | 323 | self.extrainfo = [0] * len(self.positions) |
|
324 | 324 | while i < len(self.positions): |
|
325 | 325 | if self.positions[i] >= 0: |
|
326 | 326 | cur = self.positions[i] |
|
327 | 327 | last_cut = cur |
|
328 | 328 | while True: |
|
329 | 329 | self.positions[i] = offset |
|
330 | 330 | i += 1 |
|
331 | 331 | if i == len(self.positions) or self.positions[i] < 0: |
|
332 | 332 | break |
|
333 | 333 | offset += self.positions[i] - cur |
|
334 | 334 | cur = self.positions[i] |
|
335 | 335 | end_cut = self.data.find('\n', cur) |
|
336 | 336 | if end_cut != -1: |
|
337 | 337 | end_cut += 1 |
|
338 | 338 | offset += end_cut - cur |
|
339 | 339 | l.append(self.data[last_cut:end_cut]) |
|
340 | 340 | else: |
|
341 | 341 | while i < len(self.positions) and self.positions[i] < 0: |
|
342 | 342 | cur = self.positions[i] |
|
343 | 343 | t = self.extradata[-cur - 1] |
|
344 | 344 | l.append(self._pack(t)) |
|
345 | 345 | self.positions[i] = offset |
|
346 | 346 | if len(t[1]) > 20: |
|
347 | 347 | self.extrainfo[i] = ord(t[1][21]) |
|
348 | 348 | offset += len(l[-1]) |
|
349 | 349 | i += 1 |
|
350 | 350 | self.data = ''.join(l) |
|
351 | 351 | self.extradata = [] |
|
352 | 352 | |
|
353 | 353 | def _pack(self, d): |
|
354 | 354 | return d[0] + '\x00' + d[1][:20].encode('hex') + d[2] + '\n' |
|
355 | 355 | |
|
356 | 356 | def text(self): |
|
357 | 357 | self._compact() |
|
358 | 358 | return self.data |
|
359 | 359 | |
|
360 | 360 | def diff(self, m2, clean=False): |
|
361 | 361 | '''Finds changes between the current manifest and m2.''' |
|
362 | 362 | # XXX think whether efficiency matters here |
|
363 | 363 | diff = {} |
|
364 | 364 | |
|
365 | 365 | for fn, e1, flags in self.iterentries(): |
|
366 | 366 | if fn not in m2: |
|
367 | 367 | diff[fn] = (e1, flags), (None, '') |
|
368 | 368 | else: |
|
369 | 369 | e2 = m2[fn] |
|
370 | 370 | if (e1, flags) != e2: |
|
371 | 371 | diff[fn] = (e1, flags), e2 |
|
372 | 372 | elif clean: |
|
373 | 373 | diff[fn] = None |
|
374 | 374 | |
|
375 | 375 | for fn, e2, flags in m2.iterentries(): |
|
376 | 376 | if fn not in self: |
|
377 | 377 | diff[fn] = (None, ''), (e2, flags) |
|
378 | 378 | |
|
379 | 379 | return diff |
|
380 | 380 | |
|
381 | 381 | def iterentries(self): |
|
382 | 382 | return lazymanifestiterentries(self) |
|
383 | 383 | |
|
384 | 384 | def iterkeys(self): |
|
385 | 385 | return lazymanifestiter(self) |
|
386 | 386 | |
|
387 | 387 | def __iter__(self): |
|
388 | 388 | return lazymanifestiter(self) |
|
389 | 389 | |
|
390 | 390 | def __len__(self): |
|
391 | 391 | return len(self.positions) |
|
392 | 392 | |
|
393 | 393 | def filtercopy(self, filterfn): |
|
394 | 394 | # XXX should be optimized |
|
395 | 395 | c = _lazymanifest('') |
|
396 | 396 | for f, n, fl in self.iterentries(): |
|
397 | 397 | if filterfn(f): |
|
398 | 398 | c[f] = n, fl |
|
399 | 399 | return c |
|
400 | 400 | |
|
401 | 401 | try: |
|
402 | 402 | _lazymanifest = parsers.lazymanifest |
|
403 | 403 | except AttributeError: |
|
404 | 404 | pass |
|
405 | 405 | |
|
406 | 406 | class manifestdict(object): |
|
407 | 407 | def __init__(self, data=''): |
|
408 | 408 | if data.startswith('\0'): |
|
409 | 409 | #_lazymanifest can not parse v2 |
|
410 | 410 | self._lm = _lazymanifest('') |
|
411 | 411 | for f, n, fl in _parsev2(data): |
|
412 | 412 | self._lm[f] = n, fl |
|
413 | 413 | else: |
|
414 | 414 | self._lm = _lazymanifest(data) |
|
415 | 415 | |
|
416 | 416 | def __getitem__(self, key): |
|
417 | 417 | return self._lm[key][0] |
|
418 | 418 | |
|
419 | 419 | def find(self, key): |
|
420 | 420 | return self._lm[key] |
|
421 | 421 | |
|
422 | 422 | def __len__(self): |
|
423 | 423 | return len(self._lm) |
|
424 | 424 | |
|
425 | 425 | def __nonzero__(self): |
|
426 | 426 | # nonzero is covered by the __len__ function, but implementing it here |
|
427 | 427 | # makes it easier for extensions to override. |
|
428 | 428 | return len(self._lm) != 0 |
|
429 | 429 | |
|
430 | 430 | def __setitem__(self, key, node): |
|
431 | 431 | self._lm[key] = node, self.flags(key, '') |
|
432 | 432 | |
|
433 | 433 | def __contains__(self, key): |
|
434 | 434 | return key in self._lm |
|
435 | 435 | |
|
436 | 436 | def __delitem__(self, key): |
|
437 | 437 | del self._lm[key] |
|
438 | 438 | |
|
439 | 439 | def __iter__(self): |
|
440 | 440 | return self._lm.__iter__() |
|
441 | 441 | |
|
442 | 442 | def iterkeys(self): |
|
443 | 443 | return self._lm.iterkeys() |
|
444 | 444 | |
|
445 | 445 | def keys(self): |
|
446 | 446 | return list(self.iterkeys()) |
|
447 | 447 | |
|
448 | 448 | def filesnotin(self, m2): |
|
449 | 449 | '''Set of files in this manifest that are not in the other''' |
|
450 | 450 | diff = self.diff(m2) |
|
451 | 451 | files = set(filepath |
|
452 | 452 | for filepath, hashflags in diff.iteritems() |
|
453 | 453 | if hashflags[1][0] is None) |
|
454 | 454 | return files |
|
455 | 455 | |
|
456 | 456 | @propertycache |
|
457 | 457 | def _dirs(self): |
|
458 | 458 | return util.dirs(self) |
|
459 | 459 | |
|
460 | 460 | def dirs(self): |
|
461 | 461 | return self._dirs |
|
462 | 462 | |
|
463 | 463 | def hasdir(self, dir): |
|
464 | 464 | return dir in self._dirs |
|
465 | 465 | |
|
466 | 466 | def _filesfastpath(self, match): |
|
467 | 467 | '''Checks whether we can correctly and quickly iterate over matcher |
|
468 | 468 | files instead of over manifest files.''' |
|
469 | 469 | files = match.files() |
|
470 | 470 | return (len(files) < 100 and (match.isexact() or |
|
471 | 471 | (match.prefix() and all(fn in self for fn in files)))) |
|
472 | 472 | |
|
473 | 473 | def walk(self, match): |
|
474 | 474 | '''Generates matching file names. |
|
475 | 475 | |
|
476 | 476 | Equivalent to manifest.matches(match).iterkeys(), but without creating |
|
477 | 477 | an entirely new manifest. |
|
478 | 478 | |
|
479 | 479 | It also reports nonexistent files by marking them bad with match.bad(). |
|
480 | 480 | ''' |
|
481 | 481 | if match.always(): |
|
482 | 482 | for f in iter(self): |
|
483 | 483 | yield f |
|
484 | 484 | return |
|
485 | 485 | |
|
486 | 486 | fset = set(match.files()) |
|
487 | 487 | |
|
488 | 488 | # avoid the entire walk if we're only looking for specific files |
|
489 | 489 | if self._filesfastpath(match): |
|
490 | 490 | for fn in sorted(fset): |
|
491 | 491 | yield fn |
|
492 | 492 | return |
|
493 | 493 | |
|
494 | 494 | for fn in self: |
|
495 | 495 | if fn in fset: |
|
496 | 496 | # specified pattern is the exact name |
|
497 | 497 | fset.remove(fn) |
|
498 | 498 | if match(fn): |
|
499 | 499 | yield fn |
|
500 | 500 | |
|
501 | 501 | # for dirstate.walk, files=['.'] means "walk the whole tree". |
|
502 | 502 | # follow that here, too |
|
503 | 503 | fset.discard('.') |
|
504 | 504 | |
|
505 | 505 | for fn in sorted(fset): |
|
506 | 506 | if not self.hasdir(fn): |
|
507 | 507 | match.bad(fn, None) |
|
508 | 508 | |
|
509 | 509 | def matches(self, match): |
|
510 | 510 | '''generate a new manifest filtered by the match argument''' |
|
511 | 511 | if match.always(): |
|
512 | 512 | return self.copy() |
|
513 | 513 | |
|
514 | 514 | if self._filesfastpath(match): |
|
515 | 515 | m = manifestdict() |
|
516 | 516 | lm = self._lm |
|
517 | 517 | for fn in match.files(): |
|
518 | 518 | if fn in lm: |
|
519 | 519 | m._lm[fn] = lm[fn] |
|
520 | 520 | return m |
|
521 | 521 | |
|
522 | 522 | m = manifestdict() |
|
523 | 523 | m._lm = self._lm.filtercopy(match) |
|
524 | 524 | return m |
|
525 | 525 | |
|
526 | 526 | def diff(self, m2, clean=False): |
|
527 | 527 | '''Finds changes between the current manifest and m2. |
|
528 | 528 | |
|
529 | 529 | Args: |
|
530 | 530 | m2: the manifest to which this manifest should be compared. |
|
531 | 531 | clean: if true, include files unchanged between these manifests |
|
532 | 532 | with a None value in the returned dictionary. |
|
533 | 533 | |
|
534 | 534 | The result is returned as a dict with filename as key and |
|
535 | 535 | values of the form ((n1,fl1),(n2,fl2)), where n1/n2 is the |
|
536 | 536 | nodeid in the current/other manifest and fl1/fl2 is the flag |
|
537 | 537 | in the current/other manifest. Where the file does not exist, |
|
538 | 538 | the nodeid will be None and the flags will be the empty |
|
539 | 539 | string. |
|
540 | 540 | ''' |
|
541 | 541 | return self._lm.diff(m2._lm, clean) |
|
542 | 542 | |
|
543 | 543 | def setflag(self, key, flag): |
|
544 | 544 | self._lm[key] = self[key], flag |
|
545 | 545 | |
|
546 | 546 | def get(self, key, default=None): |
|
547 | 547 | try: |
|
548 | 548 | return self._lm[key][0] |
|
549 | 549 | except KeyError: |
|
550 | 550 | return default |
|
551 | 551 | |
|
552 | 552 | def flags(self, key, default=''): |
|
553 | 553 | try: |
|
554 | 554 | return self._lm[key][1] |
|
555 | 555 | except KeyError: |
|
556 | 556 | return default |
|
557 | 557 | |
|
558 | 558 | def copy(self): |
|
559 | 559 | c = manifestdict() |
|
560 | 560 | c._lm = self._lm.copy() |
|
561 | 561 | return c |
|
562 | 562 | |
|
563 | 563 | def iteritems(self): |
|
564 | 564 | return (x[:2] for x in self._lm.iterentries()) |
|
565 | 565 | |
|
566 | 566 | def iterentries(self): |
|
567 | 567 | return self._lm.iterentries() |
|
568 | 568 | |
|
569 | 569 | def text(self, usemanifestv2=False): |
|
570 | 570 | if usemanifestv2: |
|
571 | 571 | return _textv2(self._lm.iterentries()) |
|
572 | 572 | else: |
|
573 | 573 | # use (probably) native version for v1 |
|
574 | 574 | return self._lm.text() |
|
575 | 575 | |
|
576 | 576 | def fastdelta(self, base, changes): |
|
577 | 577 | """Given a base manifest text as an array.array and a list of changes |
|
578 | 578 | relative to that text, compute a delta that can be used by revlog. |
|
579 | 579 | """ |
|
580 | 580 | delta = [] |
|
581 | 581 | dstart = None |
|
582 | 582 | dend = None |
|
583 | 583 | dline = [""] |
|
584 | 584 | start = 0 |
|
585 | 585 | # zero copy representation of base as a buffer |
|
586 | 586 | addbuf = util.buffer(base) |
|
587 | 587 | |
|
588 | 588 | changes = list(changes) |
|
589 | 589 | if len(changes) < 1000: |
|
590 | 590 | # start with a readonly loop that finds the offset of |
|
591 | 591 | # each line and creates the deltas |
|
592 | 592 | for f, todelete in changes: |
|
593 | 593 | # bs will either be the index of the item or the insert point |
|
594 | 594 | start, end = _msearch(addbuf, f, start) |
|
595 | 595 | if not todelete: |
|
596 | 596 | h, fl = self._lm[f] |
|
597 | 597 | l = "%s\0%s%s\n" % (f, revlog.hex(h), fl) |
|
598 | 598 | else: |
|
599 | 599 | if start == end: |
|
600 | 600 | # item we want to delete was not found, error out |
|
601 | 601 | raise AssertionError( |
|
602 | 602 | _("failed to remove %s from manifest") % f) |
|
603 | 603 | l = "" |
|
604 | 604 | if dstart is not None and dstart <= start and dend >= start: |
|
605 | 605 | if dend < end: |
|
606 | 606 | dend = end |
|
607 | 607 | if l: |
|
608 | 608 | dline.append(l) |
|
609 | 609 | else: |
|
610 | 610 | if dstart is not None: |
|
611 | 611 | delta.append([dstart, dend, "".join(dline)]) |
|
612 | 612 | dstart = start |
|
613 | 613 | dend = end |
|
614 | 614 | dline = [l] |
|
615 | 615 | |
|
616 | 616 | if dstart is not None: |
|
617 | 617 | delta.append([dstart, dend, "".join(dline)]) |
|
618 | 618 | # apply the delta to the base, and get a delta for addrevision |
|
619 | 619 | deltatext, arraytext = _addlistdelta(base, delta) |
|
620 | 620 | else: |
|
621 | 621 | # For large changes, it's much cheaper to just build the text and |
|
622 | 622 | # diff it. |
|
623 | 623 | arraytext = array.array('c', self.text()) |
|
624 | 624 | deltatext = mdiff.textdiff(base, arraytext) |
|
625 | 625 | |
|
626 | 626 | return arraytext, deltatext |
|
627 | 627 | |
|
628 | 628 | def _msearch(m, s, lo=0, hi=None): |
|
629 | 629 | '''return a tuple (start, end) that says where to find s within m. |
|
630 | 630 | |
|
631 | 631 | If the string is found m[start:end] are the line containing |
|
632 | 632 | that string. If start == end the string was not found and |
|
633 | 633 | they indicate the proper sorted insertion point. |
|
634 | 634 | |
|
635 | 635 | m should be a buffer or a string |
|
636 | 636 | s is a string''' |
|
637 | 637 | def advance(i, c): |
|
638 | 638 | while i < lenm and m[i] != c: |
|
639 | 639 | i += 1 |
|
640 | 640 | return i |
|
641 | 641 | if not s: |
|
642 | 642 | return (lo, lo) |
|
643 | 643 | lenm = len(m) |
|
644 | 644 | if not hi: |
|
645 | 645 | hi = lenm |
|
646 | 646 | while lo < hi: |
|
647 | 647 | mid = (lo + hi) // 2 |
|
648 | 648 | start = mid |
|
649 | 649 | while start > 0 and m[start - 1] != '\n': |
|
650 | 650 | start -= 1 |
|
651 | 651 | end = advance(start, '\0') |
|
652 | 652 | if m[start:end] < s: |
|
653 | 653 | # we know that after the null there are 40 bytes of sha1 |
|
654 | 654 | # this translates to the bisect lo = mid + 1 |
|
655 | 655 | lo = advance(end + 40, '\n') + 1 |
|
656 | 656 | else: |
|
657 | 657 | # this translates to the bisect hi = mid |
|
658 | 658 | hi = start |
|
659 | 659 | end = advance(lo, '\0') |
|
660 | 660 | found = m[lo:end] |
|
661 | 661 | if s == found: |
|
662 | 662 | # we know that after the null there are 40 bytes of sha1 |
|
663 | 663 | end = advance(end + 40, '\n') |
|
664 | 664 | return (lo, end + 1) |
|
665 | 665 | else: |
|
666 | 666 | return (lo, lo) |
|
667 | 667 | |
|
668 | 668 | def _checkforbidden(l): |
|
669 | 669 | """Check filenames for illegal characters.""" |
|
670 | 670 | for f in l: |
|
671 | 671 | if '\n' in f or '\r' in f: |
|
672 | 672 | raise error.RevlogError( |
|
673 | 673 | _("'\\n' and '\\r' disallowed in filenames: %r") % f) |
|
674 | 674 | |
|
675 | 675 | |
|
676 | 676 | # apply the changes collected during the bisect loop to our addlist |
|
677 | 677 | # return a delta suitable for addrevision |
|
678 | 678 | def _addlistdelta(addlist, x): |
|
679 | 679 | # for large addlist arrays, building a new array is cheaper |
|
680 | 680 | # than repeatedly modifying the existing one |
|
681 | 681 | currentposition = 0 |
|
682 | 682 | newaddlist = array.array('c') |
|
683 | 683 | |
|
684 | 684 | for start, end, content in x: |
|
685 | 685 | newaddlist += addlist[currentposition:start] |
|
686 | 686 | if content: |
|
687 | 687 | newaddlist += array.array('c', content) |
|
688 | 688 | |
|
689 | 689 | currentposition = end |
|
690 | 690 | |
|
691 | 691 | newaddlist += addlist[currentposition:] |
|
692 | 692 | |
|
693 | 693 | deltatext = "".join(struct.pack(">lll", start, end, len(content)) |
|
694 | 694 | + content for start, end, content in x) |
|
695 | 695 | return deltatext, newaddlist |
|
696 | 696 | |
|
697 | 697 | def _splittopdir(f): |
|
698 | 698 | if '/' in f: |
|
699 | 699 | dir, subpath = f.split('/', 1) |
|
700 | 700 | return dir + '/', subpath |
|
701 | 701 | else: |
|
702 | 702 | return '', f |
|
703 | 703 | |
|
704 | 704 | _noop = lambda s: None |
|
705 | 705 | |
|
706 | 706 | class treemanifest(object): |
|
707 | 707 | def __init__(self, dir='', text=''): |
|
708 | 708 | self._dir = dir |
|
709 | 709 | self._node = revlog.nullid |
|
710 | 710 | self._loadfunc = _noop |
|
711 | 711 | self._copyfunc = _noop |
|
712 | 712 | self._dirty = False |
|
713 | 713 | self._dirs = {} |
|
714 | 714 | # Using _lazymanifest here is a little slower than plain old dicts |
|
715 | 715 | self._files = {} |
|
716 | 716 | self._flags = {} |
|
717 | 717 | if text: |
|
718 | 718 | def readsubtree(subdir, subm): |
|
719 | 719 | raise AssertionError('treemanifest constructor only accepts ' |
|
720 | 720 | 'flat manifests') |
|
721 | 721 | self.parse(text, readsubtree) |
|
722 | 722 | self._dirty = True # Mark flat manifest dirty after parsing |
|
723 | 723 | |
|
724 | 724 | def _subpath(self, path): |
|
725 | 725 | return self._dir + path |
|
726 | 726 | |
|
727 | 727 | def __len__(self): |
|
728 | 728 | self._load() |
|
729 | 729 | size = len(self._files) |
|
730 | 730 | for m in self._dirs.values(): |
|
731 | 731 | size += m.__len__() |
|
732 | 732 | return size |
|
733 | 733 | |
|
734 | 734 | def _isempty(self): |
|
735 | 735 | self._load() # for consistency; already loaded by all callers |
|
736 | 736 | return (not self._files and (not self._dirs or |
|
737 | 737 | all(m._isempty() for m in self._dirs.values()))) |
|
738 | 738 | |
|
739 | 739 | def __repr__(self): |
|
740 | 740 | return ('<treemanifest dir=%s, node=%s, loaded=%s, dirty=%s at 0x%x>' % |
|
741 | 741 | (self._dir, revlog.hex(self._node), |
|
742 | 742 | bool(self._loadfunc is _noop), |
|
743 | 743 | self._dirty, id(self))) |
|
744 | 744 | |
|
745 | 745 | def dir(self): |
|
746 | 746 | '''The directory that this tree manifest represents, including a |
|
747 | 747 | trailing '/'. Empty string for the repo root directory.''' |
|
748 | 748 | return self._dir |
|
749 | 749 | |
|
750 | 750 | def node(self): |
|
751 | 751 | '''This node of this instance. nullid for unsaved instances. Should |
|
752 | 752 | be updated when the instance is read or written from a revlog. |
|
753 | 753 | ''' |
|
754 | 754 | assert not self._dirty |
|
755 | 755 | return self._node |
|
756 | 756 | |
|
757 | 757 | def setnode(self, node): |
|
758 | 758 | self._node = node |
|
759 | 759 | self._dirty = False |
|
760 | 760 | |
|
761 | 761 | def iterentries(self): |
|
762 | 762 | self._load() |
|
763 | 763 | for p, n in sorted(self._dirs.items() + self._files.items()): |
|
764 | 764 | if p in self._files: |
|
765 | 765 | yield self._subpath(p), n, self._flags.get(p, '') |
|
766 | 766 | else: |
|
767 | 767 | for x in n.iterentries(): |
|
768 | 768 | yield x |
|
769 | 769 | |
|
770 | 770 | def iteritems(self): |
|
771 | 771 | self._load() |
|
772 | 772 | for p, n in sorted(self._dirs.items() + self._files.items()): |
|
773 | 773 | if p in self._files: |
|
774 | 774 | yield self._subpath(p), n |
|
775 | 775 | else: |
|
776 | 776 | for f, sn in n.iteritems(): |
|
777 | 777 | yield f, sn |
|
778 | 778 | |
|
779 | 779 | def iterkeys(self): |
|
780 | 780 | self._load() |
|
781 | 781 | for p in sorted(self._dirs.keys() + self._files.keys()): |
|
782 | 782 | if p in self._files: |
|
783 | 783 | yield self._subpath(p) |
|
784 | 784 | else: |
|
785 | 785 | for f in self._dirs[p].iterkeys(): |
|
786 | 786 | yield f |
|
787 | 787 | |
|
788 | 788 | def keys(self): |
|
789 | 789 | return list(self.iterkeys()) |
|
790 | 790 | |
|
791 | 791 | def __iter__(self): |
|
792 | 792 | return self.iterkeys() |
|
793 | 793 | |
|
794 | 794 | def __contains__(self, f): |
|
795 | 795 | if f is None: |
|
796 | 796 | return False |
|
797 | 797 | self._load() |
|
798 | 798 | dir, subpath = _splittopdir(f) |
|
799 | 799 | if dir: |
|
800 | 800 | if dir not in self._dirs: |
|
801 | 801 | return False |
|
802 | 802 | return self._dirs[dir].__contains__(subpath) |
|
803 | 803 | else: |
|
804 | 804 | return f in self._files |
|
805 | 805 | |
|
806 | 806 | def get(self, f, default=None): |
|
807 | 807 | self._load() |
|
808 | 808 | dir, subpath = _splittopdir(f) |
|
809 | 809 | if dir: |
|
810 | 810 | if dir not in self._dirs: |
|
811 | 811 | return default |
|
812 | 812 | return self._dirs[dir].get(subpath, default) |
|
813 | 813 | else: |
|
814 | 814 | return self._files.get(f, default) |
|
815 | 815 | |
|
816 | 816 | def __getitem__(self, f): |
|
817 | 817 | self._load() |
|
818 | 818 | dir, subpath = _splittopdir(f) |
|
819 | 819 | if dir: |
|
820 | 820 | return self._dirs[dir].__getitem__(subpath) |
|
821 | 821 | else: |
|
822 | 822 | return self._files[f] |
|
823 | 823 | |
|
824 | 824 | def flags(self, f): |
|
825 | 825 | self._load() |
|
826 | 826 | dir, subpath = _splittopdir(f) |
|
827 | 827 | if dir: |
|
828 | 828 | if dir not in self._dirs: |
|
829 | 829 | return '' |
|
830 | 830 | return self._dirs[dir].flags(subpath) |
|
831 | 831 | else: |
|
832 | 832 | if f in self._dirs: |
|
833 | 833 | return '' |
|
834 | 834 | return self._flags.get(f, '') |
|
835 | 835 | |
|
836 | 836 | def find(self, f): |
|
837 | 837 | self._load() |
|
838 | 838 | dir, subpath = _splittopdir(f) |
|
839 | 839 | if dir: |
|
840 | 840 | return self._dirs[dir].find(subpath) |
|
841 | 841 | else: |
|
842 | 842 | return self._files[f], self._flags.get(f, '') |
|
843 | 843 | |
|
844 | 844 | def __delitem__(self, f): |
|
845 | 845 | self._load() |
|
846 | 846 | dir, subpath = _splittopdir(f) |
|
847 | 847 | if dir: |
|
848 | 848 | self._dirs[dir].__delitem__(subpath) |
|
849 | 849 | # If the directory is now empty, remove it |
|
850 | 850 | if self._dirs[dir]._isempty(): |
|
851 | 851 | del self._dirs[dir] |
|
852 | 852 | else: |
|
853 | 853 | del self._files[f] |
|
854 | 854 | if f in self._flags: |
|
855 | 855 | del self._flags[f] |
|
856 | 856 | self._dirty = True |
|
857 | 857 | |
|
858 | 858 | def __setitem__(self, f, n): |
|
859 | 859 | assert n is not None |
|
860 | 860 | self._load() |
|
861 | 861 | dir, subpath = _splittopdir(f) |
|
862 | 862 | if dir: |
|
863 | 863 | if dir not in self._dirs: |
|
864 | 864 | self._dirs[dir] = treemanifest(self._subpath(dir)) |
|
865 | 865 | self._dirs[dir].__setitem__(subpath, n) |
|
866 | 866 | else: |
|
867 | 867 | self._files[f] = n[:21] # to match manifestdict's behavior |
|
868 | 868 | self._dirty = True |
|
869 | 869 | |
|
870 | 870 | def _load(self): |
|
871 | 871 | if self._loadfunc is not _noop: |
|
872 | 872 | lf, self._loadfunc = self._loadfunc, _noop |
|
873 | 873 | lf(self) |
|
874 | 874 | elif self._copyfunc is not _noop: |
|
875 | 875 | cf, self._copyfunc = self._copyfunc, _noop |
|
876 | 876 | cf(self) |
|
877 | 877 | |
|
878 | 878 | def setflag(self, f, flags): |
|
879 | 879 | """Set the flags (symlink, executable) for path f.""" |
|
880 | 880 | self._load() |
|
881 | 881 | dir, subpath = _splittopdir(f) |
|
882 | 882 | if dir: |
|
883 | 883 | if dir not in self._dirs: |
|
884 | 884 | self._dirs[dir] = treemanifest(self._subpath(dir)) |
|
885 | 885 | self._dirs[dir].setflag(subpath, flags) |
|
886 | 886 | else: |
|
887 | 887 | self._flags[f] = flags |
|
888 | 888 | self._dirty = True |
|
889 | 889 | |
|
890 | 890 | def copy(self): |
|
891 | 891 | copy = treemanifest(self._dir) |
|
892 | 892 | copy._node = self._node |
|
893 | 893 | copy._dirty = self._dirty |
|
894 | 894 | if self._copyfunc is _noop: |
|
895 | 895 | def _copyfunc(s): |
|
896 | 896 | self._load() |
|
897 | 897 | for d in self._dirs: |
|
898 | 898 | s._dirs[d] = self._dirs[d].copy() |
|
899 | 899 | s._files = dict.copy(self._files) |
|
900 | 900 | s._flags = dict.copy(self._flags) |
|
901 | 901 | if self._loadfunc is _noop: |
|
902 | 902 | _copyfunc(copy) |
|
903 | 903 | else: |
|
904 | 904 | copy._copyfunc = _copyfunc |
|
905 | 905 | else: |
|
906 | 906 | copy._copyfunc = self._copyfunc |
|
907 | 907 | return copy |
|
908 | 908 | |
|
909 | 909 | def filesnotin(self, m2): |
|
910 | 910 | '''Set of files in this manifest that are not in the other''' |
|
911 | 911 | files = set() |
|
912 | 912 | def _filesnotin(t1, t2): |
|
913 | 913 | if t1._node == t2._node and not t1._dirty and not t2._dirty: |
|
914 | 914 | return |
|
915 | 915 | t1._load() |
|
916 | 916 | t2._load() |
|
917 | 917 | for d, m1 in t1._dirs.iteritems(): |
|
918 | 918 | if d in t2._dirs: |
|
919 | 919 | m2 = t2._dirs[d] |
|
920 | 920 | _filesnotin(m1, m2) |
|
921 | 921 | else: |
|
922 | 922 | files.update(m1.iterkeys()) |
|
923 | 923 | |
|
924 | 924 | for fn in t1._files.iterkeys(): |
|
925 | 925 | if fn not in t2._files: |
|
926 | 926 | files.add(t1._subpath(fn)) |
|
927 | 927 | |
|
928 | 928 | _filesnotin(self, m2) |
|
929 | 929 | return files |
|
930 | 930 | |
|
931 | 931 | @propertycache |
|
932 | 932 | def _alldirs(self): |
|
933 | 933 | return util.dirs(self) |
|
934 | 934 | |
|
935 | 935 | def dirs(self): |
|
936 | 936 | return self._alldirs |
|
937 | 937 | |
|
938 | 938 | def hasdir(self, dir): |
|
939 | 939 | self._load() |
|
940 | 940 | topdir, subdir = _splittopdir(dir) |
|
941 | 941 | if topdir: |
|
942 | 942 | if topdir in self._dirs: |
|
943 | 943 | return self._dirs[topdir].hasdir(subdir) |
|
944 | 944 | return False |
|
945 | 945 | return (dir + '/') in self._dirs |
|
946 | 946 | |
|
947 | 947 | def walk(self, match): |
|
948 | 948 | '''Generates matching file names. |
|
949 | 949 | |
|
950 | 950 | Equivalent to manifest.matches(match).iterkeys(), but without creating |
|
951 | 951 | an entirely new manifest. |
|
952 | 952 | |
|
953 | 953 | It also reports nonexistent files by marking them bad with match.bad(). |
|
954 | 954 | ''' |
|
955 | 955 | if match.always(): |
|
956 | 956 | for f in iter(self): |
|
957 | 957 | yield f |
|
958 | 958 | return |
|
959 | 959 | |
|
960 | 960 | fset = set(match.files()) |
|
961 | 961 | |
|
962 | 962 | for fn in self._walk(match): |
|
963 | 963 | if fn in fset: |
|
964 | 964 | # specified pattern is the exact name |
|
965 | 965 | fset.remove(fn) |
|
966 | 966 | yield fn |
|
967 | 967 | |
|
968 | 968 | # for dirstate.walk, files=['.'] means "walk the whole tree". |
|
969 | 969 | # follow that here, too |
|
970 | 970 | fset.discard('.') |
|
971 | 971 | |
|
972 | 972 | for fn in sorted(fset): |
|
973 | 973 | if not self.hasdir(fn): |
|
974 | 974 | match.bad(fn, None) |
|
975 | 975 | |
|
976 | 976 | def _walk(self, match): |
|
977 | 977 | '''Recursively generates matching file names for walk().''' |
|
978 | 978 | if not match.visitdir(self._dir[:-1] or '.'): |
|
979 | 979 | return |
|
980 | 980 | |
|
981 | 981 | # yield this dir's files and walk its submanifests |
|
982 | 982 | self._load() |
|
983 | 983 | for p in sorted(self._dirs.keys() + self._files.keys()): |
|
984 | 984 | if p in self._files: |
|
985 | 985 | fullp = self._subpath(p) |
|
986 | 986 | if match(fullp): |
|
987 | 987 | yield fullp |
|
988 | 988 | else: |
|
989 | 989 | for f in self._dirs[p]._walk(match): |
|
990 | 990 | yield f |
|
991 | 991 | |
|
992 | 992 | def matches(self, match): |
|
993 | 993 | '''generate a new manifest filtered by the match argument''' |
|
994 | 994 | if match.always(): |
|
995 | 995 | return self.copy() |
|
996 | 996 | |
|
997 | 997 | return self._matches(match) |
|
998 | 998 | |
|
999 | 999 | def _matches(self, match): |
|
1000 | 1000 | '''recursively generate a new manifest filtered by the match argument. |
|
1001 | 1001 | ''' |
|
1002 | 1002 | |
|
1003 | 1003 | visit = match.visitdir(self._dir[:-1] or '.') |
|
1004 | 1004 | if visit == 'all': |
|
1005 | 1005 | return self.copy() |
|
1006 | 1006 | ret = treemanifest(self._dir) |
|
1007 | 1007 | if not visit: |
|
1008 | 1008 | return ret |
|
1009 | 1009 | |
|
1010 | 1010 | self._load() |
|
1011 | 1011 | for fn in self._files: |
|
1012 | 1012 | fullp = self._subpath(fn) |
|
1013 | 1013 | if not match(fullp): |
|
1014 | 1014 | continue |
|
1015 | 1015 | ret._files[fn] = self._files[fn] |
|
1016 | 1016 | if fn in self._flags: |
|
1017 | 1017 | ret._flags[fn] = self._flags[fn] |
|
1018 | 1018 | |
|
1019 | 1019 | for dir, subm in self._dirs.iteritems(): |
|
1020 | 1020 | m = subm._matches(match) |
|
1021 | 1021 | if not m._isempty(): |
|
1022 | 1022 | ret._dirs[dir] = m |
|
1023 | 1023 | |
|
1024 | 1024 | if not ret._isempty(): |
|
1025 | 1025 | ret._dirty = True |
|
1026 | 1026 | return ret |
|
1027 | 1027 | |
|
1028 | 1028 | def diff(self, m2, clean=False): |
|
1029 | 1029 | '''Finds changes between the current manifest and m2. |
|
1030 | 1030 | |
|
1031 | 1031 | Args: |
|
1032 | 1032 | m2: the manifest to which this manifest should be compared. |
|
1033 | 1033 | clean: if true, include files unchanged between these manifests |
|
1034 | 1034 | with a None value in the returned dictionary. |
|
1035 | 1035 | |
|
1036 | 1036 | The result is returned as a dict with filename as key and |
|
1037 | 1037 | values of the form ((n1,fl1),(n2,fl2)), where n1/n2 is the |
|
1038 | 1038 | nodeid in the current/other manifest and fl1/fl2 is the flag |
|
1039 | 1039 | in the current/other manifest. Where the file does not exist, |
|
1040 | 1040 | the nodeid will be None and the flags will be the empty |
|
1041 | 1041 | string. |
|
1042 | 1042 | ''' |
|
1043 | 1043 | result = {} |
|
1044 | 1044 | emptytree = treemanifest() |
|
1045 | 1045 | def _diff(t1, t2): |
|
1046 | 1046 | if t1._node == t2._node and not t1._dirty and not t2._dirty: |
|
1047 | 1047 | return |
|
1048 | 1048 | t1._load() |
|
1049 | 1049 | t2._load() |
|
1050 | 1050 | for d, m1 in t1._dirs.iteritems(): |
|
1051 | 1051 | m2 = t2._dirs.get(d, emptytree) |
|
1052 | 1052 | _diff(m1, m2) |
|
1053 | 1053 | |
|
1054 | 1054 | for d, m2 in t2._dirs.iteritems(): |
|
1055 | 1055 | if d not in t1._dirs: |
|
1056 | 1056 | _diff(emptytree, m2) |
|
1057 | 1057 | |
|
1058 | 1058 | for fn, n1 in t1._files.iteritems(): |
|
1059 | 1059 | fl1 = t1._flags.get(fn, '') |
|
1060 | 1060 | n2 = t2._files.get(fn, None) |
|
1061 | 1061 | fl2 = t2._flags.get(fn, '') |
|
1062 | 1062 | if n1 != n2 or fl1 != fl2: |
|
1063 | 1063 | result[t1._subpath(fn)] = ((n1, fl1), (n2, fl2)) |
|
1064 | 1064 | elif clean: |
|
1065 | 1065 | result[t1._subpath(fn)] = None |
|
1066 | 1066 | |
|
1067 | 1067 | for fn, n2 in t2._files.iteritems(): |
|
1068 | 1068 | if fn not in t1._files: |
|
1069 | 1069 | fl2 = t2._flags.get(fn, '') |
|
1070 | 1070 | result[t2._subpath(fn)] = ((None, ''), (n2, fl2)) |
|
1071 | 1071 | |
|
1072 | 1072 | _diff(self, m2) |
|
1073 | 1073 | return result |
|
1074 | 1074 | |
|
1075 | 1075 | def unmodifiedsince(self, m2): |
|
1076 | 1076 | return not self._dirty and not m2._dirty and self._node == m2._node |
|
1077 | 1077 | |
|
1078 | 1078 | def parse(self, text, readsubtree): |
|
1079 | 1079 | for f, n, fl in _parse(text): |
|
1080 | 1080 | if fl == 't': |
|
1081 | 1081 | f = f + '/' |
|
1082 | 1082 | self._dirs[f] = readsubtree(self._subpath(f), n) |
|
1083 | 1083 | elif '/' in f: |
|
1084 | 1084 | # This is a flat manifest, so use __setitem__ and setflag rather |
|
1085 | 1085 | # than assigning directly to _files and _flags, so we can |
|
1086 | 1086 | # assign a path in a subdirectory, and to mark dirty (compared |
|
1087 | 1087 | # to nullid). |
|
1088 | 1088 | self[f] = n |
|
1089 | 1089 | if fl: |
|
1090 | 1090 | self.setflag(f, fl) |
|
1091 | 1091 | else: |
|
1092 | 1092 | # Assigning to _files and _flags avoids marking as dirty, |
|
1093 | 1093 | # and should be a little faster. |
|
1094 | 1094 | self._files[f] = n |
|
1095 | 1095 | if fl: |
|
1096 | 1096 | self._flags[f] = fl |
|
1097 | 1097 | |
|
1098 | 1098 | def text(self, usemanifestv2=False): |
|
1099 | 1099 | """Get the full data of this manifest as a bytestring.""" |
|
1100 | 1100 | self._load() |
|
1101 | 1101 | return _text(self.iterentries(), usemanifestv2) |
|
1102 | 1102 | |
|
1103 | 1103 | def dirtext(self, usemanifestv2=False): |
|
1104 | 1104 | """Get the full data of this directory as a bytestring. Make sure that |
|
1105 | 1105 | any submanifests have been written first, so their nodeids are correct. |
|
1106 | 1106 | """ |
|
1107 | 1107 | self._load() |
|
1108 | 1108 | flags = self.flags |
|
1109 | 1109 | dirs = [(d[:-1], self._dirs[d]._node, 't') for d in self._dirs] |
|
1110 | 1110 | files = [(f, self._files[f], flags(f)) for f in self._files] |
|
1111 | 1111 | return _text(sorted(dirs + files), usemanifestv2) |
|
1112 | 1112 | |
|
1113 | 1113 | def read(self, gettext, readsubtree): |
|
1114 | 1114 | def _load_for_read(s): |
|
1115 | 1115 | s.parse(gettext(), readsubtree) |
|
1116 | 1116 | s._dirty = False |
|
1117 | 1117 | self._loadfunc = _load_for_read |
|
1118 | 1118 | |
|
1119 | 1119 | def writesubtrees(self, m1, m2, writesubtree): |
|
1120 | 1120 | self._load() # for consistency; should never have any effect here |
|
1121 | 1121 | m1._load() |
|
1122 | 1122 | m2._load() |
|
1123 | 1123 | emptytree = treemanifest() |
|
1124 | 1124 | for d, subm in self._dirs.iteritems(): |
|
1125 | 1125 | subp1 = m1._dirs.get(d, emptytree)._node |
|
1126 | 1126 | subp2 = m2._dirs.get(d, emptytree)._node |
|
1127 | 1127 | if subp1 == revlog.nullid: |
|
1128 | 1128 | subp1, subp2 = subp2, subp1 |
|
1129 | 1129 | writesubtree(subm, subp1, subp2) |
|
1130 | 1130 | |
|
1131 | 1131 | class manifestrevlog(revlog.revlog): |
|
1132 | 1132 | '''A revlog that stores manifest texts. This is responsible for caching the |
|
1133 | 1133 | full-text manifest contents. |
|
1134 | 1134 | ''' |
|
1135 | 1135 | def __init__(self, opener, dir='', dirlogcache=None): |
|
1136 | 1136 | # During normal operations, we expect to deal with not more than four |
|
1137 | 1137 | # revs at a time (such as during commit --amend). When rebasing large |
|
1138 | 1138 | # stacks of commits, the number can go up, hence the config knob below. |
|
1139 | 1139 | cachesize = 4 |
|
1140 | 1140 | usetreemanifest = False |
|
1141 | 1141 | usemanifestv2 = False |
|
1142 | 1142 | opts = getattr(opener, 'options', None) |
|
1143 | 1143 | if opts is not None: |
|
1144 | 1144 | cachesize = opts.get('manifestcachesize', cachesize) |
|
1145 | 1145 | usetreemanifest = opts.get('treemanifest', usetreemanifest) |
|
1146 | 1146 | usemanifestv2 = opts.get('manifestv2', usemanifestv2) |
|
1147 | 1147 | |
|
1148 | 1148 | self._treeondisk = usetreemanifest |
|
1149 | 1149 | self._usemanifestv2 = usemanifestv2 |
|
1150 | 1150 | |
|
1151 | 1151 | self._fulltextcache = util.lrucachedict(cachesize) |
|
1152 | 1152 | |
|
1153 | 1153 | indexfile = "00manifest.i" |
|
1154 | 1154 | if dir: |
|
1155 | 1155 | assert self._treeondisk, 'opts is %r' % opts |
|
1156 | 1156 | if not dir.endswith('/'): |
|
1157 | 1157 | dir = dir + '/' |
|
1158 | 1158 | indexfile = "meta/" + dir + "00manifest.i" |
|
1159 | 1159 | self._dir = dir |
|
1160 | 1160 | # The dirlogcache is kept on the root manifest log |
|
1161 | 1161 | if dir: |
|
1162 | 1162 | self._dirlogcache = dirlogcache |
|
1163 | 1163 | else: |
|
1164 | 1164 | self._dirlogcache = {'': self} |
|
1165 | 1165 | |
|
1166 | 1166 | super(manifestrevlog, self).__init__(opener, indexfile, |
|
1167 | 1167 | checkambig=bool(dir)) |
|
1168 | 1168 | |
|
1169 | 1169 | @property |
|
1170 | 1170 | def fulltextcache(self): |
|
1171 | 1171 | return self._fulltextcache |
|
1172 | 1172 | |
|
1173 | 1173 | def clearcaches(self): |
|
1174 | 1174 | super(manifestrevlog, self).clearcaches() |
|
1175 | 1175 | self._fulltextcache.clear() |
|
1176 | 1176 | self._dirlogcache = {'': self} |
|
1177 | 1177 | |
|
1178 | 1178 | def dirlog(self, dir): |
|
1179 | 1179 | if dir: |
|
1180 | 1180 | assert self._treeondisk |
|
1181 | 1181 | if dir not in self._dirlogcache: |
|
1182 | 1182 | self._dirlogcache[dir] = manifestrevlog(self.opener, dir, |
|
1183 | 1183 | self._dirlogcache) |
|
1184 | 1184 | return self._dirlogcache[dir] |
|
1185 | 1185 | |
|
1186 | 1186 | def add(self, m, transaction, link, p1, p2, added, removed, readtree=None): |
|
1187 | 1187 | if (p1 in self.fulltextcache and util.safehasattr(m, 'fastdelta') |
|
1188 | 1188 | and not self._usemanifestv2): |
|
1189 | 1189 | # If our first parent is in the manifest cache, we can |
|
1190 | 1190 | # compute a delta here using properties we know about the |
|
1191 | 1191 | # manifest up-front, which may save time later for the |
|
1192 | 1192 | # revlog layer. |
|
1193 | 1193 | |
|
1194 | 1194 | _checkforbidden(added) |
|
1195 | 1195 | # combine the changed lists into one sorted iterator |
|
1196 | 1196 | work = heapq.merge([(x, False) for x in added], |
|
1197 | 1197 | [(x, True) for x in removed]) |
|
1198 | 1198 | |
|
1199 | 1199 | arraytext, deltatext = m.fastdelta(self.fulltextcache[p1], work) |
|
1200 | 1200 | cachedelta = self.rev(p1), deltatext |
|
1201 | 1201 | text = util.buffer(arraytext) |
|
1202 | 1202 | n = self.addrevision(text, transaction, link, p1, p2, cachedelta) |
|
1203 | 1203 | else: |
|
1204 | 1204 | # The first parent manifest isn't already loaded, so we'll |
|
1205 | 1205 | # just encode a fulltext of the manifest and pass that |
|
1206 | 1206 | # through to the revlog layer, and let it handle the delta |
|
1207 | 1207 | # process. |
|
1208 | 1208 | if self._treeondisk: |
|
1209 | 1209 | assert readtree, "readtree must be set for treemanifest writes" |
|
1210 | 1210 | m1 = readtree(self._dir, p1) |
|
1211 | 1211 | m2 = readtree(self._dir, p2) |
|
1212 | 1212 | n = self._addtree(m, transaction, link, m1, m2, readtree) |
|
1213 | 1213 | arraytext = None |
|
1214 | 1214 | else: |
|
1215 | 1215 | text = m.text(self._usemanifestv2) |
|
1216 | 1216 | n = self.addrevision(text, transaction, link, p1, p2) |
|
1217 | 1217 | arraytext = array.array('c', text) |
|
1218 | 1218 | |
|
1219 | 1219 | if arraytext is not None: |
|
1220 | 1220 | self.fulltextcache[n] = arraytext |
|
1221 | 1221 | |
|
1222 | 1222 | return n |
|
1223 | 1223 | |
|
1224 | 1224 | def _addtree(self, m, transaction, link, m1, m2, readtree): |
|
1225 | 1225 | # If the manifest is unchanged compared to one parent, |
|
1226 | 1226 | # don't write a new revision |
|
1227 | 1227 | if m.unmodifiedsince(m1) or m.unmodifiedsince(m2): |
|
1228 | 1228 | return m.node() |
|
1229 | 1229 | def writesubtree(subm, subp1, subp2): |
|
1230 | 1230 | sublog = self.dirlog(subm.dir()) |
|
1231 | 1231 | sublog.add(subm, transaction, link, subp1, subp2, None, None, |
|
1232 | 1232 | readtree=readtree) |
|
1233 | 1233 | m.writesubtrees(m1, m2, writesubtree) |
|
1234 | 1234 | text = m.dirtext(self._usemanifestv2) |
|
1235 | 1235 | # Double-check whether contents are unchanged to one parent |
|
1236 | 1236 | if text == m1.dirtext(self._usemanifestv2): |
|
1237 | 1237 | n = m1.node() |
|
1238 | 1238 | elif text == m2.dirtext(self._usemanifestv2): |
|
1239 | 1239 | n = m2.node() |
|
1240 | 1240 | else: |
|
1241 | 1241 | n = self.addrevision(text, transaction, link, m1.node(), m2.node()) |
|
1242 | 1242 | # Save nodeid so parent manifest can calculate its nodeid |
|
1243 | 1243 | m.setnode(n) |
|
1244 | 1244 | return n |
|
1245 | 1245 | |
|
1246 | 1246 | class manifestlog(object): |
|
1247 | 1247 | """A collection class representing the collection of manifest snapshots |
|
1248 | 1248 | referenced by commits in the repository. |
|
1249 | 1249 | |
|
1250 | 1250 | In this situation, 'manifest' refers to the abstract concept of a snapshot |
|
1251 | 1251 | of the list of files in the given commit. Consumers of the output of this |
|
1252 | 1252 | class do not care about the implementation details of the actual manifests |
|
1253 | 1253 | they receive (i.e. tree or flat or lazily loaded, etc).""" |
|
1254 | 1254 | def __init__(self, opener, repo): |
|
1255 | 1255 | self._repo = repo |
|
1256 | 1256 | |
|
1257 | 1257 | usetreemanifest = False |
|
1258 | 1258 | cachesize = 4 |
|
1259 | 1259 | |
|
1260 | 1260 | opts = getattr(opener, 'options', None) |
|
1261 | 1261 | if opts is not None: |
|
1262 | 1262 | usetreemanifest = opts.get('treemanifest', usetreemanifest) |
|
1263 | 1263 | cachesize = opts.get('manifestcachesize', cachesize) |
|
1264 | 1264 | self._treeinmem = usetreemanifest |
|
1265 | 1265 | |
|
1266 | 1266 | self._oldmanifest = repo._constructmanifest() |
|
1267 | 1267 | self._revlog = self._oldmanifest |
|
1268 | 1268 | |
|
1269 | 1269 | # A cache of the manifestctx or treemanifestctx for each directory |
|
1270 | 1270 | self._dirmancache = {} |
|
1271 | 1271 | self._dirmancache[''] = util.lrucachedict(cachesize) |
|
1272 | 1272 | |
|
1273 | 1273 | self.cachesize = cachesize |
|
1274 | 1274 | |
|
1275 | 1275 | def __getitem__(self, node): |
|
1276 | 1276 | """Retrieves the manifest instance for the given node. Throws a |
|
1277 | 1277 | LookupError if not found. |
|
1278 | 1278 | """ |
|
1279 | 1279 | return self.get('', node) |
|
1280 | 1280 | |
|
1281 | def get(self, dir, node): | |
|
1281 | def get(self, dir, node, verify=True): | |
|
1282 | 1282 | """Retrieves the manifest instance for the given node. Throws a |
|
1283 | 1283 | LookupError if not found. |
|
1284 | ||
|
1285 | `verify` - if True an exception will be thrown if the node is not in | |
|
1286 | the revlog | |
|
1284 | 1287 | """ |
|
1285 | 1288 | if node in self._dirmancache.get(dir, ()): |
|
1286 | 1289 | cachemf = self._dirmancache[dir][node] |
|
1287 | 1290 | # The old manifest may put non-ctx manifests in the cache, so |
|
1288 | 1291 | # skip those since they don't implement the full api. |
|
1289 | 1292 | if (isinstance(cachemf, manifestctx) or |
|
1290 | 1293 | isinstance(cachemf, treemanifestctx)): |
|
1291 | 1294 | return cachemf |
|
1292 | 1295 | |
|
1293 | 1296 | if dir: |
|
1294 | 1297 | if self._revlog._treeondisk: |
|
1298 | if verify: | |
|
1295 | 1299 | dirlog = self._revlog.dirlog(dir) |
|
1296 | 1300 | if node not in dirlog.nodemap: |
|
1297 | 1301 | raise LookupError(node, dirlog.indexfile, |
|
1298 | 1302 | _('no node')) |
|
1299 | 1303 | m = treemanifestctx(self._repo, dir, node) |
|
1300 | 1304 | else: |
|
1301 | 1305 | raise error.Abort( |
|
1302 | 1306 | _("cannot ask for manifest directory '%s' in a flat " |
|
1303 | 1307 | "manifest") % dir) |
|
1304 | 1308 | else: |
|
1309 | if verify: | |
|
1305 | 1310 | if node not in self._revlog.nodemap: |
|
1306 | 1311 | raise LookupError(node, self._revlog.indexfile, |
|
1307 | 1312 | _('no node')) |
|
1308 | 1313 | if self._treeinmem: |
|
1309 | 1314 | m = treemanifestctx(self._repo, '', node) |
|
1310 | 1315 | else: |
|
1311 | 1316 | m = manifestctx(self._repo, node) |
|
1312 | 1317 | |
|
1313 | 1318 | if node != revlog.nullid: |
|
1314 | 1319 | mancache = self._dirmancache.get(dir) |
|
1315 | 1320 | if not mancache: |
|
1316 | 1321 | mancache = util.lrucachedict(self.cachesize) |
|
1317 | 1322 | self._dirmancache[dir] = mancache |
|
1318 | 1323 | mancache[node] = m |
|
1319 | 1324 | return m |
|
1320 | 1325 | |
|
1321 | 1326 | def clearcaches(self): |
|
1322 | 1327 | self._dirmancache.clear() |
|
1323 | 1328 | self._revlog.clearcaches() |
|
1324 | 1329 | |
|
1325 | 1330 | class memmanifestctx(object): |
|
1326 | 1331 | def __init__(self, repo): |
|
1327 | 1332 | self._repo = repo |
|
1328 | 1333 | self._manifestdict = manifestdict() |
|
1329 | 1334 | |
|
1330 | 1335 | def _revlog(self): |
|
1331 | 1336 | return self._repo.manifestlog._revlog |
|
1332 | 1337 | |
|
1333 | 1338 | def new(self): |
|
1334 | 1339 | return memmanifestctx(self._repo) |
|
1335 | 1340 | |
|
1336 | 1341 | def copy(self): |
|
1337 | 1342 | memmf = memmanifestctx(self._repo) |
|
1338 | 1343 | memmf._manifestdict = self.read().copy() |
|
1339 | 1344 | return memmf |
|
1340 | 1345 | |
|
1341 | 1346 | def read(self): |
|
1342 | 1347 | return self._manifestdict |
|
1343 | 1348 | |
|
1344 | 1349 | def write(self, transaction, link, p1, p2, added, removed): |
|
1345 | 1350 | return self._revlog().add(self._manifestdict, transaction, link, p1, p2, |
|
1346 | 1351 | added, removed) |
|
1347 | 1352 | |
|
1348 | 1353 | class manifestctx(object): |
|
1349 | 1354 | """A class representing a single revision of a manifest, including its |
|
1350 | 1355 | contents, its parent revs, and its linkrev. |
|
1351 | 1356 | """ |
|
1352 | 1357 | def __init__(self, repo, node): |
|
1353 | 1358 | self._repo = repo |
|
1354 | 1359 | self._data = None |
|
1355 | 1360 | |
|
1356 | 1361 | self._node = node |
|
1357 | 1362 | |
|
1358 | 1363 | # TODO: We eventually want p1, p2, and linkrev exposed on this class, |
|
1359 | 1364 | # but let's add it later when something needs it and we can load it |
|
1360 | 1365 | # lazily. |
|
1361 | 1366 | #self.p1, self.p2 = revlog.parents(node) |
|
1362 | 1367 | #rev = revlog.rev(node) |
|
1363 | 1368 | #self.linkrev = revlog.linkrev(rev) |
|
1364 | 1369 | |
|
1365 | 1370 | def _revlog(self): |
|
1366 | 1371 | return self._repo.manifestlog._revlog |
|
1367 | 1372 | |
|
1368 | 1373 | def node(self): |
|
1369 | 1374 | return self._node |
|
1370 | 1375 | |
|
1371 | 1376 | def new(self): |
|
1372 | 1377 | return memmanifestctx(self._repo) |
|
1373 | 1378 | |
|
1374 | 1379 | def copy(self): |
|
1375 | 1380 | memmf = memmanifestctx(self._repo) |
|
1376 | 1381 | memmf._manifestdict = self.read().copy() |
|
1377 | 1382 | return memmf |
|
1378 | 1383 | |
|
1379 | 1384 | def read(self): |
|
1380 | 1385 | if not self._data: |
|
1381 | 1386 | if self._node == revlog.nullid: |
|
1382 | 1387 | self._data = manifestdict() |
|
1383 | 1388 | else: |
|
1384 | 1389 | rl = self._revlog() |
|
1385 | 1390 | text = rl.revision(self._node) |
|
1386 | 1391 | arraytext = array.array('c', text) |
|
1387 | 1392 | rl._fulltextcache[self._node] = arraytext |
|
1388 | 1393 | self._data = manifestdict(text) |
|
1389 | 1394 | return self._data |
|
1390 | 1395 | |
|
1391 | 1396 | def readfast(self, shallow=False): |
|
1392 | 1397 | '''Calls either readdelta or read, based on which would be less work. |
|
1393 | 1398 | readdelta is called if the delta is against the p1, and therefore can be |
|
1394 | 1399 | read quickly. |
|
1395 | 1400 | |
|
1396 | 1401 | If `shallow` is True, nothing changes since this is a flat manifest. |
|
1397 | 1402 | ''' |
|
1398 | 1403 | rl = self._revlog() |
|
1399 | 1404 | r = rl.rev(self._node) |
|
1400 | 1405 | deltaparent = rl.deltaparent(r) |
|
1401 | 1406 | if deltaparent != revlog.nullrev and deltaparent in rl.parentrevs(r): |
|
1402 | 1407 | return self.readdelta() |
|
1403 | 1408 | return self.read() |
|
1404 | 1409 | |
|
1405 | 1410 | def readdelta(self, shallow=False): |
|
1406 | 1411 | '''Returns a manifest containing just the entries that are present |
|
1407 | 1412 | in this manifest, but not in its p1 manifest. This is efficient to read |
|
1408 | 1413 | if the revlog delta is already p1. |
|
1409 | 1414 | |
|
1410 | 1415 | Changing the value of `shallow` has no effect on flat manifests. |
|
1411 | 1416 | ''' |
|
1412 | 1417 | revlog = self._revlog() |
|
1413 | 1418 | if revlog._usemanifestv2: |
|
1414 | 1419 | # Need to perform a slow delta |
|
1415 | 1420 | r0 = revlog.deltaparent(revlog.rev(self._node)) |
|
1416 | 1421 | m0 = manifestctx(self._repo, revlog.node(r0)).read() |
|
1417 | 1422 | m1 = self.read() |
|
1418 | 1423 | md = manifestdict() |
|
1419 | 1424 | for f, ((n0, fl0), (n1, fl1)) in m0.diff(m1).iteritems(): |
|
1420 | 1425 | if n1: |
|
1421 | 1426 | md[f] = n1 |
|
1422 | 1427 | if fl1: |
|
1423 | 1428 | md.setflag(f, fl1) |
|
1424 | 1429 | return md |
|
1425 | 1430 | |
|
1426 | 1431 | r = revlog.rev(self._node) |
|
1427 | 1432 | d = mdiff.patchtext(revlog.revdiff(revlog.deltaparent(r), r)) |
|
1428 | 1433 | return manifestdict(d) |
|
1429 | 1434 | |
|
1430 | 1435 | def find(self, key): |
|
1431 | 1436 | return self.read().find(key) |
|
1432 | 1437 | |
|
1433 | 1438 | class memtreemanifestctx(object): |
|
1434 | 1439 | def __init__(self, repo, dir=''): |
|
1435 | 1440 | self._repo = repo |
|
1436 | 1441 | self._dir = dir |
|
1437 | 1442 | self._treemanifest = treemanifest() |
|
1438 | 1443 | |
|
1439 | 1444 | def _revlog(self): |
|
1440 | 1445 | return self._repo.manifestlog._revlog |
|
1441 | 1446 | |
|
1442 | 1447 | def new(self, dir=''): |
|
1443 | 1448 | return memtreemanifestctx(self._repo, dir=dir) |
|
1444 | 1449 | |
|
1445 | 1450 | def copy(self): |
|
1446 | 1451 | memmf = memtreemanifestctx(self._repo, dir=self._dir) |
|
1447 | 1452 | memmf._treemanifest = self._treemanifest.copy() |
|
1448 | 1453 | return memmf |
|
1449 | 1454 | |
|
1450 | 1455 | def read(self): |
|
1451 | 1456 | return self._treemanifest |
|
1452 | 1457 | |
|
1453 | 1458 | def write(self, transaction, link, p1, p2, added, removed): |
|
1454 | 1459 | def readtree(dir, node): |
|
1455 | 1460 | return self._repo.manifestlog.get(dir, node).read() |
|
1456 | 1461 | return self._revlog().add(self._treemanifest, transaction, link, p1, p2, |
|
1457 | 1462 | added, removed, readtree=readtree) |
|
1458 | 1463 | |
|
1459 | 1464 | class treemanifestctx(object): |
|
1460 | 1465 | def __init__(self, repo, dir, node): |
|
1461 | 1466 | self._repo = repo |
|
1462 | 1467 | self._dir = dir |
|
1463 | 1468 | self._data = None |
|
1464 | 1469 | |
|
1465 | 1470 | self._node = node |
|
1466 | 1471 | |
|
1467 | 1472 | # TODO: Load p1/p2/linkrev lazily. They need to be lazily loaded so that |
|
1468 | 1473 | # we can instantiate treemanifestctx objects for directories we don't |
|
1469 | 1474 | # have on disk. |
|
1470 | 1475 | #self.p1, self.p2 = revlog.parents(node) |
|
1471 | 1476 | #rev = revlog.rev(node) |
|
1472 | 1477 | #self.linkrev = revlog.linkrev(rev) |
|
1473 | 1478 | |
|
1474 | 1479 | def _revlog(self): |
|
1475 | 1480 | return self._repo.manifestlog._revlog.dirlog(self._dir) |
|
1476 | 1481 | |
|
1477 | 1482 | def read(self): |
|
1478 | 1483 | if not self._data: |
|
1479 | 1484 | rl = self._revlog() |
|
1480 | 1485 | if self._node == revlog.nullid: |
|
1481 | 1486 | self._data = treemanifest() |
|
1482 | 1487 | elif rl._treeondisk: |
|
1483 | 1488 | m = treemanifest(dir=self._dir) |
|
1484 | 1489 | def gettext(): |
|
1485 | 1490 | return rl.revision(self._node) |
|
1486 | 1491 | def readsubtree(dir, subm): |
|
1487 | 1492 | return treemanifestctx(self._repo, dir, subm).read() |
|
1488 | 1493 | m.read(gettext, readsubtree) |
|
1489 | 1494 | m.setnode(self._node) |
|
1490 | 1495 | self._data = m |
|
1491 | 1496 | else: |
|
1492 | 1497 | text = rl.revision(self._node) |
|
1493 | 1498 | arraytext = array.array('c', text) |
|
1494 | 1499 | rl.fulltextcache[self._node] = arraytext |
|
1495 | 1500 | self._data = treemanifest(dir=self._dir, text=text) |
|
1496 | 1501 | |
|
1497 | 1502 | return self._data |
|
1498 | 1503 | |
|
1499 | 1504 | def node(self): |
|
1500 | 1505 | return self._node |
|
1501 | 1506 | |
|
1502 | 1507 | def new(self, dir=''): |
|
1503 | 1508 | return memtreemanifestctx(self._repo, dir=dir) |
|
1504 | 1509 | |
|
1505 | 1510 | def copy(self): |
|
1506 | 1511 | memmf = memtreemanifestctx(self._repo, dir=self._dir) |
|
1507 | 1512 | memmf._treemanifest = self.read().copy() |
|
1508 | 1513 | return memmf |
|
1509 | 1514 | |
|
1510 | 1515 | def readdelta(self, shallow=False): |
|
1511 | 1516 | '''Returns a manifest containing just the entries that are present |
|
1512 | 1517 | in this manifest, but not in its p1 manifest. This is efficient to read |
|
1513 | 1518 | if the revlog delta is already p1. |
|
1514 | 1519 | |
|
1515 | 1520 | If `shallow` is True, this will read the delta for this directory, |
|
1516 | 1521 | without recursively reading subdirectory manifests. Instead, any |
|
1517 | 1522 | subdirectory entry will be reported as it appears in the manifest, i.e. |
|
1518 | 1523 | the subdirectory will be reported among files and distinguished only by |
|
1519 | 1524 | its 't' flag. |
|
1520 | 1525 | ''' |
|
1521 | 1526 | revlog = self._revlog() |
|
1522 | 1527 | if shallow and not revlog._usemanifestv2: |
|
1523 | 1528 | r = revlog.rev(self._node) |
|
1524 | 1529 | d = mdiff.patchtext(revlog.revdiff(revlog.deltaparent(r), r)) |
|
1525 | 1530 | return manifestdict(d) |
|
1526 | 1531 | else: |
|
1527 | 1532 | # Need to perform a slow delta |
|
1528 | 1533 | r0 = revlog.deltaparent(revlog.rev(self._node)) |
|
1529 | 1534 | m0 = treemanifestctx(self._repo, self._dir, revlog.node(r0)).read() |
|
1530 | 1535 | m1 = self.read() |
|
1531 | 1536 | md = treemanifest(dir=self._dir) |
|
1532 | 1537 | for f, ((n0, fl0), (n1, fl1)) in m0.diff(m1).iteritems(): |
|
1533 | 1538 | if n1: |
|
1534 | 1539 | md[f] = n1 |
|
1535 | 1540 | if fl1: |
|
1536 | 1541 | md.setflag(f, fl1) |
|
1537 | 1542 | return md |
|
1538 | 1543 | |
|
1539 | 1544 | def readfast(self, shallow=False): |
|
1540 | 1545 | '''Calls either readdelta or read, based on which would be less work. |
|
1541 | 1546 | readdelta is called if the delta is against the p1, and therefore can be |
|
1542 | 1547 | read quickly. |
|
1543 | 1548 | |
|
1544 | 1549 | If `shallow` is True, it only returns the entries from this manifest, |
|
1545 | 1550 | and not any submanifests. |
|
1546 | 1551 | ''' |
|
1547 | 1552 | rl = self._revlog() |
|
1548 | 1553 | r = rl.rev(self._node) |
|
1549 | 1554 | deltaparent = rl.deltaparent(r) |
|
1550 | 1555 | if (deltaparent != revlog.nullrev and |
|
1551 | 1556 | deltaparent in rl.parentrevs(r)): |
|
1552 | 1557 | return self.readdelta(shallow=shallow) |
|
1553 | 1558 | |
|
1554 | 1559 | if shallow: |
|
1555 | 1560 | return manifestdict(rl.revision(self._node)) |
|
1556 | 1561 | else: |
|
1557 | 1562 | return self.read() |
|
1558 | 1563 | |
|
1559 | 1564 | def find(self, key): |
|
1560 | 1565 | return self.read().find(key) |
General Comments 0
You need to be logged in to leave comments.
Login now