##// END OF EJS Templates
revset: categorize wdir() as very fast function...
Yuya Nishihara -
r30700:323f0c4b default
parent child Browse files
Show More
@@ -1,3857 +1,3857
1 1 # revset.py - revision set queries for mercurial
2 2 #
3 3 # Copyright 2010 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 heapq
11 11 import re
12 12 import string
13 13
14 14 from .i18n import _
15 15 from . import (
16 16 destutil,
17 17 encoding,
18 18 error,
19 19 hbisect,
20 20 match as matchmod,
21 21 node,
22 22 obsolete as obsmod,
23 23 parser,
24 24 pathutil,
25 25 phases,
26 26 pycompat,
27 27 registrar,
28 28 repoview,
29 29 util,
30 30 )
31 31
32 32 def _revancestors(repo, revs, followfirst):
33 33 """Like revlog.ancestors(), but supports followfirst."""
34 34 if followfirst:
35 35 cut = 1
36 36 else:
37 37 cut = None
38 38 cl = repo.changelog
39 39
40 40 def iterate():
41 41 revs.sort(reverse=True)
42 42 irevs = iter(revs)
43 43 h = []
44 44
45 45 inputrev = next(irevs, None)
46 46 if inputrev is not None:
47 47 heapq.heappush(h, -inputrev)
48 48
49 49 seen = set()
50 50 while h:
51 51 current = -heapq.heappop(h)
52 52 if current == inputrev:
53 53 inputrev = next(irevs, None)
54 54 if inputrev is not None:
55 55 heapq.heappush(h, -inputrev)
56 56 if current not in seen:
57 57 seen.add(current)
58 58 yield current
59 59 for parent in cl.parentrevs(current)[:cut]:
60 60 if parent != node.nullrev:
61 61 heapq.heappush(h, -parent)
62 62
63 63 return generatorset(iterate(), iterasc=False)
64 64
65 65 def _revdescendants(repo, revs, followfirst):
66 66 """Like revlog.descendants() but supports followfirst."""
67 67 if followfirst:
68 68 cut = 1
69 69 else:
70 70 cut = None
71 71
72 72 def iterate():
73 73 cl = repo.changelog
74 74 # XXX this should be 'parentset.min()' assuming 'parentset' is a
75 75 # smartset (and if it is not, it should.)
76 76 first = min(revs)
77 77 nullrev = node.nullrev
78 78 if first == nullrev:
79 79 # Are there nodes with a null first parent and a non-null
80 80 # second one? Maybe. Do we care? Probably not.
81 81 for i in cl:
82 82 yield i
83 83 else:
84 84 seen = set(revs)
85 85 for i in cl.revs(first + 1):
86 86 for x in cl.parentrevs(i)[:cut]:
87 87 if x != nullrev and x in seen:
88 88 seen.add(i)
89 89 yield i
90 90 break
91 91
92 92 return generatorset(iterate(), iterasc=True)
93 93
94 94 def _reachablerootspure(repo, minroot, roots, heads, includepath):
95 95 """return (heads(::<roots> and ::<heads>))
96 96
97 97 If includepath is True, return (<roots>::<heads>)."""
98 98 if not roots:
99 99 return []
100 100 parentrevs = repo.changelog.parentrevs
101 101 roots = set(roots)
102 102 visit = list(heads)
103 103 reachable = set()
104 104 seen = {}
105 105 # prefetch all the things! (because python is slow)
106 106 reached = reachable.add
107 107 dovisit = visit.append
108 108 nextvisit = visit.pop
109 109 # open-code the post-order traversal due to the tiny size of
110 110 # sys.getrecursionlimit()
111 111 while visit:
112 112 rev = nextvisit()
113 113 if rev in roots:
114 114 reached(rev)
115 115 if not includepath:
116 116 continue
117 117 parents = parentrevs(rev)
118 118 seen[rev] = parents
119 119 for parent in parents:
120 120 if parent >= minroot and parent not in seen:
121 121 dovisit(parent)
122 122 if not reachable:
123 123 return baseset()
124 124 if not includepath:
125 125 return reachable
126 126 for rev in sorted(seen):
127 127 for parent in seen[rev]:
128 128 if parent in reachable:
129 129 reached(rev)
130 130 return reachable
131 131
132 132 def reachableroots(repo, roots, heads, includepath=False):
133 133 """return (heads(::<roots> and ::<heads>))
134 134
135 135 If includepath is True, return (<roots>::<heads>)."""
136 136 if not roots:
137 137 return baseset()
138 138 minroot = roots.min()
139 139 roots = list(roots)
140 140 heads = list(heads)
141 141 try:
142 142 revs = repo.changelog.reachableroots(minroot, heads, roots, includepath)
143 143 except AttributeError:
144 144 revs = _reachablerootspure(repo, minroot, roots, heads, includepath)
145 145 revs = baseset(revs)
146 146 revs.sort()
147 147 return revs
148 148
149 149 elements = {
150 150 # token-type: binding-strength, primary, prefix, infix, suffix
151 151 "(": (21, None, ("group", 1, ")"), ("func", 1, ")"), None),
152 152 "##": (20, None, None, ("_concat", 20), None),
153 153 "~": (18, None, None, ("ancestor", 18), None),
154 154 "^": (18, None, None, ("parent", 18), "parentpost"),
155 155 "-": (5, None, ("negate", 19), ("minus", 5), None),
156 156 "::": (17, None, ("dagrangepre", 17), ("dagrange", 17), "dagrangepost"),
157 157 "..": (17, None, ("dagrangepre", 17), ("dagrange", 17), "dagrangepost"),
158 158 ":": (15, "rangeall", ("rangepre", 15), ("range", 15), "rangepost"),
159 159 "not": (10, None, ("not", 10), None, None),
160 160 "!": (10, None, ("not", 10), None, None),
161 161 "and": (5, None, None, ("and", 5), None),
162 162 "&": (5, None, None, ("and", 5), None),
163 163 "%": (5, None, None, ("only", 5), "onlypost"),
164 164 "or": (4, None, None, ("or", 4), None),
165 165 "|": (4, None, None, ("or", 4), None),
166 166 "+": (4, None, None, ("or", 4), None),
167 167 "=": (3, None, None, ("keyvalue", 3), None),
168 168 ",": (2, None, None, ("list", 2), None),
169 169 ")": (0, None, None, None, None),
170 170 "symbol": (0, "symbol", None, None, None),
171 171 "string": (0, "string", None, None, None),
172 172 "end": (0, None, None, None, None),
173 173 }
174 174
175 175 keywords = set(['and', 'or', 'not'])
176 176
177 177 # default set of valid characters for the initial letter of symbols
178 178 _syminitletters = set(
179 179 string.ascii_letters +
180 180 string.digits + pycompat.sysstr('._@')) | set(map(chr, xrange(128, 256)))
181 181
182 182 # default set of valid characters for non-initial letters of symbols
183 183 _symletters = _syminitletters | set(pycompat.sysstr('-/'))
184 184
185 185 def tokenize(program, lookup=None, syminitletters=None, symletters=None):
186 186 '''
187 187 Parse a revset statement into a stream of tokens
188 188
189 189 ``syminitletters`` is the set of valid characters for the initial
190 190 letter of symbols.
191 191
192 192 By default, character ``c`` is recognized as valid for initial
193 193 letter of symbols, if ``c.isalnum() or c in '._@' or ord(c) > 127``.
194 194
195 195 ``symletters`` is the set of valid characters for non-initial
196 196 letters of symbols.
197 197
198 198 By default, character ``c`` is recognized as valid for non-initial
199 199 letters of symbols, if ``c.isalnum() or c in '-._/@' or ord(c) > 127``.
200 200
201 201 Check that @ is a valid unquoted token character (issue3686):
202 202 >>> list(tokenize("@::"))
203 203 [('symbol', '@', 0), ('::', None, 1), ('end', None, 3)]
204 204
205 205 '''
206 206 if syminitletters is None:
207 207 syminitletters = _syminitletters
208 208 if symletters is None:
209 209 symletters = _symletters
210 210
211 211 if program and lookup:
212 212 # attempt to parse old-style ranges first to deal with
213 213 # things like old-tag which contain query metacharacters
214 214 parts = program.split(':', 1)
215 215 if all(lookup(sym) for sym in parts if sym):
216 216 if parts[0]:
217 217 yield ('symbol', parts[0], 0)
218 218 if len(parts) > 1:
219 219 s = len(parts[0])
220 220 yield (':', None, s)
221 221 if parts[1]:
222 222 yield ('symbol', parts[1], s + 1)
223 223 yield ('end', None, len(program))
224 224 return
225 225
226 226 pos, l = 0, len(program)
227 227 while pos < l:
228 228 c = program[pos]
229 229 if c.isspace(): # skip inter-token whitespace
230 230 pass
231 231 elif c == ':' and program[pos:pos + 2] == '::': # look ahead carefully
232 232 yield ('::', None, pos)
233 233 pos += 1 # skip ahead
234 234 elif c == '.' and program[pos:pos + 2] == '..': # look ahead carefully
235 235 yield ('..', None, pos)
236 236 pos += 1 # skip ahead
237 237 elif c == '#' and program[pos:pos + 2] == '##': # look ahead carefully
238 238 yield ('##', None, pos)
239 239 pos += 1 # skip ahead
240 240 elif c in "():=,-|&+!~^%": # handle simple operators
241 241 yield (c, None, pos)
242 242 elif (c in '"\'' or c == 'r' and
243 243 program[pos:pos + 2] in ("r'", 'r"')): # handle quoted strings
244 244 if c == 'r':
245 245 pos += 1
246 246 c = program[pos]
247 247 decode = lambda x: x
248 248 else:
249 249 decode = parser.unescapestr
250 250 pos += 1
251 251 s = pos
252 252 while pos < l: # find closing quote
253 253 d = program[pos]
254 254 if d == '\\': # skip over escaped characters
255 255 pos += 2
256 256 continue
257 257 if d == c:
258 258 yield ('string', decode(program[s:pos]), s)
259 259 break
260 260 pos += 1
261 261 else:
262 262 raise error.ParseError(_("unterminated string"), s)
263 263 # gather up a symbol/keyword
264 264 elif c in syminitletters:
265 265 s = pos
266 266 pos += 1
267 267 while pos < l: # find end of symbol
268 268 d = program[pos]
269 269 if d not in symletters:
270 270 break
271 271 if d == '.' and program[pos - 1] == '.': # special case for ..
272 272 pos -= 1
273 273 break
274 274 pos += 1
275 275 sym = program[s:pos]
276 276 if sym in keywords: # operator keywords
277 277 yield (sym, None, s)
278 278 elif '-' in sym:
279 279 # some jerk gave us foo-bar-baz, try to check if it's a symbol
280 280 if lookup and lookup(sym):
281 281 # looks like a real symbol
282 282 yield ('symbol', sym, s)
283 283 else:
284 284 # looks like an expression
285 285 parts = sym.split('-')
286 286 for p in parts[:-1]:
287 287 if p: # possible consecutive -
288 288 yield ('symbol', p, s)
289 289 s += len(p)
290 290 yield ('-', None, pos)
291 291 s += 1
292 292 if parts[-1]: # possible trailing -
293 293 yield ('symbol', parts[-1], s)
294 294 else:
295 295 yield ('symbol', sym, s)
296 296 pos -= 1
297 297 else:
298 298 raise error.ParseError(_("syntax error in revset '%s'") %
299 299 program, pos)
300 300 pos += 1
301 301 yield ('end', None, pos)
302 302
303 303 # helpers
304 304
305 305 def getsymbol(x):
306 306 if x and x[0] == 'symbol':
307 307 return x[1]
308 308 raise error.ParseError(_('not a symbol'))
309 309
310 310 def getstring(x, err):
311 311 if x and (x[0] == 'string' or x[0] == 'symbol'):
312 312 return x[1]
313 313 raise error.ParseError(err)
314 314
315 315 def getlist(x):
316 316 if not x:
317 317 return []
318 318 if x[0] == 'list':
319 319 return list(x[1:])
320 320 return [x]
321 321
322 322 def getargs(x, min, max, err):
323 323 l = getlist(x)
324 324 if len(l) < min or (max >= 0 and len(l) > max):
325 325 raise error.ParseError(err)
326 326 return l
327 327
328 328 def getargsdict(x, funcname, keys):
329 329 return parser.buildargsdict(getlist(x), funcname, keys.split(),
330 330 keyvaluenode='keyvalue', keynode='symbol')
331 331
332 332 def getset(repo, subset, x):
333 333 if not x:
334 334 raise error.ParseError(_("missing argument"))
335 335 s = methods[x[0]](repo, subset, *x[1:])
336 336 if util.safehasattr(s, 'isascending'):
337 337 return s
338 338 # else case should not happen, because all non-func are internal,
339 339 # ignoring for now.
340 340 if x[0] == 'func' and x[1][0] == 'symbol' and x[1][1] in symbols:
341 341 repo.ui.deprecwarn('revset "%s" uses list instead of smartset'
342 342 % x[1][1],
343 343 '3.9')
344 344 return baseset(s)
345 345
346 346 def _getrevsource(repo, r):
347 347 extra = repo[r].extra()
348 348 for label in ('source', 'transplant_source', 'rebase_source'):
349 349 if label in extra:
350 350 try:
351 351 return repo[extra[label]].rev()
352 352 except error.RepoLookupError:
353 353 pass
354 354 return None
355 355
356 356 # operator methods
357 357
358 358 def stringset(repo, subset, x):
359 359 x = repo[x].rev()
360 360 if (x in subset
361 361 or x == node.nullrev and isinstance(subset, fullreposet)):
362 362 return baseset([x])
363 363 return baseset()
364 364
365 365 def rangeset(repo, subset, x, y, order):
366 366 m = getset(repo, fullreposet(repo), x)
367 367 n = getset(repo, fullreposet(repo), y)
368 368
369 369 if not m or not n:
370 370 return baseset()
371 371 return _makerangeset(repo, subset, m.first(), n.last(), order)
372 372
373 373 def rangepre(repo, subset, y, order):
374 374 # ':y' can't be rewritten to '0:y' since '0' may be hidden
375 375 n = getset(repo, fullreposet(repo), y)
376 376 if not n:
377 377 return baseset()
378 378 return _makerangeset(repo, subset, 0, n.last(), order)
379 379
380 380 def _makerangeset(repo, subset, m, n, order):
381 381 if m == n:
382 382 r = baseset([m])
383 383 elif n == node.wdirrev:
384 384 r = spanset(repo, m, len(repo)) + baseset([n])
385 385 elif m == node.wdirrev:
386 386 r = baseset([m]) + spanset(repo, len(repo) - 1, n - 1)
387 387 elif m < n:
388 388 r = spanset(repo, m, n + 1)
389 389 else:
390 390 r = spanset(repo, m, n - 1)
391 391
392 392 if order == defineorder:
393 393 return r & subset
394 394 else:
395 395 # carrying the sorting over when possible would be more efficient
396 396 return subset & r
397 397
398 398 def dagrange(repo, subset, x, y, order):
399 399 r = fullreposet(repo)
400 400 xs = reachableroots(repo, getset(repo, r, x), getset(repo, r, y),
401 401 includepath=True)
402 402 return subset & xs
403 403
404 404 def andset(repo, subset, x, y, order):
405 405 return getset(repo, getset(repo, subset, x), y)
406 406
407 407 def differenceset(repo, subset, x, y, order):
408 408 return getset(repo, subset, x) - getset(repo, subset, y)
409 409
410 410 def _orsetlist(repo, subset, xs):
411 411 assert xs
412 412 if len(xs) == 1:
413 413 return getset(repo, subset, xs[0])
414 414 p = len(xs) // 2
415 415 a = _orsetlist(repo, subset, xs[:p])
416 416 b = _orsetlist(repo, subset, xs[p:])
417 417 return a + b
418 418
419 419 def orset(repo, subset, x, order):
420 420 xs = getlist(x)
421 421 if order == followorder:
422 422 # slow path to take the subset order
423 423 return subset & _orsetlist(repo, fullreposet(repo), xs)
424 424 else:
425 425 return _orsetlist(repo, subset, xs)
426 426
427 427 def notset(repo, subset, x, order):
428 428 return subset - getset(repo, subset, x)
429 429
430 430 def listset(repo, subset, *xs):
431 431 raise error.ParseError(_("can't use a list in this context"),
432 432 hint=_('see hg help "revsets.x or y"'))
433 433
434 434 def keyvaluepair(repo, subset, k, v):
435 435 raise error.ParseError(_("can't use a key-value pair in this context"))
436 436
437 437 def func(repo, subset, a, b, order):
438 438 f = getsymbol(a)
439 439 if f in symbols:
440 440 func = symbols[f]
441 441 if getattr(func, '_takeorder', False):
442 442 return func(repo, subset, b, order)
443 443 return func(repo, subset, b)
444 444
445 445 keep = lambda fn: getattr(fn, '__doc__', None) is not None
446 446
447 447 syms = [s for (s, fn) in symbols.items() if keep(fn)]
448 448 raise error.UnknownIdentifier(f, syms)
449 449
450 450 # functions
451 451
452 452 # symbols are callables like:
453 453 # fn(repo, subset, x)
454 454 # with:
455 455 # repo - current repository instance
456 456 # subset - of revisions to be examined
457 457 # x - argument in tree form
458 458 symbols = {}
459 459
460 460 # symbols which can't be used for a DoS attack for any given input
461 461 # (e.g. those which accept regexes as plain strings shouldn't be included)
462 462 # functions that just return a lot of changesets (like all) don't count here
463 463 safesymbols = set()
464 464
465 465 predicate = registrar.revsetpredicate()
466 466
467 467 @predicate('_destupdate')
468 468 def _destupdate(repo, subset, x):
469 469 # experimental revset for update destination
470 470 args = getargsdict(x, 'limit', 'clean check')
471 471 return subset & baseset([destutil.destupdate(repo, **args)[0]])
472 472
473 473 @predicate('_destmerge')
474 474 def _destmerge(repo, subset, x):
475 475 # experimental revset for merge destination
476 476 sourceset = None
477 477 if x is not None:
478 478 sourceset = getset(repo, fullreposet(repo), x)
479 479 return subset & baseset([destutil.destmerge(repo, sourceset=sourceset)])
480 480
481 481 @predicate('adds(pattern)', safe=True)
482 482 def adds(repo, subset, x):
483 483 """Changesets that add a file matching pattern.
484 484
485 485 The pattern without explicit kind like ``glob:`` is expected to be
486 486 relative to the current directory and match against a file or a
487 487 directory.
488 488 """
489 489 # i18n: "adds" is a keyword
490 490 pat = getstring(x, _("adds requires a pattern"))
491 491 return checkstatus(repo, subset, pat, 1)
492 492
493 493 @predicate('ancestor(*changeset)', safe=True)
494 494 def ancestor(repo, subset, x):
495 495 """A greatest common ancestor of the changesets.
496 496
497 497 Accepts 0 or more changesets.
498 498 Will return empty list when passed no args.
499 499 Greatest common ancestor of a single changeset is that changeset.
500 500 """
501 501 # i18n: "ancestor" is a keyword
502 502 l = getlist(x)
503 503 rl = fullreposet(repo)
504 504 anc = None
505 505
506 506 # (getset(repo, rl, i) for i in l) generates a list of lists
507 507 for revs in (getset(repo, rl, i) for i in l):
508 508 for r in revs:
509 509 if anc is None:
510 510 anc = repo[r]
511 511 else:
512 512 anc = anc.ancestor(repo[r])
513 513
514 514 if anc is not None and anc.rev() in subset:
515 515 return baseset([anc.rev()])
516 516 return baseset()
517 517
518 518 def _ancestors(repo, subset, x, followfirst=False):
519 519 heads = getset(repo, fullreposet(repo), x)
520 520 if not heads:
521 521 return baseset()
522 522 s = _revancestors(repo, heads, followfirst)
523 523 return subset & s
524 524
525 525 @predicate('ancestors(set)', safe=True)
526 526 def ancestors(repo, subset, x):
527 527 """Changesets that are ancestors of a changeset in set.
528 528 """
529 529 return _ancestors(repo, subset, x)
530 530
531 531 @predicate('_firstancestors', safe=True)
532 532 def _firstancestors(repo, subset, x):
533 533 # ``_firstancestors(set)``
534 534 # Like ``ancestors(set)`` but follows only the first parents.
535 535 return _ancestors(repo, subset, x, followfirst=True)
536 536
537 537 def ancestorspec(repo, subset, x, n, order):
538 538 """``set~n``
539 539 Changesets that are the Nth ancestor (first parents only) of a changeset
540 540 in set.
541 541 """
542 542 try:
543 543 n = int(n[1])
544 544 except (TypeError, ValueError):
545 545 raise error.ParseError(_("~ expects a number"))
546 546 ps = set()
547 547 cl = repo.changelog
548 548 for r in getset(repo, fullreposet(repo), x):
549 549 for i in range(n):
550 550 r = cl.parentrevs(r)[0]
551 551 ps.add(r)
552 552 return subset & ps
553 553
554 554 @predicate('author(string)', safe=True)
555 555 def author(repo, subset, x):
556 556 """Alias for ``user(string)``.
557 557 """
558 558 # i18n: "author" is a keyword
559 559 n = encoding.lower(getstring(x, _("author requires a string")))
560 560 kind, pattern, matcher = _substringmatcher(n)
561 561 return subset.filter(lambda x: matcher(encoding.lower(repo[x].user())),
562 562 condrepr=('<user %r>', n))
563 563
564 564 @predicate('bisect(string)', safe=True)
565 565 def bisect(repo, subset, x):
566 566 """Changesets marked in the specified bisect status:
567 567
568 568 - ``good``, ``bad``, ``skip``: csets explicitly marked as good/bad/skip
569 569 - ``goods``, ``bads`` : csets topologically good/bad
570 570 - ``range`` : csets taking part in the bisection
571 571 - ``pruned`` : csets that are goods, bads or skipped
572 572 - ``untested`` : csets whose fate is yet unknown
573 573 - ``ignored`` : csets ignored due to DAG topology
574 574 - ``current`` : the cset currently being bisected
575 575 """
576 576 # i18n: "bisect" is a keyword
577 577 status = getstring(x, _("bisect requires a string")).lower()
578 578 state = set(hbisect.get(repo, status))
579 579 return subset & state
580 580
581 581 # Backward-compatibility
582 582 # - no help entry so that we do not advertise it any more
583 583 @predicate('bisected', safe=True)
584 584 def bisected(repo, subset, x):
585 585 return bisect(repo, subset, x)
586 586
587 587 @predicate('bookmark([name])', safe=True)
588 588 def bookmark(repo, subset, x):
589 589 """The named bookmark or all bookmarks.
590 590
591 591 If `name` starts with `re:`, the remainder of the name is treated as
592 592 a regular expression. To match a bookmark that actually starts with `re:`,
593 593 use the prefix `literal:`.
594 594 """
595 595 # i18n: "bookmark" is a keyword
596 596 args = getargs(x, 0, 1, _('bookmark takes one or no arguments'))
597 597 if args:
598 598 bm = getstring(args[0],
599 599 # i18n: "bookmark" is a keyword
600 600 _('the argument to bookmark must be a string'))
601 601 kind, pattern, matcher = util.stringmatcher(bm)
602 602 bms = set()
603 603 if kind == 'literal':
604 604 bmrev = repo._bookmarks.get(pattern, None)
605 605 if not bmrev:
606 606 raise error.RepoLookupError(_("bookmark '%s' does not exist")
607 607 % pattern)
608 608 bms.add(repo[bmrev].rev())
609 609 else:
610 610 matchrevs = set()
611 611 for name, bmrev in repo._bookmarks.iteritems():
612 612 if matcher(name):
613 613 matchrevs.add(bmrev)
614 614 if not matchrevs:
615 615 raise error.RepoLookupError(_("no bookmarks exist"
616 616 " that match '%s'") % pattern)
617 617 for bmrev in matchrevs:
618 618 bms.add(repo[bmrev].rev())
619 619 else:
620 620 bms = set([repo[r].rev()
621 621 for r in repo._bookmarks.values()])
622 622 bms -= set([node.nullrev])
623 623 return subset & bms
624 624
625 625 @predicate('branch(string or set)', safe=True)
626 626 def branch(repo, subset, x):
627 627 """
628 628 All changesets belonging to the given branch or the branches of the given
629 629 changesets.
630 630
631 631 If `string` starts with `re:`, the remainder of the name is treated as
632 632 a regular expression. To match a branch that actually starts with `re:`,
633 633 use the prefix `literal:`.
634 634 """
635 635 getbi = repo.revbranchcache().branchinfo
636 636
637 637 try:
638 638 b = getstring(x, '')
639 639 except error.ParseError:
640 640 # not a string, but another revspec, e.g. tip()
641 641 pass
642 642 else:
643 643 kind, pattern, matcher = util.stringmatcher(b)
644 644 if kind == 'literal':
645 645 # note: falls through to the revspec case if no branch with
646 646 # this name exists and pattern kind is not specified explicitly
647 647 if pattern in repo.branchmap():
648 648 return subset.filter(lambda r: matcher(getbi(r)[0]),
649 649 condrepr=('<branch %r>', b))
650 650 if b.startswith('literal:'):
651 651 raise error.RepoLookupError(_("branch '%s' does not exist")
652 652 % pattern)
653 653 else:
654 654 return subset.filter(lambda r: matcher(getbi(r)[0]),
655 655 condrepr=('<branch %r>', b))
656 656
657 657 s = getset(repo, fullreposet(repo), x)
658 658 b = set()
659 659 for r in s:
660 660 b.add(getbi(r)[0])
661 661 c = s.__contains__
662 662 return subset.filter(lambda r: c(r) or getbi(r)[0] in b,
663 663 condrepr=lambda: '<branch %r>' % sorted(b))
664 664
665 665 @predicate('bumped()', safe=True)
666 666 def bumped(repo, subset, x):
667 667 """Mutable changesets marked as successors of public changesets.
668 668
669 669 Only non-public and non-obsolete changesets can be `bumped`.
670 670 """
671 671 # i18n: "bumped" is a keyword
672 672 getargs(x, 0, 0, _("bumped takes no arguments"))
673 673 bumped = obsmod.getrevs(repo, 'bumped')
674 674 return subset & bumped
675 675
676 676 @predicate('bundle()', safe=True)
677 677 def bundle(repo, subset, x):
678 678 """Changesets in the bundle.
679 679
680 680 Bundle must be specified by the -R option."""
681 681
682 682 try:
683 683 bundlerevs = repo.changelog.bundlerevs
684 684 except AttributeError:
685 685 raise error.Abort(_("no bundle provided - specify with -R"))
686 686 return subset & bundlerevs
687 687
688 688 def checkstatus(repo, subset, pat, field):
689 689 hasset = matchmod.patkind(pat) == 'set'
690 690
691 691 mcache = [None]
692 692 def matches(x):
693 693 c = repo[x]
694 694 if not mcache[0] or hasset:
695 695 mcache[0] = matchmod.match(repo.root, repo.getcwd(), [pat], ctx=c)
696 696 m = mcache[0]
697 697 fname = None
698 698 if not m.anypats() and len(m.files()) == 1:
699 699 fname = m.files()[0]
700 700 if fname is not None:
701 701 if fname not in c.files():
702 702 return False
703 703 else:
704 704 for f in c.files():
705 705 if m(f):
706 706 break
707 707 else:
708 708 return False
709 709 files = repo.status(c.p1().node(), c.node())[field]
710 710 if fname is not None:
711 711 if fname in files:
712 712 return True
713 713 else:
714 714 for f in files:
715 715 if m(f):
716 716 return True
717 717
718 718 return subset.filter(matches, condrepr=('<status[%r] %r>', field, pat))
719 719
720 720 def _children(repo, subset, parentset):
721 721 if not parentset:
722 722 return baseset()
723 723 cs = set()
724 724 pr = repo.changelog.parentrevs
725 725 minrev = parentset.min()
726 726 nullrev = node.nullrev
727 727 for r in subset:
728 728 if r <= minrev:
729 729 continue
730 730 p1, p2 = pr(r)
731 731 if p1 in parentset:
732 732 cs.add(r)
733 733 if p2 != nullrev and p2 in parentset:
734 734 cs.add(r)
735 735 return baseset(cs)
736 736
737 737 @predicate('children(set)', safe=True)
738 738 def children(repo, subset, x):
739 739 """Child changesets of changesets in set.
740 740 """
741 741 s = getset(repo, fullreposet(repo), x)
742 742 cs = _children(repo, subset, s)
743 743 return subset & cs
744 744
745 745 @predicate('closed()', safe=True)
746 746 def closed(repo, subset, x):
747 747 """Changeset is closed.
748 748 """
749 749 # i18n: "closed" is a keyword
750 750 getargs(x, 0, 0, _("closed takes no arguments"))
751 751 return subset.filter(lambda r: repo[r].closesbranch(),
752 752 condrepr='<branch closed>')
753 753
754 754 @predicate('contains(pattern)')
755 755 def contains(repo, subset, x):
756 756 """The revision's manifest contains a file matching pattern (but might not
757 757 modify it). See :hg:`help patterns` for information about file patterns.
758 758
759 759 The pattern without explicit kind like ``glob:`` is expected to be
760 760 relative to the current directory and match against a file exactly
761 761 for efficiency.
762 762 """
763 763 # i18n: "contains" is a keyword
764 764 pat = getstring(x, _("contains requires a pattern"))
765 765
766 766 def matches(x):
767 767 if not matchmod.patkind(pat):
768 768 pats = pathutil.canonpath(repo.root, repo.getcwd(), pat)
769 769 if pats in repo[x]:
770 770 return True
771 771 else:
772 772 c = repo[x]
773 773 m = matchmod.match(repo.root, repo.getcwd(), [pat], ctx=c)
774 774 for f in c.manifest():
775 775 if m(f):
776 776 return True
777 777 return False
778 778
779 779 return subset.filter(matches, condrepr=('<contains %r>', pat))
780 780
781 781 @predicate('converted([id])', safe=True)
782 782 def converted(repo, subset, x):
783 783 """Changesets converted from the given identifier in the old repository if
784 784 present, or all converted changesets if no identifier is specified.
785 785 """
786 786
787 787 # There is exactly no chance of resolving the revision, so do a simple
788 788 # string compare and hope for the best
789 789
790 790 rev = None
791 791 # i18n: "converted" is a keyword
792 792 l = getargs(x, 0, 1, _('converted takes one or no arguments'))
793 793 if l:
794 794 # i18n: "converted" is a keyword
795 795 rev = getstring(l[0], _('converted requires a revision'))
796 796
797 797 def _matchvalue(r):
798 798 source = repo[r].extra().get('convert_revision', None)
799 799 return source is not None and (rev is None or source.startswith(rev))
800 800
801 801 return subset.filter(lambda r: _matchvalue(r),
802 802 condrepr=('<converted %r>', rev))
803 803
804 804 @predicate('date(interval)', safe=True)
805 805 def date(repo, subset, x):
806 806 """Changesets within the interval, see :hg:`help dates`.
807 807 """
808 808 # i18n: "date" is a keyword
809 809 ds = getstring(x, _("date requires a string"))
810 810 dm = util.matchdate(ds)
811 811 return subset.filter(lambda x: dm(repo[x].date()[0]),
812 812 condrepr=('<date %r>', ds))
813 813
814 814 @predicate('desc(string)', safe=True)
815 815 def desc(repo, subset, x):
816 816 """Search commit message for string. The match is case-insensitive.
817 817 """
818 818 # i18n: "desc" is a keyword
819 819 ds = encoding.lower(getstring(x, _("desc requires a string")))
820 820
821 821 def matches(x):
822 822 c = repo[x]
823 823 return ds in encoding.lower(c.description())
824 824
825 825 return subset.filter(matches, condrepr=('<desc %r>', ds))
826 826
827 827 def _descendants(repo, subset, x, followfirst=False):
828 828 roots = getset(repo, fullreposet(repo), x)
829 829 if not roots:
830 830 return baseset()
831 831 s = _revdescendants(repo, roots, followfirst)
832 832
833 833 # Both sets need to be ascending in order to lazily return the union
834 834 # in the correct order.
835 835 base = subset & roots
836 836 desc = subset & s
837 837 result = base + desc
838 838 if subset.isascending():
839 839 result.sort()
840 840 elif subset.isdescending():
841 841 result.sort(reverse=True)
842 842 else:
843 843 result = subset & result
844 844 return result
845 845
846 846 @predicate('descendants(set)', safe=True)
847 847 def descendants(repo, subset, x):
848 848 """Changesets which are descendants of changesets in set.
849 849 """
850 850 return _descendants(repo, subset, x)
851 851
852 852 @predicate('_firstdescendants', safe=True)
853 853 def _firstdescendants(repo, subset, x):
854 854 # ``_firstdescendants(set)``
855 855 # Like ``descendants(set)`` but follows only the first parents.
856 856 return _descendants(repo, subset, x, followfirst=True)
857 857
858 858 @predicate('destination([set])', safe=True)
859 859 def destination(repo, subset, x):
860 860 """Changesets that were created by a graft, transplant or rebase operation,
861 861 with the given revisions specified as the source. Omitting the optional set
862 862 is the same as passing all().
863 863 """
864 864 if x is not None:
865 865 sources = getset(repo, fullreposet(repo), x)
866 866 else:
867 867 sources = fullreposet(repo)
868 868
869 869 dests = set()
870 870
871 871 # subset contains all of the possible destinations that can be returned, so
872 872 # iterate over them and see if their source(s) were provided in the arg set.
873 873 # Even if the immediate src of r is not in the arg set, src's source (or
874 874 # further back) may be. Scanning back further than the immediate src allows
875 875 # transitive transplants and rebases to yield the same results as transitive
876 876 # grafts.
877 877 for r in subset:
878 878 src = _getrevsource(repo, r)
879 879 lineage = None
880 880
881 881 while src is not None:
882 882 if lineage is None:
883 883 lineage = list()
884 884
885 885 lineage.append(r)
886 886
887 887 # The visited lineage is a match if the current source is in the arg
888 888 # set. Since every candidate dest is visited by way of iterating
889 889 # subset, any dests further back in the lineage will be tested by a
890 890 # different iteration over subset. Likewise, if the src was already
891 891 # selected, the current lineage can be selected without going back
892 892 # further.
893 893 if src in sources or src in dests:
894 894 dests.update(lineage)
895 895 break
896 896
897 897 r = src
898 898 src = _getrevsource(repo, r)
899 899
900 900 return subset.filter(dests.__contains__,
901 901 condrepr=lambda: '<destination %r>' % sorted(dests))
902 902
903 903 @predicate('divergent()', safe=True)
904 904 def divergent(repo, subset, x):
905 905 """
906 906 Final successors of changesets with an alternative set of final successors.
907 907 """
908 908 # i18n: "divergent" is a keyword
909 909 getargs(x, 0, 0, _("divergent takes no arguments"))
910 910 divergent = obsmod.getrevs(repo, 'divergent')
911 911 return subset & divergent
912 912
913 913 @predicate('extinct()', safe=True)
914 914 def extinct(repo, subset, x):
915 915 """Obsolete changesets with obsolete descendants only.
916 916 """
917 917 # i18n: "extinct" is a keyword
918 918 getargs(x, 0, 0, _("extinct takes no arguments"))
919 919 extincts = obsmod.getrevs(repo, 'extinct')
920 920 return subset & extincts
921 921
922 922 @predicate('extra(label, [value])', safe=True)
923 923 def extra(repo, subset, x):
924 924 """Changesets with the given label in the extra metadata, with the given
925 925 optional value.
926 926
927 927 If `value` starts with `re:`, the remainder of the value is treated as
928 928 a regular expression. To match a value that actually starts with `re:`,
929 929 use the prefix `literal:`.
930 930 """
931 931 args = getargsdict(x, 'extra', 'label value')
932 932 if 'label' not in args:
933 933 # i18n: "extra" is a keyword
934 934 raise error.ParseError(_('extra takes at least 1 argument'))
935 935 # i18n: "extra" is a keyword
936 936 label = getstring(args['label'], _('first argument to extra must be '
937 937 'a string'))
938 938 value = None
939 939
940 940 if 'value' in args:
941 941 # i18n: "extra" is a keyword
942 942 value = getstring(args['value'], _('second argument to extra must be '
943 943 'a string'))
944 944 kind, value, matcher = util.stringmatcher(value)
945 945
946 946 def _matchvalue(r):
947 947 extra = repo[r].extra()
948 948 return label in extra and (value is None or matcher(extra[label]))
949 949
950 950 return subset.filter(lambda r: _matchvalue(r),
951 951 condrepr=('<extra[%r] %r>', label, value))
952 952
953 953 @predicate('filelog(pattern)', safe=True)
954 954 def filelog(repo, subset, x):
955 955 """Changesets connected to the specified filelog.
956 956
957 957 For performance reasons, visits only revisions mentioned in the file-level
958 958 filelog, rather than filtering through all changesets (much faster, but
959 959 doesn't include deletes or duplicate changes). For a slower, more accurate
960 960 result, use ``file()``.
961 961
962 962 The pattern without explicit kind like ``glob:`` is expected to be
963 963 relative to the current directory and match against a file exactly
964 964 for efficiency.
965 965
966 966 If some linkrev points to revisions filtered by the current repoview, we'll
967 967 work around it to return a non-filtered value.
968 968 """
969 969
970 970 # i18n: "filelog" is a keyword
971 971 pat = getstring(x, _("filelog requires a pattern"))
972 972 s = set()
973 973 cl = repo.changelog
974 974
975 975 if not matchmod.patkind(pat):
976 976 f = pathutil.canonpath(repo.root, repo.getcwd(), pat)
977 977 files = [f]
978 978 else:
979 979 m = matchmod.match(repo.root, repo.getcwd(), [pat], ctx=repo[None])
980 980 files = (f for f in repo[None] if m(f))
981 981
982 982 for f in files:
983 983 fl = repo.file(f)
984 984 known = {}
985 985 scanpos = 0
986 986 for fr in list(fl):
987 987 fn = fl.node(fr)
988 988 if fn in known:
989 989 s.add(known[fn])
990 990 continue
991 991
992 992 lr = fl.linkrev(fr)
993 993 if lr in cl:
994 994 s.add(lr)
995 995 elif scanpos is not None:
996 996 # lowest matching changeset is filtered, scan further
997 997 # ahead in changelog
998 998 start = max(lr, scanpos) + 1
999 999 scanpos = None
1000 1000 for r in cl.revs(start):
1001 1001 # minimize parsing of non-matching entries
1002 1002 if f in cl.revision(r) and f in cl.readfiles(r):
1003 1003 try:
1004 1004 # try to use manifest delta fastpath
1005 1005 n = repo[r].filenode(f)
1006 1006 if n not in known:
1007 1007 if n == fn:
1008 1008 s.add(r)
1009 1009 scanpos = r
1010 1010 break
1011 1011 else:
1012 1012 known[n] = r
1013 1013 except error.ManifestLookupError:
1014 1014 # deletion in changelog
1015 1015 continue
1016 1016
1017 1017 return subset & s
1018 1018
1019 1019 @predicate('first(set, [n])', safe=True)
1020 1020 def first(repo, subset, x):
1021 1021 """An alias for limit().
1022 1022 """
1023 1023 return limit(repo, subset, x)
1024 1024
1025 1025 def _follow(repo, subset, x, name, followfirst=False):
1026 1026 l = getargs(x, 0, 2, _("%s takes no arguments or a pattern "
1027 1027 "and an optional revset") % name)
1028 1028 c = repo['.']
1029 1029 if l:
1030 1030 x = getstring(l[0], _("%s expected a pattern") % name)
1031 1031 rev = None
1032 1032 if len(l) >= 2:
1033 1033 revs = getset(repo, fullreposet(repo), l[1])
1034 1034 if len(revs) != 1:
1035 1035 raise error.RepoLookupError(
1036 1036 _("%s expected one starting revision") % name)
1037 1037 rev = revs.last()
1038 1038 c = repo[rev]
1039 1039 matcher = matchmod.match(repo.root, repo.getcwd(), [x],
1040 1040 ctx=repo[rev], default='path')
1041 1041
1042 1042 files = c.manifest().walk(matcher)
1043 1043
1044 1044 s = set()
1045 1045 for fname in files:
1046 1046 fctx = c[fname]
1047 1047 s = s.union(set(c.rev() for c in fctx.ancestors(followfirst)))
1048 1048 # include the revision responsible for the most recent version
1049 1049 s.add(fctx.introrev())
1050 1050 else:
1051 1051 s = _revancestors(repo, baseset([c.rev()]), followfirst)
1052 1052
1053 1053 return subset & s
1054 1054
1055 1055 @predicate('follow([pattern[, startrev]])', safe=True)
1056 1056 def follow(repo, subset, x):
1057 1057 """
1058 1058 An alias for ``::.`` (ancestors of the working directory's first parent).
1059 1059 If pattern is specified, the histories of files matching given
1060 1060 pattern in the revision given by startrev are followed, including copies.
1061 1061 """
1062 1062 return _follow(repo, subset, x, 'follow')
1063 1063
1064 1064 @predicate('_followfirst', safe=True)
1065 1065 def _followfirst(repo, subset, x):
1066 1066 # ``followfirst([pattern[, startrev]])``
1067 1067 # Like ``follow([pattern[, startrev]])`` but follows only the first parent
1068 1068 # of every revisions or files revisions.
1069 1069 return _follow(repo, subset, x, '_followfirst', followfirst=True)
1070 1070
1071 1071 @predicate('all()', safe=True)
1072 1072 def getall(repo, subset, x):
1073 1073 """All changesets, the same as ``0:tip``.
1074 1074 """
1075 1075 # i18n: "all" is a keyword
1076 1076 getargs(x, 0, 0, _("all takes no arguments"))
1077 1077 return subset & spanset(repo) # drop "null" if any
1078 1078
1079 1079 @predicate('grep(regex)')
1080 1080 def grep(repo, subset, x):
1081 1081 """Like ``keyword(string)`` but accepts a regex. Use ``grep(r'...')``
1082 1082 to ensure special escape characters are handled correctly. Unlike
1083 1083 ``keyword(string)``, the match is case-sensitive.
1084 1084 """
1085 1085 try:
1086 1086 # i18n: "grep" is a keyword
1087 1087 gr = re.compile(getstring(x, _("grep requires a string")))
1088 1088 except re.error as e:
1089 1089 raise error.ParseError(_('invalid match pattern: %s') % e)
1090 1090
1091 1091 def matches(x):
1092 1092 c = repo[x]
1093 1093 for e in c.files() + [c.user(), c.description()]:
1094 1094 if gr.search(e):
1095 1095 return True
1096 1096 return False
1097 1097
1098 1098 return subset.filter(matches, condrepr=('<grep %r>', gr.pattern))
1099 1099
1100 1100 @predicate('_matchfiles', safe=True)
1101 1101 def _matchfiles(repo, subset, x):
1102 1102 # _matchfiles takes a revset list of prefixed arguments:
1103 1103 #
1104 1104 # [p:foo, i:bar, x:baz]
1105 1105 #
1106 1106 # builds a match object from them and filters subset. Allowed
1107 1107 # prefixes are 'p:' for regular patterns, 'i:' for include
1108 1108 # patterns and 'x:' for exclude patterns. Use 'r:' prefix to pass
1109 1109 # a revision identifier, or the empty string to reference the
1110 1110 # working directory, from which the match object is
1111 1111 # initialized. Use 'd:' to set the default matching mode, default
1112 1112 # to 'glob'. At most one 'r:' and 'd:' argument can be passed.
1113 1113
1114 1114 l = getargs(x, 1, -1, "_matchfiles requires at least one argument")
1115 1115 pats, inc, exc = [], [], []
1116 1116 rev, default = None, None
1117 1117 for arg in l:
1118 1118 s = getstring(arg, "_matchfiles requires string arguments")
1119 1119 prefix, value = s[:2], s[2:]
1120 1120 if prefix == 'p:':
1121 1121 pats.append(value)
1122 1122 elif prefix == 'i:':
1123 1123 inc.append(value)
1124 1124 elif prefix == 'x:':
1125 1125 exc.append(value)
1126 1126 elif prefix == 'r:':
1127 1127 if rev is not None:
1128 1128 raise error.ParseError('_matchfiles expected at most one '
1129 1129 'revision')
1130 1130 if value != '': # empty means working directory; leave rev as None
1131 1131 rev = value
1132 1132 elif prefix == 'd:':
1133 1133 if default is not None:
1134 1134 raise error.ParseError('_matchfiles expected at most one '
1135 1135 'default mode')
1136 1136 default = value
1137 1137 else:
1138 1138 raise error.ParseError('invalid _matchfiles prefix: %s' % prefix)
1139 1139 if not default:
1140 1140 default = 'glob'
1141 1141
1142 1142 m = matchmod.match(repo.root, repo.getcwd(), pats, include=inc,
1143 1143 exclude=exc, ctx=repo[rev], default=default)
1144 1144
1145 1145 # This directly read the changelog data as creating changectx for all
1146 1146 # revisions is quite expensive.
1147 1147 getfiles = repo.changelog.readfiles
1148 1148 wdirrev = node.wdirrev
1149 1149 def matches(x):
1150 1150 if x == wdirrev:
1151 1151 files = repo[x].files()
1152 1152 else:
1153 1153 files = getfiles(x)
1154 1154 for f in files:
1155 1155 if m(f):
1156 1156 return True
1157 1157 return False
1158 1158
1159 1159 return subset.filter(matches,
1160 1160 condrepr=('<matchfiles patterns=%r, include=%r '
1161 1161 'exclude=%r, default=%r, rev=%r>',
1162 1162 pats, inc, exc, default, rev))
1163 1163
1164 1164 @predicate('file(pattern)', safe=True)
1165 1165 def hasfile(repo, subset, x):
1166 1166 """Changesets affecting files matched by pattern.
1167 1167
1168 1168 For a faster but less accurate result, consider using ``filelog()``
1169 1169 instead.
1170 1170
1171 1171 This predicate uses ``glob:`` as the default kind of pattern.
1172 1172 """
1173 1173 # i18n: "file" is a keyword
1174 1174 pat = getstring(x, _("file requires a pattern"))
1175 1175 return _matchfiles(repo, subset, ('string', 'p:' + pat))
1176 1176
1177 1177 @predicate('head()', safe=True)
1178 1178 def head(repo, subset, x):
1179 1179 """Changeset is a named branch head.
1180 1180 """
1181 1181 # i18n: "head" is a keyword
1182 1182 getargs(x, 0, 0, _("head takes no arguments"))
1183 1183 hs = set()
1184 1184 cl = repo.changelog
1185 1185 for ls in repo.branchmap().itervalues():
1186 1186 hs.update(cl.rev(h) for h in ls)
1187 1187 return subset & baseset(hs)
1188 1188
1189 1189 @predicate('heads(set)', safe=True)
1190 1190 def heads(repo, subset, x):
1191 1191 """Members of set with no children in set.
1192 1192 """
1193 1193 s = getset(repo, subset, x)
1194 1194 ps = parents(repo, subset, x)
1195 1195 return s - ps
1196 1196
1197 1197 @predicate('hidden()', safe=True)
1198 1198 def hidden(repo, subset, x):
1199 1199 """Hidden changesets.
1200 1200 """
1201 1201 # i18n: "hidden" is a keyword
1202 1202 getargs(x, 0, 0, _("hidden takes no arguments"))
1203 1203 hiddenrevs = repoview.filterrevs(repo, 'visible')
1204 1204 return subset & hiddenrevs
1205 1205
1206 1206 @predicate('keyword(string)', safe=True)
1207 1207 def keyword(repo, subset, x):
1208 1208 """Search commit message, user name, and names of changed files for
1209 1209 string. The match is case-insensitive.
1210 1210 """
1211 1211 # i18n: "keyword" is a keyword
1212 1212 kw = encoding.lower(getstring(x, _("keyword requires a string")))
1213 1213
1214 1214 def matches(r):
1215 1215 c = repo[r]
1216 1216 return any(kw in encoding.lower(t)
1217 1217 for t in c.files() + [c.user(), c.description()])
1218 1218
1219 1219 return subset.filter(matches, condrepr=('<keyword %r>', kw))
1220 1220
1221 1221 @predicate('limit(set[, n[, offset]])', safe=True)
1222 1222 def limit(repo, subset, x):
1223 1223 """First n members of set, defaulting to 1, starting from offset.
1224 1224 """
1225 1225 args = getargsdict(x, 'limit', 'set n offset')
1226 1226 if 'set' not in args:
1227 1227 # i18n: "limit" is a keyword
1228 1228 raise error.ParseError(_("limit requires one to three arguments"))
1229 1229 try:
1230 1230 lim, ofs = 1, 0
1231 1231 if 'n' in args:
1232 1232 # i18n: "limit" is a keyword
1233 1233 lim = int(getstring(args['n'], _("limit requires a number")))
1234 1234 if 'offset' in args:
1235 1235 # i18n: "limit" is a keyword
1236 1236 ofs = int(getstring(args['offset'], _("limit requires a number")))
1237 1237 if ofs < 0:
1238 1238 raise error.ParseError(_("negative offset"))
1239 1239 except (TypeError, ValueError):
1240 1240 # i18n: "limit" is a keyword
1241 1241 raise error.ParseError(_("limit expects a number"))
1242 1242 os = getset(repo, fullreposet(repo), args['set'])
1243 1243 result = []
1244 1244 it = iter(os)
1245 1245 for x in xrange(ofs):
1246 1246 y = next(it, None)
1247 1247 if y is None:
1248 1248 break
1249 1249 for x in xrange(lim):
1250 1250 y = next(it, None)
1251 1251 if y is None:
1252 1252 break
1253 1253 elif y in subset:
1254 1254 result.append(y)
1255 1255 return baseset(result, datarepr=('<limit n=%d, offset=%d, %r, %r>',
1256 1256 lim, ofs, subset, os))
1257 1257
1258 1258 @predicate('last(set, [n])', safe=True)
1259 1259 def last(repo, subset, x):
1260 1260 """Last n members of set, defaulting to 1.
1261 1261 """
1262 1262 # i18n: "last" is a keyword
1263 1263 l = getargs(x, 1, 2, _("last requires one or two arguments"))
1264 1264 try:
1265 1265 lim = 1
1266 1266 if len(l) == 2:
1267 1267 # i18n: "last" is a keyword
1268 1268 lim = int(getstring(l[1], _("last requires a number")))
1269 1269 except (TypeError, ValueError):
1270 1270 # i18n: "last" is a keyword
1271 1271 raise error.ParseError(_("last expects a number"))
1272 1272 os = getset(repo, fullreposet(repo), l[0])
1273 1273 os.reverse()
1274 1274 result = []
1275 1275 it = iter(os)
1276 1276 for x in xrange(lim):
1277 1277 y = next(it, None)
1278 1278 if y is None:
1279 1279 break
1280 1280 elif y in subset:
1281 1281 result.append(y)
1282 1282 return baseset(result, datarepr=('<last n=%d, %r, %r>', lim, subset, os))
1283 1283
1284 1284 @predicate('max(set)', safe=True)
1285 1285 def maxrev(repo, subset, x):
1286 1286 """Changeset with highest revision number in set.
1287 1287 """
1288 1288 os = getset(repo, fullreposet(repo), x)
1289 1289 try:
1290 1290 m = os.max()
1291 1291 if m in subset:
1292 1292 return baseset([m], datarepr=('<max %r, %r>', subset, os))
1293 1293 except ValueError:
1294 1294 # os.max() throws a ValueError when the collection is empty.
1295 1295 # Same as python's max().
1296 1296 pass
1297 1297 return baseset(datarepr=('<max %r, %r>', subset, os))
1298 1298
1299 1299 @predicate('merge()', safe=True)
1300 1300 def merge(repo, subset, x):
1301 1301 """Changeset is a merge changeset.
1302 1302 """
1303 1303 # i18n: "merge" is a keyword
1304 1304 getargs(x, 0, 0, _("merge takes no arguments"))
1305 1305 cl = repo.changelog
1306 1306 return subset.filter(lambda r: cl.parentrevs(r)[1] != -1,
1307 1307 condrepr='<merge>')
1308 1308
1309 1309 @predicate('branchpoint()', safe=True)
1310 1310 def branchpoint(repo, subset, x):
1311 1311 """Changesets with more than one child.
1312 1312 """
1313 1313 # i18n: "branchpoint" is a keyword
1314 1314 getargs(x, 0, 0, _("branchpoint takes no arguments"))
1315 1315 cl = repo.changelog
1316 1316 if not subset:
1317 1317 return baseset()
1318 1318 # XXX this should be 'parentset.min()' assuming 'parentset' is a smartset
1319 1319 # (and if it is not, it should.)
1320 1320 baserev = min(subset)
1321 1321 parentscount = [0]*(len(repo) - baserev)
1322 1322 for r in cl.revs(start=baserev + 1):
1323 1323 for p in cl.parentrevs(r):
1324 1324 if p >= baserev:
1325 1325 parentscount[p - baserev] += 1
1326 1326 return subset.filter(lambda r: parentscount[r - baserev] > 1,
1327 1327 condrepr='<branchpoint>')
1328 1328
1329 1329 @predicate('min(set)', safe=True)
1330 1330 def minrev(repo, subset, x):
1331 1331 """Changeset with lowest revision number in set.
1332 1332 """
1333 1333 os = getset(repo, fullreposet(repo), x)
1334 1334 try:
1335 1335 m = os.min()
1336 1336 if m in subset:
1337 1337 return baseset([m], datarepr=('<min %r, %r>', subset, os))
1338 1338 except ValueError:
1339 1339 # os.min() throws a ValueError when the collection is empty.
1340 1340 # Same as python's min().
1341 1341 pass
1342 1342 return baseset(datarepr=('<min %r, %r>', subset, os))
1343 1343
1344 1344 @predicate('modifies(pattern)', safe=True)
1345 1345 def modifies(repo, subset, x):
1346 1346 """Changesets modifying files matched by pattern.
1347 1347
1348 1348 The pattern without explicit kind like ``glob:`` is expected to be
1349 1349 relative to the current directory and match against a file or a
1350 1350 directory.
1351 1351 """
1352 1352 # i18n: "modifies" is a keyword
1353 1353 pat = getstring(x, _("modifies requires a pattern"))
1354 1354 return checkstatus(repo, subset, pat, 0)
1355 1355
1356 1356 @predicate('named(namespace)')
1357 1357 def named(repo, subset, x):
1358 1358 """The changesets in a given namespace.
1359 1359
1360 1360 If `namespace` starts with `re:`, the remainder of the string is treated as
1361 1361 a regular expression. To match a namespace that actually starts with `re:`,
1362 1362 use the prefix `literal:`.
1363 1363 """
1364 1364 # i18n: "named" is a keyword
1365 1365 args = getargs(x, 1, 1, _('named requires a namespace argument'))
1366 1366
1367 1367 ns = getstring(args[0],
1368 1368 # i18n: "named" is a keyword
1369 1369 _('the argument to named must be a string'))
1370 1370 kind, pattern, matcher = util.stringmatcher(ns)
1371 1371 namespaces = set()
1372 1372 if kind == 'literal':
1373 1373 if pattern not in repo.names:
1374 1374 raise error.RepoLookupError(_("namespace '%s' does not exist")
1375 1375 % ns)
1376 1376 namespaces.add(repo.names[pattern])
1377 1377 else:
1378 1378 for name, ns in repo.names.iteritems():
1379 1379 if matcher(name):
1380 1380 namespaces.add(ns)
1381 1381 if not namespaces:
1382 1382 raise error.RepoLookupError(_("no namespace exists"
1383 1383 " that match '%s'") % pattern)
1384 1384
1385 1385 names = set()
1386 1386 for ns in namespaces:
1387 1387 for name in ns.listnames(repo):
1388 1388 if name not in ns.deprecated:
1389 1389 names.update(repo[n].rev() for n in ns.nodes(repo, name))
1390 1390
1391 1391 names -= set([node.nullrev])
1392 1392 return subset & names
1393 1393
1394 1394 @predicate('id(string)', safe=True)
1395 1395 def node_(repo, subset, x):
1396 1396 """Revision non-ambiguously specified by the given hex string prefix.
1397 1397 """
1398 1398 # i18n: "id" is a keyword
1399 1399 l = getargs(x, 1, 1, _("id requires one argument"))
1400 1400 # i18n: "id" is a keyword
1401 1401 n = getstring(l[0], _("id requires a string"))
1402 1402 if len(n) == 40:
1403 1403 try:
1404 1404 rn = repo.changelog.rev(node.bin(n))
1405 1405 except (LookupError, TypeError):
1406 1406 rn = None
1407 1407 else:
1408 1408 rn = None
1409 1409 pm = repo.changelog._partialmatch(n)
1410 1410 if pm is not None:
1411 1411 rn = repo.changelog.rev(pm)
1412 1412
1413 1413 if rn is None:
1414 1414 return baseset()
1415 1415 result = baseset([rn])
1416 1416 return result & subset
1417 1417
1418 1418 @predicate('obsolete()', safe=True)
1419 1419 def obsolete(repo, subset, x):
1420 1420 """Mutable changeset with a newer version."""
1421 1421 # i18n: "obsolete" is a keyword
1422 1422 getargs(x, 0, 0, _("obsolete takes no arguments"))
1423 1423 obsoletes = obsmod.getrevs(repo, 'obsolete')
1424 1424 return subset & obsoletes
1425 1425
1426 1426 @predicate('only(set, [set])', safe=True)
1427 1427 def only(repo, subset, x):
1428 1428 """Changesets that are ancestors of the first set that are not ancestors
1429 1429 of any other head in the repo. If a second set is specified, the result
1430 1430 is ancestors of the first set that are not ancestors of the second set
1431 1431 (i.e. ::<set1> - ::<set2>).
1432 1432 """
1433 1433 cl = repo.changelog
1434 1434 # i18n: "only" is a keyword
1435 1435 args = getargs(x, 1, 2, _('only takes one or two arguments'))
1436 1436 include = getset(repo, fullreposet(repo), args[0])
1437 1437 if len(args) == 1:
1438 1438 if not include:
1439 1439 return baseset()
1440 1440
1441 1441 descendants = set(_revdescendants(repo, include, False))
1442 1442 exclude = [rev for rev in cl.headrevs()
1443 1443 if not rev in descendants and not rev in include]
1444 1444 else:
1445 1445 exclude = getset(repo, fullreposet(repo), args[1])
1446 1446
1447 1447 results = set(cl.findmissingrevs(common=exclude, heads=include))
1448 1448 # XXX we should turn this into a baseset instead of a set, smartset may do
1449 1449 # some optimizations from the fact this is a baseset.
1450 1450 return subset & results
1451 1451
1452 1452 @predicate('origin([set])', safe=True)
1453 1453 def origin(repo, subset, x):
1454 1454 """
1455 1455 Changesets that were specified as a source for the grafts, transplants or
1456 1456 rebases that created the given revisions. Omitting the optional set is the
1457 1457 same as passing all(). If a changeset created by these operations is itself
1458 1458 specified as a source for one of these operations, only the source changeset
1459 1459 for the first operation is selected.
1460 1460 """
1461 1461 if x is not None:
1462 1462 dests = getset(repo, fullreposet(repo), x)
1463 1463 else:
1464 1464 dests = fullreposet(repo)
1465 1465
1466 1466 def _firstsrc(rev):
1467 1467 src = _getrevsource(repo, rev)
1468 1468 if src is None:
1469 1469 return None
1470 1470
1471 1471 while True:
1472 1472 prev = _getrevsource(repo, src)
1473 1473
1474 1474 if prev is None:
1475 1475 return src
1476 1476 src = prev
1477 1477
1478 1478 o = set([_firstsrc(r) for r in dests])
1479 1479 o -= set([None])
1480 1480 # XXX we should turn this into a baseset instead of a set, smartset may do
1481 1481 # some optimizations from the fact this is a baseset.
1482 1482 return subset & o
1483 1483
1484 1484 @predicate('outgoing([path])', safe=True)
1485 1485 def outgoing(repo, subset, x):
1486 1486 """Changesets not found in the specified destination repository, or the
1487 1487 default push location.
1488 1488 """
1489 1489 # Avoid cycles.
1490 1490 from . import (
1491 1491 discovery,
1492 1492 hg,
1493 1493 )
1494 1494 # i18n: "outgoing" is a keyword
1495 1495 l = getargs(x, 0, 1, _("outgoing takes one or no arguments"))
1496 1496 # i18n: "outgoing" is a keyword
1497 1497 dest = l and getstring(l[0], _("outgoing requires a repository path")) or ''
1498 1498 dest = repo.ui.expandpath(dest or 'default-push', dest or 'default')
1499 1499 dest, branches = hg.parseurl(dest)
1500 1500 revs, checkout = hg.addbranchrevs(repo, repo, branches, [])
1501 1501 if revs:
1502 1502 revs = [repo.lookup(rev) for rev in revs]
1503 1503 other = hg.peer(repo, {}, dest)
1504 1504 repo.ui.pushbuffer()
1505 1505 outgoing = discovery.findcommonoutgoing(repo, other, onlyheads=revs)
1506 1506 repo.ui.popbuffer()
1507 1507 cl = repo.changelog
1508 1508 o = set([cl.rev(r) for r in outgoing.missing])
1509 1509 return subset & o
1510 1510
1511 1511 @predicate('p1([set])', safe=True)
1512 1512 def p1(repo, subset, x):
1513 1513 """First parent of changesets in set, or the working directory.
1514 1514 """
1515 1515 if x is None:
1516 1516 p = repo[x].p1().rev()
1517 1517 if p >= 0:
1518 1518 return subset & baseset([p])
1519 1519 return baseset()
1520 1520
1521 1521 ps = set()
1522 1522 cl = repo.changelog
1523 1523 for r in getset(repo, fullreposet(repo), x):
1524 1524 ps.add(cl.parentrevs(r)[0])
1525 1525 ps -= set([node.nullrev])
1526 1526 # XXX we should turn this into a baseset instead of a set, smartset may do
1527 1527 # some optimizations from the fact this is a baseset.
1528 1528 return subset & ps
1529 1529
1530 1530 @predicate('p2([set])', safe=True)
1531 1531 def p2(repo, subset, x):
1532 1532 """Second parent of changesets in set, or the working directory.
1533 1533 """
1534 1534 if x is None:
1535 1535 ps = repo[x].parents()
1536 1536 try:
1537 1537 p = ps[1].rev()
1538 1538 if p >= 0:
1539 1539 return subset & baseset([p])
1540 1540 return baseset()
1541 1541 except IndexError:
1542 1542 return baseset()
1543 1543
1544 1544 ps = set()
1545 1545 cl = repo.changelog
1546 1546 for r in getset(repo, fullreposet(repo), x):
1547 1547 ps.add(cl.parentrevs(r)[1])
1548 1548 ps -= set([node.nullrev])
1549 1549 # XXX we should turn this into a baseset instead of a set, smartset may do
1550 1550 # some optimizations from the fact this is a baseset.
1551 1551 return subset & ps
1552 1552
1553 1553 def parentpost(repo, subset, x, order):
1554 1554 return p1(repo, subset, x)
1555 1555
1556 1556 @predicate('parents([set])', safe=True)
1557 1557 def parents(repo, subset, x):
1558 1558 """
1559 1559 The set of all parents for all changesets in set, or the working directory.
1560 1560 """
1561 1561 if x is None:
1562 1562 ps = set(p.rev() for p in repo[x].parents())
1563 1563 else:
1564 1564 ps = set()
1565 1565 cl = repo.changelog
1566 1566 up = ps.update
1567 1567 parentrevs = cl.parentrevs
1568 1568 for r in getset(repo, fullreposet(repo), x):
1569 1569 if r == node.wdirrev:
1570 1570 up(p.rev() for p in repo[r].parents())
1571 1571 else:
1572 1572 up(parentrevs(r))
1573 1573 ps -= set([node.nullrev])
1574 1574 return subset & ps
1575 1575
1576 1576 def _phase(repo, subset, target):
1577 1577 """helper to select all rev in phase <target>"""
1578 1578 repo._phasecache.loadphaserevs(repo) # ensure phase's sets are loaded
1579 1579 if repo._phasecache._phasesets:
1580 1580 s = repo._phasecache._phasesets[target] - repo.changelog.filteredrevs
1581 1581 s = baseset(s)
1582 1582 s.sort() # set are non ordered, so we enforce ascending
1583 1583 return subset & s
1584 1584 else:
1585 1585 phase = repo._phasecache.phase
1586 1586 condition = lambda r: phase(repo, r) == target
1587 1587 return subset.filter(condition, condrepr=('<phase %r>', target),
1588 1588 cache=False)
1589 1589
1590 1590 @predicate('draft()', safe=True)
1591 1591 def draft(repo, subset, x):
1592 1592 """Changeset in draft phase."""
1593 1593 # i18n: "draft" is a keyword
1594 1594 getargs(x, 0, 0, _("draft takes no arguments"))
1595 1595 target = phases.draft
1596 1596 return _phase(repo, subset, target)
1597 1597
1598 1598 @predicate('secret()', safe=True)
1599 1599 def secret(repo, subset, x):
1600 1600 """Changeset in secret phase."""
1601 1601 # i18n: "secret" is a keyword
1602 1602 getargs(x, 0, 0, _("secret takes no arguments"))
1603 1603 target = phases.secret
1604 1604 return _phase(repo, subset, target)
1605 1605
1606 1606 def parentspec(repo, subset, x, n, order):
1607 1607 """``set^0``
1608 1608 The set.
1609 1609 ``set^1`` (or ``set^``), ``set^2``
1610 1610 First or second parent, respectively, of all changesets in set.
1611 1611 """
1612 1612 try:
1613 1613 n = int(n[1])
1614 1614 if n not in (0, 1, 2):
1615 1615 raise ValueError
1616 1616 except (TypeError, ValueError):
1617 1617 raise error.ParseError(_("^ expects a number 0, 1, or 2"))
1618 1618 ps = set()
1619 1619 cl = repo.changelog
1620 1620 for r in getset(repo, fullreposet(repo), x):
1621 1621 if n == 0:
1622 1622 ps.add(r)
1623 1623 elif n == 1:
1624 1624 ps.add(cl.parentrevs(r)[0])
1625 1625 elif n == 2:
1626 1626 parents = cl.parentrevs(r)
1627 1627 if parents[1] != node.nullrev:
1628 1628 ps.add(parents[1])
1629 1629 return subset & ps
1630 1630
1631 1631 @predicate('present(set)', safe=True)
1632 1632 def present(repo, subset, x):
1633 1633 """An empty set, if any revision in set isn't found; otherwise,
1634 1634 all revisions in set.
1635 1635
1636 1636 If any of specified revisions is not present in the local repository,
1637 1637 the query is normally aborted. But this predicate allows the query
1638 1638 to continue even in such cases.
1639 1639 """
1640 1640 try:
1641 1641 return getset(repo, subset, x)
1642 1642 except error.RepoLookupError:
1643 1643 return baseset()
1644 1644
1645 1645 # for internal use
1646 1646 @predicate('_notpublic', safe=True)
1647 1647 def _notpublic(repo, subset, x):
1648 1648 getargs(x, 0, 0, "_notpublic takes no arguments")
1649 1649 repo._phasecache.loadphaserevs(repo) # ensure phase's sets are loaded
1650 1650 if repo._phasecache._phasesets:
1651 1651 s = set()
1652 1652 for u in repo._phasecache._phasesets[1:]:
1653 1653 s.update(u)
1654 1654 s = baseset(s - repo.changelog.filteredrevs)
1655 1655 s.sort()
1656 1656 return subset & s
1657 1657 else:
1658 1658 phase = repo._phasecache.phase
1659 1659 target = phases.public
1660 1660 condition = lambda r: phase(repo, r) != target
1661 1661 return subset.filter(condition, condrepr=('<phase %r>', target),
1662 1662 cache=False)
1663 1663
1664 1664 @predicate('public()', safe=True)
1665 1665 def public(repo, subset, x):
1666 1666 """Changeset in public phase."""
1667 1667 # i18n: "public" is a keyword
1668 1668 getargs(x, 0, 0, _("public takes no arguments"))
1669 1669 phase = repo._phasecache.phase
1670 1670 target = phases.public
1671 1671 condition = lambda r: phase(repo, r) == target
1672 1672 return subset.filter(condition, condrepr=('<phase %r>', target),
1673 1673 cache=False)
1674 1674
1675 1675 @predicate('remote([id [,path]])', safe=True)
1676 1676 def remote(repo, subset, x):
1677 1677 """Local revision that corresponds to the given identifier in a
1678 1678 remote repository, if present. Here, the '.' identifier is a
1679 1679 synonym for the current local branch.
1680 1680 """
1681 1681
1682 1682 from . import hg # avoid start-up nasties
1683 1683 # i18n: "remote" is a keyword
1684 1684 l = getargs(x, 0, 2, _("remote takes zero, one, or two arguments"))
1685 1685
1686 1686 q = '.'
1687 1687 if len(l) > 0:
1688 1688 # i18n: "remote" is a keyword
1689 1689 q = getstring(l[0], _("remote requires a string id"))
1690 1690 if q == '.':
1691 1691 q = repo['.'].branch()
1692 1692
1693 1693 dest = ''
1694 1694 if len(l) > 1:
1695 1695 # i18n: "remote" is a keyword
1696 1696 dest = getstring(l[1], _("remote requires a repository path"))
1697 1697 dest = repo.ui.expandpath(dest or 'default')
1698 1698 dest, branches = hg.parseurl(dest)
1699 1699 revs, checkout = hg.addbranchrevs(repo, repo, branches, [])
1700 1700 if revs:
1701 1701 revs = [repo.lookup(rev) for rev in revs]
1702 1702 other = hg.peer(repo, {}, dest)
1703 1703 n = other.lookup(q)
1704 1704 if n in repo:
1705 1705 r = repo[n].rev()
1706 1706 if r in subset:
1707 1707 return baseset([r])
1708 1708 return baseset()
1709 1709
1710 1710 @predicate('removes(pattern)', safe=True)
1711 1711 def removes(repo, subset, x):
1712 1712 """Changesets which remove files matching pattern.
1713 1713
1714 1714 The pattern without explicit kind like ``glob:`` is expected to be
1715 1715 relative to the current directory and match against a file or a
1716 1716 directory.
1717 1717 """
1718 1718 # i18n: "removes" is a keyword
1719 1719 pat = getstring(x, _("removes requires a pattern"))
1720 1720 return checkstatus(repo, subset, pat, 2)
1721 1721
1722 1722 @predicate('rev(number)', safe=True)
1723 1723 def rev(repo, subset, x):
1724 1724 """Revision with the given numeric identifier.
1725 1725 """
1726 1726 # i18n: "rev" is a keyword
1727 1727 l = getargs(x, 1, 1, _("rev requires one argument"))
1728 1728 try:
1729 1729 # i18n: "rev" is a keyword
1730 1730 l = int(getstring(l[0], _("rev requires a number")))
1731 1731 except (TypeError, ValueError):
1732 1732 # i18n: "rev" is a keyword
1733 1733 raise error.ParseError(_("rev expects a number"))
1734 1734 if l not in repo.changelog and l != node.nullrev:
1735 1735 return baseset()
1736 1736 return subset & baseset([l])
1737 1737
1738 1738 @predicate('matching(revision [, field])', safe=True)
1739 1739 def matching(repo, subset, x):
1740 1740 """Changesets in which a given set of fields match the set of fields in the
1741 1741 selected revision or set.
1742 1742
1743 1743 To match more than one field pass the list of fields to match separated
1744 1744 by spaces (e.g. ``author description``).
1745 1745
1746 1746 Valid fields are most regular revision fields and some special fields.
1747 1747
1748 1748 Regular revision fields are ``description``, ``author``, ``branch``,
1749 1749 ``date``, ``files``, ``phase``, ``parents``, ``substate``, ``user``
1750 1750 and ``diff``.
1751 1751 Note that ``author`` and ``user`` are synonyms. ``diff`` refers to the
1752 1752 contents of the revision. Two revisions matching their ``diff`` will
1753 1753 also match their ``files``.
1754 1754
1755 1755 Special fields are ``summary`` and ``metadata``:
1756 1756 ``summary`` matches the first line of the description.
1757 1757 ``metadata`` is equivalent to matching ``description user date``
1758 1758 (i.e. it matches the main metadata fields).
1759 1759
1760 1760 ``metadata`` is the default field which is used when no fields are
1761 1761 specified. You can match more than one field at a time.
1762 1762 """
1763 1763 # i18n: "matching" is a keyword
1764 1764 l = getargs(x, 1, 2, _("matching takes 1 or 2 arguments"))
1765 1765
1766 1766 revs = getset(repo, fullreposet(repo), l[0])
1767 1767
1768 1768 fieldlist = ['metadata']
1769 1769 if len(l) > 1:
1770 1770 fieldlist = getstring(l[1],
1771 1771 # i18n: "matching" is a keyword
1772 1772 _("matching requires a string "
1773 1773 "as its second argument")).split()
1774 1774
1775 1775 # Make sure that there are no repeated fields,
1776 1776 # expand the 'special' 'metadata' field type
1777 1777 # and check the 'files' whenever we check the 'diff'
1778 1778 fields = []
1779 1779 for field in fieldlist:
1780 1780 if field == 'metadata':
1781 1781 fields += ['user', 'description', 'date']
1782 1782 elif field == 'diff':
1783 1783 # a revision matching the diff must also match the files
1784 1784 # since matching the diff is very costly, make sure to
1785 1785 # also match the files first
1786 1786 fields += ['files', 'diff']
1787 1787 else:
1788 1788 if field == 'author':
1789 1789 field = 'user'
1790 1790 fields.append(field)
1791 1791 fields = set(fields)
1792 1792 if 'summary' in fields and 'description' in fields:
1793 1793 # If a revision matches its description it also matches its summary
1794 1794 fields.discard('summary')
1795 1795
1796 1796 # We may want to match more than one field
1797 1797 # Not all fields take the same amount of time to be matched
1798 1798 # Sort the selected fields in order of increasing matching cost
1799 1799 fieldorder = ['phase', 'parents', 'user', 'date', 'branch', 'summary',
1800 1800 'files', 'description', 'substate', 'diff']
1801 1801 def fieldkeyfunc(f):
1802 1802 try:
1803 1803 return fieldorder.index(f)
1804 1804 except ValueError:
1805 1805 # assume an unknown field is very costly
1806 1806 return len(fieldorder)
1807 1807 fields = list(fields)
1808 1808 fields.sort(key=fieldkeyfunc)
1809 1809
1810 1810 # Each field will be matched with its own "getfield" function
1811 1811 # which will be added to the getfieldfuncs array of functions
1812 1812 getfieldfuncs = []
1813 1813 _funcs = {
1814 1814 'user': lambda r: repo[r].user(),
1815 1815 'branch': lambda r: repo[r].branch(),
1816 1816 'date': lambda r: repo[r].date(),
1817 1817 'description': lambda r: repo[r].description(),
1818 1818 'files': lambda r: repo[r].files(),
1819 1819 'parents': lambda r: repo[r].parents(),
1820 1820 'phase': lambda r: repo[r].phase(),
1821 1821 'substate': lambda r: repo[r].substate,
1822 1822 'summary': lambda r: repo[r].description().splitlines()[0],
1823 1823 'diff': lambda r: list(repo[r].diff(git=True),)
1824 1824 }
1825 1825 for info in fields:
1826 1826 getfield = _funcs.get(info, None)
1827 1827 if getfield is None:
1828 1828 raise error.ParseError(
1829 1829 # i18n: "matching" is a keyword
1830 1830 _("unexpected field name passed to matching: %s") % info)
1831 1831 getfieldfuncs.append(getfield)
1832 1832 # convert the getfield array of functions into a "getinfo" function
1833 1833 # which returns an array of field values (or a single value if there
1834 1834 # is only one field to match)
1835 1835 getinfo = lambda r: [f(r) for f in getfieldfuncs]
1836 1836
1837 1837 def matches(x):
1838 1838 for rev in revs:
1839 1839 target = getinfo(rev)
1840 1840 match = True
1841 1841 for n, f in enumerate(getfieldfuncs):
1842 1842 if target[n] != f(x):
1843 1843 match = False
1844 1844 if match:
1845 1845 return True
1846 1846 return False
1847 1847
1848 1848 return subset.filter(matches, condrepr=('<matching%r %r>', fields, revs))
1849 1849
1850 1850 @predicate('reverse(set)', safe=True, takeorder=True)
1851 1851 def reverse(repo, subset, x, order):
1852 1852 """Reverse order of set.
1853 1853 """
1854 1854 l = getset(repo, subset, x)
1855 1855 if order == defineorder:
1856 1856 l.reverse()
1857 1857 return l
1858 1858
1859 1859 @predicate('roots(set)', safe=True)
1860 1860 def roots(repo, subset, x):
1861 1861 """Changesets in set with no parent changeset in set.
1862 1862 """
1863 1863 s = getset(repo, fullreposet(repo), x)
1864 1864 parents = repo.changelog.parentrevs
1865 1865 def filter(r):
1866 1866 for p in parents(r):
1867 1867 if 0 <= p and p in s:
1868 1868 return False
1869 1869 return True
1870 1870 return subset & s.filter(filter, condrepr='<roots>')
1871 1871
1872 1872 _sortkeyfuncs = {
1873 1873 'rev': lambda c: c.rev(),
1874 1874 'branch': lambda c: c.branch(),
1875 1875 'desc': lambda c: c.description(),
1876 1876 'user': lambda c: c.user(),
1877 1877 'author': lambda c: c.user(),
1878 1878 'date': lambda c: c.date()[0],
1879 1879 }
1880 1880
1881 1881 def _getsortargs(x):
1882 1882 """Parse sort options into (set, [(key, reverse)], opts)"""
1883 1883 args = getargsdict(x, 'sort', 'set keys topo.firstbranch')
1884 1884 if 'set' not in args:
1885 1885 # i18n: "sort" is a keyword
1886 1886 raise error.ParseError(_('sort requires one or two arguments'))
1887 1887 keys = "rev"
1888 1888 if 'keys' in args:
1889 1889 # i18n: "sort" is a keyword
1890 1890 keys = getstring(args['keys'], _("sort spec must be a string"))
1891 1891
1892 1892 keyflags = []
1893 1893 for k in keys.split():
1894 1894 fk = k
1895 1895 reverse = (k[0] == '-')
1896 1896 if reverse:
1897 1897 k = k[1:]
1898 1898 if k not in _sortkeyfuncs and k != 'topo':
1899 1899 raise error.ParseError(_("unknown sort key %r") % fk)
1900 1900 keyflags.append((k, reverse))
1901 1901
1902 1902 if len(keyflags) > 1 and any(k == 'topo' for k, reverse in keyflags):
1903 1903 # i18n: "topo" is a keyword
1904 1904 raise error.ParseError(_('topo sort order cannot be combined '
1905 1905 'with other sort keys'))
1906 1906
1907 1907 opts = {}
1908 1908 if 'topo.firstbranch' in args:
1909 1909 if any(k == 'topo' for k, reverse in keyflags):
1910 1910 opts['topo.firstbranch'] = args['topo.firstbranch']
1911 1911 else:
1912 1912 # i18n: "topo" and "topo.firstbranch" are keywords
1913 1913 raise error.ParseError(_('topo.firstbranch can only be used '
1914 1914 'when using the topo sort key'))
1915 1915
1916 1916 return args['set'], keyflags, opts
1917 1917
1918 1918 @predicate('sort(set[, [-]key... [, ...]])', safe=True, takeorder=True)
1919 1919 def sort(repo, subset, x, order):
1920 1920 """Sort set by keys. The default sort order is ascending, specify a key
1921 1921 as ``-key`` to sort in descending order.
1922 1922
1923 1923 The keys can be:
1924 1924
1925 1925 - ``rev`` for the revision number,
1926 1926 - ``branch`` for the branch name,
1927 1927 - ``desc`` for the commit message (description),
1928 1928 - ``user`` for user name (``author`` can be used as an alias),
1929 1929 - ``date`` for the commit date
1930 1930 - ``topo`` for a reverse topographical sort
1931 1931
1932 1932 The ``topo`` sort order cannot be combined with other sort keys. This sort
1933 1933 takes one optional argument, ``topo.firstbranch``, which takes a revset that
1934 1934 specifies what topographical branches to prioritize in the sort.
1935 1935
1936 1936 """
1937 1937 s, keyflags, opts = _getsortargs(x)
1938 1938 revs = getset(repo, subset, s)
1939 1939
1940 1940 if not keyflags or order != defineorder:
1941 1941 return revs
1942 1942 if len(keyflags) == 1 and keyflags[0][0] == "rev":
1943 1943 revs.sort(reverse=keyflags[0][1])
1944 1944 return revs
1945 1945 elif keyflags[0][0] == "topo":
1946 1946 firstbranch = ()
1947 1947 if 'topo.firstbranch' in opts:
1948 1948 firstbranch = getset(repo, subset, opts['topo.firstbranch'])
1949 1949 revs = baseset(_toposort(revs, repo.changelog.parentrevs, firstbranch),
1950 1950 istopo=True)
1951 1951 if keyflags[0][1]:
1952 1952 revs.reverse()
1953 1953 return revs
1954 1954
1955 1955 # sort() is guaranteed to be stable
1956 1956 ctxs = [repo[r] for r in revs]
1957 1957 for k, reverse in reversed(keyflags):
1958 1958 ctxs.sort(key=_sortkeyfuncs[k], reverse=reverse)
1959 1959 return baseset([c.rev() for c in ctxs])
1960 1960
1961 1961 def _toposort(revs, parentsfunc, firstbranch=()):
1962 1962 """Yield revisions from heads to roots one (topo) branch at a time.
1963 1963
1964 1964 This function aims to be used by a graph generator that wishes to minimize
1965 1965 the number of parallel branches and their interleaving.
1966 1966
1967 1967 Example iteration order (numbers show the "true" order in a changelog):
1968 1968
1969 1969 o 4
1970 1970 |
1971 1971 o 1
1972 1972 |
1973 1973 | o 3
1974 1974 | |
1975 1975 | o 2
1976 1976 |/
1977 1977 o 0
1978 1978
1979 1979 Note that the ancestors of merges are understood by the current
1980 1980 algorithm to be on the same branch. This means no reordering will
1981 1981 occur behind a merge.
1982 1982 """
1983 1983
1984 1984 ### Quick summary of the algorithm
1985 1985 #
1986 1986 # This function is based around a "retention" principle. We keep revisions
1987 1987 # in memory until we are ready to emit a whole branch that immediately
1988 1988 # "merges" into an existing one. This reduces the number of parallel
1989 1989 # branches with interleaved revisions.
1990 1990 #
1991 1991 # During iteration revs are split into two groups:
1992 1992 # A) revision already emitted
1993 1993 # B) revision in "retention". They are stored as different subgroups.
1994 1994 #
1995 1995 # for each REV, we do the following logic:
1996 1996 #
1997 1997 # 1) if REV is a parent of (A), we will emit it. If there is a
1998 1998 # retention group ((B) above) that is blocked on REV being
1999 1999 # available, we emit all the revisions out of that retention
2000 2000 # group first.
2001 2001 #
2002 2002 # 2) else, we'll search for a subgroup in (B) awaiting for REV to be
2003 2003 # available, if such subgroup exist, we add REV to it and the subgroup is
2004 2004 # now awaiting for REV.parents() to be available.
2005 2005 #
2006 2006 # 3) finally if no such group existed in (B), we create a new subgroup.
2007 2007 #
2008 2008 #
2009 2009 # To bootstrap the algorithm, we emit the tipmost revision (which
2010 2010 # puts it in group (A) from above).
2011 2011
2012 2012 revs.sort(reverse=True)
2013 2013
2014 2014 # Set of parents of revision that have been emitted. They can be considered
2015 2015 # unblocked as the graph generator is already aware of them so there is no
2016 2016 # need to delay the revisions that reference them.
2017 2017 #
2018 2018 # If someone wants to prioritize a branch over the others, pre-filling this
2019 2019 # set will force all other branches to wait until this branch is ready to be
2020 2020 # emitted.
2021 2021 unblocked = set(firstbranch)
2022 2022
2023 2023 # list of groups waiting to be displayed, each group is defined by:
2024 2024 #
2025 2025 # (revs: lists of revs waiting to be displayed,
2026 2026 # blocked: set of that cannot be displayed before those in 'revs')
2027 2027 #
2028 2028 # The second value ('blocked') correspond to parents of any revision in the
2029 2029 # group ('revs') that is not itself contained in the group. The main idea
2030 2030 # of this algorithm is to delay as much as possible the emission of any
2031 2031 # revision. This means waiting for the moment we are about to display
2032 2032 # these parents to display the revs in a group.
2033 2033 #
2034 2034 # This first implementation is smart until it encounters a merge: it will
2035 2035 # emit revs as soon as any parent is about to be emitted and can grow an
2036 2036 # arbitrary number of revs in 'blocked'. In practice this mean we properly
2037 2037 # retains new branches but gives up on any special ordering for ancestors
2038 2038 # of merges. The implementation can be improved to handle this better.
2039 2039 #
2040 2040 # The first subgroup is special. It corresponds to all the revision that
2041 2041 # were already emitted. The 'revs' lists is expected to be empty and the
2042 2042 # 'blocked' set contains the parents revisions of already emitted revision.
2043 2043 #
2044 2044 # You could pre-seed the <parents> set of groups[0] to a specific
2045 2045 # changesets to select what the first emitted branch should be.
2046 2046 groups = [([], unblocked)]
2047 2047 pendingheap = []
2048 2048 pendingset = set()
2049 2049
2050 2050 heapq.heapify(pendingheap)
2051 2051 heappop = heapq.heappop
2052 2052 heappush = heapq.heappush
2053 2053 for currentrev in revs:
2054 2054 # Heap works with smallest element, we want highest so we invert
2055 2055 if currentrev not in pendingset:
2056 2056 heappush(pendingheap, -currentrev)
2057 2057 pendingset.add(currentrev)
2058 2058 # iterates on pending rev until after the current rev have been
2059 2059 # processed.
2060 2060 rev = None
2061 2061 while rev != currentrev:
2062 2062 rev = -heappop(pendingheap)
2063 2063 pendingset.remove(rev)
2064 2064
2065 2065 # Seek for a subgroup blocked, waiting for the current revision.
2066 2066 matching = [i for i, g in enumerate(groups) if rev in g[1]]
2067 2067
2068 2068 if matching:
2069 2069 # The main idea is to gather together all sets that are blocked
2070 2070 # on the same revision.
2071 2071 #
2072 2072 # Groups are merged when a common blocking ancestor is
2073 2073 # observed. For example, given two groups:
2074 2074 #
2075 2075 # revs [5, 4] waiting for 1
2076 2076 # revs [3, 2] waiting for 1
2077 2077 #
2078 2078 # These two groups will be merged when we process
2079 2079 # 1. In theory, we could have merged the groups when
2080 2080 # we added 2 to the group it is now in (we could have
2081 2081 # noticed the groups were both blocked on 1 then), but
2082 2082 # the way it works now makes the algorithm simpler.
2083 2083 #
2084 2084 # We also always keep the oldest subgroup first. We can
2085 2085 # probably improve the behavior by having the longest set
2086 2086 # first. That way, graph algorithms could minimise the length
2087 2087 # of parallel lines their drawing. This is currently not done.
2088 2088 targetidx = matching.pop(0)
2089 2089 trevs, tparents = groups[targetidx]
2090 2090 for i in matching:
2091 2091 gr = groups[i]
2092 2092 trevs.extend(gr[0])
2093 2093 tparents |= gr[1]
2094 2094 # delete all merged subgroups (except the one we kept)
2095 2095 # (starting from the last subgroup for performance and
2096 2096 # sanity reasons)
2097 2097 for i in reversed(matching):
2098 2098 del groups[i]
2099 2099 else:
2100 2100 # This is a new head. We create a new subgroup for it.
2101 2101 targetidx = len(groups)
2102 2102 groups.append(([], set([rev])))
2103 2103
2104 2104 gr = groups[targetidx]
2105 2105
2106 2106 # We now add the current nodes to this subgroups. This is done
2107 2107 # after the subgroup merging because all elements from a subgroup
2108 2108 # that relied on this rev must precede it.
2109 2109 #
2110 2110 # we also update the <parents> set to include the parents of the
2111 2111 # new nodes.
2112 2112 if rev == currentrev: # only display stuff in rev
2113 2113 gr[0].append(rev)
2114 2114 gr[1].remove(rev)
2115 2115 parents = [p for p in parentsfunc(rev) if p > node.nullrev]
2116 2116 gr[1].update(parents)
2117 2117 for p in parents:
2118 2118 if p not in pendingset:
2119 2119 pendingset.add(p)
2120 2120 heappush(pendingheap, -p)
2121 2121
2122 2122 # Look for a subgroup to display
2123 2123 #
2124 2124 # When unblocked is empty (if clause), we were not waiting for any
2125 2125 # revisions during the first iteration (if no priority was given) or
2126 2126 # if we emitted a whole disconnected set of the graph (reached a
2127 2127 # root). In that case we arbitrarily take the oldest known
2128 2128 # subgroup. The heuristic could probably be better.
2129 2129 #
2130 2130 # Otherwise (elif clause) if the subgroup is blocked on
2131 2131 # a revision we just emitted, we can safely emit it as
2132 2132 # well.
2133 2133 if not unblocked:
2134 2134 if len(groups) > 1: # display other subset
2135 2135 targetidx = 1
2136 2136 gr = groups[1]
2137 2137 elif not gr[1] & unblocked:
2138 2138 gr = None
2139 2139
2140 2140 if gr is not None:
2141 2141 # update the set of awaited revisions with the one from the
2142 2142 # subgroup
2143 2143 unblocked |= gr[1]
2144 2144 # output all revisions in the subgroup
2145 2145 for r in gr[0]:
2146 2146 yield r
2147 2147 # delete the subgroup that you just output
2148 2148 # unless it is groups[0] in which case you just empty it.
2149 2149 if targetidx:
2150 2150 del groups[targetidx]
2151 2151 else:
2152 2152 gr[0][:] = []
2153 2153 # Check if we have some subgroup waiting for revisions we are not going to
2154 2154 # iterate over
2155 2155 for g in groups:
2156 2156 for r in g[0]:
2157 2157 yield r
2158 2158
2159 2159 @predicate('subrepo([pattern])')
2160 2160 def subrepo(repo, subset, x):
2161 2161 """Changesets that add, modify or remove the given subrepo. If no subrepo
2162 2162 pattern is named, any subrepo changes are returned.
2163 2163 """
2164 2164 # i18n: "subrepo" is a keyword
2165 2165 args = getargs(x, 0, 1, _('subrepo takes at most one argument'))
2166 2166 pat = None
2167 2167 if len(args) != 0:
2168 2168 pat = getstring(args[0], _("subrepo requires a pattern"))
2169 2169
2170 2170 m = matchmod.exact(repo.root, repo.root, ['.hgsubstate'])
2171 2171
2172 2172 def submatches(names):
2173 2173 k, p, m = util.stringmatcher(pat)
2174 2174 for name in names:
2175 2175 if m(name):
2176 2176 yield name
2177 2177
2178 2178 def matches(x):
2179 2179 c = repo[x]
2180 2180 s = repo.status(c.p1().node(), c.node(), match=m)
2181 2181
2182 2182 if pat is None:
2183 2183 return s.added or s.modified or s.removed
2184 2184
2185 2185 if s.added:
2186 2186 return any(submatches(c.substate.keys()))
2187 2187
2188 2188 if s.modified:
2189 2189 subs = set(c.p1().substate.keys())
2190 2190 subs.update(c.substate.keys())
2191 2191
2192 2192 for path in submatches(subs):
2193 2193 if c.p1().substate.get(path) != c.substate.get(path):
2194 2194 return True
2195 2195
2196 2196 if s.removed:
2197 2197 return any(submatches(c.p1().substate.keys()))
2198 2198
2199 2199 return False
2200 2200
2201 2201 return subset.filter(matches, condrepr=('<subrepo %r>', pat))
2202 2202
2203 2203 def _substringmatcher(pattern):
2204 2204 kind, pattern, matcher = util.stringmatcher(pattern)
2205 2205 if kind == 'literal':
2206 2206 matcher = lambda s: pattern in s
2207 2207 return kind, pattern, matcher
2208 2208
2209 2209 @predicate('tag([name])', safe=True)
2210 2210 def tag(repo, subset, x):
2211 2211 """The specified tag by name, or all tagged revisions if no name is given.
2212 2212
2213 2213 If `name` starts with `re:`, the remainder of the name is treated as
2214 2214 a regular expression. To match a tag that actually starts with `re:`,
2215 2215 use the prefix `literal:`.
2216 2216 """
2217 2217 # i18n: "tag" is a keyword
2218 2218 args = getargs(x, 0, 1, _("tag takes one or no arguments"))
2219 2219 cl = repo.changelog
2220 2220 if args:
2221 2221 pattern = getstring(args[0],
2222 2222 # i18n: "tag" is a keyword
2223 2223 _('the argument to tag must be a string'))
2224 2224 kind, pattern, matcher = util.stringmatcher(pattern)
2225 2225 if kind == 'literal':
2226 2226 # avoid resolving all tags
2227 2227 tn = repo._tagscache.tags.get(pattern, None)
2228 2228 if tn is None:
2229 2229 raise error.RepoLookupError(_("tag '%s' does not exist")
2230 2230 % pattern)
2231 2231 s = set([repo[tn].rev()])
2232 2232 else:
2233 2233 s = set([cl.rev(n) for t, n in repo.tagslist() if matcher(t)])
2234 2234 else:
2235 2235 s = set([cl.rev(n) for t, n in repo.tagslist() if t != 'tip'])
2236 2236 return subset & s
2237 2237
2238 2238 @predicate('tagged', safe=True)
2239 2239 def tagged(repo, subset, x):
2240 2240 return tag(repo, subset, x)
2241 2241
2242 2242 @predicate('unstable()', safe=True)
2243 2243 def unstable(repo, subset, x):
2244 2244 """Non-obsolete changesets with obsolete ancestors.
2245 2245 """
2246 2246 # i18n: "unstable" is a keyword
2247 2247 getargs(x, 0, 0, _("unstable takes no arguments"))
2248 2248 unstables = obsmod.getrevs(repo, 'unstable')
2249 2249 return subset & unstables
2250 2250
2251 2251
2252 2252 @predicate('user(string)', safe=True)
2253 2253 def user(repo, subset, x):
2254 2254 """User name contains string. The match is case-insensitive.
2255 2255
2256 2256 If `string` starts with `re:`, the remainder of the string is treated as
2257 2257 a regular expression. To match a user that actually contains `re:`, use
2258 2258 the prefix `literal:`.
2259 2259 """
2260 2260 return author(repo, subset, x)
2261 2261
2262 2262 # experimental
2263 2263 @predicate('wdir', safe=True)
2264 2264 def wdir(repo, subset, x):
2265 2265 # i18n: "wdir" is a keyword
2266 2266 getargs(x, 0, 0, _("wdir takes no arguments"))
2267 2267 if node.wdirrev in subset or isinstance(subset, fullreposet):
2268 2268 return baseset([node.wdirrev])
2269 2269 return baseset()
2270 2270
2271 2271 def _orderedlist(repo, subset, x):
2272 2272 s = getstring(x, "internal error")
2273 2273 if not s:
2274 2274 return baseset()
2275 2275 # remove duplicates here. it's difficult for caller to deduplicate sets
2276 2276 # because different symbols can point to the same rev.
2277 2277 cl = repo.changelog
2278 2278 ls = []
2279 2279 seen = set()
2280 2280 for t in s.split('\0'):
2281 2281 try:
2282 2282 # fast path for integer revision
2283 2283 r = int(t)
2284 2284 if str(r) != t or r not in cl:
2285 2285 raise ValueError
2286 2286 revs = [r]
2287 2287 except ValueError:
2288 2288 revs = stringset(repo, subset, t)
2289 2289
2290 2290 for r in revs:
2291 2291 if r in seen:
2292 2292 continue
2293 2293 if (r in subset
2294 2294 or r == node.nullrev and isinstance(subset, fullreposet)):
2295 2295 ls.append(r)
2296 2296 seen.add(r)
2297 2297 return baseset(ls)
2298 2298
2299 2299 # for internal use
2300 2300 @predicate('_list', safe=True, takeorder=True)
2301 2301 def _list(repo, subset, x, order):
2302 2302 if order == followorder:
2303 2303 # slow path to take the subset order
2304 2304 return subset & _orderedlist(repo, fullreposet(repo), x)
2305 2305 else:
2306 2306 return _orderedlist(repo, subset, x)
2307 2307
2308 2308 def _orderedintlist(repo, subset, x):
2309 2309 s = getstring(x, "internal error")
2310 2310 if not s:
2311 2311 return baseset()
2312 2312 ls = [int(r) for r in s.split('\0')]
2313 2313 s = subset
2314 2314 return baseset([r for r in ls if r in s])
2315 2315
2316 2316 # for internal use
2317 2317 @predicate('_intlist', safe=True, takeorder=True)
2318 2318 def _intlist(repo, subset, x, order):
2319 2319 if order == followorder:
2320 2320 # slow path to take the subset order
2321 2321 return subset & _orderedintlist(repo, fullreposet(repo), x)
2322 2322 else:
2323 2323 return _orderedintlist(repo, subset, x)
2324 2324
2325 2325 def _orderedhexlist(repo, subset, x):
2326 2326 s = getstring(x, "internal error")
2327 2327 if not s:
2328 2328 return baseset()
2329 2329 cl = repo.changelog
2330 2330 ls = [cl.rev(node.bin(r)) for r in s.split('\0')]
2331 2331 s = subset
2332 2332 return baseset([r for r in ls if r in s])
2333 2333
2334 2334 # for internal use
2335 2335 @predicate('_hexlist', safe=True, takeorder=True)
2336 2336 def _hexlist(repo, subset, x, order):
2337 2337 if order == followorder:
2338 2338 # slow path to take the subset order
2339 2339 return subset & _orderedhexlist(repo, fullreposet(repo), x)
2340 2340 else:
2341 2341 return _orderedhexlist(repo, subset, x)
2342 2342
2343 2343 methods = {
2344 2344 "range": rangeset,
2345 2345 "rangepre": rangepre,
2346 2346 "dagrange": dagrange,
2347 2347 "string": stringset,
2348 2348 "symbol": stringset,
2349 2349 "and": andset,
2350 2350 "or": orset,
2351 2351 "not": notset,
2352 2352 "difference": differenceset,
2353 2353 "list": listset,
2354 2354 "keyvalue": keyvaluepair,
2355 2355 "func": func,
2356 2356 "ancestor": ancestorspec,
2357 2357 "parent": parentspec,
2358 2358 "parentpost": parentpost,
2359 2359 }
2360 2360
2361 2361 # Constants for ordering requirement, used in _analyze():
2362 2362 #
2363 2363 # If 'define', any nested functions and operations can change the ordering of
2364 2364 # the entries in the set. If 'follow', any nested functions and operations
2365 2365 # should take the ordering specified by the first operand to the '&' operator.
2366 2366 #
2367 2367 # For instance,
2368 2368 #
2369 2369 # X & (Y | Z)
2370 2370 # ^ ^^^^^^^
2371 2371 # | follow
2372 2372 # define
2373 2373 #
2374 2374 # will be evaluated as 'or(y(x()), z(x()))', where 'x()' can change the order
2375 2375 # of the entries in the set, but 'y()', 'z()' and 'or()' shouldn't.
2376 2376 #
2377 2377 # 'any' means the order doesn't matter. For instance,
2378 2378 #
2379 2379 # X & !Y
2380 2380 # ^
2381 2381 # any
2382 2382 #
2383 2383 # 'y()' can either enforce its ordering requirement or take the ordering
2384 2384 # specified by 'x()' because 'not()' doesn't care the order.
2385 2385 #
2386 2386 # Transition of ordering requirement:
2387 2387 #
2388 2388 # 1. starts with 'define'
2389 2389 # 2. shifts to 'follow' by 'x & y'
2390 2390 # 3. changes back to 'define' on function call 'f(x)' or function-like
2391 2391 # operation 'x (f) y' because 'f' may have its own ordering requirement
2392 2392 # for 'x' and 'y' (e.g. 'first(x)')
2393 2393 #
2394 2394 anyorder = 'any' # don't care the order
2395 2395 defineorder = 'define' # should define the order
2396 2396 followorder = 'follow' # must follow the current order
2397 2397
2398 2398 # transition table for 'x & y', from the current expression 'x' to 'y'
2399 2399 _tofolloworder = {
2400 2400 anyorder: anyorder,
2401 2401 defineorder: followorder,
2402 2402 followorder: followorder,
2403 2403 }
2404 2404
2405 2405 def _matchonly(revs, bases):
2406 2406 """
2407 2407 >>> f = lambda *args: _matchonly(*map(parse, args))
2408 2408 >>> f('ancestors(A)', 'not ancestors(B)')
2409 2409 ('list', ('symbol', 'A'), ('symbol', 'B'))
2410 2410 """
2411 2411 if (revs is not None
2412 2412 and revs[0] == 'func'
2413 2413 and getsymbol(revs[1]) == 'ancestors'
2414 2414 and bases is not None
2415 2415 and bases[0] == 'not'
2416 2416 and bases[1][0] == 'func'
2417 2417 and getsymbol(bases[1][1]) == 'ancestors'):
2418 2418 return ('list', revs[2], bases[1][2])
2419 2419
2420 2420 def _fixops(x):
2421 2421 """Rewrite raw parsed tree to resolve ambiguous syntax which cannot be
2422 2422 handled well by our simple top-down parser"""
2423 2423 if not isinstance(x, tuple):
2424 2424 return x
2425 2425
2426 2426 op = x[0]
2427 2427 if op == 'parent':
2428 2428 # x^:y means (x^) : y, not x ^ (:y)
2429 2429 # x^: means (x^) :, not x ^ (:)
2430 2430 post = ('parentpost', x[1])
2431 2431 if x[2][0] == 'dagrangepre':
2432 2432 return _fixops(('dagrange', post, x[2][1]))
2433 2433 elif x[2][0] == 'rangepre':
2434 2434 return _fixops(('range', post, x[2][1]))
2435 2435 elif x[2][0] == 'rangeall':
2436 2436 return _fixops(('rangepost', post))
2437 2437 elif op == 'or':
2438 2438 # make number of arguments deterministic:
2439 2439 # x + y + z -> (or x y z) -> (or (list x y z))
2440 2440 return (op, _fixops(('list',) + x[1:]))
2441 2441
2442 2442 return (op,) + tuple(_fixops(y) for y in x[1:])
2443 2443
2444 2444 def _analyze(x, order):
2445 2445 if x is None:
2446 2446 return x
2447 2447
2448 2448 op = x[0]
2449 2449 if op == 'minus':
2450 2450 return _analyze(('and', x[1], ('not', x[2])), order)
2451 2451 elif op == 'only':
2452 2452 t = ('func', ('symbol', 'only'), ('list', x[1], x[2]))
2453 2453 return _analyze(t, order)
2454 2454 elif op == 'onlypost':
2455 2455 return _analyze(('func', ('symbol', 'only'), x[1]), order)
2456 2456 elif op == 'dagrangepre':
2457 2457 return _analyze(('func', ('symbol', 'ancestors'), x[1]), order)
2458 2458 elif op == 'dagrangepost':
2459 2459 return _analyze(('func', ('symbol', 'descendants'), x[1]), order)
2460 2460 elif op == 'rangeall':
2461 2461 return _analyze(('rangepre', ('string', 'tip')), order)
2462 2462 elif op == 'rangepost':
2463 2463 return _analyze(('range', x[1], ('string', 'tip')), order)
2464 2464 elif op == 'negate':
2465 2465 s = getstring(x[1], _("can't negate that"))
2466 2466 return _analyze(('string', '-' + s), order)
2467 2467 elif op in ('string', 'symbol'):
2468 2468 return x
2469 2469 elif op == 'and':
2470 2470 ta = _analyze(x[1], order)
2471 2471 tb = _analyze(x[2], _tofolloworder[order])
2472 2472 return (op, ta, tb, order)
2473 2473 elif op == 'or':
2474 2474 return (op, _analyze(x[1], order), order)
2475 2475 elif op == 'not':
2476 2476 return (op, _analyze(x[1], anyorder), order)
2477 2477 elif op in ('rangepre', 'parentpost'):
2478 2478 return (op, _analyze(x[1], defineorder), order)
2479 2479 elif op == 'group':
2480 2480 return _analyze(x[1], order)
2481 2481 elif op in ('dagrange', 'range', 'parent', 'ancestor'):
2482 2482 ta = _analyze(x[1], defineorder)
2483 2483 tb = _analyze(x[2], defineorder)
2484 2484 return (op, ta, tb, order)
2485 2485 elif op == 'list':
2486 2486 return (op,) + tuple(_analyze(y, order) for y in x[1:])
2487 2487 elif op == 'keyvalue':
2488 2488 return (op, x[1], _analyze(x[2], order))
2489 2489 elif op == 'func':
2490 2490 f = getsymbol(x[1])
2491 2491 d = defineorder
2492 2492 if f == 'present':
2493 2493 # 'present(set)' is known to return the argument set with no
2494 2494 # modification, so forward the current order to its argument
2495 2495 d = order
2496 2496 return (op, x[1], _analyze(x[2], d), order)
2497 2497 raise ValueError('invalid operator %r' % op)
2498 2498
2499 2499 def analyze(x, order=defineorder):
2500 2500 """Transform raw parsed tree to evaluatable tree which can be fed to
2501 2501 optimize() or getset()
2502 2502
2503 2503 All pseudo operations should be mapped to real operations or functions
2504 2504 defined in methods or symbols table respectively.
2505 2505
2506 2506 'order' specifies how the current expression 'x' is ordered (see the
2507 2507 constants defined above.)
2508 2508 """
2509 2509 return _analyze(x, order)
2510 2510
2511 2511 def _optimize(x, small):
2512 2512 if x is None:
2513 2513 return 0, x
2514 2514
2515 2515 smallbonus = 1
2516 2516 if small:
2517 2517 smallbonus = .5
2518 2518
2519 2519 op = x[0]
2520 2520 if op in ('string', 'symbol'):
2521 2521 return smallbonus, x # single revisions are small
2522 2522 elif op == 'and':
2523 2523 wa, ta = _optimize(x[1], True)
2524 2524 wb, tb = _optimize(x[2], True)
2525 2525 order = x[3]
2526 2526 w = min(wa, wb)
2527 2527
2528 2528 # (::x and not ::y)/(not ::y and ::x) have a fast path
2529 2529 tm = _matchonly(ta, tb) or _matchonly(tb, ta)
2530 2530 if tm:
2531 2531 return w, ('func', ('symbol', 'only'), tm, order)
2532 2532
2533 2533 if tb is not None and tb[0] == 'not':
2534 2534 return wa, ('difference', ta, tb[1], order)
2535 2535
2536 2536 if wa > wb:
2537 2537 return w, (op, tb, ta, order)
2538 2538 return w, (op, ta, tb, order)
2539 2539 elif op == 'or':
2540 2540 # fast path for machine-generated expression, that is likely to have
2541 2541 # lots of trivial revisions: 'a + b + c()' to '_list(a b) + c()'
2542 2542 order = x[2]
2543 2543 ws, ts, ss = [], [], []
2544 2544 def flushss():
2545 2545 if not ss:
2546 2546 return
2547 2547 if len(ss) == 1:
2548 2548 w, t = ss[0]
2549 2549 else:
2550 2550 s = '\0'.join(t[1] for w, t in ss)
2551 2551 y = ('func', ('symbol', '_list'), ('string', s), order)
2552 2552 w, t = _optimize(y, False)
2553 2553 ws.append(w)
2554 2554 ts.append(t)
2555 2555 del ss[:]
2556 2556 for y in getlist(x[1]):
2557 2557 w, t = _optimize(y, False)
2558 2558 if t is not None and (t[0] == 'string' or t[0] == 'symbol'):
2559 2559 ss.append((w, t))
2560 2560 continue
2561 2561 flushss()
2562 2562 ws.append(w)
2563 2563 ts.append(t)
2564 2564 flushss()
2565 2565 if len(ts) == 1:
2566 2566 return ws[0], ts[0] # 'or' operation is fully optimized out
2567 2567 # we can't reorder trees by weight because it would change the order.
2568 2568 # ("sort(a + b)" == "sort(b + a)", but "a + b" != "b + a")
2569 2569 # ts = tuple(t for w, t in sorted(zip(ws, ts), key=lambda wt: wt[0]))
2570 2570 return max(ws), (op, ('list',) + tuple(ts), order)
2571 2571 elif op == 'not':
2572 2572 # Optimize not public() to _notpublic() because we have a fast version
2573 2573 if x[1][:3] == ('func', ('symbol', 'public'), None):
2574 2574 order = x[1][3]
2575 2575 newsym = ('func', ('symbol', '_notpublic'), None, order)
2576 2576 o = _optimize(newsym, not small)
2577 2577 return o[0], o[1]
2578 2578 else:
2579 2579 o = _optimize(x[1], not small)
2580 2580 order = x[2]
2581 2581 return o[0], (op, o[1], order)
2582 2582 elif op in ('rangepre', 'parentpost'):
2583 2583 o = _optimize(x[1], small)
2584 2584 order = x[2]
2585 2585 return o[0], (op, o[1], order)
2586 2586 elif op in ('dagrange', 'range', 'parent', 'ancestor'):
2587 2587 wa, ta = _optimize(x[1], small)
2588 2588 wb, tb = _optimize(x[2], small)
2589 2589 order = x[3]
2590 2590 return wa + wb, (op, ta, tb, order)
2591 2591 elif op == 'list':
2592 2592 ws, ts = zip(*(_optimize(y, small) for y in x[1:]))
2593 2593 return sum(ws), (op,) + ts
2594 2594 elif op == 'keyvalue':
2595 2595 w, t = _optimize(x[2], small)
2596 2596 return w, (op, x[1], t)
2597 2597 elif op == 'func':
2598 2598 f = getsymbol(x[1])
2599 2599 wa, ta = _optimize(x[2], small)
2600 2600 if f in ('author', 'branch', 'closed', 'date', 'desc', 'file', 'grep',
2601 2601 'keyword', 'outgoing', 'user', 'destination'):
2602 2602 w = 10 # slow
2603 2603 elif f in ('modifies', 'adds', 'removes'):
2604 2604 w = 30 # slower
2605 2605 elif f == "contains":
2606 2606 w = 100 # very slow
2607 2607 elif f == "ancestor":
2608 2608 w = 1 * smallbonus
2609 elif f in ('reverse', 'limit', 'first', '_intlist'):
2609 elif f in ('reverse', 'limit', 'first', 'wdir', '_intlist'):
2610 2610 w = 0
2611 2611 elif f == "sort":
2612 2612 w = 10 # assume most sorts look at changelog
2613 2613 else:
2614 2614 w = 1
2615 2615 order = x[3]
2616 2616 return w + wa, (op, x[1], ta, order)
2617 2617 raise ValueError('invalid operator %r' % op)
2618 2618
2619 2619 def optimize(tree):
2620 2620 """Optimize evaluatable tree
2621 2621
2622 2622 All pseudo operations should be transformed beforehand.
2623 2623 """
2624 2624 _weight, newtree = _optimize(tree, small=True)
2625 2625 return newtree
2626 2626
2627 2627 # the set of valid characters for the initial letter of symbols in
2628 2628 # alias declarations and definitions
2629 2629 _aliassyminitletters = _syminitletters | set(pycompat.sysstr('$'))
2630 2630
2631 2631 def _parsewith(spec, lookup=None, syminitletters=None):
2632 2632 """Generate a parse tree of given spec with given tokenizing options
2633 2633
2634 2634 >>> _parsewith('foo($1)', syminitletters=_aliassyminitletters)
2635 2635 ('func', ('symbol', 'foo'), ('symbol', '$1'))
2636 2636 >>> _parsewith('$1')
2637 2637 Traceback (most recent call last):
2638 2638 ...
2639 2639 ParseError: ("syntax error in revset '$1'", 0)
2640 2640 >>> _parsewith('foo bar')
2641 2641 Traceback (most recent call last):
2642 2642 ...
2643 2643 ParseError: ('invalid token', 4)
2644 2644 """
2645 2645 p = parser.parser(elements)
2646 2646 tree, pos = p.parse(tokenize(spec, lookup=lookup,
2647 2647 syminitletters=syminitletters))
2648 2648 if pos != len(spec):
2649 2649 raise error.ParseError(_('invalid token'), pos)
2650 2650 return _fixops(parser.simplifyinfixops(tree, ('list', 'or')))
2651 2651
2652 2652 class _aliasrules(parser.basealiasrules):
2653 2653 """Parsing and expansion rule set of revset aliases"""
2654 2654 _section = _('revset alias')
2655 2655
2656 2656 @staticmethod
2657 2657 def _parse(spec):
2658 2658 """Parse alias declaration/definition ``spec``
2659 2659
2660 2660 This allows symbol names to use also ``$`` as an initial letter
2661 2661 (for backward compatibility), and callers of this function should
2662 2662 examine whether ``$`` is used also for unexpected symbols or not.
2663 2663 """
2664 2664 return _parsewith(spec, syminitletters=_aliassyminitletters)
2665 2665
2666 2666 @staticmethod
2667 2667 def _trygetfunc(tree):
2668 2668 if tree[0] == 'func' and tree[1][0] == 'symbol':
2669 2669 return tree[1][1], getlist(tree[2])
2670 2670
2671 2671 def expandaliases(ui, tree):
2672 2672 aliases = _aliasrules.buildmap(ui.configitems('revsetalias'))
2673 2673 tree = _aliasrules.expand(aliases, tree)
2674 2674 # warn about problematic (but not referred) aliases
2675 2675 for name, alias in sorted(aliases.iteritems()):
2676 2676 if alias.error and not alias.warned:
2677 2677 ui.warn(_('warning: %s\n') % (alias.error))
2678 2678 alias.warned = True
2679 2679 return tree
2680 2680
2681 2681 def foldconcat(tree):
2682 2682 """Fold elements to be concatenated by `##`
2683 2683 """
2684 2684 if not isinstance(tree, tuple) or tree[0] in ('string', 'symbol'):
2685 2685 return tree
2686 2686 if tree[0] == '_concat':
2687 2687 pending = [tree]
2688 2688 l = []
2689 2689 while pending:
2690 2690 e = pending.pop()
2691 2691 if e[0] == '_concat':
2692 2692 pending.extend(reversed(e[1:]))
2693 2693 elif e[0] in ('string', 'symbol'):
2694 2694 l.append(e[1])
2695 2695 else:
2696 2696 msg = _("\"##\" can't concatenate \"%s\" element") % (e[0])
2697 2697 raise error.ParseError(msg)
2698 2698 return ('string', ''.join(l))
2699 2699 else:
2700 2700 return tuple(foldconcat(t) for t in tree)
2701 2701
2702 2702 def parse(spec, lookup=None):
2703 2703 return _parsewith(spec, lookup=lookup)
2704 2704
2705 2705 def posttreebuilthook(tree, repo):
2706 2706 # hook for extensions to execute code on the optimized tree
2707 2707 pass
2708 2708
2709 2709 def match(ui, spec, repo=None, order=defineorder):
2710 2710 """Create a matcher for a single revision spec
2711 2711
2712 2712 If order=followorder, a matcher takes the ordering specified by the input
2713 2713 set.
2714 2714 """
2715 2715 return matchany(ui, [spec], repo=repo, order=order)
2716 2716
2717 2717 def matchany(ui, specs, repo=None, order=defineorder):
2718 2718 """Create a matcher that will include any revisions matching one of the
2719 2719 given specs
2720 2720
2721 2721 If order=followorder, a matcher takes the ordering specified by the input
2722 2722 set.
2723 2723 """
2724 2724 if not specs:
2725 2725 def mfunc(repo, subset=None):
2726 2726 return baseset()
2727 2727 return mfunc
2728 2728 if not all(specs):
2729 2729 raise error.ParseError(_("empty query"))
2730 2730 lookup = None
2731 2731 if repo:
2732 2732 lookup = repo.__contains__
2733 2733 if len(specs) == 1:
2734 2734 tree = parse(specs[0], lookup)
2735 2735 else:
2736 2736 tree = ('or', ('list',) + tuple(parse(s, lookup) for s in specs))
2737 2737
2738 2738 if ui:
2739 2739 tree = expandaliases(ui, tree)
2740 2740 tree = foldconcat(tree)
2741 2741 tree = analyze(tree, order)
2742 2742 tree = optimize(tree)
2743 2743 posttreebuilthook(tree, repo)
2744 2744 return makematcher(tree)
2745 2745
2746 2746 def makematcher(tree):
2747 2747 """Create a matcher from an evaluatable tree"""
2748 2748 def mfunc(repo, subset=None):
2749 2749 if subset is None:
2750 2750 subset = fullreposet(repo)
2751 2751 if util.safehasattr(subset, 'isascending'):
2752 2752 result = getset(repo, subset, tree)
2753 2753 else:
2754 2754 result = getset(repo, baseset(subset), tree)
2755 2755 return result
2756 2756 return mfunc
2757 2757
2758 2758 def formatspec(expr, *args):
2759 2759 '''
2760 2760 This is a convenience function for using revsets internally, and
2761 2761 escapes arguments appropriately. Aliases are intentionally ignored
2762 2762 so that intended expression behavior isn't accidentally subverted.
2763 2763
2764 2764 Supported arguments:
2765 2765
2766 2766 %r = revset expression, parenthesized
2767 2767 %d = int(arg), no quoting
2768 2768 %s = string(arg), escaped and single-quoted
2769 2769 %b = arg.branch(), escaped and single-quoted
2770 2770 %n = hex(arg), single-quoted
2771 2771 %% = a literal '%'
2772 2772
2773 2773 Prefixing the type with 'l' specifies a parenthesized list of that type.
2774 2774
2775 2775 >>> formatspec('%r:: and %lr', '10 or 11', ("this()", "that()"))
2776 2776 '(10 or 11):: and ((this()) or (that()))'
2777 2777 >>> formatspec('%d:: and not %d::', 10, 20)
2778 2778 '10:: and not 20::'
2779 2779 >>> formatspec('%ld or %ld', [], [1])
2780 2780 "_list('') or 1"
2781 2781 >>> formatspec('keyword(%s)', 'foo\\xe9')
2782 2782 "keyword('foo\\\\xe9')"
2783 2783 >>> b = lambda: 'default'
2784 2784 >>> b.branch = b
2785 2785 >>> formatspec('branch(%b)', b)
2786 2786 "branch('default')"
2787 2787 >>> formatspec('root(%ls)', ['a', 'b', 'c', 'd'])
2788 2788 "root(_list('a\\x00b\\x00c\\x00d'))"
2789 2789 '''
2790 2790
2791 2791 def quote(s):
2792 2792 return repr(str(s))
2793 2793
2794 2794 def argtype(c, arg):
2795 2795 if c == 'd':
2796 2796 return str(int(arg))
2797 2797 elif c == 's':
2798 2798 return quote(arg)
2799 2799 elif c == 'r':
2800 2800 parse(arg) # make sure syntax errors are confined
2801 2801 return '(%s)' % arg
2802 2802 elif c == 'n':
2803 2803 return quote(node.hex(arg))
2804 2804 elif c == 'b':
2805 2805 return quote(arg.branch())
2806 2806
2807 2807 def listexp(s, t):
2808 2808 l = len(s)
2809 2809 if l == 0:
2810 2810 return "_list('')"
2811 2811 elif l == 1:
2812 2812 return argtype(t, s[0])
2813 2813 elif t == 'd':
2814 2814 return "_intlist('%s')" % "\0".join(str(int(a)) for a in s)
2815 2815 elif t == 's':
2816 2816 return "_list('%s')" % "\0".join(s)
2817 2817 elif t == 'n':
2818 2818 return "_hexlist('%s')" % "\0".join(node.hex(a) for a in s)
2819 2819 elif t == 'b':
2820 2820 return "_list('%s')" % "\0".join(a.branch() for a in s)
2821 2821
2822 2822 m = l // 2
2823 2823 return '(%s or %s)' % (listexp(s[:m], t), listexp(s[m:], t))
2824 2824
2825 2825 ret = ''
2826 2826 pos = 0
2827 2827 arg = 0
2828 2828 while pos < len(expr):
2829 2829 c = expr[pos]
2830 2830 if c == '%':
2831 2831 pos += 1
2832 2832 d = expr[pos]
2833 2833 if d == '%':
2834 2834 ret += d
2835 2835 elif d in 'dsnbr':
2836 2836 ret += argtype(d, args[arg])
2837 2837 arg += 1
2838 2838 elif d == 'l':
2839 2839 # a list of some type
2840 2840 pos += 1
2841 2841 d = expr[pos]
2842 2842 ret += listexp(list(args[arg]), d)
2843 2843 arg += 1
2844 2844 else:
2845 2845 raise error.Abort(_('unexpected revspec format character %s')
2846 2846 % d)
2847 2847 else:
2848 2848 ret += c
2849 2849 pos += 1
2850 2850
2851 2851 return ret
2852 2852
2853 2853 def prettyformat(tree):
2854 2854 return parser.prettyformat(tree, ('string', 'symbol'))
2855 2855
2856 2856 def depth(tree):
2857 2857 if isinstance(tree, tuple):
2858 2858 return max(map(depth, tree)) + 1
2859 2859 else:
2860 2860 return 0
2861 2861
2862 2862 def funcsused(tree):
2863 2863 if not isinstance(tree, tuple) or tree[0] in ('string', 'symbol'):
2864 2864 return set()
2865 2865 else:
2866 2866 funcs = set()
2867 2867 for s in tree[1:]:
2868 2868 funcs |= funcsused(s)
2869 2869 if tree[0] == 'func':
2870 2870 funcs.add(tree[1][1])
2871 2871 return funcs
2872 2872
2873 2873 def _formatsetrepr(r):
2874 2874 """Format an optional printable representation of a set
2875 2875
2876 2876 ======== =================================
2877 2877 type(r) example
2878 2878 ======== =================================
2879 2879 tuple ('<not %r>', other)
2880 2880 str '<branch closed>'
2881 2881 callable lambda: '<branch %r>' % sorted(b)
2882 2882 object other
2883 2883 ======== =================================
2884 2884 """
2885 2885 if r is None:
2886 2886 return ''
2887 2887 elif isinstance(r, tuple):
2888 2888 return r[0] % r[1:]
2889 2889 elif isinstance(r, str):
2890 2890 return r
2891 2891 elif callable(r):
2892 2892 return r()
2893 2893 else:
2894 2894 return repr(r)
2895 2895
2896 2896 class abstractsmartset(object):
2897 2897
2898 2898 def __nonzero__(self):
2899 2899 """True if the smartset is not empty"""
2900 2900 raise NotImplementedError()
2901 2901
2902 2902 def __contains__(self, rev):
2903 2903 """provide fast membership testing"""
2904 2904 raise NotImplementedError()
2905 2905
2906 2906 def __iter__(self):
2907 2907 """iterate the set in the order it is supposed to be iterated"""
2908 2908 raise NotImplementedError()
2909 2909
2910 2910 # Attributes containing a function to perform a fast iteration in a given
2911 2911 # direction. A smartset can have none, one, or both defined.
2912 2912 #
2913 2913 # Default value is None instead of a function returning None to avoid
2914 2914 # initializing an iterator just for testing if a fast method exists.
2915 2915 fastasc = None
2916 2916 fastdesc = None
2917 2917
2918 2918 def isascending(self):
2919 2919 """True if the set will iterate in ascending order"""
2920 2920 raise NotImplementedError()
2921 2921
2922 2922 def isdescending(self):
2923 2923 """True if the set will iterate in descending order"""
2924 2924 raise NotImplementedError()
2925 2925
2926 2926 def istopo(self):
2927 2927 """True if the set will iterate in topographical order"""
2928 2928 raise NotImplementedError()
2929 2929
2930 2930 def min(self):
2931 2931 """return the minimum element in the set"""
2932 2932 if self.fastasc is None:
2933 2933 v = min(self)
2934 2934 else:
2935 2935 for v in self.fastasc():
2936 2936 break
2937 2937 else:
2938 2938 raise ValueError('arg is an empty sequence')
2939 2939 self.min = lambda: v
2940 2940 return v
2941 2941
2942 2942 def max(self):
2943 2943 """return the maximum element in the set"""
2944 2944 if self.fastdesc is None:
2945 2945 return max(self)
2946 2946 else:
2947 2947 for v in self.fastdesc():
2948 2948 break
2949 2949 else:
2950 2950 raise ValueError('arg is an empty sequence')
2951 2951 self.max = lambda: v
2952 2952 return v
2953 2953
2954 2954 def first(self):
2955 2955 """return the first element in the set (user iteration perspective)
2956 2956
2957 2957 Return None if the set is empty"""
2958 2958 raise NotImplementedError()
2959 2959
2960 2960 def last(self):
2961 2961 """return the last element in the set (user iteration perspective)
2962 2962
2963 2963 Return None if the set is empty"""
2964 2964 raise NotImplementedError()
2965 2965
2966 2966 def __len__(self):
2967 2967 """return the length of the smartsets
2968 2968
2969 2969 This can be expensive on smartset that could be lazy otherwise."""
2970 2970 raise NotImplementedError()
2971 2971
2972 2972 def reverse(self):
2973 2973 """reverse the expected iteration order"""
2974 2974 raise NotImplementedError()
2975 2975
2976 2976 def sort(self, reverse=True):
2977 2977 """get the set to iterate in an ascending or descending order"""
2978 2978 raise NotImplementedError()
2979 2979
2980 2980 def __and__(self, other):
2981 2981 """Returns a new object with the intersection of the two collections.
2982 2982
2983 2983 This is part of the mandatory API for smartset."""
2984 2984 if isinstance(other, fullreposet):
2985 2985 return self
2986 2986 return self.filter(other.__contains__, condrepr=other, cache=False)
2987 2987
2988 2988 def __add__(self, other):
2989 2989 """Returns a new object with the union of the two collections.
2990 2990
2991 2991 This is part of the mandatory API for smartset."""
2992 2992 return addset(self, other)
2993 2993
2994 2994 def __sub__(self, other):
2995 2995 """Returns a new object with the substraction of the two collections.
2996 2996
2997 2997 This is part of the mandatory API for smartset."""
2998 2998 c = other.__contains__
2999 2999 return self.filter(lambda r: not c(r), condrepr=('<not %r>', other),
3000 3000 cache=False)
3001 3001
3002 3002 def filter(self, condition, condrepr=None, cache=True):
3003 3003 """Returns this smartset filtered by condition as a new smartset.
3004 3004
3005 3005 `condition` is a callable which takes a revision number and returns a
3006 3006 boolean. Optional `condrepr` provides a printable representation of
3007 3007 the given `condition`.
3008 3008
3009 3009 This is part of the mandatory API for smartset."""
3010 3010 # builtin cannot be cached. but do not needs to
3011 3011 if cache and util.safehasattr(condition, 'func_code'):
3012 3012 condition = util.cachefunc(condition)
3013 3013 return filteredset(self, condition, condrepr)
3014 3014
3015 3015 class baseset(abstractsmartset):
3016 3016 """Basic data structure that represents a revset and contains the basic
3017 3017 operation that it should be able to perform.
3018 3018
3019 3019 Every method in this class should be implemented by any smartset class.
3020 3020 """
3021 3021 def __init__(self, data=(), datarepr=None, istopo=False):
3022 3022 """
3023 3023 datarepr: a tuple of (format, obj, ...), a function or an object that
3024 3024 provides a printable representation of the given data.
3025 3025 """
3026 3026 self._ascending = None
3027 3027 self._istopo = istopo
3028 3028 if not isinstance(data, list):
3029 3029 if isinstance(data, set):
3030 3030 self._set = data
3031 3031 # set has no order we pick one for stability purpose
3032 3032 self._ascending = True
3033 3033 data = list(data)
3034 3034 self._list = data
3035 3035 self._datarepr = datarepr
3036 3036
3037 3037 @util.propertycache
3038 3038 def _set(self):
3039 3039 return set(self._list)
3040 3040
3041 3041 @util.propertycache
3042 3042 def _asclist(self):
3043 3043 asclist = self._list[:]
3044 3044 asclist.sort()
3045 3045 return asclist
3046 3046
3047 3047 def __iter__(self):
3048 3048 if self._ascending is None:
3049 3049 return iter(self._list)
3050 3050 elif self._ascending:
3051 3051 return iter(self._asclist)
3052 3052 else:
3053 3053 return reversed(self._asclist)
3054 3054
3055 3055 def fastasc(self):
3056 3056 return iter(self._asclist)
3057 3057
3058 3058 def fastdesc(self):
3059 3059 return reversed(self._asclist)
3060 3060
3061 3061 @util.propertycache
3062 3062 def __contains__(self):
3063 3063 return self._set.__contains__
3064 3064
3065 3065 def __nonzero__(self):
3066 3066 return bool(self._list)
3067 3067
3068 3068 def sort(self, reverse=False):
3069 3069 self._ascending = not bool(reverse)
3070 3070 self._istopo = False
3071 3071
3072 3072 def reverse(self):
3073 3073 if self._ascending is None:
3074 3074 self._list.reverse()
3075 3075 else:
3076 3076 self._ascending = not self._ascending
3077 3077 self._istopo = False
3078 3078
3079 3079 def __len__(self):
3080 3080 return len(self._list)
3081 3081
3082 3082 def isascending(self):
3083 3083 """Returns True if the collection is ascending order, False if not.
3084 3084
3085 3085 This is part of the mandatory API for smartset."""
3086 3086 if len(self) <= 1:
3087 3087 return True
3088 3088 return self._ascending is not None and self._ascending
3089 3089
3090 3090 def isdescending(self):
3091 3091 """Returns True if the collection is descending order, False if not.
3092 3092
3093 3093 This is part of the mandatory API for smartset."""
3094 3094 if len(self) <= 1:
3095 3095 return True
3096 3096 return self._ascending is not None and not self._ascending
3097 3097
3098 3098 def istopo(self):
3099 3099 """Is the collection is in topographical order or not.
3100 3100
3101 3101 This is part of the mandatory API for smartset."""
3102 3102 if len(self) <= 1:
3103 3103 return True
3104 3104 return self._istopo
3105 3105
3106 3106 def first(self):
3107 3107 if self:
3108 3108 if self._ascending is None:
3109 3109 return self._list[0]
3110 3110 elif self._ascending:
3111 3111 return self._asclist[0]
3112 3112 else:
3113 3113 return self._asclist[-1]
3114 3114 return None
3115 3115
3116 3116 def last(self):
3117 3117 if self:
3118 3118 if self._ascending is None:
3119 3119 return self._list[-1]
3120 3120 elif self._ascending:
3121 3121 return self._asclist[-1]
3122 3122 else:
3123 3123 return self._asclist[0]
3124 3124 return None
3125 3125
3126 3126 def __repr__(self):
3127 3127 d = {None: '', False: '-', True: '+'}[self._ascending]
3128 3128 s = _formatsetrepr(self._datarepr)
3129 3129 if not s:
3130 3130 l = self._list
3131 3131 # if _list has been built from a set, it might have a different
3132 3132 # order from one python implementation to another.
3133 3133 # We fallback to the sorted version for a stable output.
3134 3134 if self._ascending is not None:
3135 3135 l = self._asclist
3136 3136 s = repr(l)
3137 3137 return '<%s%s %s>' % (type(self).__name__, d, s)
3138 3138
3139 3139 class filteredset(abstractsmartset):
3140 3140 """Duck type for baseset class which iterates lazily over the revisions in
3141 3141 the subset and contains a function which tests for membership in the
3142 3142 revset
3143 3143 """
3144 3144 def __init__(self, subset, condition=lambda x: True, condrepr=None):
3145 3145 """
3146 3146 condition: a function that decide whether a revision in the subset
3147 3147 belongs to the revset or not.
3148 3148 condrepr: a tuple of (format, obj, ...), a function or an object that
3149 3149 provides a printable representation of the given condition.
3150 3150 """
3151 3151 self._subset = subset
3152 3152 self._condition = condition
3153 3153 self._condrepr = condrepr
3154 3154
3155 3155 def __contains__(self, x):
3156 3156 return x in self._subset and self._condition(x)
3157 3157
3158 3158 def __iter__(self):
3159 3159 return self._iterfilter(self._subset)
3160 3160
3161 3161 def _iterfilter(self, it):
3162 3162 cond = self._condition
3163 3163 for x in it:
3164 3164 if cond(x):
3165 3165 yield x
3166 3166
3167 3167 @property
3168 3168 def fastasc(self):
3169 3169 it = self._subset.fastasc
3170 3170 if it is None:
3171 3171 return None
3172 3172 return lambda: self._iterfilter(it())
3173 3173
3174 3174 @property
3175 3175 def fastdesc(self):
3176 3176 it = self._subset.fastdesc
3177 3177 if it is None:
3178 3178 return None
3179 3179 return lambda: self._iterfilter(it())
3180 3180
3181 3181 def __nonzero__(self):
3182 3182 fast = None
3183 3183 candidates = [self.fastasc if self.isascending() else None,
3184 3184 self.fastdesc if self.isdescending() else None,
3185 3185 self.fastasc,
3186 3186 self.fastdesc]
3187 3187 for candidate in candidates:
3188 3188 if candidate is not None:
3189 3189 fast = candidate
3190 3190 break
3191 3191
3192 3192 if fast is not None:
3193 3193 it = fast()
3194 3194 else:
3195 3195 it = self
3196 3196
3197 3197 for r in it:
3198 3198 return True
3199 3199 return False
3200 3200
3201 3201 def __len__(self):
3202 3202 # Basic implementation to be changed in future patches.
3203 3203 # until this gets improved, we use generator expression
3204 3204 # here, since list comprehensions are free to call __len__ again
3205 3205 # causing infinite recursion
3206 3206 l = baseset(r for r in self)
3207 3207 return len(l)
3208 3208
3209 3209 def sort(self, reverse=False):
3210 3210 self._subset.sort(reverse=reverse)
3211 3211
3212 3212 def reverse(self):
3213 3213 self._subset.reverse()
3214 3214
3215 3215 def isascending(self):
3216 3216 return self._subset.isascending()
3217 3217
3218 3218 def isdescending(self):
3219 3219 return self._subset.isdescending()
3220 3220
3221 3221 def istopo(self):
3222 3222 return self._subset.istopo()
3223 3223
3224 3224 def first(self):
3225 3225 for x in self:
3226 3226 return x
3227 3227 return None
3228 3228
3229 3229 def last(self):
3230 3230 it = None
3231 3231 if self.isascending():
3232 3232 it = self.fastdesc
3233 3233 elif self.isdescending():
3234 3234 it = self.fastasc
3235 3235 if it is not None:
3236 3236 for x in it():
3237 3237 return x
3238 3238 return None #empty case
3239 3239 else:
3240 3240 x = None
3241 3241 for x in self:
3242 3242 pass
3243 3243 return x
3244 3244
3245 3245 def __repr__(self):
3246 3246 xs = [repr(self._subset)]
3247 3247 s = _formatsetrepr(self._condrepr)
3248 3248 if s:
3249 3249 xs.append(s)
3250 3250 return '<%s %s>' % (type(self).__name__, ', '.join(xs))
3251 3251
3252 3252 def _iterordered(ascending, iter1, iter2):
3253 3253 """produce an ordered iteration from two iterators with the same order
3254 3254
3255 3255 The ascending is used to indicated the iteration direction.
3256 3256 """
3257 3257 choice = max
3258 3258 if ascending:
3259 3259 choice = min
3260 3260
3261 3261 val1 = None
3262 3262 val2 = None
3263 3263 try:
3264 3264 # Consume both iterators in an ordered way until one is empty
3265 3265 while True:
3266 3266 if val1 is None:
3267 3267 val1 = next(iter1)
3268 3268 if val2 is None:
3269 3269 val2 = next(iter2)
3270 3270 n = choice(val1, val2)
3271 3271 yield n
3272 3272 if val1 == n:
3273 3273 val1 = None
3274 3274 if val2 == n:
3275 3275 val2 = None
3276 3276 except StopIteration:
3277 3277 # Flush any remaining values and consume the other one
3278 3278 it = iter2
3279 3279 if val1 is not None:
3280 3280 yield val1
3281 3281 it = iter1
3282 3282 elif val2 is not None:
3283 3283 # might have been equality and both are empty
3284 3284 yield val2
3285 3285 for val in it:
3286 3286 yield val
3287 3287
3288 3288 class addset(abstractsmartset):
3289 3289 """Represent the addition of two sets
3290 3290
3291 3291 Wrapper structure for lazily adding two structures without losing much
3292 3292 performance on the __contains__ method
3293 3293
3294 3294 If the ascending attribute is set, that means the two structures are
3295 3295 ordered in either an ascending or descending way. Therefore, we can add
3296 3296 them maintaining the order by iterating over both at the same time
3297 3297
3298 3298 >>> xs = baseset([0, 3, 2])
3299 3299 >>> ys = baseset([5, 2, 4])
3300 3300
3301 3301 >>> rs = addset(xs, ys)
3302 3302 >>> bool(rs), 0 in rs, 1 in rs, 5 in rs, rs.first(), rs.last()
3303 3303 (True, True, False, True, 0, 4)
3304 3304 >>> rs = addset(xs, baseset([]))
3305 3305 >>> bool(rs), 0 in rs, 1 in rs, rs.first(), rs.last()
3306 3306 (True, True, False, 0, 2)
3307 3307 >>> rs = addset(baseset([]), baseset([]))
3308 3308 >>> bool(rs), 0 in rs, rs.first(), rs.last()
3309 3309 (False, False, None, None)
3310 3310
3311 3311 iterate unsorted:
3312 3312 >>> rs = addset(xs, ys)
3313 3313 >>> # (use generator because pypy could call len())
3314 3314 >>> list(x for x in rs) # without _genlist
3315 3315 [0, 3, 2, 5, 4]
3316 3316 >>> assert not rs._genlist
3317 3317 >>> len(rs)
3318 3318 5
3319 3319 >>> [x for x in rs] # with _genlist
3320 3320 [0, 3, 2, 5, 4]
3321 3321 >>> assert rs._genlist
3322 3322
3323 3323 iterate ascending:
3324 3324 >>> rs = addset(xs, ys, ascending=True)
3325 3325 >>> # (use generator because pypy could call len())
3326 3326 >>> list(x for x in rs), list(x for x in rs.fastasc()) # without _asclist
3327 3327 ([0, 2, 3, 4, 5], [0, 2, 3, 4, 5])
3328 3328 >>> assert not rs._asclist
3329 3329 >>> len(rs)
3330 3330 5
3331 3331 >>> [x for x in rs], [x for x in rs.fastasc()]
3332 3332 ([0, 2, 3, 4, 5], [0, 2, 3, 4, 5])
3333 3333 >>> assert rs._asclist
3334 3334
3335 3335 iterate descending:
3336 3336 >>> rs = addset(xs, ys, ascending=False)
3337 3337 >>> # (use generator because pypy could call len())
3338 3338 >>> list(x for x in rs), list(x for x in rs.fastdesc()) # without _asclist
3339 3339 ([5, 4, 3, 2, 0], [5, 4, 3, 2, 0])
3340 3340 >>> assert not rs._asclist
3341 3341 >>> len(rs)
3342 3342 5
3343 3343 >>> [x for x in rs], [x for x in rs.fastdesc()]
3344 3344 ([5, 4, 3, 2, 0], [5, 4, 3, 2, 0])
3345 3345 >>> assert rs._asclist
3346 3346
3347 3347 iterate ascending without fastasc:
3348 3348 >>> rs = addset(xs, generatorset(ys), ascending=True)
3349 3349 >>> assert rs.fastasc is None
3350 3350 >>> [x for x in rs]
3351 3351 [0, 2, 3, 4, 5]
3352 3352
3353 3353 iterate descending without fastdesc:
3354 3354 >>> rs = addset(generatorset(xs), ys, ascending=False)
3355 3355 >>> assert rs.fastdesc is None
3356 3356 >>> [x for x in rs]
3357 3357 [5, 4, 3, 2, 0]
3358 3358 """
3359 3359 def __init__(self, revs1, revs2, ascending=None):
3360 3360 self._r1 = revs1
3361 3361 self._r2 = revs2
3362 3362 self._iter = None
3363 3363 self._ascending = ascending
3364 3364 self._genlist = None
3365 3365 self._asclist = None
3366 3366
3367 3367 def __len__(self):
3368 3368 return len(self._list)
3369 3369
3370 3370 def __nonzero__(self):
3371 3371 return bool(self._r1) or bool(self._r2)
3372 3372
3373 3373 @util.propertycache
3374 3374 def _list(self):
3375 3375 if not self._genlist:
3376 3376 self._genlist = baseset(iter(self))
3377 3377 return self._genlist
3378 3378
3379 3379 def __iter__(self):
3380 3380 """Iterate over both collections without repeating elements
3381 3381
3382 3382 If the ascending attribute is not set, iterate over the first one and
3383 3383 then over the second one checking for membership on the first one so we
3384 3384 dont yield any duplicates.
3385 3385
3386 3386 If the ascending attribute is set, iterate over both collections at the
3387 3387 same time, yielding only one value at a time in the given order.
3388 3388 """
3389 3389 if self._ascending is None:
3390 3390 if self._genlist:
3391 3391 return iter(self._genlist)
3392 3392 def arbitraryordergen():
3393 3393 for r in self._r1:
3394 3394 yield r
3395 3395 inr1 = self._r1.__contains__
3396 3396 for r in self._r2:
3397 3397 if not inr1(r):
3398 3398 yield r
3399 3399 return arbitraryordergen()
3400 3400 # try to use our own fast iterator if it exists
3401 3401 self._trysetasclist()
3402 3402 if self._ascending:
3403 3403 attr = 'fastasc'
3404 3404 else:
3405 3405 attr = 'fastdesc'
3406 3406 it = getattr(self, attr)
3407 3407 if it is not None:
3408 3408 return it()
3409 3409 # maybe half of the component supports fast
3410 3410 # get iterator for _r1
3411 3411 iter1 = getattr(self._r1, attr)
3412 3412 if iter1 is None:
3413 3413 # let's avoid side effect (not sure it matters)
3414 3414 iter1 = iter(sorted(self._r1, reverse=not self._ascending))
3415 3415 else:
3416 3416 iter1 = iter1()
3417 3417 # get iterator for _r2
3418 3418 iter2 = getattr(self._r2, attr)
3419 3419 if iter2 is None:
3420 3420 # let's avoid side effect (not sure it matters)
3421 3421 iter2 = iter(sorted(self._r2, reverse=not self._ascending))
3422 3422 else:
3423 3423 iter2 = iter2()
3424 3424 return _iterordered(self._ascending, iter1, iter2)
3425 3425
3426 3426 def _trysetasclist(self):
3427 3427 """populate the _asclist attribute if possible and necessary"""
3428 3428 if self._genlist is not None and self._asclist is None:
3429 3429 self._asclist = sorted(self._genlist)
3430 3430
3431 3431 @property
3432 3432 def fastasc(self):
3433 3433 self._trysetasclist()
3434 3434 if self._asclist is not None:
3435 3435 return self._asclist.__iter__
3436 3436 iter1 = self._r1.fastasc
3437 3437 iter2 = self._r2.fastasc
3438 3438 if None in (iter1, iter2):
3439 3439 return None
3440 3440 return lambda: _iterordered(True, iter1(), iter2())
3441 3441
3442 3442 @property
3443 3443 def fastdesc(self):
3444 3444 self._trysetasclist()
3445 3445 if self._asclist is not None:
3446 3446 return self._asclist.__reversed__
3447 3447 iter1 = self._r1.fastdesc
3448 3448 iter2 = self._r2.fastdesc
3449 3449 if None in (iter1, iter2):
3450 3450 return None
3451 3451 return lambda: _iterordered(False, iter1(), iter2())
3452 3452
3453 3453 def __contains__(self, x):
3454 3454 return x in self._r1 or x in self._r2
3455 3455
3456 3456 def sort(self, reverse=False):
3457 3457 """Sort the added set
3458 3458
3459 3459 For this we use the cached list with all the generated values and if we
3460 3460 know they are ascending or descending we can sort them in a smart way.
3461 3461 """
3462 3462 self._ascending = not reverse
3463 3463
3464 3464 def isascending(self):
3465 3465 return self._ascending is not None and self._ascending
3466 3466
3467 3467 def isdescending(self):
3468 3468 return self._ascending is not None and not self._ascending
3469 3469
3470 3470 def istopo(self):
3471 3471 # not worth the trouble asserting if the two sets combined are still
3472 3472 # in topographical order. Use the sort() predicate to explicitly sort
3473 3473 # again instead.
3474 3474 return False
3475 3475
3476 3476 def reverse(self):
3477 3477 if self._ascending is None:
3478 3478 self._list.reverse()
3479 3479 else:
3480 3480 self._ascending = not self._ascending
3481 3481
3482 3482 def first(self):
3483 3483 for x in self:
3484 3484 return x
3485 3485 return None
3486 3486
3487 3487 def last(self):
3488 3488 self.reverse()
3489 3489 val = self.first()
3490 3490 self.reverse()
3491 3491 return val
3492 3492
3493 3493 def __repr__(self):
3494 3494 d = {None: '', False: '-', True: '+'}[self._ascending]
3495 3495 return '<%s%s %r, %r>' % (type(self).__name__, d, self._r1, self._r2)
3496 3496
3497 3497 class generatorset(abstractsmartset):
3498 3498 """Wrap a generator for lazy iteration
3499 3499
3500 3500 Wrapper structure for generators that provides lazy membership and can
3501 3501 be iterated more than once.
3502 3502 When asked for membership it generates values until either it finds the
3503 3503 requested one or has gone through all the elements in the generator
3504 3504 """
3505 3505 def __init__(self, gen, iterasc=None):
3506 3506 """
3507 3507 gen: a generator producing the values for the generatorset.
3508 3508 """
3509 3509 self._gen = gen
3510 3510 self._asclist = None
3511 3511 self._cache = {}
3512 3512 self._genlist = []
3513 3513 self._finished = False
3514 3514 self._ascending = True
3515 3515 if iterasc is not None:
3516 3516 if iterasc:
3517 3517 self.fastasc = self._iterator
3518 3518 self.__contains__ = self._asccontains
3519 3519 else:
3520 3520 self.fastdesc = self._iterator
3521 3521 self.__contains__ = self._desccontains
3522 3522
3523 3523 def __nonzero__(self):
3524 3524 # Do not use 'for r in self' because it will enforce the iteration
3525 3525 # order (default ascending), possibly unrolling a whole descending
3526 3526 # iterator.
3527 3527 if self._genlist:
3528 3528 return True
3529 3529 for r in self._consumegen():
3530 3530 return True
3531 3531 return False
3532 3532
3533 3533 def __contains__(self, x):
3534 3534 if x in self._cache:
3535 3535 return self._cache[x]
3536 3536
3537 3537 # Use new values only, as existing values would be cached.
3538 3538 for l in self._consumegen():
3539 3539 if l == x:
3540 3540 return True
3541 3541
3542 3542 self._cache[x] = False
3543 3543 return False
3544 3544
3545 3545 def _asccontains(self, x):
3546 3546 """version of contains optimised for ascending generator"""
3547 3547 if x in self._cache:
3548 3548 return self._cache[x]
3549 3549
3550 3550 # Use new values only, as existing values would be cached.
3551 3551 for l in self._consumegen():
3552 3552 if l == x:
3553 3553 return True
3554 3554 if l > x:
3555 3555 break
3556 3556
3557 3557 self._cache[x] = False
3558 3558 return False
3559 3559
3560 3560 def _desccontains(self, x):
3561 3561 """version of contains optimised for descending generator"""
3562 3562 if x in self._cache:
3563 3563 return self._cache[x]
3564 3564
3565 3565 # Use new values only, as existing values would be cached.
3566 3566 for l in self._consumegen():
3567 3567 if l == x:
3568 3568 return True
3569 3569 if l < x:
3570 3570 break
3571 3571
3572 3572 self._cache[x] = False
3573 3573 return False
3574 3574
3575 3575 def __iter__(self):
3576 3576 if self._ascending:
3577 3577 it = self.fastasc
3578 3578 else:
3579 3579 it = self.fastdesc
3580 3580 if it is not None:
3581 3581 return it()
3582 3582 # we need to consume the iterator
3583 3583 for x in self._consumegen():
3584 3584 pass
3585 3585 # recall the same code
3586 3586 return iter(self)
3587 3587
3588 3588 def _iterator(self):
3589 3589 if self._finished:
3590 3590 return iter(self._genlist)
3591 3591
3592 3592 # We have to use this complex iteration strategy to allow multiple
3593 3593 # iterations at the same time. We need to be able to catch revision
3594 3594 # removed from _consumegen and added to genlist in another instance.
3595 3595 #
3596 3596 # Getting rid of it would provide an about 15% speed up on this
3597 3597 # iteration.
3598 3598 genlist = self._genlist
3599 3599 nextrev = self._consumegen().next
3600 3600 _len = len # cache global lookup
3601 3601 def gen():
3602 3602 i = 0
3603 3603 while True:
3604 3604 if i < _len(genlist):
3605 3605 yield genlist[i]
3606 3606 else:
3607 3607 yield nextrev()
3608 3608 i += 1
3609 3609 return gen()
3610 3610
3611 3611 def _consumegen(self):
3612 3612 cache = self._cache
3613 3613 genlist = self._genlist.append
3614 3614 for item in self._gen:
3615 3615 cache[item] = True
3616 3616 genlist(item)
3617 3617 yield item
3618 3618 if not self._finished:
3619 3619 self._finished = True
3620 3620 asc = self._genlist[:]
3621 3621 asc.sort()
3622 3622 self._asclist = asc
3623 3623 self.fastasc = asc.__iter__
3624 3624 self.fastdesc = asc.__reversed__
3625 3625
3626 3626 def __len__(self):
3627 3627 for x in self._consumegen():
3628 3628 pass
3629 3629 return len(self._genlist)
3630 3630
3631 3631 def sort(self, reverse=False):
3632 3632 self._ascending = not reverse
3633 3633
3634 3634 def reverse(self):
3635 3635 self._ascending = not self._ascending
3636 3636
3637 3637 def isascending(self):
3638 3638 return self._ascending
3639 3639
3640 3640 def isdescending(self):
3641 3641 return not self._ascending
3642 3642
3643 3643 def istopo(self):
3644 3644 # not worth the trouble asserting if the two sets combined are still
3645 3645 # in topographical order. Use the sort() predicate to explicitly sort
3646 3646 # again instead.
3647 3647 return False
3648 3648
3649 3649 def first(self):
3650 3650 if self._ascending:
3651 3651 it = self.fastasc
3652 3652 else:
3653 3653 it = self.fastdesc
3654 3654 if it is None:
3655 3655 # we need to consume all and try again
3656 3656 for x in self._consumegen():
3657 3657 pass
3658 3658 return self.first()
3659 3659 return next(it(), None)
3660 3660
3661 3661 def last(self):
3662 3662 if self._ascending:
3663 3663 it = self.fastdesc
3664 3664 else:
3665 3665 it = self.fastasc
3666 3666 if it is None:
3667 3667 # we need to consume all and try again
3668 3668 for x in self._consumegen():
3669 3669 pass
3670 3670 return self.first()
3671 3671 return next(it(), None)
3672 3672
3673 3673 def __repr__(self):
3674 3674 d = {False: '-', True: '+'}[self._ascending]
3675 3675 return '<%s%s>' % (type(self).__name__, d)
3676 3676
3677 3677 class spanset(abstractsmartset):
3678 3678 """Duck type for baseset class which represents a range of revisions and
3679 3679 can work lazily and without having all the range in memory
3680 3680
3681 3681 Note that spanset(x, y) behave almost like xrange(x, y) except for two
3682 3682 notable points:
3683 3683 - when x < y it will be automatically descending,
3684 3684 - revision filtered with this repoview will be skipped.
3685 3685
3686 3686 """
3687 3687 def __init__(self, repo, start=0, end=None):
3688 3688 """
3689 3689 start: first revision included the set
3690 3690 (default to 0)
3691 3691 end: first revision excluded (last+1)
3692 3692 (default to len(repo)
3693 3693
3694 3694 Spanset will be descending if `end` < `start`.
3695 3695 """
3696 3696 if end is None:
3697 3697 end = len(repo)
3698 3698 self._ascending = start <= end
3699 3699 if not self._ascending:
3700 3700 start, end = end + 1, start +1
3701 3701 self._start = start
3702 3702 self._end = end
3703 3703 self._hiddenrevs = repo.changelog.filteredrevs
3704 3704
3705 3705 def sort(self, reverse=False):
3706 3706 self._ascending = not reverse
3707 3707
3708 3708 def reverse(self):
3709 3709 self._ascending = not self._ascending
3710 3710
3711 3711 def istopo(self):
3712 3712 # not worth the trouble asserting if the two sets combined are still
3713 3713 # in topographical order. Use the sort() predicate to explicitly sort
3714 3714 # again instead.
3715 3715 return False
3716 3716
3717 3717 def _iterfilter(self, iterrange):
3718 3718 s = self._hiddenrevs
3719 3719 for r in iterrange:
3720 3720 if r not in s:
3721 3721 yield r
3722 3722
3723 3723 def __iter__(self):
3724 3724 if self._ascending:
3725 3725 return self.fastasc()
3726 3726 else:
3727 3727 return self.fastdesc()
3728 3728
3729 3729 def fastasc(self):
3730 3730 iterrange = xrange(self._start, self._end)
3731 3731 if self._hiddenrevs:
3732 3732 return self._iterfilter(iterrange)
3733 3733 return iter(iterrange)
3734 3734
3735 3735 def fastdesc(self):
3736 3736 iterrange = xrange(self._end - 1, self._start - 1, -1)
3737 3737 if self._hiddenrevs:
3738 3738 return self._iterfilter(iterrange)
3739 3739 return iter(iterrange)
3740 3740
3741 3741 def __contains__(self, rev):
3742 3742 hidden = self._hiddenrevs
3743 3743 return ((self._start <= rev < self._end)
3744 3744 and not (hidden and rev in hidden))
3745 3745
3746 3746 def __nonzero__(self):
3747 3747 for r in self:
3748 3748 return True
3749 3749 return False
3750 3750
3751 3751 def __len__(self):
3752 3752 if not self._hiddenrevs:
3753 3753 return abs(self._end - self._start)
3754 3754 else:
3755 3755 count = 0
3756 3756 start = self._start
3757 3757 end = self._end
3758 3758 for rev in self._hiddenrevs:
3759 3759 if (end < rev <= start) or (start <= rev < end):
3760 3760 count += 1
3761 3761 return abs(self._end - self._start) - count
3762 3762
3763 3763 def isascending(self):
3764 3764 return self._ascending
3765 3765
3766 3766 def isdescending(self):
3767 3767 return not self._ascending
3768 3768
3769 3769 def first(self):
3770 3770 if self._ascending:
3771 3771 it = self.fastasc
3772 3772 else:
3773 3773 it = self.fastdesc
3774 3774 for x in it():
3775 3775 return x
3776 3776 return None
3777 3777
3778 3778 def last(self):
3779 3779 if self._ascending:
3780 3780 it = self.fastdesc
3781 3781 else:
3782 3782 it = self.fastasc
3783 3783 for x in it():
3784 3784 return x
3785 3785 return None
3786 3786
3787 3787 def __repr__(self):
3788 3788 d = {False: '-', True: '+'}[self._ascending]
3789 3789 return '<%s%s %d:%d>' % (type(self).__name__, d,
3790 3790 self._start, self._end - 1)
3791 3791
3792 3792 class fullreposet(spanset):
3793 3793 """a set containing all revisions in the repo
3794 3794
3795 3795 This class exists to host special optimization and magic to handle virtual
3796 3796 revisions such as "null".
3797 3797 """
3798 3798
3799 3799 def __init__(self, repo):
3800 3800 super(fullreposet, self).__init__(repo)
3801 3801
3802 3802 def __and__(self, other):
3803 3803 """As self contains the whole repo, all of the other set should also be
3804 3804 in self. Therefore `self & other = other`.
3805 3805
3806 3806 This boldly assumes the other contains valid revs only.
3807 3807 """
3808 3808 # other not a smartset, make is so
3809 3809 if not util.safehasattr(other, 'isascending'):
3810 3810 # filter out hidden revision
3811 3811 # (this boldly assumes all smartset are pure)
3812 3812 #
3813 3813 # `other` was used with "&", let's assume this is a set like
3814 3814 # object.
3815 3815 other = baseset(other - self._hiddenrevs)
3816 3816
3817 3817 # XXX As fullreposet is also used as bootstrap, this is wrong.
3818 3818 #
3819 3819 # With a giveme312() revset returning [3,1,2], this makes
3820 3820 # 'hg log -r "giveme312()"' -> 1, 2, 3 (wrong)
3821 3821 # We cannot just drop it because other usage still need to sort it:
3822 3822 # 'hg log -r "all() and giveme312()"' -> 1, 2, 3 (right)
3823 3823 #
3824 3824 # There is also some faulty revset implementations that rely on it
3825 3825 # (eg: children as of its state in e8075329c5fb)
3826 3826 #
3827 3827 # When we fix the two points above we can move this into the if clause
3828 3828 other.sort(reverse=self.isdescending())
3829 3829 return other
3830 3830
3831 3831 def prettyformatset(revs):
3832 3832 lines = []
3833 3833 rs = repr(revs)
3834 3834 p = 0
3835 3835 while p < len(rs):
3836 3836 q = rs.find('<', p + 1)
3837 3837 if q < 0:
3838 3838 q = len(rs)
3839 3839 l = rs.count('<', 0, p) - rs.count('>', 0, p)
3840 3840 assert l >= 0
3841 3841 lines.append((l, rs[p:q].rstrip()))
3842 3842 p = q
3843 3843 return '\n'.join(' ' * l + s for l, s in lines)
3844 3844
3845 3845 def loadpredicate(ui, extname, registrarobj):
3846 3846 """Load revset predicates from specified registrarobj
3847 3847 """
3848 3848 for name, func in registrarobj._table.iteritems():
3849 3849 symbols[name] = func
3850 3850 if func._safe:
3851 3851 safesymbols.add(name)
3852 3852
3853 3853 # load built-in predicates explicitly to setup safesymbols
3854 3854 loadpredicate(None, None, predicate)
3855 3855
3856 3856 # tell hggettext to extract docstrings from these functions:
3857 3857 i18nfunctions = symbols.values()
General Comments 0
You need to be logged in to leave comments. Login now