##// END OF EJS Templates
parser: factor out _trygetfunc() that extracts function name and arguments...
Yuya Nishihara -
r28910:1203159c default
parent child Browse files
Show More
@@ -1,542 +1,542
1 1 # parser.py - simple top-down operator precedence parser 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 # see http://effbot.org/zone/simple-top-down-parsing.htm and
9 9 # http://eli.thegreenplace.net/2010/01/02/top-down-operator-precedence-parsing/
10 10 # for background
11 11
12 12 # takes a tokenizer and elements
13 13 # tokenizer is an iterator that returns (type, value, pos) tuples
14 14 # elements is a mapping of types to binding strength, primary, prefix, infix
15 15 # and suffix actions
16 16 # an action is a tree node name, a tree label, and an optional match
17 17 # __call__(program) parses program into a labeled tree
18 18
19 19 from __future__ import absolute_import
20 20
21 21 from .i18n import _
22 22 from . import error
23 23
24 24 class parser(object):
25 25 def __init__(self, elements, methods=None):
26 26 self._elements = elements
27 27 self._methods = methods
28 28 self.current = None
29 29 def _advance(self):
30 30 'advance the tokenizer'
31 31 t = self.current
32 32 self.current = next(self._iter, None)
33 33 return t
34 34 def _hasnewterm(self):
35 35 'True if next token may start new term'
36 36 return any(self._elements[self.current[0]][1:3])
37 37 def _match(self, m):
38 38 'make sure the tokenizer matches an end condition'
39 39 if self.current[0] != m:
40 40 raise error.ParseError(_("unexpected token: %s") % self.current[0],
41 41 self.current[2])
42 42 self._advance()
43 43 def _parseoperand(self, bind, m=None):
44 44 'gather right-hand-side operand until an end condition or binding met'
45 45 if m and self.current[0] == m:
46 46 expr = None
47 47 else:
48 48 expr = self._parse(bind)
49 49 if m:
50 50 self._match(m)
51 51 return expr
52 52 def _parse(self, bind=0):
53 53 token, value, pos = self._advance()
54 54 # handle prefix rules on current token, take as primary if unambiguous
55 55 primary, prefix = self._elements[token][1:3]
56 56 if primary and not (prefix and self._hasnewterm()):
57 57 expr = (primary, value)
58 58 elif prefix:
59 59 expr = (prefix[0], self._parseoperand(*prefix[1:]))
60 60 else:
61 61 raise error.ParseError(_("not a prefix: %s") % token, pos)
62 62 # gather tokens until we meet a lower binding strength
63 63 while bind < self._elements[self.current[0]][0]:
64 64 token, value, pos = self._advance()
65 65 # handle infix rules, take as suffix if unambiguous
66 66 infix, suffix = self._elements[token][3:]
67 67 if suffix and not (infix and self._hasnewterm()):
68 68 expr = (suffix[0], expr)
69 69 elif infix:
70 70 expr = (infix[0], expr, self._parseoperand(*infix[1:]))
71 71 else:
72 72 raise error.ParseError(_("not an infix: %s") % token, pos)
73 73 return expr
74 74 def parse(self, tokeniter):
75 75 'generate a parse tree from tokens'
76 76 self._iter = tokeniter
77 77 self._advance()
78 78 res = self._parse()
79 79 token, value, pos = self.current
80 80 return res, pos
81 81 def eval(self, tree):
82 82 'recursively evaluate a parse tree using node methods'
83 83 if not isinstance(tree, tuple):
84 84 return tree
85 85 return self._methods[tree[0]](*[self.eval(t) for t in tree[1:]])
86 86 def __call__(self, tokeniter):
87 87 'parse tokens into a parse tree and evaluate if methods given'
88 88 t = self.parse(tokeniter)
89 89 if self._methods:
90 90 return self.eval(t)
91 91 return t
92 92
93 93 def buildargsdict(trees, funcname, keys, keyvaluenode, keynode):
94 94 """Build dict from list containing positional and keyword arguments
95 95
96 96 Invalid keywords or too many positional arguments are rejected, but
97 97 missing arguments are just omitted.
98 98 """
99 99 if len(trees) > len(keys):
100 100 raise error.ParseError(_("%(func)s takes at most %(nargs)d arguments")
101 101 % {'func': funcname, 'nargs': len(keys)})
102 102 args = {}
103 103 # consume positional arguments
104 104 for k, x in zip(keys, trees):
105 105 if x[0] == keyvaluenode:
106 106 break
107 107 args[k] = x
108 108 # remainder should be keyword arguments
109 109 for x in trees[len(args):]:
110 110 if x[0] != keyvaluenode or x[1][0] != keynode:
111 111 raise error.ParseError(_("%(func)s got an invalid argument")
112 112 % {'func': funcname})
113 113 k = x[1][1]
114 114 if k not in keys:
115 115 raise error.ParseError(_("%(func)s got an unexpected keyword "
116 116 "argument '%(key)s'")
117 117 % {'func': funcname, 'key': k})
118 118 if k in args:
119 119 raise error.ParseError(_("%(func)s got multiple values for keyword "
120 120 "argument '%(key)s'")
121 121 % {'func': funcname, 'key': k})
122 122 args[k] = x[2]
123 123 return args
124 124
125 125 def unescapestr(s):
126 126 try:
127 127 return s.decode("string_escape")
128 128 except ValueError as e:
129 129 # mangle Python's exception into our format
130 130 raise error.ParseError(str(e).lower())
131 131
132 132 def _prettyformat(tree, leafnodes, level, lines):
133 133 if not isinstance(tree, tuple) or tree[0] in leafnodes:
134 134 lines.append((level, str(tree)))
135 135 else:
136 136 lines.append((level, '(%s' % tree[0]))
137 137 for s in tree[1:]:
138 138 _prettyformat(s, leafnodes, level + 1, lines)
139 139 lines[-1:] = [(lines[-1][0], lines[-1][1] + ')')]
140 140
141 141 def prettyformat(tree, leafnodes):
142 142 lines = []
143 143 _prettyformat(tree, leafnodes, 0, lines)
144 144 output = '\n'.join((' ' * l + s) for l, s in lines)
145 145 return output
146 146
147 147 def simplifyinfixops(tree, targetnodes):
148 148 """Flatten chained infix operations to reduce usage of Python stack
149 149
150 150 >>> def f(tree):
151 151 ... print prettyformat(simplifyinfixops(tree, ('or',)), ('symbol',))
152 152 >>> f(('or',
153 153 ... ('or',
154 154 ... ('symbol', '1'),
155 155 ... ('symbol', '2')),
156 156 ... ('symbol', '3')))
157 157 (or
158 158 ('symbol', '1')
159 159 ('symbol', '2')
160 160 ('symbol', '3'))
161 161 >>> f(('func',
162 162 ... ('symbol', 'p1'),
163 163 ... ('or',
164 164 ... ('or',
165 165 ... ('func',
166 166 ... ('symbol', 'sort'),
167 167 ... ('list',
168 168 ... ('or',
169 169 ... ('or',
170 170 ... ('symbol', '1'),
171 171 ... ('symbol', '2')),
172 172 ... ('symbol', '3')),
173 173 ... ('negate',
174 174 ... ('symbol', 'rev')))),
175 175 ... ('and',
176 176 ... ('symbol', '4'),
177 177 ... ('group',
178 178 ... ('or',
179 179 ... ('or',
180 180 ... ('symbol', '5'),
181 181 ... ('symbol', '6')),
182 182 ... ('symbol', '7'))))),
183 183 ... ('symbol', '8'))))
184 184 (func
185 185 ('symbol', 'p1')
186 186 (or
187 187 (func
188 188 ('symbol', 'sort')
189 189 (list
190 190 (or
191 191 ('symbol', '1')
192 192 ('symbol', '2')
193 193 ('symbol', '3'))
194 194 (negate
195 195 ('symbol', 'rev'))))
196 196 (and
197 197 ('symbol', '4')
198 198 (group
199 199 (or
200 200 ('symbol', '5')
201 201 ('symbol', '6')
202 202 ('symbol', '7'))))
203 203 ('symbol', '8')))
204 204 """
205 205 if not isinstance(tree, tuple):
206 206 return tree
207 207 op = tree[0]
208 208 if op not in targetnodes:
209 209 return (op,) + tuple(simplifyinfixops(x, targetnodes) for x in tree[1:])
210 210
211 211 # walk down left nodes taking each right node. no recursion to left nodes
212 212 # because infix operators are left-associative, i.e. left tree is deep.
213 213 # e.g. '1 + 2 + 3' -> (+ (+ 1 2) 3) -> (+ 1 2 3)
214 214 simplified = []
215 215 x = tree
216 216 while x[0] == op:
217 217 l, r = x[1:]
218 218 simplified.append(simplifyinfixops(r, targetnodes))
219 219 x = l
220 220 simplified.append(simplifyinfixops(x, targetnodes))
221 221 simplified.append(op)
222 222 return tuple(reversed(simplified))
223 223
224 224 def parseerrordetail(inst):
225 225 """Compose error message from specified ParseError object
226 226 """
227 227 if len(inst.args) > 1:
228 228 return _('at %s: %s') % (inst.args[1], inst.args[0])
229 229 else:
230 230 return inst.args[0]
231 231
232 232 class alias(object):
233 233 """Parsed result of alias"""
234 234
235 235 def __init__(self, name, args, err, replacement):
236 236 self.name = name
237 237 self.args = args
238 238 self.error = err
239 239 self.replacement = replacement
240 240 # whether own `error` information is already shown or not.
241 241 # this avoids showing same warning multiple times at each
242 242 # `expandaliases`.
243 243 self.warned = False
244 244
245 245 class basealiasrules(object):
246 246 """Parsing and expansion rule set of aliases
247 247
248 248 This is a helper for fileset/revset/template aliases. A concrete rule set
249 249 should be made by sub-classing this and implementing class/static methods.
250 250
251 251 It supports alias expansion of symbol and funciton-call styles::
252 252
253 253 # decl = defn
254 254 h = heads(default)
255 255 b($1) = ancestors($1) - ancestors(default)
256 256 """
257 257 # typically a config section, which will be included in error messages
258 258 _section = None
259 # tags of symbol and function nodes
259 # tag of symbol node
260 260 _symbolnode = 'symbol'
261 _funcnode = 'func'
262 261
263 262 def __new__(cls):
264 263 raise TypeError("'%s' is not instantiatable" % cls.__name__)
265 264
266 265 @staticmethod
267 266 def _parse(spec):
268 267 """Parse an alias name, arguments and definition"""
269 268 raise NotImplementedError
270 269
271 270 @staticmethod
272 def _getlist(tree):
273 """Extract a list of arguments from parsed tree"""
271 def _trygetfunc(tree):
272 """Return (name, args) if tree is a function; otherwise None"""
274 273 raise NotImplementedError
275 274
276 275 @classmethod
277 276 def _builddecl(cls, decl):
278 277 """Parse an alias declaration into ``(name, args, errorstr)``
279 278
280 279 This function analyzes the parsed tree. The parsing rule is provided
281 280 by ``_parse()``.
282 281
283 282 - ``name``: of declared alias (may be ``decl`` itself at error)
284 283 - ``args``: list of argument names (or None for symbol declaration)
285 284 - ``errorstr``: detail about detected error (or None)
286 285
287 286 >>> sym = lambda x: ('symbol', x)
288 287 >>> symlist = lambda *xs: ('list',) + tuple(sym(x) for x in xs)
289 288 >>> func = lambda n, a: ('func', sym(n), a)
290 289 >>> parsemap = {
291 290 ... 'foo': sym('foo'),
292 291 ... '$foo': sym('$foo'),
293 292 ... 'foo::bar': ('dagrange', sym('foo'), sym('bar')),
294 293 ... 'foo()': func('foo', None),
295 294 ... '$foo()': func('$foo', None),
296 295 ... 'foo($1, $2)': func('foo', symlist('$1', '$2')),
297 296 ... 'foo(bar_bar, baz.baz)':
298 297 ... func('foo', symlist('bar_bar', 'baz.baz')),
299 298 ... 'foo(bar($1, $2))':
300 299 ... func('foo', func('bar', symlist('$1', '$2'))),
301 300 ... 'foo($1, $2, nested($1, $2))':
302 301 ... func('foo', (symlist('$1', '$2') +
303 302 ... (func('nested', symlist('$1', '$2')),))),
304 303 ... 'foo("bar")': func('foo', ('string', 'bar')),
305 304 ... 'foo($1, $2': error.ParseError('unexpected token: end', 10),
306 305 ... 'foo("bar': error.ParseError('unterminated string', 5),
307 306 ... 'foo($1, $2, $1)': func('foo', symlist('$1', '$2', '$1')),
308 307 ... }
309 308 >>> def parse(expr):
310 309 ... x = parsemap[expr]
311 310 ... if isinstance(x, Exception):
312 311 ... raise x
313 312 ... return x
314 >>> def getlist(tree):
315 ... if not tree:
316 ... return []
317 ... if tree[0] == 'list':
318 ... return list(tree[1:])
319 ... return [tree]
313 >>> def trygetfunc(tree):
314 ... if not tree or tree[0] != 'func' or tree[1][0] != 'symbol':
315 ... return None
316 ... if not tree[2]:
317 ... return tree[1][1], []
318 ... if tree[2][0] == 'list':
319 ... return tree[1][1], list(tree[2][1:])
320 ... return tree[1][1], [tree[2]]
320 321 >>> class aliasrules(basealiasrules):
321 322 ... _parse = staticmethod(parse)
322 ... _getlist = staticmethod(getlist)
323 ... _trygetfunc = staticmethod(trygetfunc)
323 324 >>> builddecl = aliasrules._builddecl
324 325 >>> builddecl('foo')
325 326 ('foo', None, None)
326 327 >>> builddecl('$foo')
327 328 ('$foo', None, "'$' not for alias arguments")
328 329 >>> builddecl('foo::bar')
329 330 ('foo::bar', None, 'invalid format')
330 331 >>> builddecl('foo()')
331 332 ('foo', [], None)
332 333 >>> builddecl('$foo()')
333 334 ('$foo()', None, "'$' not for alias arguments")
334 335 >>> builddecl('foo($1, $2)')
335 336 ('foo', ['$1', '$2'], None)
336 337 >>> builddecl('foo(bar_bar, baz.baz)')
337 338 ('foo', ['bar_bar', 'baz.baz'], None)
338 339 >>> builddecl('foo($1, $2, nested($1, $2))')
339 340 ('foo($1, $2, nested($1, $2))', None, 'invalid argument list')
340 341 >>> builddecl('foo(bar($1, $2))')
341 342 ('foo(bar($1, $2))', None, 'invalid argument list')
342 343 >>> builddecl('foo("bar")')
343 344 ('foo("bar")', None, 'invalid argument list')
344 345 >>> builddecl('foo($1, $2')
345 346 ('foo($1, $2', None, 'at 10: unexpected token: end')
346 347 >>> builddecl('foo("bar')
347 348 ('foo("bar', None, 'at 5: unterminated string')
348 349 >>> builddecl('foo($1, $2, $1)')
349 350 ('foo', None, 'argument names collide with each other')
350 351 """
351 352 try:
352 353 tree = cls._parse(decl)
353 354 except error.ParseError as inst:
354 355 return (decl, None, parseerrordetail(inst))
355 356
356 357 if tree[0] == cls._symbolnode:
357 358 # "name = ...." style
358 359 name = tree[1]
359 360 if name.startswith('$'):
360 361 return (decl, None, _("'$' not for alias arguments"))
361 362 return (name, None, None)
362 363
363 if tree[0] == cls._funcnode and tree[1][0] == cls._symbolnode:
364 func = cls._trygetfunc(tree)
365 if func:
364 366 # "name(arg, ....) = ...." style
365 name = tree[1][1]
367 name, args = func
366 368 if name.startswith('$'):
367 369 return (decl, None, _("'$' not for alias arguments"))
368 args = []
369 for arg in cls._getlist(tree[2]):
370 if arg[0] != cls._symbolnode:
370 if any(t[0] != cls._symbolnode for t in args):
371 371 return (decl, None, _("invalid argument list"))
372 args.append(arg[1])
373 372 if len(args) != len(set(args)):
374 373 return (name, None, _("argument names collide with each other"))
375 return (name, args, None)
374 return (name, [t[1] for t in args], None)
376 375
377 376 return (decl, None, _("invalid format"))
378 377
379 378 @classmethod
380 379 def _relabelargs(cls, tree, args):
381 380 """Mark alias arguments as ``_aliasarg``"""
382 381 if not isinstance(tree, tuple):
383 382 return tree
384 383 op = tree[0]
385 384 if op != cls._symbolnode:
386 385 return (op,) + tuple(cls._relabelargs(x, args) for x in tree[1:])
387 386
388 387 assert len(tree) == 2
389 388 sym = tree[1]
390 389 if sym in args:
391 390 op = '_aliasarg'
392 391 elif sym.startswith('$'):
393 392 raise error.ParseError(_("'$' not for alias arguments"))
394 393 return (op, sym)
395 394
396 395 @classmethod
397 396 def _builddefn(cls, defn, args):
398 397 """Parse an alias definition into a tree and marks substitutions
399 398
400 399 This function marks alias argument references as ``_aliasarg``. The
401 400 parsing rule is provided by ``_parse()``.
402 401
403 402 ``args`` is a list of alias argument names, or None if the alias
404 403 is declared as a symbol.
405 404
406 405 >>> parsemap = {
407 406 ... '$1 or foo': ('or', ('symbol', '$1'), ('symbol', 'foo')),
408 407 ... '$1 or $bar': ('or', ('symbol', '$1'), ('symbol', '$bar')),
409 408 ... '$10 or baz': ('or', ('symbol', '$10'), ('symbol', 'baz')),
410 409 ... '"$1" or "foo"': ('or', ('string', '$1'), ('string', 'foo')),
411 410 ... }
412 411 >>> class aliasrules(basealiasrules):
413 412 ... _parse = staticmethod(parsemap.__getitem__)
414 ... _getlist = staticmethod(lambda x: [])
413 ... _trygetfunc = staticmethod(lambda x: None)
415 414 >>> builddefn = aliasrules._builddefn
416 415 >>> def pprint(tree):
417 416 ... print prettyformat(tree, ('_aliasarg', 'string', 'symbol'))
418 417 >>> args = ['$1', '$2', 'foo']
419 418 >>> pprint(builddefn('$1 or foo', args))
420 419 (or
421 420 ('_aliasarg', '$1')
422 421 ('_aliasarg', 'foo'))
423 422 >>> try:
424 423 ... builddefn('$1 or $bar', args)
425 424 ... except error.ParseError as inst:
426 425 ... print parseerrordetail(inst)
427 426 '$' not for alias arguments
428 427 >>> args = ['$1', '$10', 'foo']
429 428 >>> pprint(builddefn('$10 or baz', args))
430 429 (or
431 430 ('_aliasarg', '$10')
432 431 ('symbol', 'baz'))
433 432 >>> pprint(builddefn('"$1" or "foo"', args))
434 433 (or
435 434 ('string', '$1')
436 435 ('string', 'foo'))
437 436 """
438 437 tree = cls._parse(defn)
439 438 if args:
440 439 args = set(args)
441 440 else:
442 441 args = set()
443 442 return cls._relabelargs(tree, args)
444 443
445 444 @classmethod
446 445 def build(cls, decl, defn):
447 446 """Parse an alias declaration and definition into an alias object"""
448 447 repl = efmt = None
449 448 name, args, err = cls._builddecl(decl)
450 449 if err:
451 450 efmt = _('failed to parse the declaration of %(section)s '
452 451 '"%(name)s": %(error)s')
453 452 else:
454 453 try:
455 454 repl = cls._builddefn(defn, args)
456 455 except error.ParseError as inst:
457 456 err = parseerrordetail(inst)
458 457 efmt = _('failed to parse the definition of %(section)s '
459 458 '"%(name)s": %(error)s')
460 459 if err:
461 460 err = efmt % {'section': cls._section, 'name': name, 'error': err}
462 461 return alias(name, args, err, repl)
463 462
464 463 @classmethod
465 464 def buildmap(cls, items):
466 465 """Parse a list of alias (name, replacement) pairs into a dict of
467 466 alias objects"""
468 467 aliases = {}
469 468 for decl, defn in items:
470 469 a = cls.build(decl, defn)
471 470 aliases[a.name] = a
472 471 return aliases
473 472
474 473 @classmethod
475 474 def _getalias(cls, aliases, tree):
476 475 """If tree looks like an unexpanded alias, return (alias, pattern-args)
477 476 pair. Return None otherwise.
478 477 """
479 478 if not isinstance(tree, tuple):
480 479 return None
481 480 if tree[0] == cls._symbolnode:
482 481 name = tree[1]
483 482 a = aliases.get(name)
484 483 if a and a.args is None:
485 484 return a, None
486 if tree[0] == cls._funcnode and tree[1][0] == cls._symbolnode:
487 name = tree[1][1]
485 func = cls._trygetfunc(tree)
486 if func:
487 name, args = func
488 488 a = aliases.get(name)
489 489 if a and a.args is not None:
490 return a, cls._getlist(tree[2])
490 return a, args
491 491 return None
492 492
493 493 @classmethod
494 494 def _expandargs(cls, tree, args):
495 495 """Replace _aliasarg instances with the substitution value of the
496 496 same name in args, recursively.
497 497 """
498 498 if not isinstance(tree, tuple):
499 499 return tree
500 500 if tree[0] == '_aliasarg':
501 501 sym = tree[1]
502 502 return args[sym]
503 503 return tuple(cls._expandargs(t, args) for t in tree)
504 504
505 505 @classmethod
506 506 def _expand(cls, aliases, tree, expanding, cache):
507 507 if not isinstance(tree, tuple):
508 508 return tree
509 509 r = cls._getalias(aliases, tree)
510 510 if r is None:
511 511 return tuple(cls._expand(aliases, t, expanding, cache)
512 512 for t in tree)
513 513 a, l = r
514 514 if a.error:
515 515 raise error.Abort(a.error)
516 516 if a in expanding:
517 517 raise error.ParseError(_('infinite expansion of %(section)s '
518 518 '"%(name)s" detected')
519 519 % {'section': cls._section, 'name': a.name})
520 520 # get cacheable replacement tree by expanding aliases recursively
521 521 expanding.append(a)
522 522 if a.name not in cache:
523 523 cache[a.name] = cls._expand(aliases, a.replacement, expanding,
524 524 cache)
525 525 result = cache[a.name]
526 526 expanding.pop()
527 527 if a.args is None:
528 528 return result
529 529 # substitute function arguments in replacement tree
530 530 if len(l) != len(a.args):
531 531 raise error.ParseError(_('invalid number of arguments: %d')
532 532 % len(l))
533 533 l = [cls._expand(aliases, t, [], cache) for t in l]
534 534 return cls._expandargs(result, dict(zip(a.args, l)))
535 535
536 536 @classmethod
537 537 def expand(cls, aliases, tree):
538 538 """Expand aliases in tree, recursively.
539 539
540 540 'aliases' is a dictionary mapping user defined aliases to alias objects.
541 541 """
542 542 return cls._expand(aliases, tree, [], {})
@@ -1,3394 +1,3398
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
13 13 from .i18n import _
14 14 from . import (
15 15 destutil,
16 16 encoding,
17 17 error,
18 18 hbisect,
19 19 match as matchmod,
20 20 node,
21 21 obsolete as obsmod,
22 22 parser,
23 23 pathutil,
24 24 phases,
25 25 registrar,
26 26 repoview,
27 27 util,
28 28 )
29 29
30 30 def _revancestors(repo, revs, followfirst):
31 31 """Like revlog.ancestors(), but supports followfirst."""
32 32 if followfirst:
33 33 cut = 1
34 34 else:
35 35 cut = None
36 36 cl = repo.changelog
37 37
38 38 def iterate():
39 39 revs.sort(reverse=True)
40 40 irevs = iter(revs)
41 41 h = []
42 42
43 43 inputrev = next(irevs, None)
44 44 if inputrev is not None:
45 45 heapq.heappush(h, -inputrev)
46 46
47 47 seen = set()
48 48 while h:
49 49 current = -heapq.heappop(h)
50 50 if current == inputrev:
51 51 inputrev = next(irevs, None)
52 52 if inputrev is not None:
53 53 heapq.heappush(h, -inputrev)
54 54 if current not in seen:
55 55 seen.add(current)
56 56 yield current
57 57 for parent in cl.parentrevs(current)[:cut]:
58 58 if parent != node.nullrev:
59 59 heapq.heappush(h, -parent)
60 60
61 61 return generatorset(iterate(), iterasc=False)
62 62
63 63 def _revdescendants(repo, revs, followfirst):
64 64 """Like revlog.descendants() but supports followfirst."""
65 65 if followfirst:
66 66 cut = 1
67 67 else:
68 68 cut = None
69 69
70 70 def iterate():
71 71 cl = repo.changelog
72 72 # XXX this should be 'parentset.min()' assuming 'parentset' is a
73 73 # smartset (and if it is not, it should.)
74 74 first = min(revs)
75 75 nullrev = node.nullrev
76 76 if first == nullrev:
77 77 # Are there nodes with a null first parent and a non-null
78 78 # second one? Maybe. Do we care? Probably not.
79 79 for i in cl:
80 80 yield i
81 81 else:
82 82 seen = set(revs)
83 83 for i in cl.revs(first + 1):
84 84 for x in cl.parentrevs(i)[:cut]:
85 85 if x != nullrev and x in seen:
86 86 seen.add(i)
87 87 yield i
88 88 break
89 89
90 90 return generatorset(iterate(), iterasc=True)
91 91
92 92 def _reachablerootspure(repo, minroot, roots, heads, includepath):
93 93 """return (heads(::<roots> and ::<heads>))
94 94
95 95 If includepath is True, return (<roots>::<heads>)."""
96 96 if not roots:
97 97 return []
98 98 parentrevs = repo.changelog.parentrevs
99 99 roots = set(roots)
100 100 visit = list(heads)
101 101 reachable = set()
102 102 seen = {}
103 103 # prefetch all the things! (because python is slow)
104 104 reached = reachable.add
105 105 dovisit = visit.append
106 106 nextvisit = visit.pop
107 107 # open-code the post-order traversal due to the tiny size of
108 108 # sys.getrecursionlimit()
109 109 while visit:
110 110 rev = nextvisit()
111 111 if rev in roots:
112 112 reached(rev)
113 113 if not includepath:
114 114 continue
115 115 parents = parentrevs(rev)
116 116 seen[rev] = parents
117 117 for parent in parents:
118 118 if parent >= minroot and parent not in seen:
119 119 dovisit(parent)
120 120 if not reachable:
121 121 return baseset()
122 122 if not includepath:
123 123 return reachable
124 124 for rev in sorted(seen):
125 125 for parent in seen[rev]:
126 126 if parent in reachable:
127 127 reached(rev)
128 128 return reachable
129 129
130 130 def reachableroots(repo, roots, heads, includepath=False):
131 131 """return (heads(::<roots> and ::<heads>))
132 132
133 133 If includepath is True, return (<roots>::<heads>)."""
134 134 if not roots:
135 135 return baseset()
136 136 minroot = roots.min()
137 137 roots = list(roots)
138 138 heads = list(heads)
139 139 try:
140 140 revs = repo.changelog.reachableroots(minroot, heads, roots, includepath)
141 141 except AttributeError:
142 142 revs = _reachablerootspure(repo, minroot, roots, heads, includepath)
143 143 revs = baseset(revs)
144 144 revs.sort()
145 145 return revs
146 146
147 147 elements = {
148 148 # token-type: binding-strength, primary, prefix, infix, suffix
149 149 "(": (21, None, ("group", 1, ")"), ("func", 1, ")"), None),
150 150 "##": (20, None, None, ("_concat", 20), None),
151 151 "~": (18, None, None, ("ancestor", 18), None),
152 152 "^": (18, None, None, ("parent", 18), ("parentpost", 18)),
153 153 "-": (5, None, ("negate", 19), ("minus", 5), None),
154 154 "::": (17, None, ("dagrangepre", 17), ("dagrange", 17),
155 155 ("dagrangepost", 17)),
156 156 "..": (17, None, ("dagrangepre", 17), ("dagrange", 17),
157 157 ("dagrangepost", 17)),
158 158 ":": (15, "rangeall", ("rangepre", 15), ("range", 15), ("rangepost", 15)),
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", 5)),
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(c for c in [chr(i) for i in xrange(256)]
179 179 if c.isalnum() or c in '._@' or ord(c) > 127)
180 180
181 181 # default set of valid characters for non-initial letters of symbols
182 182 _symletters = set(c for c in [chr(i) for i in xrange(256)]
183 183 if c.isalnum() or c in '-._/@' or ord(c) > 127)
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 getstring(x, err):
306 306 if x and (x[0] == 'string' or x[0] == 'symbol'):
307 307 return x[1]
308 308 raise error.ParseError(err)
309 309
310 310 def getlist(x):
311 311 if not x:
312 312 return []
313 313 if x[0] == 'list':
314 314 return list(x[1:])
315 315 return [x]
316 316
317 317 def getargs(x, min, max, err):
318 318 l = getlist(x)
319 319 if len(l) < min or (max >= 0 and len(l) > max):
320 320 raise error.ParseError(err)
321 321 return l
322 322
323 323 def getargsdict(x, funcname, keys):
324 324 return parser.buildargsdict(getlist(x), funcname, keys.split(),
325 325 keyvaluenode='keyvalue', keynode='symbol')
326 326
327 327 def getset(repo, subset, x):
328 328 if not x:
329 329 raise error.ParseError(_("missing argument"))
330 330 s = methods[x[0]](repo, subset, *x[1:])
331 331 if util.safehasattr(s, 'isascending'):
332 332 return s
333 333 if (repo.ui.configbool('devel', 'all-warnings')
334 334 or repo.ui.configbool('devel', 'old-revset')):
335 335 # else case should not happen, because all non-func are internal,
336 336 # ignoring for now.
337 337 if x[0] == 'func' and x[1][0] == 'symbol' and x[1][1] in symbols:
338 338 repo.ui.develwarn('revset "%s" use list instead of smartset, '
339 339 '(upgrade your code)' % x[1][1])
340 340 return baseset(s)
341 341
342 342 def _getrevsource(repo, r):
343 343 extra = repo[r].extra()
344 344 for label in ('source', 'transplant_source', 'rebase_source'):
345 345 if label in extra:
346 346 try:
347 347 return repo[extra[label]].rev()
348 348 except error.RepoLookupError:
349 349 pass
350 350 return None
351 351
352 352 # operator methods
353 353
354 354 def stringset(repo, subset, x):
355 355 x = repo[x].rev()
356 356 if (x in subset
357 357 or x == node.nullrev and isinstance(subset, fullreposet)):
358 358 return baseset([x])
359 359 return baseset()
360 360
361 361 def rangeset(repo, subset, x, y):
362 362 m = getset(repo, fullreposet(repo), x)
363 363 n = getset(repo, fullreposet(repo), y)
364 364
365 365 if not m or not n:
366 366 return baseset()
367 367 m, n = m.first(), n.last()
368 368
369 369 if m == n:
370 370 r = baseset([m])
371 371 elif n == node.wdirrev:
372 372 r = spanset(repo, m, len(repo)) + baseset([n])
373 373 elif m == node.wdirrev:
374 374 r = baseset([m]) + spanset(repo, len(repo) - 1, n - 1)
375 375 elif m < n:
376 376 r = spanset(repo, m, n + 1)
377 377 else:
378 378 r = spanset(repo, m, n - 1)
379 379 # XXX We should combine with subset first: 'subset & baseset(...)'. This is
380 380 # necessary to ensure we preserve the order in subset.
381 381 #
382 382 # This has performance implication, carrying the sorting over when possible
383 383 # would be more efficient.
384 384 return r & subset
385 385
386 386 def dagrange(repo, subset, x, y):
387 387 r = fullreposet(repo)
388 388 xs = reachableroots(repo, getset(repo, r, x), getset(repo, r, y),
389 389 includepath=True)
390 390 # XXX We should combine with subset first: 'subset & baseset(...)'. This is
391 391 # necessary to ensure we preserve the order in subset.
392 392 return xs & subset
393 393
394 394 def andset(repo, subset, x, y):
395 395 return getset(repo, getset(repo, subset, x), y)
396 396
397 397 def differenceset(repo, subset, x, y):
398 398 return getset(repo, subset, x) - getset(repo, subset, y)
399 399
400 400 def orset(repo, subset, *xs):
401 401 assert xs
402 402 if len(xs) == 1:
403 403 return getset(repo, subset, xs[0])
404 404 p = len(xs) // 2
405 405 a = orset(repo, subset, *xs[:p])
406 406 b = orset(repo, subset, *xs[p:])
407 407 return a + b
408 408
409 409 def notset(repo, subset, x):
410 410 return subset - getset(repo, subset, x)
411 411
412 412 def listset(repo, subset, *xs):
413 413 raise error.ParseError(_("can't use a list in this context"),
414 414 hint=_('see hg help "revsets.x or y"'))
415 415
416 416 def keyvaluepair(repo, subset, k, v):
417 417 raise error.ParseError(_("can't use a key-value pair in this context"))
418 418
419 419 def func(repo, subset, a, b):
420 420 if a[0] == 'symbol' and a[1] in symbols:
421 421 return symbols[a[1]](repo, subset, b)
422 422
423 423 keep = lambda fn: getattr(fn, '__doc__', None) is not None
424 424
425 425 syms = [s for (s, fn) in symbols.items() if keep(fn)]
426 426 raise error.UnknownIdentifier(a[1], syms)
427 427
428 428 # functions
429 429
430 430 # symbols are callables like:
431 431 # fn(repo, subset, x)
432 432 # with:
433 433 # repo - current repository instance
434 434 # subset - of revisions to be examined
435 435 # x - argument in tree form
436 436 symbols = {}
437 437
438 438 # symbols which can't be used for a DoS attack for any given input
439 439 # (e.g. those which accept regexes as plain strings shouldn't be included)
440 440 # functions that just return a lot of changesets (like all) don't count here
441 441 safesymbols = set()
442 442
443 443 predicate = registrar.revsetpredicate()
444 444
445 445 @predicate('_destupdate')
446 446 def _destupdate(repo, subset, x):
447 447 # experimental revset for update destination
448 448 args = getargsdict(x, 'limit', 'clean check')
449 449 return subset & baseset([destutil.destupdate(repo, **args)[0]])
450 450
451 451 @predicate('_destmerge')
452 452 def _destmerge(repo, subset, x):
453 453 # experimental revset for merge destination
454 454 sourceset = None
455 455 if x is not None:
456 456 sourceset = getset(repo, fullreposet(repo), x)
457 457 return subset & baseset([destutil.destmerge(repo, sourceset=sourceset)])
458 458
459 459 @predicate('adds(pattern)', safe=True)
460 460 def adds(repo, subset, x):
461 461 """Changesets that add a file matching pattern.
462 462
463 463 The pattern without explicit kind like ``glob:`` is expected to be
464 464 relative to the current directory and match against a file or a
465 465 directory.
466 466 """
467 467 # i18n: "adds" is a keyword
468 468 pat = getstring(x, _("adds requires a pattern"))
469 469 return checkstatus(repo, subset, pat, 1)
470 470
471 471 @predicate('ancestor(*changeset)', safe=True)
472 472 def ancestor(repo, subset, x):
473 473 """A greatest common ancestor of the changesets.
474 474
475 475 Accepts 0 or more changesets.
476 476 Will return empty list when passed no args.
477 477 Greatest common ancestor of a single changeset is that changeset.
478 478 """
479 479 # i18n: "ancestor" is a keyword
480 480 l = getlist(x)
481 481 rl = fullreposet(repo)
482 482 anc = None
483 483
484 484 # (getset(repo, rl, i) for i in l) generates a list of lists
485 485 for revs in (getset(repo, rl, i) for i in l):
486 486 for r in revs:
487 487 if anc is None:
488 488 anc = repo[r]
489 489 else:
490 490 anc = anc.ancestor(repo[r])
491 491
492 492 if anc is not None and anc.rev() in subset:
493 493 return baseset([anc.rev()])
494 494 return baseset()
495 495
496 496 def _ancestors(repo, subset, x, followfirst=False):
497 497 heads = getset(repo, fullreposet(repo), x)
498 498 if not heads:
499 499 return baseset()
500 500 s = _revancestors(repo, heads, followfirst)
501 501 return subset & s
502 502
503 503 @predicate('ancestors(set)', safe=True)
504 504 def ancestors(repo, subset, x):
505 505 """Changesets that are ancestors of a changeset in set.
506 506 """
507 507 return _ancestors(repo, subset, x)
508 508
509 509 @predicate('_firstancestors', safe=True)
510 510 def _firstancestors(repo, subset, x):
511 511 # ``_firstancestors(set)``
512 512 # Like ``ancestors(set)`` but follows only the first parents.
513 513 return _ancestors(repo, subset, x, followfirst=True)
514 514
515 515 def ancestorspec(repo, subset, x, n):
516 516 """``set~n``
517 517 Changesets that are the Nth ancestor (first parents only) of a changeset
518 518 in set.
519 519 """
520 520 try:
521 521 n = int(n[1])
522 522 except (TypeError, ValueError):
523 523 raise error.ParseError(_("~ expects a number"))
524 524 ps = set()
525 525 cl = repo.changelog
526 526 for r in getset(repo, fullreposet(repo), x):
527 527 for i in range(n):
528 528 r = cl.parentrevs(r)[0]
529 529 ps.add(r)
530 530 return subset & ps
531 531
532 532 @predicate('author(string)', safe=True)
533 533 def author(repo, subset, x):
534 534 """Alias for ``user(string)``.
535 535 """
536 536 # i18n: "author" is a keyword
537 537 n = encoding.lower(getstring(x, _("author requires a string")))
538 538 kind, pattern, matcher = _substringmatcher(n)
539 539 return subset.filter(lambda x: matcher(encoding.lower(repo[x].user())),
540 540 condrepr=('<user %r>', n))
541 541
542 542 @predicate('bisect(string)', safe=True)
543 543 def bisect(repo, subset, x):
544 544 """Changesets marked in the specified bisect status:
545 545
546 546 - ``good``, ``bad``, ``skip``: csets explicitly marked as good/bad/skip
547 547 - ``goods``, ``bads`` : csets topologically good/bad
548 548 - ``range`` : csets taking part in the bisection
549 549 - ``pruned`` : csets that are goods, bads or skipped
550 550 - ``untested`` : csets whose fate is yet unknown
551 551 - ``ignored`` : csets ignored due to DAG topology
552 552 - ``current`` : the cset currently being bisected
553 553 """
554 554 # i18n: "bisect" is a keyword
555 555 status = getstring(x, _("bisect requires a string")).lower()
556 556 state = set(hbisect.get(repo, status))
557 557 return subset & state
558 558
559 559 # Backward-compatibility
560 560 # - no help entry so that we do not advertise it any more
561 561 @predicate('bisected', safe=True)
562 562 def bisected(repo, subset, x):
563 563 return bisect(repo, subset, x)
564 564
565 565 @predicate('bookmark([name])', safe=True)
566 566 def bookmark(repo, subset, x):
567 567 """The named bookmark or all bookmarks.
568 568
569 569 If `name` starts with `re:`, the remainder of the name is treated as
570 570 a regular expression. To match a bookmark that actually starts with `re:`,
571 571 use the prefix `literal:`.
572 572 """
573 573 # i18n: "bookmark" is a keyword
574 574 args = getargs(x, 0, 1, _('bookmark takes one or no arguments'))
575 575 if args:
576 576 bm = getstring(args[0],
577 577 # i18n: "bookmark" is a keyword
578 578 _('the argument to bookmark must be a string'))
579 579 kind, pattern, matcher = util.stringmatcher(bm)
580 580 bms = set()
581 581 if kind == 'literal':
582 582 bmrev = repo._bookmarks.get(pattern, None)
583 583 if not bmrev:
584 584 raise error.RepoLookupError(_("bookmark '%s' does not exist")
585 585 % pattern)
586 586 bms.add(repo[bmrev].rev())
587 587 else:
588 588 matchrevs = set()
589 589 for name, bmrev in repo._bookmarks.iteritems():
590 590 if matcher(name):
591 591 matchrevs.add(bmrev)
592 592 if not matchrevs:
593 593 raise error.RepoLookupError(_("no bookmarks exist"
594 594 " that match '%s'") % pattern)
595 595 for bmrev in matchrevs:
596 596 bms.add(repo[bmrev].rev())
597 597 else:
598 598 bms = set([repo[r].rev()
599 599 for r in repo._bookmarks.values()])
600 600 bms -= set([node.nullrev])
601 601 return subset & bms
602 602
603 603 @predicate('branch(string or set)', safe=True)
604 604 def branch(repo, subset, x):
605 605 """
606 606 All changesets belonging to the given branch or the branches of the given
607 607 changesets.
608 608
609 609 If `string` starts with `re:`, the remainder of the name is treated as
610 610 a regular expression. To match a branch that actually starts with `re:`,
611 611 use the prefix `literal:`.
612 612 """
613 613 getbi = repo.revbranchcache().branchinfo
614 614
615 615 try:
616 616 b = getstring(x, '')
617 617 except error.ParseError:
618 618 # not a string, but another revspec, e.g. tip()
619 619 pass
620 620 else:
621 621 kind, pattern, matcher = util.stringmatcher(b)
622 622 if kind == 'literal':
623 623 # note: falls through to the revspec case if no branch with
624 624 # this name exists and pattern kind is not specified explicitly
625 625 if pattern in repo.branchmap():
626 626 return subset.filter(lambda r: matcher(getbi(r)[0]),
627 627 condrepr=('<branch %r>', b))
628 628 if b.startswith('literal:'):
629 629 raise error.RepoLookupError(_("branch '%s' does not exist")
630 630 % pattern)
631 631 else:
632 632 return subset.filter(lambda r: matcher(getbi(r)[0]),
633 633 condrepr=('<branch %r>', b))
634 634
635 635 s = getset(repo, fullreposet(repo), x)
636 636 b = set()
637 637 for r in s:
638 638 b.add(getbi(r)[0])
639 639 c = s.__contains__
640 640 return subset.filter(lambda r: c(r) or getbi(r)[0] in b,
641 641 condrepr=lambda: '<branch %r>' % sorted(b))
642 642
643 643 @predicate('bumped()', safe=True)
644 644 def bumped(repo, subset, x):
645 645 """Mutable changesets marked as successors of public changesets.
646 646
647 647 Only non-public and non-obsolete changesets can be `bumped`.
648 648 """
649 649 # i18n: "bumped" is a keyword
650 650 getargs(x, 0, 0, _("bumped takes no arguments"))
651 651 bumped = obsmod.getrevs(repo, 'bumped')
652 652 return subset & bumped
653 653
654 654 @predicate('bundle()', safe=True)
655 655 def bundle(repo, subset, x):
656 656 """Changesets in the bundle.
657 657
658 658 Bundle must be specified by the -R option."""
659 659
660 660 try:
661 661 bundlerevs = repo.changelog.bundlerevs
662 662 except AttributeError:
663 663 raise error.Abort(_("no bundle provided - specify with -R"))
664 664 return subset & bundlerevs
665 665
666 666 def checkstatus(repo, subset, pat, field):
667 667 hasset = matchmod.patkind(pat) == 'set'
668 668
669 669 mcache = [None]
670 670 def matches(x):
671 671 c = repo[x]
672 672 if not mcache[0] or hasset:
673 673 mcache[0] = matchmod.match(repo.root, repo.getcwd(), [pat], ctx=c)
674 674 m = mcache[0]
675 675 fname = None
676 676 if not m.anypats() and len(m.files()) == 1:
677 677 fname = m.files()[0]
678 678 if fname is not None:
679 679 if fname not in c.files():
680 680 return False
681 681 else:
682 682 for f in c.files():
683 683 if m(f):
684 684 break
685 685 else:
686 686 return False
687 687 files = repo.status(c.p1().node(), c.node())[field]
688 688 if fname is not None:
689 689 if fname in files:
690 690 return True
691 691 else:
692 692 for f in files:
693 693 if m(f):
694 694 return True
695 695
696 696 return subset.filter(matches, condrepr=('<status[%r] %r>', field, pat))
697 697
698 698 def _children(repo, narrow, parentset):
699 699 if not parentset:
700 700 return baseset()
701 701 cs = set()
702 702 pr = repo.changelog.parentrevs
703 703 minrev = parentset.min()
704 704 for r in narrow:
705 705 if r <= minrev:
706 706 continue
707 707 for p in pr(r):
708 708 if p in parentset:
709 709 cs.add(r)
710 710 # XXX using a set to feed the baseset is wrong. Sets are not ordered.
711 711 # This does not break because of other fullreposet misbehavior.
712 712 return baseset(cs)
713 713
714 714 @predicate('children(set)', safe=True)
715 715 def children(repo, subset, x):
716 716 """Child changesets of changesets in set.
717 717 """
718 718 s = getset(repo, fullreposet(repo), x)
719 719 cs = _children(repo, subset, s)
720 720 return subset & cs
721 721
722 722 @predicate('closed()', safe=True)
723 723 def closed(repo, subset, x):
724 724 """Changeset is closed.
725 725 """
726 726 # i18n: "closed" is a keyword
727 727 getargs(x, 0, 0, _("closed takes no arguments"))
728 728 return subset.filter(lambda r: repo[r].closesbranch(),
729 729 condrepr='<branch closed>')
730 730
731 731 @predicate('contains(pattern)')
732 732 def contains(repo, subset, x):
733 733 """The revision's manifest contains a file matching pattern (but might not
734 734 modify it). See :hg:`help patterns` for information about file patterns.
735 735
736 736 The pattern without explicit kind like ``glob:`` is expected to be
737 737 relative to the current directory and match against a file exactly
738 738 for efficiency.
739 739 """
740 740 # i18n: "contains" is a keyword
741 741 pat = getstring(x, _("contains requires a pattern"))
742 742
743 743 def matches(x):
744 744 if not matchmod.patkind(pat):
745 745 pats = pathutil.canonpath(repo.root, repo.getcwd(), pat)
746 746 if pats in repo[x]:
747 747 return True
748 748 else:
749 749 c = repo[x]
750 750 m = matchmod.match(repo.root, repo.getcwd(), [pat], ctx=c)
751 751 for f in c.manifest():
752 752 if m(f):
753 753 return True
754 754 return False
755 755
756 756 return subset.filter(matches, condrepr=('<contains %r>', pat))
757 757
758 758 @predicate('converted([id])', safe=True)
759 759 def converted(repo, subset, x):
760 760 """Changesets converted from the given identifier in the old repository if
761 761 present, or all converted changesets if no identifier is specified.
762 762 """
763 763
764 764 # There is exactly no chance of resolving the revision, so do a simple
765 765 # string compare and hope for the best
766 766
767 767 rev = None
768 768 # i18n: "converted" is a keyword
769 769 l = getargs(x, 0, 1, _('converted takes one or no arguments'))
770 770 if l:
771 771 # i18n: "converted" is a keyword
772 772 rev = getstring(l[0], _('converted requires a revision'))
773 773
774 774 def _matchvalue(r):
775 775 source = repo[r].extra().get('convert_revision', None)
776 776 return source is not None and (rev is None or source.startswith(rev))
777 777
778 778 return subset.filter(lambda r: _matchvalue(r),
779 779 condrepr=('<converted %r>', rev))
780 780
781 781 @predicate('date(interval)', safe=True)
782 782 def date(repo, subset, x):
783 783 """Changesets within the interval, see :hg:`help dates`.
784 784 """
785 785 # i18n: "date" is a keyword
786 786 ds = getstring(x, _("date requires a string"))
787 787 dm = util.matchdate(ds)
788 788 return subset.filter(lambda x: dm(repo[x].date()[0]),
789 789 condrepr=('<date %r>', ds))
790 790
791 791 @predicate('desc(string)', safe=True)
792 792 def desc(repo, subset, x):
793 793 """Search commit message for string. The match is case-insensitive.
794 794 """
795 795 # i18n: "desc" is a keyword
796 796 ds = encoding.lower(getstring(x, _("desc requires a string")))
797 797
798 798 def matches(x):
799 799 c = repo[x]
800 800 return ds in encoding.lower(c.description())
801 801
802 802 return subset.filter(matches, condrepr=('<desc %r>', ds))
803 803
804 804 def _descendants(repo, subset, x, followfirst=False):
805 805 roots = getset(repo, fullreposet(repo), x)
806 806 if not roots:
807 807 return baseset()
808 808 s = _revdescendants(repo, roots, followfirst)
809 809
810 810 # Both sets need to be ascending in order to lazily return the union
811 811 # in the correct order.
812 812 base = subset & roots
813 813 desc = subset & s
814 814 result = base + desc
815 815 if subset.isascending():
816 816 result.sort()
817 817 elif subset.isdescending():
818 818 result.sort(reverse=True)
819 819 else:
820 820 result = subset & result
821 821 return result
822 822
823 823 @predicate('descendants(set)', safe=True)
824 824 def descendants(repo, subset, x):
825 825 """Changesets which are descendants of changesets in set.
826 826 """
827 827 return _descendants(repo, subset, x)
828 828
829 829 @predicate('_firstdescendants', safe=True)
830 830 def _firstdescendants(repo, subset, x):
831 831 # ``_firstdescendants(set)``
832 832 # Like ``descendants(set)`` but follows only the first parents.
833 833 return _descendants(repo, subset, x, followfirst=True)
834 834
835 835 @predicate('destination([set])', safe=True)
836 836 def destination(repo, subset, x):
837 837 """Changesets that were created by a graft, transplant or rebase operation,
838 838 with the given revisions specified as the source. Omitting the optional set
839 839 is the same as passing all().
840 840 """
841 841 if x is not None:
842 842 sources = getset(repo, fullreposet(repo), x)
843 843 else:
844 844 sources = fullreposet(repo)
845 845
846 846 dests = set()
847 847
848 848 # subset contains all of the possible destinations that can be returned, so
849 849 # iterate over them and see if their source(s) were provided in the arg set.
850 850 # Even if the immediate src of r is not in the arg set, src's source (or
851 851 # further back) may be. Scanning back further than the immediate src allows
852 852 # transitive transplants and rebases to yield the same results as transitive
853 853 # grafts.
854 854 for r in subset:
855 855 src = _getrevsource(repo, r)
856 856 lineage = None
857 857
858 858 while src is not None:
859 859 if lineage is None:
860 860 lineage = list()
861 861
862 862 lineage.append(r)
863 863
864 864 # The visited lineage is a match if the current source is in the arg
865 865 # set. Since every candidate dest is visited by way of iterating
866 866 # subset, any dests further back in the lineage will be tested by a
867 867 # different iteration over subset. Likewise, if the src was already
868 868 # selected, the current lineage can be selected without going back
869 869 # further.
870 870 if src in sources or src in dests:
871 871 dests.update(lineage)
872 872 break
873 873
874 874 r = src
875 875 src = _getrevsource(repo, r)
876 876
877 877 return subset.filter(dests.__contains__,
878 878 condrepr=lambda: '<destination %r>' % sorted(dests))
879 879
880 880 @predicate('divergent()', safe=True)
881 881 def divergent(repo, subset, x):
882 882 """
883 883 Final successors of changesets with an alternative set of final successors.
884 884 """
885 885 # i18n: "divergent" is a keyword
886 886 getargs(x, 0, 0, _("divergent takes no arguments"))
887 887 divergent = obsmod.getrevs(repo, 'divergent')
888 888 return subset & divergent
889 889
890 890 @predicate('extinct()', safe=True)
891 891 def extinct(repo, subset, x):
892 892 """Obsolete changesets with obsolete descendants only.
893 893 """
894 894 # i18n: "extinct" is a keyword
895 895 getargs(x, 0, 0, _("extinct takes no arguments"))
896 896 extincts = obsmod.getrevs(repo, 'extinct')
897 897 return subset & extincts
898 898
899 899 @predicate('extra(label, [value])', safe=True)
900 900 def extra(repo, subset, x):
901 901 """Changesets with the given label in the extra metadata, with the given
902 902 optional value.
903 903
904 904 If `value` starts with `re:`, the remainder of the value is treated as
905 905 a regular expression. To match a value that actually starts with `re:`,
906 906 use the prefix `literal:`.
907 907 """
908 908 args = getargsdict(x, 'extra', 'label value')
909 909 if 'label' not in args:
910 910 # i18n: "extra" is a keyword
911 911 raise error.ParseError(_('extra takes at least 1 argument'))
912 912 # i18n: "extra" is a keyword
913 913 label = getstring(args['label'], _('first argument to extra must be '
914 914 'a string'))
915 915 value = None
916 916
917 917 if 'value' in args:
918 918 # i18n: "extra" is a keyword
919 919 value = getstring(args['value'], _('second argument to extra must be '
920 920 'a string'))
921 921 kind, value, matcher = util.stringmatcher(value)
922 922
923 923 def _matchvalue(r):
924 924 extra = repo[r].extra()
925 925 return label in extra and (value is None or matcher(extra[label]))
926 926
927 927 return subset.filter(lambda r: _matchvalue(r),
928 928 condrepr=('<extra[%r] %r>', label, value))
929 929
930 930 @predicate('filelog(pattern)', safe=True)
931 931 def filelog(repo, subset, x):
932 932 """Changesets connected to the specified filelog.
933 933
934 934 For performance reasons, visits only revisions mentioned in the file-level
935 935 filelog, rather than filtering through all changesets (much faster, but
936 936 doesn't include deletes or duplicate changes). For a slower, more accurate
937 937 result, use ``file()``.
938 938
939 939 The pattern without explicit kind like ``glob:`` is expected to be
940 940 relative to the current directory and match against a file exactly
941 941 for efficiency.
942 942
943 943 If some linkrev points to revisions filtered by the current repoview, we'll
944 944 work around it to return a non-filtered value.
945 945 """
946 946
947 947 # i18n: "filelog" is a keyword
948 948 pat = getstring(x, _("filelog requires a pattern"))
949 949 s = set()
950 950 cl = repo.changelog
951 951
952 952 if not matchmod.patkind(pat):
953 953 f = pathutil.canonpath(repo.root, repo.getcwd(), pat)
954 954 files = [f]
955 955 else:
956 956 m = matchmod.match(repo.root, repo.getcwd(), [pat], ctx=repo[None])
957 957 files = (f for f in repo[None] if m(f))
958 958
959 959 for f in files:
960 960 fl = repo.file(f)
961 961 known = {}
962 962 scanpos = 0
963 963 for fr in list(fl):
964 964 fn = fl.node(fr)
965 965 if fn in known:
966 966 s.add(known[fn])
967 967 continue
968 968
969 969 lr = fl.linkrev(fr)
970 970 if lr in cl:
971 971 s.add(lr)
972 972 elif scanpos is not None:
973 973 # lowest matching changeset is filtered, scan further
974 974 # ahead in changelog
975 975 start = max(lr, scanpos) + 1
976 976 scanpos = None
977 977 for r in cl.revs(start):
978 978 # minimize parsing of non-matching entries
979 979 if f in cl.revision(r) and f in cl.readfiles(r):
980 980 try:
981 981 # try to use manifest delta fastpath
982 982 n = repo[r].filenode(f)
983 983 if n not in known:
984 984 if n == fn:
985 985 s.add(r)
986 986 scanpos = r
987 987 break
988 988 else:
989 989 known[n] = r
990 990 except error.ManifestLookupError:
991 991 # deletion in changelog
992 992 continue
993 993
994 994 return subset & s
995 995
996 996 @predicate('first(set, [n])', safe=True)
997 997 def first(repo, subset, x):
998 998 """An alias for limit().
999 999 """
1000 1000 return limit(repo, subset, x)
1001 1001
1002 1002 def _follow(repo, subset, x, name, followfirst=False):
1003 1003 l = getargs(x, 0, 1, _("%s takes no arguments or a pattern") % name)
1004 1004 c = repo['.']
1005 1005 if l:
1006 1006 x = getstring(l[0], _("%s expected a pattern") % name)
1007 1007 matcher = matchmod.match(repo.root, repo.getcwd(), [x],
1008 1008 ctx=repo[None], default='path')
1009 1009
1010 1010 files = c.manifest().walk(matcher)
1011 1011
1012 1012 s = set()
1013 1013 for fname in files:
1014 1014 fctx = c[fname]
1015 1015 s = s.union(set(c.rev() for c in fctx.ancestors(followfirst)))
1016 1016 # include the revision responsible for the most recent version
1017 1017 s.add(fctx.introrev())
1018 1018 else:
1019 1019 s = _revancestors(repo, baseset([c.rev()]), followfirst)
1020 1020
1021 1021 return subset & s
1022 1022
1023 1023 @predicate('follow([pattern])', safe=True)
1024 1024 def follow(repo, subset, x):
1025 1025 """
1026 1026 An alias for ``::.`` (ancestors of the working directory's first parent).
1027 1027 If pattern is specified, the histories of files matching given
1028 1028 pattern is followed, including copies.
1029 1029 """
1030 1030 return _follow(repo, subset, x, 'follow')
1031 1031
1032 1032 @predicate('_followfirst', safe=True)
1033 1033 def _followfirst(repo, subset, x):
1034 1034 # ``followfirst([pattern])``
1035 1035 # Like ``follow([pattern])`` but follows only the first parent of
1036 1036 # every revisions or files revisions.
1037 1037 return _follow(repo, subset, x, '_followfirst', followfirst=True)
1038 1038
1039 1039 @predicate('all()', safe=True)
1040 1040 def getall(repo, subset, x):
1041 1041 """All changesets, the same as ``0:tip``.
1042 1042 """
1043 1043 # i18n: "all" is a keyword
1044 1044 getargs(x, 0, 0, _("all takes no arguments"))
1045 1045 return subset & spanset(repo) # drop "null" if any
1046 1046
1047 1047 @predicate('grep(regex)')
1048 1048 def grep(repo, subset, x):
1049 1049 """Like ``keyword(string)`` but accepts a regex. Use ``grep(r'...')``
1050 1050 to ensure special escape characters are handled correctly. Unlike
1051 1051 ``keyword(string)``, the match is case-sensitive.
1052 1052 """
1053 1053 try:
1054 1054 # i18n: "grep" is a keyword
1055 1055 gr = re.compile(getstring(x, _("grep requires a string")))
1056 1056 except re.error as e:
1057 1057 raise error.ParseError(_('invalid match pattern: %s') % e)
1058 1058
1059 1059 def matches(x):
1060 1060 c = repo[x]
1061 1061 for e in c.files() + [c.user(), c.description()]:
1062 1062 if gr.search(e):
1063 1063 return True
1064 1064 return False
1065 1065
1066 1066 return subset.filter(matches, condrepr=('<grep %r>', gr.pattern))
1067 1067
1068 1068 @predicate('_matchfiles', safe=True)
1069 1069 def _matchfiles(repo, subset, x):
1070 1070 # _matchfiles takes a revset list of prefixed arguments:
1071 1071 #
1072 1072 # [p:foo, i:bar, x:baz]
1073 1073 #
1074 1074 # builds a match object from them and filters subset. Allowed
1075 1075 # prefixes are 'p:' for regular patterns, 'i:' for include
1076 1076 # patterns and 'x:' for exclude patterns. Use 'r:' prefix to pass
1077 1077 # a revision identifier, or the empty string to reference the
1078 1078 # working directory, from which the match object is
1079 1079 # initialized. Use 'd:' to set the default matching mode, default
1080 1080 # to 'glob'. At most one 'r:' and 'd:' argument can be passed.
1081 1081
1082 1082 l = getargs(x, 1, -1, "_matchfiles requires at least one argument")
1083 1083 pats, inc, exc = [], [], []
1084 1084 rev, default = None, None
1085 1085 for arg in l:
1086 1086 s = getstring(arg, "_matchfiles requires string arguments")
1087 1087 prefix, value = s[:2], s[2:]
1088 1088 if prefix == 'p:':
1089 1089 pats.append(value)
1090 1090 elif prefix == 'i:':
1091 1091 inc.append(value)
1092 1092 elif prefix == 'x:':
1093 1093 exc.append(value)
1094 1094 elif prefix == 'r:':
1095 1095 if rev is not None:
1096 1096 raise error.ParseError('_matchfiles expected at most one '
1097 1097 'revision')
1098 1098 if value != '': # empty means working directory; leave rev as None
1099 1099 rev = value
1100 1100 elif prefix == 'd:':
1101 1101 if default is not None:
1102 1102 raise error.ParseError('_matchfiles expected at most one '
1103 1103 'default mode')
1104 1104 default = value
1105 1105 else:
1106 1106 raise error.ParseError('invalid _matchfiles prefix: %s' % prefix)
1107 1107 if not default:
1108 1108 default = 'glob'
1109 1109
1110 1110 m = matchmod.match(repo.root, repo.getcwd(), pats, include=inc,
1111 1111 exclude=exc, ctx=repo[rev], default=default)
1112 1112
1113 1113 # This directly read the changelog data as creating changectx for all
1114 1114 # revisions is quite expensive.
1115 1115 getfiles = repo.changelog.readfiles
1116 1116 wdirrev = node.wdirrev
1117 1117 def matches(x):
1118 1118 if x == wdirrev:
1119 1119 files = repo[x].files()
1120 1120 else:
1121 1121 files = getfiles(x)
1122 1122 for f in files:
1123 1123 if m(f):
1124 1124 return True
1125 1125 return False
1126 1126
1127 1127 return subset.filter(matches,
1128 1128 condrepr=('<matchfiles patterns=%r, include=%r '
1129 1129 'exclude=%r, default=%r, rev=%r>',
1130 1130 pats, inc, exc, default, rev))
1131 1131
1132 1132 @predicate('file(pattern)', safe=True)
1133 1133 def hasfile(repo, subset, x):
1134 1134 """Changesets affecting files matched by pattern.
1135 1135
1136 1136 For a faster but less accurate result, consider using ``filelog()``
1137 1137 instead.
1138 1138
1139 1139 This predicate uses ``glob:`` as the default kind of pattern.
1140 1140 """
1141 1141 # i18n: "file" is a keyword
1142 1142 pat = getstring(x, _("file requires a pattern"))
1143 1143 return _matchfiles(repo, subset, ('string', 'p:' + pat))
1144 1144
1145 1145 @predicate('head()', safe=True)
1146 1146 def head(repo, subset, x):
1147 1147 """Changeset is a named branch head.
1148 1148 """
1149 1149 # i18n: "head" is a keyword
1150 1150 getargs(x, 0, 0, _("head takes no arguments"))
1151 1151 hs = set()
1152 1152 cl = repo.changelog
1153 1153 for b, ls in repo.branchmap().iteritems():
1154 1154 hs.update(cl.rev(h) for h in ls)
1155 1155 # XXX using a set to feed the baseset is wrong. Sets are not ordered.
1156 1156 # This does not break because of other fullreposet misbehavior.
1157 1157 # XXX We should combine with subset first: 'subset & baseset(...)'. This is
1158 1158 # necessary to ensure we preserve the order in subset.
1159 1159 return baseset(hs) & subset
1160 1160
1161 1161 @predicate('heads(set)', safe=True)
1162 1162 def heads(repo, subset, x):
1163 1163 """Members of set with no children in set.
1164 1164 """
1165 1165 s = getset(repo, subset, x)
1166 1166 ps = parents(repo, subset, x)
1167 1167 return s - ps
1168 1168
1169 1169 @predicate('hidden()', safe=True)
1170 1170 def hidden(repo, subset, x):
1171 1171 """Hidden changesets.
1172 1172 """
1173 1173 # i18n: "hidden" is a keyword
1174 1174 getargs(x, 0, 0, _("hidden takes no arguments"))
1175 1175 hiddenrevs = repoview.filterrevs(repo, 'visible')
1176 1176 return subset & hiddenrevs
1177 1177
1178 1178 @predicate('keyword(string)', safe=True)
1179 1179 def keyword(repo, subset, x):
1180 1180 """Search commit message, user name, and names of changed files for
1181 1181 string. The match is case-insensitive.
1182 1182 """
1183 1183 # i18n: "keyword" is a keyword
1184 1184 kw = encoding.lower(getstring(x, _("keyword requires a string")))
1185 1185
1186 1186 def matches(r):
1187 1187 c = repo[r]
1188 1188 return any(kw in encoding.lower(t)
1189 1189 for t in c.files() + [c.user(), c.description()])
1190 1190
1191 1191 return subset.filter(matches, condrepr=('<keyword %r>', kw))
1192 1192
1193 1193 @predicate('limit(set[, n[, offset]])', safe=True)
1194 1194 def limit(repo, subset, x):
1195 1195 """First n members of set, defaulting to 1, starting from offset.
1196 1196 """
1197 1197 args = getargsdict(x, 'limit', 'set n offset')
1198 1198 if 'set' not in args:
1199 1199 # i18n: "limit" is a keyword
1200 1200 raise error.ParseError(_("limit requires one to three arguments"))
1201 1201 try:
1202 1202 lim, ofs = 1, 0
1203 1203 if 'n' in args:
1204 1204 # i18n: "limit" is a keyword
1205 1205 lim = int(getstring(args['n'], _("limit requires a number")))
1206 1206 if 'offset' in args:
1207 1207 # i18n: "limit" is a keyword
1208 1208 ofs = int(getstring(args['offset'], _("limit requires a number")))
1209 1209 if ofs < 0:
1210 1210 raise error.ParseError(_("negative offset"))
1211 1211 except (TypeError, ValueError):
1212 1212 # i18n: "limit" is a keyword
1213 1213 raise error.ParseError(_("limit expects a number"))
1214 1214 os = getset(repo, fullreposet(repo), args['set'])
1215 1215 result = []
1216 1216 it = iter(os)
1217 1217 for x in xrange(ofs):
1218 1218 y = next(it, None)
1219 1219 if y is None:
1220 1220 break
1221 1221 for x in xrange(lim):
1222 1222 y = next(it, None)
1223 1223 if y is None:
1224 1224 break
1225 1225 elif y in subset:
1226 1226 result.append(y)
1227 1227 return baseset(result, datarepr=('<limit n=%d, offset=%d, %r, %r>',
1228 1228 lim, ofs, subset, os))
1229 1229
1230 1230 @predicate('last(set, [n])', safe=True)
1231 1231 def last(repo, subset, x):
1232 1232 """Last n members of set, defaulting to 1.
1233 1233 """
1234 1234 # i18n: "last" is a keyword
1235 1235 l = getargs(x, 1, 2, _("last requires one or two arguments"))
1236 1236 try:
1237 1237 lim = 1
1238 1238 if len(l) == 2:
1239 1239 # i18n: "last" is a keyword
1240 1240 lim = int(getstring(l[1], _("last requires a number")))
1241 1241 except (TypeError, ValueError):
1242 1242 # i18n: "last" is a keyword
1243 1243 raise error.ParseError(_("last expects a number"))
1244 1244 os = getset(repo, fullreposet(repo), l[0])
1245 1245 os.reverse()
1246 1246 result = []
1247 1247 it = iter(os)
1248 1248 for x in xrange(lim):
1249 1249 y = next(it, None)
1250 1250 if y is None:
1251 1251 break
1252 1252 elif y in subset:
1253 1253 result.append(y)
1254 1254 return baseset(result, datarepr=('<last n=%d, %r, %r>', lim, subset, os))
1255 1255
1256 1256 @predicate('max(set)', safe=True)
1257 1257 def maxrev(repo, subset, x):
1258 1258 """Changeset with highest revision number in set.
1259 1259 """
1260 1260 os = getset(repo, fullreposet(repo), x)
1261 1261 try:
1262 1262 m = os.max()
1263 1263 if m in subset:
1264 1264 return baseset([m], datarepr=('<max %r, %r>', subset, os))
1265 1265 except ValueError:
1266 1266 # os.max() throws a ValueError when the collection is empty.
1267 1267 # Same as python's max().
1268 1268 pass
1269 1269 return baseset(datarepr=('<max %r, %r>', subset, os))
1270 1270
1271 1271 @predicate('merge()', safe=True)
1272 1272 def merge(repo, subset, x):
1273 1273 """Changeset is a merge changeset.
1274 1274 """
1275 1275 # i18n: "merge" is a keyword
1276 1276 getargs(x, 0, 0, _("merge takes no arguments"))
1277 1277 cl = repo.changelog
1278 1278 return subset.filter(lambda r: cl.parentrevs(r)[1] != -1,
1279 1279 condrepr='<merge>')
1280 1280
1281 1281 @predicate('branchpoint()', safe=True)
1282 1282 def branchpoint(repo, subset, x):
1283 1283 """Changesets with more than one child.
1284 1284 """
1285 1285 # i18n: "branchpoint" is a keyword
1286 1286 getargs(x, 0, 0, _("branchpoint takes no arguments"))
1287 1287 cl = repo.changelog
1288 1288 if not subset:
1289 1289 return baseset()
1290 1290 # XXX this should be 'parentset.min()' assuming 'parentset' is a smartset
1291 1291 # (and if it is not, it should.)
1292 1292 baserev = min(subset)
1293 1293 parentscount = [0]*(len(repo) - baserev)
1294 1294 for r in cl.revs(start=baserev + 1):
1295 1295 for p in cl.parentrevs(r):
1296 1296 if p >= baserev:
1297 1297 parentscount[p - baserev] += 1
1298 1298 return subset.filter(lambda r: parentscount[r - baserev] > 1,
1299 1299 condrepr='<branchpoint>')
1300 1300
1301 1301 @predicate('min(set)', safe=True)
1302 1302 def minrev(repo, subset, x):
1303 1303 """Changeset with lowest revision number in set.
1304 1304 """
1305 1305 os = getset(repo, fullreposet(repo), x)
1306 1306 try:
1307 1307 m = os.min()
1308 1308 if m in subset:
1309 1309 return baseset([m], datarepr=('<min %r, %r>', subset, os))
1310 1310 except ValueError:
1311 1311 # os.min() throws a ValueError when the collection is empty.
1312 1312 # Same as python's min().
1313 1313 pass
1314 1314 return baseset(datarepr=('<min %r, %r>', subset, os))
1315 1315
1316 1316 @predicate('modifies(pattern)', safe=True)
1317 1317 def modifies(repo, subset, x):
1318 1318 """Changesets modifying files matched by pattern.
1319 1319
1320 1320 The pattern without explicit kind like ``glob:`` is expected to be
1321 1321 relative to the current directory and match against a file or a
1322 1322 directory.
1323 1323 """
1324 1324 # i18n: "modifies" is a keyword
1325 1325 pat = getstring(x, _("modifies requires a pattern"))
1326 1326 return checkstatus(repo, subset, pat, 0)
1327 1327
1328 1328 @predicate('named(namespace)')
1329 1329 def named(repo, subset, x):
1330 1330 """The changesets in a given namespace.
1331 1331
1332 1332 If `namespace` starts with `re:`, the remainder of the string is treated as
1333 1333 a regular expression. To match a namespace that actually starts with `re:`,
1334 1334 use the prefix `literal:`.
1335 1335 """
1336 1336 # i18n: "named" is a keyword
1337 1337 args = getargs(x, 1, 1, _('named requires a namespace argument'))
1338 1338
1339 1339 ns = getstring(args[0],
1340 1340 # i18n: "named" is a keyword
1341 1341 _('the argument to named must be a string'))
1342 1342 kind, pattern, matcher = util.stringmatcher(ns)
1343 1343 namespaces = set()
1344 1344 if kind == 'literal':
1345 1345 if pattern not in repo.names:
1346 1346 raise error.RepoLookupError(_("namespace '%s' does not exist")
1347 1347 % ns)
1348 1348 namespaces.add(repo.names[pattern])
1349 1349 else:
1350 1350 for name, ns in repo.names.iteritems():
1351 1351 if matcher(name):
1352 1352 namespaces.add(ns)
1353 1353 if not namespaces:
1354 1354 raise error.RepoLookupError(_("no namespace exists"
1355 1355 " that match '%s'") % pattern)
1356 1356
1357 1357 names = set()
1358 1358 for ns in namespaces:
1359 1359 for name in ns.listnames(repo):
1360 1360 if name not in ns.deprecated:
1361 1361 names.update(repo[n].rev() for n in ns.nodes(repo, name))
1362 1362
1363 1363 names -= set([node.nullrev])
1364 1364 return subset & names
1365 1365
1366 1366 @predicate('id(string)', safe=True)
1367 1367 def node_(repo, subset, x):
1368 1368 """Revision non-ambiguously specified by the given hex string prefix.
1369 1369 """
1370 1370 # i18n: "id" is a keyword
1371 1371 l = getargs(x, 1, 1, _("id requires one argument"))
1372 1372 # i18n: "id" is a keyword
1373 1373 n = getstring(l[0], _("id requires a string"))
1374 1374 if len(n) == 40:
1375 1375 try:
1376 1376 rn = repo.changelog.rev(node.bin(n))
1377 1377 except (LookupError, TypeError):
1378 1378 rn = None
1379 1379 else:
1380 1380 rn = None
1381 1381 pm = repo.changelog._partialmatch(n)
1382 1382 if pm is not None:
1383 1383 rn = repo.changelog.rev(pm)
1384 1384
1385 1385 if rn is None:
1386 1386 return baseset()
1387 1387 result = baseset([rn])
1388 1388 return result & subset
1389 1389
1390 1390 @predicate('obsolete()', safe=True)
1391 1391 def obsolete(repo, subset, x):
1392 1392 """Mutable changeset with a newer version."""
1393 1393 # i18n: "obsolete" is a keyword
1394 1394 getargs(x, 0, 0, _("obsolete takes no arguments"))
1395 1395 obsoletes = obsmod.getrevs(repo, 'obsolete')
1396 1396 return subset & obsoletes
1397 1397
1398 1398 @predicate('only(set, [set])', safe=True)
1399 1399 def only(repo, subset, x):
1400 1400 """Changesets that are ancestors of the first set that are not ancestors
1401 1401 of any other head in the repo. If a second set is specified, the result
1402 1402 is ancestors of the first set that are not ancestors of the second set
1403 1403 (i.e. ::<set1> - ::<set2>).
1404 1404 """
1405 1405 cl = repo.changelog
1406 1406 # i18n: "only" is a keyword
1407 1407 args = getargs(x, 1, 2, _('only takes one or two arguments'))
1408 1408 include = getset(repo, fullreposet(repo), args[0])
1409 1409 if len(args) == 1:
1410 1410 if not include:
1411 1411 return baseset()
1412 1412
1413 1413 descendants = set(_revdescendants(repo, include, False))
1414 1414 exclude = [rev for rev in cl.headrevs()
1415 1415 if not rev in descendants and not rev in include]
1416 1416 else:
1417 1417 exclude = getset(repo, fullreposet(repo), args[1])
1418 1418
1419 1419 results = set(cl.findmissingrevs(common=exclude, heads=include))
1420 1420 # XXX we should turn this into a baseset instead of a set, smartset may do
1421 1421 # some optimisations from the fact this is a baseset.
1422 1422 return subset & results
1423 1423
1424 1424 @predicate('origin([set])', safe=True)
1425 1425 def origin(repo, subset, x):
1426 1426 """
1427 1427 Changesets that were specified as a source for the grafts, transplants or
1428 1428 rebases that created the given revisions. Omitting the optional set is the
1429 1429 same as passing all(). If a changeset created by these operations is itself
1430 1430 specified as a source for one of these operations, only the source changeset
1431 1431 for the first operation is selected.
1432 1432 """
1433 1433 if x is not None:
1434 1434 dests = getset(repo, fullreposet(repo), x)
1435 1435 else:
1436 1436 dests = fullreposet(repo)
1437 1437
1438 1438 def _firstsrc(rev):
1439 1439 src = _getrevsource(repo, rev)
1440 1440 if src is None:
1441 1441 return None
1442 1442
1443 1443 while True:
1444 1444 prev = _getrevsource(repo, src)
1445 1445
1446 1446 if prev is None:
1447 1447 return src
1448 1448 src = prev
1449 1449
1450 1450 o = set([_firstsrc(r) for r in dests])
1451 1451 o -= set([None])
1452 1452 # XXX we should turn this into a baseset instead of a set, smartset may do
1453 1453 # some optimisations from the fact this is a baseset.
1454 1454 return subset & o
1455 1455
1456 1456 @predicate('outgoing([path])', safe=True)
1457 1457 def outgoing(repo, subset, x):
1458 1458 """Changesets not found in the specified destination repository, or the
1459 1459 default push location.
1460 1460 """
1461 1461 # Avoid cycles.
1462 1462 from . import (
1463 1463 discovery,
1464 1464 hg,
1465 1465 )
1466 1466 # i18n: "outgoing" is a keyword
1467 1467 l = getargs(x, 0, 1, _("outgoing takes one or no arguments"))
1468 1468 # i18n: "outgoing" is a keyword
1469 1469 dest = l and getstring(l[0], _("outgoing requires a repository path")) or ''
1470 1470 dest = repo.ui.expandpath(dest or 'default-push', dest or 'default')
1471 1471 dest, branches = hg.parseurl(dest)
1472 1472 revs, checkout = hg.addbranchrevs(repo, repo, branches, [])
1473 1473 if revs:
1474 1474 revs = [repo.lookup(rev) for rev in revs]
1475 1475 other = hg.peer(repo, {}, dest)
1476 1476 repo.ui.pushbuffer()
1477 1477 outgoing = discovery.findcommonoutgoing(repo, other, onlyheads=revs)
1478 1478 repo.ui.popbuffer()
1479 1479 cl = repo.changelog
1480 1480 o = set([cl.rev(r) for r in outgoing.missing])
1481 1481 return subset & o
1482 1482
1483 1483 @predicate('p1([set])', safe=True)
1484 1484 def p1(repo, subset, x):
1485 1485 """First parent of changesets in set, or the working directory.
1486 1486 """
1487 1487 if x is None:
1488 1488 p = repo[x].p1().rev()
1489 1489 if p >= 0:
1490 1490 return subset & baseset([p])
1491 1491 return baseset()
1492 1492
1493 1493 ps = set()
1494 1494 cl = repo.changelog
1495 1495 for r in getset(repo, fullreposet(repo), x):
1496 1496 ps.add(cl.parentrevs(r)[0])
1497 1497 ps -= set([node.nullrev])
1498 1498 # XXX we should turn this into a baseset instead of a set, smartset may do
1499 1499 # some optimisations from the fact this is a baseset.
1500 1500 return subset & ps
1501 1501
1502 1502 @predicate('p2([set])', safe=True)
1503 1503 def p2(repo, subset, x):
1504 1504 """Second parent of changesets in set, or the working directory.
1505 1505 """
1506 1506 if x is None:
1507 1507 ps = repo[x].parents()
1508 1508 try:
1509 1509 p = ps[1].rev()
1510 1510 if p >= 0:
1511 1511 return subset & baseset([p])
1512 1512 return baseset()
1513 1513 except IndexError:
1514 1514 return baseset()
1515 1515
1516 1516 ps = set()
1517 1517 cl = repo.changelog
1518 1518 for r in getset(repo, fullreposet(repo), x):
1519 1519 ps.add(cl.parentrevs(r)[1])
1520 1520 ps -= set([node.nullrev])
1521 1521 # XXX we should turn this into a baseset instead of a set, smartset may do
1522 1522 # some optimisations from the fact this is a baseset.
1523 1523 return subset & ps
1524 1524
1525 1525 @predicate('parents([set])', safe=True)
1526 1526 def parents(repo, subset, x):
1527 1527 """
1528 1528 The set of all parents for all changesets in set, or the working directory.
1529 1529 """
1530 1530 if x is None:
1531 1531 ps = set(p.rev() for p in repo[x].parents())
1532 1532 else:
1533 1533 ps = set()
1534 1534 cl = repo.changelog
1535 1535 up = ps.update
1536 1536 parentrevs = cl.parentrevs
1537 1537 for r in getset(repo, fullreposet(repo), x):
1538 1538 if r == node.wdirrev:
1539 1539 up(p.rev() for p in repo[r].parents())
1540 1540 else:
1541 1541 up(parentrevs(r))
1542 1542 ps -= set([node.nullrev])
1543 1543 return subset & ps
1544 1544
1545 1545 def _phase(repo, subset, target):
1546 1546 """helper to select all rev in phase <target>"""
1547 1547 repo._phasecache.loadphaserevs(repo) # ensure phase's sets are loaded
1548 1548 if repo._phasecache._phasesets:
1549 1549 s = repo._phasecache._phasesets[target] - repo.changelog.filteredrevs
1550 1550 s = baseset(s)
1551 1551 s.sort() # set are non ordered, so we enforce ascending
1552 1552 return subset & s
1553 1553 else:
1554 1554 phase = repo._phasecache.phase
1555 1555 condition = lambda r: phase(repo, r) == target
1556 1556 return subset.filter(condition, condrepr=('<phase %r>', target),
1557 1557 cache=False)
1558 1558
1559 1559 @predicate('draft()', safe=True)
1560 1560 def draft(repo, subset, x):
1561 1561 """Changeset in draft phase."""
1562 1562 # i18n: "draft" is a keyword
1563 1563 getargs(x, 0, 0, _("draft takes no arguments"))
1564 1564 target = phases.draft
1565 1565 return _phase(repo, subset, target)
1566 1566
1567 1567 @predicate('secret()', safe=True)
1568 1568 def secret(repo, subset, x):
1569 1569 """Changeset in secret phase."""
1570 1570 # i18n: "secret" is a keyword
1571 1571 getargs(x, 0, 0, _("secret takes no arguments"))
1572 1572 target = phases.secret
1573 1573 return _phase(repo, subset, target)
1574 1574
1575 1575 def parentspec(repo, subset, x, n):
1576 1576 """``set^0``
1577 1577 The set.
1578 1578 ``set^1`` (or ``set^``), ``set^2``
1579 1579 First or second parent, respectively, of all changesets in set.
1580 1580 """
1581 1581 try:
1582 1582 n = int(n[1])
1583 1583 if n not in (0, 1, 2):
1584 1584 raise ValueError
1585 1585 except (TypeError, ValueError):
1586 1586 raise error.ParseError(_("^ expects a number 0, 1, or 2"))
1587 1587 ps = set()
1588 1588 cl = repo.changelog
1589 1589 for r in getset(repo, fullreposet(repo), x):
1590 1590 if n == 0:
1591 1591 ps.add(r)
1592 1592 elif n == 1:
1593 1593 ps.add(cl.parentrevs(r)[0])
1594 1594 elif n == 2:
1595 1595 parents = cl.parentrevs(r)
1596 1596 if len(parents) > 1:
1597 1597 ps.add(parents[1])
1598 1598 return subset & ps
1599 1599
1600 1600 @predicate('present(set)', safe=True)
1601 1601 def present(repo, subset, x):
1602 1602 """An empty set, if any revision in set isn't found; otherwise,
1603 1603 all revisions in set.
1604 1604
1605 1605 If any of specified revisions is not present in the local repository,
1606 1606 the query is normally aborted. But this predicate allows the query
1607 1607 to continue even in such cases.
1608 1608 """
1609 1609 try:
1610 1610 return getset(repo, subset, x)
1611 1611 except error.RepoLookupError:
1612 1612 return baseset()
1613 1613
1614 1614 # for internal use
1615 1615 @predicate('_notpublic', safe=True)
1616 1616 def _notpublic(repo, subset, x):
1617 1617 getargs(x, 0, 0, "_notpublic takes no arguments")
1618 1618 repo._phasecache.loadphaserevs(repo) # ensure phase's sets are loaded
1619 1619 if repo._phasecache._phasesets:
1620 1620 s = set()
1621 1621 for u in repo._phasecache._phasesets[1:]:
1622 1622 s.update(u)
1623 1623 s = baseset(s - repo.changelog.filteredrevs)
1624 1624 s.sort()
1625 1625 return subset & s
1626 1626 else:
1627 1627 phase = repo._phasecache.phase
1628 1628 target = phases.public
1629 1629 condition = lambda r: phase(repo, r) != target
1630 1630 return subset.filter(condition, condrepr=('<phase %r>', target),
1631 1631 cache=False)
1632 1632
1633 1633 @predicate('public()', safe=True)
1634 1634 def public(repo, subset, x):
1635 1635 """Changeset in public phase."""
1636 1636 # i18n: "public" is a keyword
1637 1637 getargs(x, 0, 0, _("public takes no arguments"))
1638 1638 phase = repo._phasecache.phase
1639 1639 target = phases.public
1640 1640 condition = lambda r: phase(repo, r) == target
1641 1641 return subset.filter(condition, condrepr=('<phase %r>', target),
1642 1642 cache=False)
1643 1643
1644 1644 @predicate('remote([id [,path]])', safe=True)
1645 1645 def remote(repo, subset, x):
1646 1646 """Local revision that corresponds to the given identifier in a
1647 1647 remote repository, if present. Here, the '.' identifier is a
1648 1648 synonym for the current local branch.
1649 1649 """
1650 1650
1651 1651 from . import hg # avoid start-up nasties
1652 1652 # i18n: "remote" is a keyword
1653 1653 l = getargs(x, 0, 2, _("remote takes zero, one, or two arguments"))
1654 1654
1655 1655 q = '.'
1656 1656 if len(l) > 0:
1657 1657 # i18n: "remote" is a keyword
1658 1658 q = getstring(l[0], _("remote requires a string id"))
1659 1659 if q == '.':
1660 1660 q = repo['.'].branch()
1661 1661
1662 1662 dest = ''
1663 1663 if len(l) > 1:
1664 1664 # i18n: "remote" is a keyword
1665 1665 dest = getstring(l[1], _("remote requires a repository path"))
1666 1666 dest = repo.ui.expandpath(dest or 'default')
1667 1667 dest, branches = hg.parseurl(dest)
1668 1668 revs, checkout = hg.addbranchrevs(repo, repo, branches, [])
1669 1669 if revs:
1670 1670 revs = [repo.lookup(rev) for rev in revs]
1671 1671 other = hg.peer(repo, {}, dest)
1672 1672 n = other.lookup(q)
1673 1673 if n in repo:
1674 1674 r = repo[n].rev()
1675 1675 if r in subset:
1676 1676 return baseset([r])
1677 1677 return baseset()
1678 1678
1679 1679 @predicate('removes(pattern)', safe=True)
1680 1680 def removes(repo, subset, x):
1681 1681 """Changesets which remove files matching pattern.
1682 1682
1683 1683 The pattern without explicit kind like ``glob:`` is expected to be
1684 1684 relative to the current directory and match against a file or a
1685 1685 directory.
1686 1686 """
1687 1687 # i18n: "removes" is a keyword
1688 1688 pat = getstring(x, _("removes requires a pattern"))
1689 1689 return checkstatus(repo, subset, pat, 2)
1690 1690
1691 1691 @predicate('rev(number)', safe=True)
1692 1692 def rev(repo, subset, x):
1693 1693 """Revision with the given numeric identifier.
1694 1694 """
1695 1695 # i18n: "rev" is a keyword
1696 1696 l = getargs(x, 1, 1, _("rev requires one argument"))
1697 1697 try:
1698 1698 # i18n: "rev" is a keyword
1699 1699 l = int(getstring(l[0], _("rev requires a number")))
1700 1700 except (TypeError, ValueError):
1701 1701 # i18n: "rev" is a keyword
1702 1702 raise error.ParseError(_("rev expects a number"))
1703 1703 if l not in repo.changelog and l != node.nullrev:
1704 1704 return baseset()
1705 1705 return subset & baseset([l])
1706 1706
1707 1707 @predicate('matching(revision [, field])', safe=True)
1708 1708 def matching(repo, subset, x):
1709 1709 """Changesets in which a given set of fields match the set of fields in the
1710 1710 selected revision or set.
1711 1711
1712 1712 To match more than one field pass the list of fields to match separated
1713 1713 by spaces (e.g. ``author description``).
1714 1714
1715 1715 Valid fields are most regular revision fields and some special fields.
1716 1716
1717 1717 Regular revision fields are ``description``, ``author``, ``branch``,
1718 1718 ``date``, ``files``, ``phase``, ``parents``, ``substate``, ``user``
1719 1719 and ``diff``.
1720 1720 Note that ``author`` and ``user`` are synonyms. ``diff`` refers to the
1721 1721 contents of the revision. Two revisions matching their ``diff`` will
1722 1722 also match their ``files``.
1723 1723
1724 1724 Special fields are ``summary`` and ``metadata``:
1725 1725 ``summary`` matches the first line of the description.
1726 1726 ``metadata`` is equivalent to matching ``description user date``
1727 1727 (i.e. it matches the main metadata fields).
1728 1728
1729 1729 ``metadata`` is the default field which is used when no fields are
1730 1730 specified. You can match more than one field at a time.
1731 1731 """
1732 1732 # i18n: "matching" is a keyword
1733 1733 l = getargs(x, 1, 2, _("matching takes 1 or 2 arguments"))
1734 1734
1735 1735 revs = getset(repo, fullreposet(repo), l[0])
1736 1736
1737 1737 fieldlist = ['metadata']
1738 1738 if len(l) > 1:
1739 1739 fieldlist = getstring(l[1],
1740 1740 # i18n: "matching" is a keyword
1741 1741 _("matching requires a string "
1742 1742 "as its second argument")).split()
1743 1743
1744 1744 # Make sure that there are no repeated fields,
1745 1745 # expand the 'special' 'metadata' field type
1746 1746 # and check the 'files' whenever we check the 'diff'
1747 1747 fields = []
1748 1748 for field in fieldlist:
1749 1749 if field == 'metadata':
1750 1750 fields += ['user', 'description', 'date']
1751 1751 elif field == 'diff':
1752 1752 # a revision matching the diff must also match the files
1753 1753 # since matching the diff is very costly, make sure to
1754 1754 # also match the files first
1755 1755 fields += ['files', 'diff']
1756 1756 else:
1757 1757 if field == 'author':
1758 1758 field = 'user'
1759 1759 fields.append(field)
1760 1760 fields = set(fields)
1761 1761 if 'summary' in fields and 'description' in fields:
1762 1762 # If a revision matches its description it also matches its summary
1763 1763 fields.discard('summary')
1764 1764
1765 1765 # We may want to match more than one field
1766 1766 # Not all fields take the same amount of time to be matched
1767 1767 # Sort the selected fields in order of increasing matching cost
1768 1768 fieldorder = ['phase', 'parents', 'user', 'date', 'branch', 'summary',
1769 1769 'files', 'description', 'substate', 'diff']
1770 1770 def fieldkeyfunc(f):
1771 1771 try:
1772 1772 return fieldorder.index(f)
1773 1773 except ValueError:
1774 1774 # assume an unknown field is very costly
1775 1775 return len(fieldorder)
1776 1776 fields = list(fields)
1777 1777 fields.sort(key=fieldkeyfunc)
1778 1778
1779 1779 # Each field will be matched with its own "getfield" function
1780 1780 # which will be added to the getfieldfuncs array of functions
1781 1781 getfieldfuncs = []
1782 1782 _funcs = {
1783 1783 'user': lambda r: repo[r].user(),
1784 1784 'branch': lambda r: repo[r].branch(),
1785 1785 'date': lambda r: repo[r].date(),
1786 1786 'description': lambda r: repo[r].description(),
1787 1787 'files': lambda r: repo[r].files(),
1788 1788 'parents': lambda r: repo[r].parents(),
1789 1789 'phase': lambda r: repo[r].phase(),
1790 1790 'substate': lambda r: repo[r].substate,
1791 1791 'summary': lambda r: repo[r].description().splitlines()[0],
1792 1792 'diff': lambda r: list(repo[r].diff(git=True),)
1793 1793 }
1794 1794 for info in fields:
1795 1795 getfield = _funcs.get(info, None)
1796 1796 if getfield is None:
1797 1797 raise error.ParseError(
1798 1798 # i18n: "matching" is a keyword
1799 1799 _("unexpected field name passed to matching: %s") % info)
1800 1800 getfieldfuncs.append(getfield)
1801 1801 # convert the getfield array of functions into a "getinfo" function
1802 1802 # which returns an array of field values (or a single value if there
1803 1803 # is only one field to match)
1804 1804 getinfo = lambda r: [f(r) for f in getfieldfuncs]
1805 1805
1806 1806 def matches(x):
1807 1807 for rev in revs:
1808 1808 target = getinfo(rev)
1809 1809 match = True
1810 1810 for n, f in enumerate(getfieldfuncs):
1811 1811 if target[n] != f(x):
1812 1812 match = False
1813 1813 if match:
1814 1814 return True
1815 1815 return False
1816 1816
1817 1817 return subset.filter(matches, condrepr=('<matching%r %r>', fields, revs))
1818 1818
1819 1819 @predicate('reverse(set)', safe=True)
1820 1820 def reverse(repo, subset, x):
1821 1821 """Reverse order of set.
1822 1822 """
1823 1823 l = getset(repo, subset, x)
1824 1824 l.reverse()
1825 1825 return l
1826 1826
1827 1827 @predicate('roots(set)', safe=True)
1828 1828 def roots(repo, subset, x):
1829 1829 """Changesets in set with no parent changeset in set.
1830 1830 """
1831 1831 s = getset(repo, fullreposet(repo), x)
1832 1832 parents = repo.changelog.parentrevs
1833 1833 def filter(r):
1834 1834 for p in parents(r):
1835 1835 if 0 <= p and p in s:
1836 1836 return False
1837 1837 return True
1838 1838 return subset & s.filter(filter, condrepr='<roots>')
1839 1839
1840 1840 @predicate('sort(set[, [-]key...])', safe=True)
1841 1841 def sort(repo, subset, x):
1842 1842 """Sort set by keys. The default sort order is ascending, specify a key
1843 1843 as ``-key`` to sort in descending order.
1844 1844
1845 1845 The keys can be:
1846 1846
1847 1847 - ``rev`` for the revision number,
1848 1848 - ``branch`` for the branch name,
1849 1849 - ``desc`` for the commit message (description),
1850 1850 - ``user`` for user name (``author`` can be used as an alias),
1851 1851 - ``date`` for the commit date
1852 1852 """
1853 1853 # i18n: "sort" is a keyword
1854 1854 l = getargs(x, 1, 2, _("sort requires one or two arguments"))
1855 1855 keys = "rev"
1856 1856 if len(l) == 2:
1857 1857 # i18n: "sort" is a keyword
1858 1858 keys = getstring(l[1], _("sort spec must be a string"))
1859 1859
1860 1860 s = l[0]
1861 1861 keys = keys.split()
1862 1862 l = []
1863 1863 def invert(s):
1864 1864 return "".join(chr(255 - ord(c)) for c in s)
1865 1865 revs = getset(repo, subset, s)
1866 1866 if keys == ["rev"]:
1867 1867 revs.sort()
1868 1868 return revs
1869 1869 elif keys == ["-rev"]:
1870 1870 revs.sort(reverse=True)
1871 1871 return revs
1872 1872 for r in revs:
1873 1873 c = repo[r]
1874 1874 e = []
1875 1875 for k in keys:
1876 1876 if k == 'rev':
1877 1877 e.append(r)
1878 1878 elif k == '-rev':
1879 1879 e.append(-r)
1880 1880 elif k == 'branch':
1881 1881 e.append(c.branch())
1882 1882 elif k == '-branch':
1883 1883 e.append(invert(c.branch()))
1884 1884 elif k == 'desc':
1885 1885 e.append(c.description())
1886 1886 elif k == '-desc':
1887 1887 e.append(invert(c.description()))
1888 1888 elif k in 'user author':
1889 1889 e.append(c.user())
1890 1890 elif k in '-user -author':
1891 1891 e.append(invert(c.user()))
1892 1892 elif k == 'date':
1893 1893 e.append(c.date()[0])
1894 1894 elif k == '-date':
1895 1895 e.append(-c.date()[0])
1896 1896 else:
1897 1897 raise error.ParseError(_("unknown sort key %r") % k)
1898 1898 e.append(r)
1899 1899 l.append(e)
1900 1900 l.sort()
1901 1901 return baseset([e[-1] for e in l])
1902 1902
1903 1903 @predicate('subrepo([pattern])')
1904 1904 def subrepo(repo, subset, x):
1905 1905 """Changesets that add, modify or remove the given subrepo. If no subrepo
1906 1906 pattern is named, any subrepo changes are returned.
1907 1907 """
1908 1908 # i18n: "subrepo" is a keyword
1909 1909 args = getargs(x, 0, 1, _('subrepo takes at most one argument'))
1910 1910 pat = None
1911 1911 if len(args) != 0:
1912 1912 pat = getstring(args[0], _("subrepo requires a pattern"))
1913 1913
1914 1914 m = matchmod.exact(repo.root, repo.root, ['.hgsubstate'])
1915 1915
1916 1916 def submatches(names):
1917 1917 k, p, m = util.stringmatcher(pat)
1918 1918 for name in names:
1919 1919 if m(name):
1920 1920 yield name
1921 1921
1922 1922 def matches(x):
1923 1923 c = repo[x]
1924 1924 s = repo.status(c.p1().node(), c.node(), match=m)
1925 1925
1926 1926 if pat is None:
1927 1927 return s.added or s.modified or s.removed
1928 1928
1929 1929 if s.added:
1930 1930 return any(submatches(c.substate.keys()))
1931 1931
1932 1932 if s.modified:
1933 1933 subs = set(c.p1().substate.keys())
1934 1934 subs.update(c.substate.keys())
1935 1935
1936 1936 for path in submatches(subs):
1937 1937 if c.p1().substate.get(path) != c.substate.get(path):
1938 1938 return True
1939 1939
1940 1940 if s.removed:
1941 1941 return any(submatches(c.p1().substate.keys()))
1942 1942
1943 1943 return False
1944 1944
1945 1945 return subset.filter(matches, condrepr=('<subrepo %r>', pat))
1946 1946
1947 1947 def _substringmatcher(pattern):
1948 1948 kind, pattern, matcher = util.stringmatcher(pattern)
1949 1949 if kind == 'literal':
1950 1950 matcher = lambda s: pattern in s
1951 1951 return kind, pattern, matcher
1952 1952
1953 1953 @predicate('tag([name])', safe=True)
1954 1954 def tag(repo, subset, x):
1955 1955 """The specified tag by name, or all tagged revisions if no name is given.
1956 1956
1957 1957 If `name` starts with `re:`, the remainder of the name is treated as
1958 1958 a regular expression. To match a tag that actually starts with `re:`,
1959 1959 use the prefix `literal:`.
1960 1960 """
1961 1961 # i18n: "tag" is a keyword
1962 1962 args = getargs(x, 0, 1, _("tag takes one or no arguments"))
1963 1963 cl = repo.changelog
1964 1964 if args:
1965 1965 pattern = getstring(args[0],
1966 1966 # i18n: "tag" is a keyword
1967 1967 _('the argument to tag must be a string'))
1968 1968 kind, pattern, matcher = util.stringmatcher(pattern)
1969 1969 if kind == 'literal':
1970 1970 # avoid resolving all tags
1971 1971 tn = repo._tagscache.tags.get(pattern, None)
1972 1972 if tn is None:
1973 1973 raise error.RepoLookupError(_("tag '%s' does not exist")
1974 1974 % pattern)
1975 1975 s = set([repo[tn].rev()])
1976 1976 else:
1977 1977 s = set([cl.rev(n) for t, n in repo.tagslist() if matcher(t)])
1978 1978 else:
1979 1979 s = set([cl.rev(n) for t, n in repo.tagslist() if t != 'tip'])
1980 1980 return subset & s
1981 1981
1982 1982 @predicate('tagged', safe=True)
1983 1983 def tagged(repo, subset, x):
1984 1984 return tag(repo, subset, x)
1985 1985
1986 1986 @predicate('unstable()', safe=True)
1987 1987 def unstable(repo, subset, x):
1988 1988 """Non-obsolete changesets with obsolete ancestors.
1989 1989 """
1990 1990 # i18n: "unstable" is a keyword
1991 1991 getargs(x, 0, 0, _("unstable takes no arguments"))
1992 1992 unstables = obsmod.getrevs(repo, 'unstable')
1993 1993 return subset & unstables
1994 1994
1995 1995
1996 1996 @predicate('user(string)', safe=True)
1997 1997 def user(repo, subset, x):
1998 1998 """User name contains string. The match is case-insensitive.
1999 1999
2000 2000 If `string` starts with `re:`, the remainder of the string is treated as
2001 2001 a regular expression. To match a user that actually contains `re:`, use
2002 2002 the prefix `literal:`.
2003 2003 """
2004 2004 return author(repo, subset, x)
2005 2005
2006 2006 # experimental
2007 2007 @predicate('wdir', safe=True)
2008 2008 def wdir(repo, subset, x):
2009 2009 # i18n: "wdir" is a keyword
2010 2010 getargs(x, 0, 0, _("wdir takes no arguments"))
2011 2011 if node.wdirrev in subset or isinstance(subset, fullreposet):
2012 2012 return baseset([node.wdirrev])
2013 2013 return baseset()
2014 2014
2015 2015 # for internal use
2016 2016 @predicate('_list', safe=True)
2017 2017 def _list(repo, subset, x):
2018 2018 s = getstring(x, "internal error")
2019 2019 if not s:
2020 2020 return baseset()
2021 2021 # remove duplicates here. it's difficult for caller to deduplicate sets
2022 2022 # because different symbols can point to the same rev.
2023 2023 cl = repo.changelog
2024 2024 ls = []
2025 2025 seen = set()
2026 2026 for t in s.split('\0'):
2027 2027 try:
2028 2028 # fast path for integer revision
2029 2029 r = int(t)
2030 2030 if str(r) != t or r not in cl:
2031 2031 raise ValueError
2032 2032 revs = [r]
2033 2033 except ValueError:
2034 2034 revs = stringset(repo, subset, t)
2035 2035
2036 2036 for r in revs:
2037 2037 if r in seen:
2038 2038 continue
2039 2039 if (r in subset
2040 2040 or r == node.nullrev and isinstance(subset, fullreposet)):
2041 2041 ls.append(r)
2042 2042 seen.add(r)
2043 2043 return baseset(ls)
2044 2044
2045 2045 # for internal use
2046 2046 @predicate('_intlist', safe=True)
2047 2047 def _intlist(repo, subset, x):
2048 2048 s = getstring(x, "internal error")
2049 2049 if not s:
2050 2050 return baseset()
2051 2051 ls = [int(r) for r in s.split('\0')]
2052 2052 s = subset
2053 2053 return baseset([r for r in ls if r in s])
2054 2054
2055 2055 # for internal use
2056 2056 @predicate('_hexlist', safe=True)
2057 2057 def _hexlist(repo, subset, x):
2058 2058 s = getstring(x, "internal error")
2059 2059 if not s:
2060 2060 return baseset()
2061 2061 cl = repo.changelog
2062 2062 ls = [cl.rev(node.bin(r)) for r in s.split('\0')]
2063 2063 s = subset
2064 2064 return baseset([r for r in ls if r in s])
2065 2065
2066 2066 methods = {
2067 2067 "range": rangeset,
2068 2068 "dagrange": dagrange,
2069 2069 "string": stringset,
2070 2070 "symbol": stringset,
2071 2071 "and": andset,
2072 2072 "or": orset,
2073 2073 "not": notset,
2074 2074 "difference": differenceset,
2075 2075 "list": listset,
2076 2076 "keyvalue": keyvaluepair,
2077 2077 "func": func,
2078 2078 "ancestor": ancestorspec,
2079 2079 "parent": parentspec,
2080 2080 "parentpost": p1,
2081 2081 }
2082 2082
2083 2083 def optimize(x, small):
2084 2084 if x is None:
2085 2085 return 0, x
2086 2086
2087 2087 smallbonus = 1
2088 2088 if small:
2089 2089 smallbonus = .5
2090 2090
2091 2091 op = x[0]
2092 2092 if op == 'minus':
2093 2093 return optimize(('and', x[1], ('not', x[2])), small)
2094 2094 elif op == 'only':
2095 2095 return optimize(('func', ('symbol', 'only'),
2096 2096 ('list', x[1], x[2])), small)
2097 2097 elif op == 'onlypost':
2098 2098 return optimize(('func', ('symbol', 'only'), x[1]), small)
2099 2099 elif op == 'dagrangepre':
2100 2100 return optimize(('func', ('symbol', 'ancestors'), x[1]), small)
2101 2101 elif op == 'dagrangepost':
2102 2102 return optimize(('func', ('symbol', 'descendants'), x[1]), small)
2103 2103 elif op == 'rangeall':
2104 2104 return optimize(('range', ('string', '0'), ('string', 'tip')), small)
2105 2105 elif op == 'rangepre':
2106 2106 return optimize(('range', ('string', '0'), x[1]), small)
2107 2107 elif op == 'rangepost':
2108 2108 return optimize(('range', x[1], ('string', 'tip')), small)
2109 2109 elif op == 'negate':
2110 2110 return optimize(('string',
2111 2111 '-' + getstring(x[1], _("can't negate that"))), small)
2112 2112 elif op in 'string symbol negate':
2113 2113 return smallbonus, x # single revisions are small
2114 2114 elif op == 'and':
2115 2115 wa, ta = optimize(x[1], True)
2116 2116 wb, tb = optimize(x[2], True)
2117 2117
2118 2118 # (::x and not ::y)/(not ::y and ::x) have a fast path
2119 2119 def isonly(revs, bases):
2120 2120 return (
2121 2121 revs is not None
2122 2122 and revs[0] == 'func'
2123 2123 and getstring(revs[1], _('not a symbol')) == 'ancestors'
2124 2124 and bases is not None
2125 2125 and bases[0] == 'not'
2126 2126 and bases[1][0] == 'func'
2127 2127 and getstring(bases[1][1], _('not a symbol')) == 'ancestors')
2128 2128
2129 2129 w = min(wa, wb)
2130 2130 if isonly(ta, tb):
2131 2131 return w, ('func', ('symbol', 'only'), ('list', ta[2], tb[1][2]))
2132 2132 if isonly(tb, ta):
2133 2133 return w, ('func', ('symbol', 'only'), ('list', tb[2], ta[1][2]))
2134 2134
2135 2135 if tb is not None and tb[0] == 'not':
2136 2136 return wa, ('difference', ta, tb[1])
2137 2137
2138 2138 if wa > wb:
2139 2139 return w, (op, tb, ta)
2140 2140 return w, (op, ta, tb)
2141 2141 elif op == 'or':
2142 2142 # fast path for machine-generated expression, that is likely to have
2143 2143 # lots of trivial revisions: 'a + b + c()' to '_list(a b) + c()'
2144 2144 ws, ts, ss = [], [], []
2145 2145 def flushss():
2146 2146 if not ss:
2147 2147 return
2148 2148 if len(ss) == 1:
2149 2149 w, t = ss[0]
2150 2150 else:
2151 2151 s = '\0'.join(t[1] for w, t in ss)
2152 2152 y = ('func', ('symbol', '_list'), ('string', s))
2153 2153 w, t = optimize(y, False)
2154 2154 ws.append(w)
2155 2155 ts.append(t)
2156 2156 del ss[:]
2157 2157 for y in x[1:]:
2158 2158 w, t = optimize(y, False)
2159 2159 if t is not None and (t[0] == 'string' or t[0] == 'symbol'):
2160 2160 ss.append((w, t))
2161 2161 continue
2162 2162 flushss()
2163 2163 ws.append(w)
2164 2164 ts.append(t)
2165 2165 flushss()
2166 2166 if len(ts) == 1:
2167 2167 return ws[0], ts[0] # 'or' operation is fully optimized out
2168 2168 # we can't reorder trees by weight because it would change the order.
2169 2169 # ("sort(a + b)" == "sort(b + a)", but "a + b" != "b + a")
2170 2170 # ts = tuple(t for w, t in sorted(zip(ws, ts), key=lambda wt: wt[0]))
2171 2171 return max(ws), (op,) + tuple(ts)
2172 2172 elif op == 'not':
2173 2173 # Optimize not public() to _notpublic() because we have a fast version
2174 2174 if x[1] == ('func', ('symbol', 'public'), None):
2175 2175 newsym = ('func', ('symbol', '_notpublic'), None)
2176 2176 o = optimize(newsym, not small)
2177 2177 return o[0], o[1]
2178 2178 else:
2179 2179 o = optimize(x[1], not small)
2180 2180 return o[0], (op, o[1])
2181 2181 elif op == 'parentpost':
2182 2182 o = optimize(x[1], small)
2183 2183 return o[0], (op, o[1])
2184 2184 elif op == 'group':
2185 2185 return optimize(x[1], small)
2186 2186 elif op in 'dagrange range parent ancestorspec':
2187 2187 if op == 'parent':
2188 2188 # x^:y means (x^) : y, not x ^ (:y)
2189 2189 post = ('parentpost', x[1])
2190 2190 if x[2][0] == 'dagrangepre':
2191 2191 return optimize(('dagrange', post, x[2][1]), small)
2192 2192 elif x[2][0] == 'rangepre':
2193 2193 return optimize(('range', post, x[2][1]), small)
2194 2194
2195 2195 wa, ta = optimize(x[1], small)
2196 2196 wb, tb = optimize(x[2], small)
2197 2197 return wa + wb, (op, ta, tb)
2198 2198 elif op == 'list':
2199 2199 ws, ts = zip(*(optimize(y, small) for y in x[1:]))
2200 2200 return sum(ws), (op,) + ts
2201 2201 elif op == 'func':
2202 2202 f = getstring(x[1], _("not a symbol"))
2203 2203 wa, ta = optimize(x[2], small)
2204 2204 if f in ("author branch closed date desc file grep keyword "
2205 2205 "outgoing user"):
2206 2206 w = 10 # slow
2207 2207 elif f in "modifies adds removes":
2208 2208 w = 30 # slower
2209 2209 elif f == "contains":
2210 2210 w = 100 # very slow
2211 2211 elif f == "ancestor":
2212 2212 w = 1 * smallbonus
2213 2213 elif f in "reverse limit first _intlist":
2214 2214 w = 0
2215 2215 elif f in "sort":
2216 2216 w = 10 # assume most sorts look at changelog
2217 2217 else:
2218 2218 w = 1
2219 2219 return w + wa, (op, x[1], ta)
2220 2220 return 1, x
2221 2221
2222 2222 # the set of valid characters for the initial letter of symbols in
2223 2223 # alias declarations and definitions
2224 2224 _aliassyminitletters = set(c for c in [chr(i) for i in xrange(256)]
2225 2225 if c.isalnum() or c in '._@$' or ord(c) > 127)
2226 2226
2227 2227 def _tokenizealias(program, lookup=None):
2228 2228 """Parse alias declaration/definition into a stream of tokens
2229 2229
2230 2230 This allows symbol names to use also ``$`` as an initial letter
2231 2231 (for backward compatibility), and callers of this function should
2232 2232 examine whether ``$`` is used also for unexpected symbols or not.
2233 2233 """
2234 2234 return tokenize(program, lookup=lookup,
2235 2235 syminitletters=_aliassyminitletters)
2236 2236
2237 2237 def _parsealias(spec):
2238 2238 """Parse alias declaration/definition ``spec``
2239 2239
2240 2240 >>> _parsealias('foo($1)')
2241 2241 ('func', ('symbol', 'foo'), ('symbol', '$1'))
2242 2242 >>> _parsealias('foo bar')
2243 2243 Traceback (most recent call last):
2244 2244 ...
2245 2245 ParseError: ('invalid token', 4)
2246 2246 """
2247 2247 p = parser.parser(elements)
2248 2248 tree, pos = p.parse(_tokenizealias(spec))
2249 2249 if pos != len(spec):
2250 2250 raise error.ParseError(_('invalid token'), pos)
2251 2251 return parser.simplifyinfixops(tree, ('list', 'or'))
2252 2252
2253 2253 class _aliasrules(parser.basealiasrules):
2254 2254 """Parsing and expansion rule set of revset aliases"""
2255 2255 _section = _('revset alias')
2256 2256 _parse = staticmethod(_parsealias)
2257 _getlist = staticmethod(getlist)
2257
2258 @staticmethod
2259 def _trygetfunc(tree):
2260 if tree[0] == 'func' and tree[1][0] == 'symbol':
2261 return tree[1][1], getlist(tree[2])
2258 2262
2259 2263 def expandaliases(ui, tree, showwarning=None):
2260 2264 aliases = _aliasrules.buildmap(ui.configitems('revsetalias'))
2261 2265 tree = _aliasrules.expand(aliases, tree)
2262 2266 if showwarning:
2263 2267 # warn about problematic (but not referred) aliases
2264 2268 for name, alias in sorted(aliases.iteritems()):
2265 2269 if alias.error and not alias.warned:
2266 2270 showwarning(_('warning: %s\n') % (alias.error))
2267 2271 alias.warned = True
2268 2272 return tree
2269 2273
2270 2274 def foldconcat(tree):
2271 2275 """Fold elements to be concatenated by `##`
2272 2276 """
2273 2277 if not isinstance(tree, tuple) or tree[0] in ('string', 'symbol'):
2274 2278 return tree
2275 2279 if tree[0] == '_concat':
2276 2280 pending = [tree]
2277 2281 l = []
2278 2282 while pending:
2279 2283 e = pending.pop()
2280 2284 if e[0] == '_concat':
2281 2285 pending.extend(reversed(e[1:]))
2282 2286 elif e[0] in ('string', 'symbol'):
2283 2287 l.append(e[1])
2284 2288 else:
2285 2289 msg = _("\"##\" can't concatenate \"%s\" element") % (e[0])
2286 2290 raise error.ParseError(msg)
2287 2291 return ('string', ''.join(l))
2288 2292 else:
2289 2293 return tuple(foldconcat(t) for t in tree)
2290 2294
2291 2295 def parse(spec, lookup=None):
2292 2296 p = parser.parser(elements)
2293 2297 tree, pos = p.parse(tokenize(spec, lookup=lookup))
2294 2298 if pos != len(spec):
2295 2299 raise error.ParseError(_("invalid token"), pos)
2296 2300 return parser.simplifyinfixops(tree, ('list', 'or'))
2297 2301
2298 2302 def posttreebuilthook(tree, repo):
2299 2303 # hook for extensions to execute code on the optimized tree
2300 2304 pass
2301 2305
2302 2306 def match(ui, spec, repo=None):
2303 2307 if not spec:
2304 2308 raise error.ParseError(_("empty query"))
2305 2309 lookup = None
2306 2310 if repo:
2307 2311 lookup = repo.__contains__
2308 2312 tree = parse(spec, lookup)
2309 2313 return _makematcher(ui, tree, repo)
2310 2314
2311 2315 def matchany(ui, specs, repo=None):
2312 2316 """Create a matcher that will include any revisions matching one of the
2313 2317 given specs"""
2314 2318 if not specs:
2315 2319 def mfunc(repo, subset=None):
2316 2320 return baseset()
2317 2321 return mfunc
2318 2322 if not all(specs):
2319 2323 raise error.ParseError(_("empty query"))
2320 2324 lookup = None
2321 2325 if repo:
2322 2326 lookup = repo.__contains__
2323 2327 if len(specs) == 1:
2324 2328 tree = parse(specs[0], lookup)
2325 2329 else:
2326 2330 tree = ('or',) + tuple(parse(s, lookup) for s in specs)
2327 2331 return _makematcher(ui, tree, repo)
2328 2332
2329 2333 def _makematcher(ui, tree, repo):
2330 2334 if ui:
2331 2335 tree = expandaliases(ui, tree, showwarning=ui.warn)
2332 2336 tree = foldconcat(tree)
2333 2337 weight, tree = optimize(tree, True)
2334 2338 posttreebuilthook(tree, repo)
2335 2339 def mfunc(repo, subset=None):
2336 2340 if subset is None:
2337 2341 subset = fullreposet(repo)
2338 2342 if util.safehasattr(subset, 'isascending'):
2339 2343 result = getset(repo, subset, tree)
2340 2344 else:
2341 2345 result = getset(repo, baseset(subset), tree)
2342 2346 return result
2343 2347 return mfunc
2344 2348
2345 2349 def formatspec(expr, *args):
2346 2350 '''
2347 2351 This is a convenience function for using revsets internally, and
2348 2352 escapes arguments appropriately. Aliases are intentionally ignored
2349 2353 so that intended expression behavior isn't accidentally subverted.
2350 2354
2351 2355 Supported arguments:
2352 2356
2353 2357 %r = revset expression, parenthesized
2354 2358 %d = int(arg), no quoting
2355 2359 %s = string(arg), escaped and single-quoted
2356 2360 %b = arg.branch(), escaped and single-quoted
2357 2361 %n = hex(arg), single-quoted
2358 2362 %% = a literal '%'
2359 2363
2360 2364 Prefixing the type with 'l' specifies a parenthesized list of that type.
2361 2365
2362 2366 >>> formatspec('%r:: and %lr', '10 or 11', ("this()", "that()"))
2363 2367 '(10 or 11):: and ((this()) or (that()))'
2364 2368 >>> formatspec('%d:: and not %d::', 10, 20)
2365 2369 '10:: and not 20::'
2366 2370 >>> formatspec('%ld or %ld', [], [1])
2367 2371 "_list('') or 1"
2368 2372 >>> formatspec('keyword(%s)', 'foo\\xe9')
2369 2373 "keyword('foo\\\\xe9')"
2370 2374 >>> b = lambda: 'default'
2371 2375 >>> b.branch = b
2372 2376 >>> formatspec('branch(%b)', b)
2373 2377 "branch('default')"
2374 2378 >>> formatspec('root(%ls)', ['a', 'b', 'c', 'd'])
2375 2379 "root(_list('a\\x00b\\x00c\\x00d'))"
2376 2380 '''
2377 2381
2378 2382 def quote(s):
2379 2383 return repr(str(s))
2380 2384
2381 2385 def argtype(c, arg):
2382 2386 if c == 'd':
2383 2387 return str(int(arg))
2384 2388 elif c == 's':
2385 2389 return quote(arg)
2386 2390 elif c == 'r':
2387 2391 parse(arg) # make sure syntax errors are confined
2388 2392 return '(%s)' % arg
2389 2393 elif c == 'n':
2390 2394 return quote(node.hex(arg))
2391 2395 elif c == 'b':
2392 2396 return quote(arg.branch())
2393 2397
2394 2398 def listexp(s, t):
2395 2399 l = len(s)
2396 2400 if l == 0:
2397 2401 return "_list('')"
2398 2402 elif l == 1:
2399 2403 return argtype(t, s[0])
2400 2404 elif t == 'd':
2401 2405 return "_intlist('%s')" % "\0".join(str(int(a)) for a in s)
2402 2406 elif t == 's':
2403 2407 return "_list('%s')" % "\0".join(s)
2404 2408 elif t == 'n':
2405 2409 return "_hexlist('%s')" % "\0".join(node.hex(a) for a in s)
2406 2410 elif t == 'b':
2407 2411 return "_list('%s')" % "\0".join(a.branch() for a in s)
2408 2412
2409 2413 m = l // 2
2410 2414 return '(%s or %s)' % (listexp(s[:m], t), listexp(s[m:], t))
2411 2415
2412 2416 ret = ''
2413 2417 pos = 0
2414 2418 arg = 0
2415 2419 while pos < len(expr):
2416 2420 c = expr[pos]
2417 2421 if c == '%':
2418 2422 pos += 1
2419 2423 d = expr[pos]
2420 2424 if d == '%':
2421 2425 ret += d
2422 2426 elif d in 'dsnbr':
2423 2427 ret += argtype(d, args[arg])
2424 2428 arg += 1
2425 2429 elif d == 'l':
2426 2430 # a list of some type
2427 2431 pos += 1
2428 2432 d = expr[pos]
2429 2433 ret += listexp(list(args[arg]), d)
2430 2434 arg += 1
2431 2435 else:
2432 2436 raise error.Abort('unexpected revspec format character %s' % d)
2433 2437 else:
2434 2438 ret += c
2435 2439 pos += 1
2436 2440
2437 2441 return ret
2438 2442
2439 2443 def prettyformat(tree):
2440 2444 return parser.prettyformat(tree, ('string', 'symbol'))
2441 2445
2442 2446 def depth(tree):
2443 2447 if isinstance(tree, tuple):
2444 2448 return max(map(depth, tree)) + 1
2445 2449 else:
2446 2450 return 0
2447 2451
2448 2452 def funcsused(tree):
2449 2453 if not isinstance(tree, tuple) or tree[0] in ('string', 'symbol'):
2450 2454 return set()
2451 2455 else:
2452 2456 funcs = set()
2453 2457 for s in tree[1:]:
2454 2458 funcs |= funcsused(s)
2455 2459 if tree[0] == 'func':
2456 2460 funcs.add(tree[1][1])
2457 2461 return funcs
2458 2462
2459 2463 def _formatsetrepr(r):
2460 2464 """Format an optional printable representation of a set
2461 2465
2462 2466 ======== =================================
2463 2467 type(r) example
2464 2468 ======== =================================
2465 2469 tuple ('<not %r>', other)
2466 2470 str '<branch closed>'
2467 2471 callable lambda: '<branch %r>' % sorted(b)
2468 2472 object other
2469 2473 ======== =================================
2470 2474 """
2471 2475 if r is None:
2472 2476 return ''
2473 2477 elif isinstance(r, tuple):
2474 2478 return r[0] % r[1:]
2475 2479 elif isinstance(r, str):
2476 2480 return r
2477 2481 elif callable(r):
2478 2482 return r()
2479 2483 else:
2480 2484 return repr(r)
2481 2485
2482 2486 class abstractsmartset(object):
2483 2487
2484 2488 def __nonzero__(self):
2485 2489 """True if the smartset is not empty"""
2486 2490 raise NotImplementedError()
2487 2491
2488 2492 def __contains__(self, rev):
2489 2493 """provide fast membership testing"""
2490 2494 raise NotImplementedError()
2491 2495
2492 2496 def __iter__(self):
2493 2497 """iterate the set in the order it is supposed to be iterated"""
2494 2498 raise NotImplementedError()
2495 2499
2496 2500 # Attributes containing a function to perform a fast iteration in a given
2497 2501 # direction. A smartset can have none, one, or both defined.
2498 2502 #
2499 2503 # Default value is None instead of a function returning None to avoid
2500 2504 # initializing an iterator just for testing if a fast method exists.
2501 2505 fastasc = None
2502 2506 fastdesc = None
2503 2507
2504 2508 def isascending(self):
2505 2509 """True if the set will iterate in ascending order"""
2506 2510 raise NotImplementedError()
2507 2511
2508 2512 def isdescending(self):
2509 2513 """True if the set will iterate in descending order"""
2510 2514 raise NotImplementedError()
2511 2515
2512 2516 @util.cachefunc
2513 2517 def min(self):
2514 2518 """return the minimum element in the set"""
2515 2519 if self.fastasc is not None:
2516 2520 for r in self.fastasc():
2517 2521 return r
2518 2522 raise ValueError('arg is an empty sequence')
2519 2523 return min(self)
2520 2524
2521 2525 @util.cachefunc
2522 2526 def max(self):
2523 2527 """return the maximum element in the set"""
2524 2528 if self.fastdesc is not None:
2525 2529 for r in self.fastdesc():
2526 2530 return r
2527 2531 raise ValueError('arg is an empty sequence')
2528 2532 return max(self)
2529 2533
2530 2534 def first(self):
2531 2535 """return the first element in the set (user iteration perspective)
2532 2536
2533 2537 Return None if the set is empty"""
2534 2538 raise NotImplementedError()
2535 2539
2536 2540 def last(self):
2537 2541 """return the last element in the set (user iteration perspective)
2538 2542
2539 2543 Return None if the set is empty"""
2540 2544 raise NotImplementedError()
2541 2545
2542 2546 def __len__(self):
2543 2547 """return the length of the smartsets
2544 2548
2545 2549 This can be expensive on smartset that could be lazy otherwise."""
2546 2550 raise NotImplementedError()
2547 2551
2548 2552 def reverse(self):
2549 2553 """reverse the expected iteration order"""
2550 2554 raise NotImplementedError()
2551 2555
2552 2556 def sort(self, reverse=True):
2553 2557 """get the set to iterate in an ascending or descending order"""
2554 2558 raise NotImplementedError()
2555 2559
2556 2560 def __and__(self, other):
2557 2561 """Returns a new object with the intersection of the two collections.
2558 2562
2559 2563 This is part of the mandatory API for smartset."""
2560 2564 if isinstance(other, fullreposet):
2561 2565 return self
2562 2566 return self.filter(other.__contains__, condrepr=other, cache=False)
2563 2567
2564 2568 def __add__(self, other):
2565 2569 """Returns a new object with the union of the two collections.
2566 2570
2567 2571 This is part of the mandatory API for smartset."""
2568 2572 return addset(self, other)
2569 2573
2570 2574 def __sub__(self, other):
2571 2575 """Returns a new object with the substraction of the two collections.
2572 2576
2573 2577 This is part of the mandatory API for smartset."""
2574 2578 c = other.__contains__
2575 2579 return self.filter(lambda r: not c(r), condrepr=('<not %r>', other),
2576 2580 cache=False)
2577 2581
2578 2582 def filter(self, condition, condrepr=None, cache=True):
2579 2583 """Returns this smartset filtered by condition as a new smartset.
2580 2584
2581 2585 `condition` is a callable which takes a revision number and returns a
2582 2586 boolean. Optional `condrepr` provides a printable representation of
2583 2587 the given `condition`.
2584 2588
2585 2589 This is part of the mandatory API for smartset."""
2586 2590 # builtin cannot be cached. but do not needs to
2587 2591 if cache and util.safehasattr(condition, 'func_code'):
2588 2592 condition = util.cachefunc(condition)
2589 2593 return filteredset(self, condition, condrepr)
2590 2594
2591 2595 class baseset(abstractsmartset):
2592 2596 """Basic data structure that represents a revset and contains the basic
2593 2597 operation that it should be able to perform.
2594 2598
2595 2599 Every method in this class should be implemented by any smartset class.
2596 2600 """
2597 2601 def __init__(self, data=(), datarepr=None):
2598 2602 """
2599 2603 datarepr: a tuple of (format, obj, ...), a function or an object that
2600 2604 provides a printable representation of the given data.
2601 2605 """
2602 2606 self._ascending = None
2603 2607 if not isinstance(data, list):
2604 2608 if isinstance(data, set):
2605 2609 self._set = data
2606 2610 # set has no order we pick one for stability purpose
2607 2611 self._ascending = True
2608 2612 data = list(data)
2609 2613 self._list = data
2610 2614 self._datarepr = datarepr
2611 2615
2612 2616 @util.propertycache
2613 2617 def _set(self):
2614 2618 return set(self._list)
2615 2619
2616 2620 @util.propertycache
2617 2621 def _asclist(self):
2618 2622 asclist = self._list[:]
2619 2623 asclist.sort()
2620 2624 return asclist
2621 2625
2622 2626 def __iter__(self):
2623 2627 if self._ascending is None:
2624 2628 return iter(self._list)
2625 2629 elif self._ascending:
2626 2630 return iter(self._asclist)
2627 2631 else:
2628 2632 return reversed(self._asclist)
2629 2633
2630 2634 def fastasc(self):
2631 2635 return iter(self._asclist)
2632 2636
2633 2637 def fastdesc(self):
2634 2638 return reversed(self._asclist)
2635 2639
2636 2640 @util.propertycache
2637 2641 def __contains__(self):
2638 2642 return self._set.__contains__
2639 2643
2640 2644 def __nonzero__(self):
2641 2645 return bool(self._list)
2642 2646
2643 2647 def sort(self, reverse=False):
2644 2648 self._ascending = not bool(reverse)
2645 2649
2646 2650 def reverse(self):
2647 2651 if self._ascending is None:
2648 2652 self._list.reverse()
2649 2653 else:
2650 2654 self._ascending = not self._ascending
2651 2655
2652 2656 def __len__(self):
2653 2657 return len(self._list)
2654 2658
2655 2659 def isascending(self):
2656 2660 """Returns True if the collection is ascending order, False if not.
2657 2661
2658 2662 This is part of the mandatory API for smartset."""
2659 2663 if len(self) <= 1:
2660 2664 return True
2661 2665 return self._ascending is not None and self._ascending
2662 2666
2663 2667 def isdescending(self):
2664 2668 """Returns True if the collection is descending order, False if not.
2665 2669
2666 2670 This is part of the mandatory API for smartset."""
2667 2671 if len(self) <= 1:
2668 2672 return True
2669 2673 return self._ascending is not None and not self._ascending
2670 2674
2671 2675 def first(self):
2672 2676 if self:
2673 2677 if self._ascending is None:
2674 2678 return self._list[0]
2675 2679 elif self._ascending:
2676 2680 return self._asclist[0]
2677 2681 else:
2678 2682 return self._asclist[-1]
2679 2683 return None
2680 2684
2681 2685 def last(self):
2682 2686 if self:
2683 2687 if self._ascending is None:
2684 2688 return self._list[-1]
2685 2689 elif self._ascending:
2686 2690 return self._asclist[-1]
2687 2691 else:
2688 2692 return self._asclist[0]
2689 2693 return None
2690 2694
2691 2695 def __repr__(self):
2692 2696 d = {None: '', False: '-', True: '+'}[self._ascending]
2693 2697 s = _formatsetrepr(self._datarepr)
2694 2698 if not s:
2695 2699 l = self._list
2696 2700 # if _list has been built from a set, it might have a different
2697 2701 # order from one python implementation to another.
2698 2702 # We fallback to the sorted version for a stable output.
2699 2703 if self._ascending is not None:
2700 2704 l = self._asclist
2701 2705 s = repr(l)
2702 2706 return '<%s%s %s>' % (type(self).__name__, d, s)
2703 2707
2704 2708 class filteredset(abstractsmartset):
2705 2709 """Duck type for baseset class which iterates lazily over the revisions in
2706 2710 the subset and contains a function which tests for membership in the
2707 2711 revset
2708 2712 """
2709 2713 def __init__(self, subset, condition=lambda x: True, condrepr=None):
2710 2714 """
2711 2715 condition: a function that decide whether a revision in the subset
2712 2716 belongs to the revset or not.
2713 2717 condrepr: a tuple of (format, obj, ...), a function or an object that
2714 2718 provides a printable representation of the given condition.
2715 2719 """
2716 2720 self._subset = subset
2717 2721 self._condition = condition
2718 2722 self._condrepr = condrepr
2719 2723
2720 2724 def __contains__(self, x):
2721 2725 return x in self._subset and self._condition(x)
2722 2726
2723 2727 def __iter__(self):
2724 2728 return self._iterfilter(self._subset)
2725 2729
2726 2730 def _iterfilter(self, it):
2727 2731 cond = self._condition
2728 2732 for x in it:
2729 2733 if cond(x):
2730 2734 yield x
2731 2735
2732 2736 @property
2733 2737 def fastasc(self):
2734 2738 it = self._subset.fastasc
2735 2739 if it is None:
2736 2740 return None
2737 2741 return lambda: self._iterfilter(it())
2738 2742
2739 2743 @property
2740 2744 def fastdesc(self):
2741 2745 it = self._subset.fastdesc
2742 2746 if it is None:
2743 2747 return None
2744 2748 return lambda: self._iterfilter(it())
2745 2749
2746 2750 def __nonzero__(self):
2747 2751 fast = self.fastasc
2748 2752 if fast is None:
2749 2753 fast = self.fastdesc
2750 2754 if fast is not None:
2751 2755 it = fast()
2752 2756 else:
2753 2757 it = self
2754 2758
2755 2759 for r in it:
2756 2760 return True
2757 2761 return False
2758 2762
2759 2763 def __len__(self):
2760 2764 # Basic implementation to be changed in future patches.
2761 2765 # until this gets improved, we use generator expression
2762 2766 # here, since list compr is free to call __len__ again
2763 2767 # causing infinite recursion
2764 2768 l = baseset(r for r in self)
2765 2769 return len(l)
2766 2770
2767 2771 def sort(self, reverse=False):
2768 2772 self._subset.sort(reverse=reverse)
2769 2773
2770 2774 def reverse(self):
2771 2775 self._subset.reverse()
2772 2776
2773 2777 def isascending(self):
2774 2778 return self._subset.isascending()
2775 2779
2776 2780 def isdescending(self):
2777 2781 return self._subset.isdescending()
2778 2782
2779 2783 def first(self):
2780 2784 for x in self:
2781 2785 return x
2782 2786 return None
2783 2787
2784 2788 def last(self):
2785 2789 it = None
2786 2790 if self.isascending():
2787 2791 it = self.fastdesc
2788 2792 elif self.isdescending():
2789 2793 it = self.fastasc
2790 2794 if it is not None:
2791 2795 for x in it():
2792 2796 return x
2793 2797 return None #empty case
2794 2798 else:
2795 2799 x = None
2796 2800 for x in self:
2797 2801 pass
2798 2802 return x
2799 2803
2800 2804 def __repr__(self):
2801 2805 xs = [repr(self._subset)]
2802 2806 s = _formatsetrepr(self._condrepr)
2803 2807 if s:
2804 2808 xs.append(s)
2805 2809 return '<%s %s>' % (type(self).__name__, ', '.join(xs))
2806 2810
2807 2811 def _iterordered(ascending, iter1, iter2):
2808 2812 """produce an ordered iteration from two iterators with the same order
2809 2813
2810 2814 The ascending is used to indicated the iteration direction.
2811 2815 """
2812 2816 choice = max
2813 2817 if ascending:
2814 2818 choice = min
2815 2819
2816 2820 val1 = None
2817 2821 val2 = None
2818 2822 try:
2819 2823 # Consume both iterators in an ordered way until one is empty
2820 2824 while True:
2821 2825 if val1 is None:
2822 2826 val1 = iter1.next()
2823 2827 if val2 is None:
2824 2828 val2 = iter2.next()
2825 2829 next = choice(val1, val2)
2826 2830 yield next
2827 2831 if val1 == next:
2828 2832 val1 = None
2829 2833 if val2 == next:
2830 2834 val2 = None
2831 2835 except StopIteration:
2832 2836 # Flush any remaining values and consume the other one
2833 2837 it = iter2
2834 2838 if val1 is not None:
2835 2839 yield val1
2836 2840 it = iter1
2837 2841 elif val2 is not None:
2838 2842 # might have been equality and both are empty
2839 2843 yield val2
2840 2844 for val in it:
2841 2845 yield val
2842 2846
2843 2847 class addset(abstractsmartset):
2844 2848 """Represent the addition of two sets
2845 2849
2846 2850 Wrapper structure for lazily adding two structures without losing much
2847 2851 performance on the __contains__ method
2848 2852
2849 2853 If the ascending attribute is set, that means the two structures are
2850 2854 ordered in either an ascending or descending way. Therefore, we can add
2851 2855 them maintaining the order by iterating over both at the same time
2852 2856
2853 2857 >>> xs = baseset([0, 3, 2])
2854 2858 >>> ys = baseset([5, 2, 4])
2855 2859
2856 2860 >>> rs = addset(xs, ys)
2857 2861 >>> bool(rs), 0 in rs, 1 in rs, 5 in rs, rs.first(), rs.last()
2858 2862 (True, True, False, True, 0, 4)
2859 2863 >>> rs = addset(xs, baseset([]))
2860 2864 >>> bool(rs), 0 in rs, 1 in rs, rs.first(), rs.last()
2861 2865 (True, True, False, 0, 2)
2862 2866 >>> rs = addset(baseset([]), baseset([]))
2863 2867 >>> bool(rs), 0 in rs, rs.first(), rs.last()
2864 2868 (False, False, None, None)
2865 2869
2866 2870 iterate unsorted:
2867 2871 >>> rs = addset(xs, ys)
2868 2872 >>> # (use generator because pypy could call len())
2869 2873 >>> list(x for x in rs) # without _genlist
2870 2874 [0, 3, 2, 5, 4]
2871 2875 >>> assert not rs._genlist
2872 2876 >>> len(rs)
2873 2877 5
2874 2878 >>> [x for x in rs] # with _genlist
2875 2879 [0, 3, 2, 5, 4]
2876 2880 >>> assert rs._genlist
2877 2881
2878 2882 iterate ascending:
2879 2883 >>> rs = addset(xs, ys, ascending=True)
2880 2884 >>> # (use generator because pypy could call len())
2881 2885 >>> list(x for x in rs), list(x for x in rs.fastasc()) # without _asclist
2882 2886 ([0, 2, 3, 4, 5], [0, 2, 3, 4, 5])
2883 2887 >>> assert not rs._asclist
2884 2888 >>> len(rs)
2885 2889 5
2886 2890 >>> [x for x in rs], [x for x in rs.fastasc()]
2887 2891 ([0, 2, 3, 4, 5], [0, 2, 3, 4, 5])
2888 2892 >>> assert rs._asclist
2889 2893
2890 2894 iterate descending:
2891 2895 >>> rs = addset(xs, ys, ascending=False)
2892 2896 >>> # (use generator because pypy could call len())
2893 2897 >>> list(x for x in rs), list(x for x in rs.fastdesc()) # without _asclist
2894 2898 ([5, 4, 3, 2, 0], [5, 4, 3, 2, 0])
2895 2899 >>> assert not rs._asclist
2896 2900 >>> len(rs)
2897 2901 5
2898 2902 >>> [x for x in rs], [x for x in rs.fastdesc()]
2899 2903 ([5, 4, 3, 2, 0], [5, 4, 3, 2, 0])
2900 2904 >>> assert rs._asclist
2901 2905
2902 2906 iterate ascending without fastasc:
2903 2907 >>> rs = addset(xs, generatorset(ys), ascending=True)
2904 2908 >>> assert rs.fastasc is None
2905 2909 >>> [x for x in rs]
2906 2910 [0, 2, 3, 4, 5]
2907 2911
2908 2912 iterate descending without fastdesc:
2909 2913 >>> rs = addset(generatorset(xs), ys, ascending=False)
2910 2914 >>> assert rs.fastdesc is None
2911 2915 >>> [x for x in rs]
2912 2916 [5, 4, 3, 2, 0]
2913 2917 """
2914 2918 def __init__(self, revs1, revs2, ascending=None):
2915 2919 self._r1 = revs1
2916 2920 self._r2 = revs2
2917 2921 self._iter = None
2918 2922 self._ascending = ascending
2919 2923 self._genlist = None
2920 2924 self._asclist = None
2921 2925
2922 2926 def __len__(self):
2923 2927 return len(self._list)
2924 2928
2925 2929 def __nonzero__(self):
2926 2930 return bool(self._r1) or bool(self._r2)
2927 2931
2928 2932 @util.propertycache
2929 2933 def _list(self):
2930 2934 if not self._genlist:
2931 2935 self._genlist = baseset(iter(self))
2932 2936 return self._genlist
2933 2937
2934 2938 def __iter__(self):
2935 2939 """Iterate over both collections without repeating elements
2936 2940
2937 2941 If the ascending attribute is not set, iterate over the first one and
2938 2942 then over the second one checking for membership on the first one so we
2939 2943 dont yield any duplicates.
2940 2944
2941 2945 If the ascending attribute is set, iterate over both collections at the
2942 2946 same time, yielding only one value at a time in the given order.
2943 2947 """
2944 2948 if self._ascending is None:
2945 2949 if self._genlist:
2946 2950 return iter(self._genlist)
2947 2951 def arbitraryordergen():
2948 2952 for r in self._r1:
2949 2953 yield r
2950 2954 inr1 = self._r1.__contains__
2951 2955 for r in self._r2:
2952 2956 if not inr1(r):
2953 2957 yield r
2954 2958 return arbitraryordergen()
2955 2959 # try to use our own fast iterator if it exists
2956 2960 self._trysetasclist()
2957 2961 if self._ascending:
2958 2962 attr = 'fastasc'
2959 2963 else:
2960 2964 attr = 'fastdesc'
2961 2965 it = getattr(self, attr)
2962 2966 if it is not None:
2963 2967 return it()
2964 2968 # maybe half of the component supports fast
2965 2969 # get iterator for _r1
2966 2970 iter1 = getattr(self._r1, attr)
2967 2971 if iter1 is None:
2968 2972 # let's avoid side effect (not sure it matters)
2969 2973 iter1 = iter(sorted(self._r1, reverse=not self._ascending))
2970 2974 else:
2971 2975 iter1 = iter1()
2972 2976 # get iterator for _r2
2973 2977 iter2 = getattr(self._r2, attr)
2974 2978 if iter2 is None:
2975 2979 # let's avoid side effect (not sure it matters)
2976 2980 iter2 = iter(sorted(self._r2, reverse=not self._ascending))
2977 2981 else:
2978 2982 iter2 = iter2()
2979 2983 return _iterordered(self._ascending, iter1, iter2)
2980 2984
2981 2985 def _trysetasclist(self):
2982 2986 """populate the _asclist attribute if possible and necessary"""
2983 2987 if self._genlist is not None and self._asclist is None:
2984 2988 self._asclist = sorted(self._genlist)
2985 2989
2986 2990 @property
2987 2991 def fastasc(self):
2988 2992 self._trysetasclist()
2989 2993 if self._asclist is not None:
2990 2994 return self._asclist.__iter__
2991 2995 iter1 = self._r1.fastasc
2992 2996 iter2 = self._r2.fastasc
2993 2997 if None in (iter1, iter2):
2994 2998 return None
2995 2999 return lambda: _iterordered(True, iter1(), iter2())
2996 3000
2997 3001 @property
2998 3002 def fastdesc(self):
2999 3003 self._trysetasclist()
3000 3004 if self._asclist is not None:
3001 3005 return self._asclist.__reversed__
3002 3006 iter1 = self._r1.fastdesc
3003 3007 iter2 = self._r2.fastdesc
3004 3008 if None in (iter1, iter2):
3005 3009 return None
3006 3010 return lambda: _iterordered(False, iter1(), iter2())
3007 3011
3008 3012 def __contains__(self, x):
3009 3013 return x in self._r1 or x in self._r2
3010 3014
3011 3015 def sort(self, reverse=False):
3012 3016 """Sort the added set
3013 3017
3014 3018 For this we use the cached list with all the generated values and if we
3015 3019 know they are ascending or descending we can sort them in a smart way.
3016 3020 """
3017 3021 self._ascending = not reverse
3018 3022
3019 3023 def isascending(self):
3020 3024 return self._ascending is not None and self._ascending
3021 3025
3022 3026 def isdescending(self):
3023 3027 return self._ascending is not None and not self._ascending
3024 3028
3025 3029 def reverse(self):
3026 3030 if self._ascending is None:
3027 3031 self._list.reverse()
3028 3032 else:
3029 3033 self._ascending = not self._ascending
3030 3034
3031 3035 def first(self):
3032 3036 for x in self:
3033 3037 return x
3034 3038 return None
3035 3039
3036 3040 def last(self):
3037 3041 self.reverse()
3038 3042 val = self.first()
3039 3043 self.reverse()
3040 3044 return val
3041 3045
3042 3046 def __repr__(self):
3043 3047 d = {None: '', False: '-', True: '+'}[self._ascending]
3044 3048 return '<%s%s %r, %r>' % (type(self).__name__, d, self._r1, self._r2)
3045 3049
3046 3050 class generatorset(abstractsmartset):
3047 3051 """Wrap a generator for lazy iteration
3048 3052
3049 3053 Wrapper structure for generators that provides lazy membership and can
3050 3054 be iterated more than once.
3051 3055 When asked for membership it generates values until either it finds the
3052 3056 requested one or has gone through all the elements in the generator
3053 3057 """
3054 3058 def __init__(self, gen, iterasc=None):
3055 3059 """
3056 3060 gen: a generator producing the values for the generatorset.
3057 3061 """
3058 3062 self._gen = gen
3059 3063 self._asclist = None
3060 3064 self._cache = {}
3061 3065 self._genlist = []
3062 3066 self._finished = False
3063 3067 self._ascending = True
3064 3068 if iterasc is not None:
3065 3069 if iterasc:
3066 3070 self.fastasc = self._iterator
3067 3071 self.__contains__ = self._asccontains
3068 3072 else:
3069 3073 self.fastdesc = self._iterator
3070 3074 self.__contains__ = self._desccontains
3071 3075
3072 3076 def __nonzero__(self):
3073 3077 # Do not use 'for r in self' because it will enforce the iteration
3074 3078 # order (default ascending), possibly unrolling a whole descending
3075 3079 # iterator.
3076 3080 if self._genlist:
3077 3081 return True
3078 3082 for r in self._consumegen():
3079 3083 return True
3080 3084 return False
3081 3085
3082 3086 def __contains__(self, x):
3083 3087 if x in self._cache:
3084 3088 return self._cache[x]
3085 3089
3086 3090 # Use new values only, as existing values would be cached.
3087 3091 for l in self._consumegen():
3088 3092 if l == x:
3089 3093 return True
3090 3094
3091 3095 self._cache[x] = False
3092 3096 return False
3093 3097
3094 3098 def _asccontains(self, x):
3095 3099 """version of contains optimised for ascending generator"""
3096 3100 if x in self._cache:
3097 3101 return self._cache[x]
3098 3102
3099 3103 # Use new values only, as existing values would be cached.
3100 3104 for l in self._consumegen():
3101 3105 if l == x:
3102 3106 return True
3103 3107 if l > x:
3104 3108 break
3105 3109
3106 3110 self._cache[x] = False
3107 3111 return False
3108 3112
3109 3113 def _desccontains(self, x):
3110 3114 """version of contains optimised for descending generator"""
3111 3115 if x in self._cache:
3112 3116 return self._cache[x]
3113 3117
3114 3118 # Use new values only, as existing values would be cached.
3115 3119 for l in self._consumegen():
3116 3120 if l == x:
3117 3121 return True
3118 3122 if l < x:
3119 3123 break
3120 3124
3121 3125 self._cache[x] = False
3122 3126 return False
3123 3127
3124 3128 def __iter__(self):
3125 3129 if self._ascending:
3126 3130 it = self.fastasc
3127 3131 else:
3128 3132 it = self.fastdesc
3129 3133 if it is not None:
3130 3134 return it()
3131 3135 # we need to consume the iterator
3132 3136 for x in self._consumegen():
3133 3137 pass
3134 3138 # recall the same code
3135 3139 return iter(self)
3136 3140
3137 3141 def _iterator(self):
3138 3142 if self._finished:
3139 3143 return iter(self._genlist)
3140 3144
3141 3145 # We have to use this complex iteration strategy to allow multiple
3142 3146 # iterations at the same time. We need to be able to catch revision
3143 3147 # removed from _consumegen and added to genlist in another instance.
3144 3148 #
3145 3149 # Getting rid of it would provide an about 15% speed up on this
3146 3150 # iteration.
3147 3151 genlist = self._genlist
3148 3152 nextrev = self._consumegen().next
3149 3153 _len = len # cache global lookup
3150 3154 def gen():
3151 3155 i = 0
3152 3156 while True:
3153 3157 if i < _len(genlist):
3154 3158 yield genlist[i]
3155 3159 else:
3156 3160 yield nextrev()
3157 3161 i += 1
3158 3162 return gen()
3159 3163
3160 3164 def _consumegen(self):
3161 3165 cache = self._cache
3162 3166 genlist = self._genlist.append
3163 3167 for item in self._gen:
3164 3168 cache[item] = True
3165 3169 genlist(item)
3166 3170 yield item
3167 3171 if not self._finished:
3168 3172 self._finished = True
3169 3173 asc = self._genlist[:]
3170 3174 asc.sort()
3171 3175 self._asclist = asc
3172 3176 self.fastasc = asc.__iter__
3173 3177 self.fastdesc = asc.__reversed__
3174 3178
3175 3179 def __len__(self):
3176 3180 for x in self._consumegen():
3177 3181 pass
3178 3182 return len(self._genlist)
3179 3183
3180 3184 def sort(self, reverse=False):
3181 3185 self._ascending = not reverse
3182 3186
3183 3187 def reverse(self):
3184 3188 self._ascending = not self._ascending
3185 3189
3186 3190 def isascending(self):
3187 3191 return self._ascending
3188 3192
3189 3193 def isdescending(self):
3190 3194 return not self._ascending
3191 3195
3192 3196 def first(self):
3193 3197 if self._ascending:
3194 3198 it = self.fastasc
3195 3199 else:
3196 3200 it = self.fastdesc
3197 3201 if it is None:
3198 3202 # we need to consume all and try again
3199 3203 for x in self._consumegen():
3200 3204 pass
3201 3205 return self.first()
3202 3206 return next(it(), None)
3203 3207
3204 3208 def last(self):
3205 3209 if self._ascending:
3206 3210 it = self.fastdesc
3207 3211 else:
3208 3212 it = self.fastasc
3209 3213 if it is None:
3210 3214 # we need to consume all and try again
3211 3215 for x in self._consumegen():
3212 3216 pass
3213 3217 return self.first()
3214 3218 return next(it(), None)
3215 3219
3216 3220 def __repr__(self):
3217 3221 d = {False: '-', True: '+'}[self._ascending]
3218 3222 return '<%s%s>' % (type(self).__name__, d)
3219 3223
3220 3224 class spanset(abstractsmartset):
3221 3225 """Duck type for baseset class which represents a range of revisions and
3222 3226 can work lazily and without having all the range in memory
3223 3227
3224 3228 Note that spanset(x, y) behave almost like xrange(x, y) except for two
3225 3229 notable points:
3226 3230 - when x < y it will be automatically descending,
3227 3231 - revision filtered with this repoview will be skipped.
3228 3232
3229 3233 """
3230 3234 def __init__(self, repo, start=0, end=None):
3231 3235 """
3232 3236 start: first revision included the set
3233 3237 (default to 0)
3234 3238 end: first revision excluded (last+1)
3235 3239 (default to len(repo)
3236 3240
3237 3241 Spanset will be descending if `end` < `start`.
3238 3242 """
3239 3243 if end is None:
3240 3244 end = len(repo)
3241 3245 self._ascending = start <= end
3242 3246 if not self._ascending:
3243 3247 start, end = end + 1, start +1
3244 3248 self._start = start
3245 3249 self._end = end
3246 3250 self._hiddenrevs = repo.changelog.filteredrevs
3247 3251
3248 3252 def sort(self, reverse=False):
3249 3253 self._ascending = not reverse
3250 3254
3251 3255 def reverse(self):
3252 3256 self._ascending = not self._ascending
3253 3257
3254 3258 def _iterfilter(self, iterrange):
3255 3259 s = self._hiddenrevs
3256 3260 for r in iterrange:
3257 3261 if r not in s:
3258 3262 yield r
3259 3263
3260 3264 def __iter__(self):
3261 3265 if self._ascending:
3262 3266 return self.fastasc()
3263 3267 else:
3264 3268 return self.fastdesc()
3265 3269
3266 3270 def fastasc(self):
3267 3271 iterrange = xrange(self._start, self._end)
3268 3272 if self._hiddenrevs:
3269 3273 return self._iterfilter(iterrange)
3270 3274 return iter(iterrange)
3271 3275
3272 3276 def fastdesc(self):
3273 3277 iterrange = xrange(self._end - 1, self._start - 1, -1)
3274 3278 if self._hiddenrevs:
3275 3279 return self._iterfilter(iterrange)
3276 3280 return iter(iterrange)
3277 3281
3278 3282 def __contains__(self, rev):
3279 3283 hidden = self._hiddenrevs
3280 3284 return ((self._start <= rev < self._end)
3281 3285 and not (hidden and rev in hidden))
3282 3286
3283 3287 def __nonzero__(self):
3284 3288 for r in self:
3285 3289 return True
3286 3290 return False
3287 3291
3288 3292 def __len__(self):
3289 3293 if not self._hiddenrevs:
3290 3294 return abs(self._end - self._start)
3291 3295 else:
3292 3296 count = 0
3293 3297 start = self._start
3294 3298 end = self._end
3295 3299 for rev in self._hiddenrevs:
3296 3300 if (end < rev <= start) or (start <= rev < end):
3297 3301 count += 1
3298 3302 return abs(self._end - self._start) - count
3299 3303
3300 3304 def isascending(self):
3301 3305 return self._ascending
3302 3306
3303 3307 def isdescending(self):
3304 3308 return not self._ascending
3305 3309
3306 3310 def first(self):
3307 3311 if self._ascending:
3308 3312 it = self.fastasc
3309 3313 else:
3310 3314 it = self.fastdesc
3311 3315 for x in it():
3312 3316 return x
3313 3317 return None
3314 3318
3315 3319 def last(self):
3316 3320 if self._ascending:
3317 3321 it = self.fastdesc
3318 3322 else:
3319 3323 it = self.fastasc
3320 3324 for x in it():
3321 3325 return x
3322 3326 return None
3323 3327
3324 3328 def __repr__(self):
3325 3329 d = {False: '-', True: '+'}[self._ascending]
3326 3330 return '<%s%s %d:%d>' % (type(self).__name__, d,
3327 3331 self._start, self._end - 1)
3328 3332
3329 3333 class fullreposet(spanset):
3330 3334 """a set containing all revisions in the repo
3331 3335
3332 3336 This class exists to host special optimization and magic to handle virtual
3333 3337 revisions such as "null".
3334 3338 """
3335 3339
3336 3340 def __init__(self, repo):
3337 3341 super(fullreposet, self).__init__(repo)
3338 3342
3339 3343 def __and__(self, other):
3340 3344 """As self contains the whole repo, all of the other set should also be
3341 3345 in self. Therefore `self & other = other`.
3342 3346
3343 3347 This boldly assumes the other contains valid revs only.
3344 3348 """
3345 3349 # other not a smartset, make is so
3346 3350 if not util.safehasattr(other, 'isascending'):
3347 3351 # filter out hidden revision
3348 3352 # (this boldly assumes all smartset are pure)
3349 3353 #
3350 3354 # `other` was used with "&", let's assume this is a set like
3351 3355 # object.
3352 3356 other = baseset(other - self._hiddenrevs)
3353 3357
3354 3358 # XXX As fullreposet is also used as bootstrap, this is wrong.
3355 3359 #
3356 3360 # With a giveme312() revset returning [3,1,2], this makes
3357 3361 # 'hg log -r "giveme312()"' -> 1, 2, 3 (wrong)
3358 3362 # We cannot just drop it because other usage still need to sort it:
3359 3363 # 'hg log -r "all() and giveme312()"' -> 1, 2, 3 (right)
3360 3364 #
3361 3365 # There is also some faulty revset implementations that rely on it
3362 3366 # (eg: children as of its state in e8075329c5fb)
3363 3367 #
3364 3368 # When we fix the two points above we can move this into the if clause
3365 3369 other.sort(reverse=self.isdescending())
3366 3370 return other
3367 3371
3368 3372 def prettyformatset(revs):
3369 3373 lines = []
3370 3374 rs = repr(revs)
3371 3375 p = 0
3372 3376 while p < len(rs):
3373 3377 q = rs.find('<', p + 1)
3374 3378 if q < 0:
3375 3379 q = len(rs)
3376 3380 l = rs.count('<', 0, p) - rs.count('>', 0, p)
3377 3381 assert l >= 0
3378 3382 lines.append((l, rs[p:q].rstrip()))
3379 3383 p = q
3380 3384 return '\n'.join(' ' * l + s for l, s in lines)
3381 3385
3382 3386 def loadpredicate(ui, extname, registrarobj):
3383 3387 """Load revset predicates from specified registrarobj
3384 3388 """
3385 3389 for name, func in registrarobj._table.iteritems():
3386 3390 symbols[name] = func
3387 3391 if func._safe:
3388 3392 safesymbols.add(name)
3389 3393
3390 3394 # load built-in predicates explicitly to setup safesymbols
3391 3395 loadpredicate(None, None, predicate)
3392 3396
3393 3397 # tell hggettext to extract docstrings from these functions:
3394 3398 i18nfunctions = symbols.values()
General Comments 0
You need to be logged in to leave comments. Login now