mercurial-scm/hg: mercurial/match.py annotate

annotate mercurial/match.py @ 8586:347fe1ac4f21

match: add exact flag to match() to unify all match forms

author	Matt Mackall <mpm@selenic.com>
date	Sun, 24 May 2009 02:56:20 -0500
parents	bbcd0da50e96
children	8f15d54437b9

rev	line source
8231 5d4d88a4f5e6 match: add copyright and license header Martin Geisler <mg@lazybytes.net> parents: 8152 diff changeset	1 # match.py - file name matching
5d4d88a4f5e6 match: add copyright and license header Martin Geisler <mg@lazybytes.net> parents: 8152 diff changeset	2 #
5d4d88a4f5e6 match: add copyright and license header Martin Geisler <mg@lazybytes.net> parents: 8152 diff changeset	3 # Copyright 2008, 2009 Matt Mackall <mpm@selenic.com> and others
5d4d88a4f5e6 match: add copyright and license header Martin Geisler <mg@lazybytes.net> parents: 8152 diff changeset	4 #
5d4d88a4f5e6 match: add copyright and license header Martin Geisler <mg@lazybytes.net> parents: 8152 diff changeset	5 # This software may be used and distributed according to the terms of the
5d4d88a4f5e6 match: add copyright and license header Martin Geisler <mg@lazybytes.net> parents: 8152 diff changeset	6 # GNU General Public License version 2, incorporated herein by reference.
5d4d88a4f5e6 match: add copyright and license header Martin Geisler <mg@lazybytes.net> parents: 8152 diff changeset	7
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	8 import util, re
6576 69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	9
6604 98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	10 class _match(object):
98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	11 def __init__(self, root, cwd, files, mf, ap):
6576 69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	12 self._root = root
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	13 self._cwd = cwd
6604 98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	14 self._files = files
8152 08e1baf924ca replace set-like dictionaries with real sets Martin Geisler <mg@lazybytes.net> parents: 6834 diff changeset	15 self._fmap = set(files)
6834 cbdfd08eabc9 dirstate.walk: speed up calling match function Matt Mackall <mpm@selenic.com> parents: 6604 diff changeset	16 self.matchfn = mf
6576 69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	17 self._anypats = ap
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	18 def __call__(self, fn):
6834 cbdfd08eabc9 dirstate.walk: speed up calling match function Matt Mackall <mpm@selenic.com> parents: 6604 diff changeset	19 return self.matchfn(fn)
6576 69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	20 def __iter__(self):
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	21 for f in self._files:
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	22 yield f
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	23 def bad(self, f, msg):
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	24 return True
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	25 def dir(self, f):
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	26 pass
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	27 def missing(self, f):
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	28 pass
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	29 def exact(self, f):
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	30 return f in self._fmap
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	31 def rel(self, f):
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	32 return util.pathto(self._root, self._cwd, f)
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	33 def files(self):
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	34 return self._files
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	35 def anypats(self):
69f3e9ac7c56 walk: introduce match objects Matt Mackall <mpm@selenic.com> parents: diff changeset	36 return self._anypats
6596 7fe4610cf920 match: add always, never, and exact methods Matt Mackall <mpm@selenic.com> parents: 6578 diff changeset	37
6604 98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	38 class match(_match):
8567 fea40a677d43 match: add some default args Matt Mackall <mpm@selenic.com> parents: 8566 diff changeset	39 def __init__(self, root, cwd, patterns, include=[], exclude=[],
8586 347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	40 default='glob', exact=False):
8581 101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	41 """build an object to match a set of file patterns
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	42
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	43 arguments:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	44 root - the canonical root of the tree you're matching against
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	45 cwd - the current working directory, if relevant
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	46 patterns - patterns to find
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	47 include - patterns to include
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	48 exclude - patterns to exclude
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	49 default - if a pattern in names has no explicit type, assume this one
8586 347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	50 exact - patterns are actually literals
8581 101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	51
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	52 a pattern is one of:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	53 'glob:<glob>' - a glob relative to cwd
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	54 're:<regexp>' - a regular expression
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	55 'path:<path>' - a path relative to canonroot
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	56 'relglob:<glob>' - an unrooted glob (*.c matches C files in all dirs)
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	57 'relpath:<path>' - a path relative to cwd
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	58 'relre:<regexp>' - a regexp that doesn't have to match the start of a name
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	59 '<something>' - one of the cases above, selected by the dflt_pat argument
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	60 """
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	61
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	62 roots = []
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	63 anypats = bool(include or exclude)
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	64
8586 347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	65 if include:
347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	66 im = _buildmatch(_normalize(include, 'glob', root, cwd), '(?:/\|$)')
347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	67 if exclude:
347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	68 em = _buildmatch(_normalize(exclude, 'glob', root, cwd), '(?:/\|$)')
347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	69 if exact:
347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	70 roots = patterns
347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	71 pm = self.exact
347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	72 elif patterns:
8581 101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	73 pats = _normalize(patterns, default, root, cwd)
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	74 roots = _roots(pats)
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	75 anypats = anypats or _anypats(pats)
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	76 pm = _buildmatch(pats, '$')
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	77
8586 347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	78 if patterns or exact:
8581 101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	79 if include:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	80 if exclude:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	81 m = lambda f: im(f) and not em(f) and pm(f)
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	82 else:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	83 m = lambda f: im(f) and pm(f)
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	84 else:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	85 if exclude:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	86 m = lambda f: not em(f) and pm(f)
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	87 else:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	88 m = pm
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	89 else:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	90 if include:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	91 if exclude:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	92 m = lambda f: im(f) and not em(f)
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	93 else:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	94 m = im
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	95 else:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	96 if exclude:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	97 m = lambda f: not em(f)
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	98 else:
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	99 m = lambda f: True
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	100
101d305c1d0b match: fold _matcher into match.__init__ Matt Mackall <mpm@selenic.com> parents: 8580 diff changeset	101 _match.__init__(self, root, cwd, roots, m, anypats)
8568 4fa1618bf495 match: refactor patkind Matt Mackall <mpm@selenic.com> parents: 8567 diff changeset	102
8586 347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	103 class exact(match):
8585 bbcd0da50e96 match: redefine always and never in terms of match and exact Matt Mackall <mpm@selenic.com> parents: 8584 diff changeset	104 def __init__(self, root, cwd, files):
8586 347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	105 match.__init__(self, root, cwd, files, exact = True)
8585 bbcd0da50e96 match: redefine always and never in terms of match and exact Matt Mackall <mpm@selenic.com> parents: 8584 diff changeset	106
bbcd0da50e96 match: redefine always and never in terms of match and exact Matt Mackall <mpm@selenic.com> parents: 8584 diff changeset	107 class always(match):
bbcd0da50e96 match: redefine always and never in terms of match and exact Matt Mackall <mpm@selenic.com> parents: 8584 diff changeset	108 def __init__(self, root, cwd):
bbcd0da50e96 match: redefine always and never in terms of match and exact Matt Mackall <mpm@selenic.com> parents: 8584 diff changeset	109 match.__init__(self, root, cwd, [])
bbcd0da50e96 match: redefine always and never in terms of match and exact Matt Mackall <mpm@selenic.com> parents: 8584 diff changeset	110
8586 347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	111 class never(match):
8585 bbcd0da50e96 match: redefine always and never in terms of match and exact Matt Mackall <mpm@selenic.com> parents: 8584 diff changeset	112 def __init__(self, root, cwd):
8586 347fe1ac4f21 match: add exact flag to match() to unify all match forms Matt Mackall <mpm@selenic.com> parents: 8585 diff changeset	113 match.__init__(self, root, cwd, [], exact = True)
8585 bbcd0da50e96 match: redefine always and never in terms of match and exact Matt Mackall <mpm@selenic.com> parents: 8584 diff changeset	114
8568 4fa1618bf495 match: refactor patkind Matt Mackall <mpm@selenic.com> parents: 8567 diff changeset	115 def patkind(pat):
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	116 return _patsplit(pat, None)[0]
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	117
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	118 def _patsplit(pat, default):
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	119 """Split a string into an optional pattern kind prefix and the
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	120 actual pattern."""
8579 aff7f83c365b match: optimize _patsplit Matt Mackall <mpm@selenic.com> parents: 8578 diff changeset	121 if ':' in pat:
aff7f83c365b match: optimize _patsplit Matt Mackall <mpm@selenic.com> parents: 8578 diff changeset	122 pat, val = pat.split(':', 1)
aff7f83c365b match: optimize _patsplit Matt Mackall <mpm@selenic.com> parents: 8578 diff changeset	123 if pat in ('re', 'glob', 'path', 'relglob', 'relpath', 'relre'):
aff7f83c365b match: optimize _patsplit Matt Mackall <mpm@selenic.com> parents: 8578 diff changeset	124 return pat, val
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	125 return default, pat
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	126
8582 a4c199e12b5a match: remove head and tail args from _globre Matt Mackall <mpm@selenic.com> parents: 8581 diff changeset	127 def _globre(pat):
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	128 "convert a glob pattern into a regexp"
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	129 i, n = 0, len(pat)
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	130 res = ''
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	131 group = 0
8583 19d1b2aec562 match: optimize escaping in _globre Matt Mackall <mpm@selenic.com> parents: 8582 diff changeset	132 escape = re.escape
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	133 def peek(): return i < n and pat[i]
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	134 while i < n:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	135 c = pat[i]
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	136 i = i+1
8583 19d1b2aec562 match: optimize escaping in _globre Matt Mackall <mpm@selenic.com> parents: 8582 diff changeset	137 if c not in '*?[{},\\':
19d1b2aec562 match: optimize escaping in _globre Matt Mackall <mpm@selenic.com> parents: 8582 diff changeset	138 res += escape(c)
19d1b2aec562 match: optimize escaping in _globre Matt Mackall <mpm@selenic.com> parents: 8582 diff changeset	139 elif c == '*':
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	140 if peek() == '*':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	141 i += 1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	142 res += '.*'
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	143 else:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	144 res += '[^/]*'
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	145 elif c == '?':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	146 res += '.'
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	147 elif c == '[':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	148 j = i
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	149 if j < n and pat[j] in '!]':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	150 j += 1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	151 while j < n and pat[j] != ']':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	152 j += 1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	153 if j >= n:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	154 res += '\\['
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	155 else:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	156 stuff = pat[i:j].replace('\\','\\\\')
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	157 i = j + 1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	158 if stuff[0] == '!':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	159 stuff = '^' + stuff[1:]
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	160 elif stuff[0] == '^':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	161 stuff = '\\' + stuff
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	162 res = '%s[%s]' % (res, stuff)
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	163 elif c == '{':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	164 group += 1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	165 res += '(?:'
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	166 elif c == '}' and group:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	167 res += ')'
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	168 group -= 1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	169 elif c == ',' and group:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	170 res += '\|'
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	171 elif c == '\\':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	172 p = peek()
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	173 if p:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	174 i += 1
8583 19d1b2aec562 match: optimize escaping in _globre Matt Mackall <mpm@selenic.com> parents: 8582 diff changeset	175 res += escape(p)
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	176 else:
8583 19d1b2aec562 match: optimize escaping in _globre Matt Mackall <mpm@selenic.com> parents: 8582 diff changeset	177 res += escape(c)
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	178 else:
8583 19d1b2aec562 match: optimize escaping in _globre Matt Mackall <mpm@selenic.com> parents: 8582 diff changeset	179 res += escape(c)
8582 a4c199e12b5a match: remove head and tail args from _globre Matt Mackall <mpm@selenic.com> parents: 8581 diff changeset	180 return res
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	181
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	182 def _regex(kind, name, tail):
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	183 '''convert a pattern into a regular expression'''
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	184 if not name:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	185 return ''
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	186 if kind == 're':
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	187 return name
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	188 elif kind == 'path':
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	189 return '^' + re.escape(name) + '(?:/\|$)'
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	190 elif kind == 'relglob':
8582 a4c199e12b5a match: remove head and tail args from _globre Matt Mackall <mpm@selenic.com> parents: 8581 diff changeset	191 return '(?:\|.*/)' + _globre(name) + tail
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	192 elif kind == 'relpath':
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	193 return re.escape(name) + '(?:/\|$)'
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	194 elif kind == 'relre':
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	195 if name.startswith('^'):
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	196 return name
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	197 return '.*' + name
8582 a4c199e12b5a match: remove head and tail args from _globre Matt Mackall <mpm@selenic.com> parents: 8581 diff changeset	198 return _globre(name) + tail
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	199
8580 f648c096f6d0 match: rename _matchfn to _buildmatch Matt Mackall <mpm@selenic.com> parents: 8579 diff changeset	200 def _buildmatch(pats, tail):
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	201 """build a matching function from a set of patterns"""
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	202 try:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	203 pat = '(?:%s)' % '\|'.join([_regex(k, p, tail) for (k, p) in pats])
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	204 if len(pat) > 20000:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	205 raise OverflowError()
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	206 return re.compile(pat).match
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	207 except OverflowError:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	208 # We're using a Python with a tiny regex engine and we
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	209 # made it explode, so we'll divide the pattern list in two
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	210 # until it works
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	211 l = len(pats)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	212 if l < 2:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	213 raise
8580 f648c096f6d0 match: rename _matchfn to _buildmatch Matt Mackall <mpm@selenic.com> parents: 8579 diff changeset	214 a, b = _buildmatch(pats[:l//2], tail), _buildmatch(pats[l//2:], tail)
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	215 return lambda s: a(s) or b(s)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	216 except re.error:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	217 for k, p in pats:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	218 try:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	219 re.compile('(?:%s)' % _regex(k, p, tail))
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	220 except re.error:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	221 raise util.Abort("invalid pattern (%s): %s" % (k, p))
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	222 raise util.Abort("invalid pattern")
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	223
8578 8388ef8d21cd match: tweak some names Matt Mackall <mpm@selenic.com> parents: 8577 diff changeset	224 def _normalize(names, default, root, cwd):
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	225 pats = []
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	226 for kind, name in [_patsplit(p, default) for p in names]:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	227 if kind in ('glob', 'relpath'):
8578 8388ef8d21cd match: tweak some names Matt Mackall <mpm@selenic.com> parents: 8577 diff changeset	228 name = util.canonpath(root, cwd, name)
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	229 elif kind in ('relglob', 'path'):
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	230 name = util.normpath(name)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	231
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	232 pats.append((kind, name))
8576 ec4ed21db4b2 match: split up _normalizepats Matt Mackall <mpm@selenic.com> parents: 8575 diff changeset	233 return pats
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	234
8576 ec4ed21db4b2 match: split up _normalizepats Matt Mackall <mpm@selenic.com> parents: 8575 diff changeset	235 def _roots(patterns):
ec4ed21db4b2 match: split up _normalizepats Matt Mackall <mpm@selenic.com> parents: 8575 diff changeset	236 r = []
ec4ed21db4b2 match: split up _normalizepats Matt Mackall <mpm@selenic.com> parents: 8575 diff changeset	237 for kind, name in patterns:
8584 0f06e72abfdc match: fold _globprefix into _roots Matt Mackall <mpm@selenic.com> parents: 8583 diff changeset	238 if kind == 'glob': # find the non-glob prefix
0f06e72abfdc match: fold _globprefix into _roots Matt Mackall <mpm@selenic.com> parents: 8583 diff changeset	239 root = []
0f06e72abfdc match: fold _globprefix into _roots Matt Mackall <mpm@selenic.com> parents: 8583 diff changeset	240 for p in name.split('/'):
0f06e72abfdc match: fold _globprefix into _roots Matt Mackall <mpm@selenic.com> parents: 8583 diff changeset	241 if '[' in p or '{' in p or '*' in p or '?' in p:
0f06e72abfdc match: fold _globprefix into _roots Matt Mackall <mpm@selenic.com> parents: 8583 diff changeset	242 break
0f06e72abfdc match: fold _globprefix into _roots Matt Mackall <mpm@selenic.com> parents: 8583 diff changeset	243 root.append(p)
0f06e72abfdc match: fold _globprefix into _roots Matt Mackall <mpm@selenic.com> parents: 8583 diff changeset	244 r.append('/'.join(root) or '.')
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	245 elif kind in ('relpath', 'path'):
8576 ec4ed21db4b2 match: split up _normalizepats Matt Mackall <mpm@selenic.com> parents: 8575 diff changeset	246 r.append(name or '.')
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	247 elif kind == 'relglob':
8576 ec4ed21db4b2 match: split up _normalizepats Matt Mackall <mpm@selenic.com> parents: 8575 diff changeset	248 r.append('.')
ec4ed21db4b2 match: split up _normalizepats Matt Mackall <mpm@selenic.com> parents: 8575 diff changeset	249 return r
ec4ed21db4b2 match: split up _normalizepats Matt Mackall <mpm@selenic.com> parents: 8575 diff changeset	250
ec4ed21db4b2 match: split up _normalizepats Matt Mackall <mpm@selenic.com> parents: 8575 diff changeset	251 def _anypats(patterns):
ec4ed21db4b2 match: split up _normalizepats Matt Mackall <mpm@selenic.com> parents: 8575 diff changeset	252 for kind, name in patterns:
ec4ed21db4b2 match: split up _normalizepats Matt Mackall <mpm@selenic.com> parents: 8575 diff changeset	253 if kind in ('glob', 're', 'relglob', 'relre'):
ec4ed21db4b2 match: split up _normalizepats Matt Mackall <mpm@selenic.com> parents: 8575 diff changeset	254 return True

Mercurial > public > mercurial-scm > hg

annotate mercurial/match.py @ 8586:347fe1ac4f21