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

annotate mercurial/match.py @ 8575:aa4fcb5c46f1

match: optimize _globprefix

author	Matt Mackall <mpm@selenic.com>
date	Sun, 24 May 2009 02:56:14 -0500
parents	63a7ed2128d5
children	ec4ed21db4b2

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 always(_match):
98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	39 def __init__(self, root, cwd):
98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	40 _match.__init__(self, root, cwd, [], lambda f: True, False)
98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	41
98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	42 class never(_match):
98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	43 def __init__(self, root, cwd):
98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	44 _match.__init__(self, root, cwd, [], lambda f: False, False)
6596 7fe4610cf920 match: add always, never, and exact methods Matt Mackall <mpm@selenic.com> parents: 6578 diff changeset	45
6604 98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	46 class exact(_match):
98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	47 def __init__(self, root, cwd, files):
8522 39fd67552297 match: use self.exact instead of lambda Simon Heimberg <simohe@besonet.ch> parents: 8231 diff changeset	48 _match.__init__(self, root, cwd, files, self.exact, False)
6596 7fe4610cf920 match: add always, never, and exact methods Matt Mackall <mpm@selenic.com> parents: 6578 diff changeset	49
6604 98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	50 class match(_match):
8567 fea40a677d43 match: add some default args Matt Mackall <mpm@selenic.com> parents: 8566 diff changeset	51 def __init__(self, root, cwd, patterns, include=[], exclude=[],
fea40a677d43 match: add some default args Matt Mackall <mpm@selenic.com> parents: 8566 diff changeset	52 default='glob'):
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	53 f, mf, ap = _matcher(root, cwd, patterns, include, exclude, default)
6604 98b6e6f0e49b match: cleanup match classes a bit Matt Mackall <mpm@selenic.com> parents: 6596 diff changeset	54 _match.__init__(self, root, cwd, f, mf, ap)
8568 4fa1618bf495 match: refactor patkind Matt Mackall <mpm@selenic.com> parents: 8567 diff changeset	55
4fa1618bf495 match: refactor patkind Matt Mackall <mpm@selenic.com> parents: 8567 diff changeset	56 def patkind(pat):
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	57 return _patsplit(pat, None)[0]
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	58
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	59 def _patsplit(pat, default):
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	60 """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	61 actual pattern."""
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	62 for prefix in 're', 'glob', 'path', 'relglob', 'relpath', 'relre':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	63 if pat.startswith(prefix + ':'): return pat.split(':', 1)
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	64 return default, pat
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	65
8573 62c996d543e6 match: kill unused defaults on _globre Matt Mackall <mpm@selenic.com> parents: 8572 diff changeset	66 def _globre(pat, head, tail):
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	67 "convert a glob pattern into a regexp"
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	68 i, n = 0, len(pat)
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	69 res = ''
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	70 group = 0
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	71 def peek(): return i < n and pat[i]
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	72 while i < n:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	73 c = pat[i]
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	74 i = i+1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	75 if c == '*':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	76 if peek() == '*':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	77 i += 1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	78 res += '.*'
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	79 else:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	80 res += '[^/]*'
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	81 elif c == '?':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	82 res += '.'
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	83 elif c == '[':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	84 j = i
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	85 if j < n and pat[j] in '!]':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	86 j += 1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	87 while j < n and pat[j] != ']':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	88 j += 1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	89 if j >= n:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	90 res += '\\['
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	91 else:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	92 stuff = pat[i:j].replace('\\','\\\\')
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	93 i = j + 1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	94 if stuff[0] == '!':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	95 stuff = '^' + stuff[1:]
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	96 elif stuff[0] == '^':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	97 stuff = '\\' + stuff
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	98 res = '%s[%s]' % (res, stuff)
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	99 elif c == '{':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	100 group += 1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	101 res += '(?:'
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	102 elif c == '}' and group:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	103 res += ')'
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	104 group -= 1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	105 elif c == ',' and group:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	106 res += '\|'
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	107 elif c == '\\':
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	108 p = peek()
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	109 if p:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	110 i += 1
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	111 res += re.escape(p)
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	112 else:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	113 res += re.escape(c)
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	114 else:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	115 res += re.escape(c)
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	116 return head + res + tail
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	117
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	118 def _regex(kind, name, tail):
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	119 '''convert a pattern into a regular expression'''
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	120 if not name:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	121 return ''
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	122 if kind == 're':
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	123 return name
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	124 elif kind == 'path':
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	125 return '^' + re.escape(name) + '(?:/\|$)'
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	126 elif kind == 'relglob':
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	127 return _globre(name, '(?:\|.*/)', tail)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	128 elif kind == 'relpath':
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	129 return re.escape(name) + '(?:/\|$)'
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	130 elif kind == 'relre':
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	131 if name.startswith('^'):
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	132 return name
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	133 return '.*' + name
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	134 return _globre(name, '', tail)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	135
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	136 def _matchfn(pats, tail):
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	137 """build a matching function from a set of patterns"""
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	138 try:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	139 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	140 if len(pat) > 20000:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	141 raise OverflowError()
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	142 return re.compile(pat).match
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	143 except OverflowError:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	144 # 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	145 # 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	146 # until it works
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	147 l = len(pats)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	148 if l < 2:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	149 raise
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	150 a, b = _matchfn(pats[:l//2], tail), matchfn(pats[l//2:], tail)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	151 return lambda s: a(s) or b(s)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	152 except re.error:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	153 for k, p in pats:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	154 try:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	155 re.compile('(?:%s)' % _regex(k, p, tail))
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	156 except re.error:
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	157 raise util.Abort("invalid pattern (%s): %s" % (k, p))
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	158 raise util.Abort("invalid pattern")
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	159
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	160 def _globprefix(pat):
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	161 '''return the non-glob prefix of a path, e.g. foo/* -> foo'''
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	162 root = []
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	163 for p in pat.split('/'):
8575 aa4fcb5c46f1 match: optimize _globprefix Matt Mackall <mpm@selenic.com> parents: 8574 diff changeset	164 if '[' in p or '{' in p or '*' in p or '?' in p:
aa4fcb5c46f1 match: optimize _globprefix Matt Mackall <mpm@selenic.com> parents: 8574 diff changeset	165 break
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	166 root.append(p)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	167 return '/'.join(root) or '.'
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	168
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	169 def _normalizepats(names, default, canonroot, cwd):
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	170 pats = []
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	171 roots = []
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	172 anypats = False
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	173 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	174 if kind in ('glob', 'relpath'):
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	175 name = util.canonpath(canonroot, cwd, name)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	176 elif kind in ('relglob', 'path'):
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	177 name = util.normpath(name)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	178
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	179 pats.append((kind, name))
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	180
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	181 if kind in ('glob', 're', 'relglob', 'relre'):
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	182 anypats = True
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	183
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	184 if kind == 'glob':
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	185 root = _globprefix(name)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	186 roots.append(root)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	187 elif kind in ('relpath', 'path'):
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	188 roots.append(name or '.')
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	189 elif kind == 'relglob':
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	190 roots.append('.')
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	191 return roots, pats, anypats
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	192
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	193 def _matcher(root, cwd='', names=[], inc=[], exc=[], dflt_pat='glob'):
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	194 """build a function to match a set of file patterns
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	195
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	196 arguments:
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	197 root - the canonical root of the tree you're matching against
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	198 cwd - the current working directory, if relevant
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	199 names - patterns to find
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	200 inc - patterns to include
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	201 exc - patterns to exclude
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	202 dflt_pat - if a pattern in names has no explicit type, assume this one
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	203
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	204 a pattern is one of:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	205 'glob:<glob>' - a glob relative to cwd
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	206 're:<regexp>' - a regular expression
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	207 'path:<path>' - a path relative to canonroot
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	208 'relglob:<glob>' - an unrooted glob (*.c matches C files in all dirs)
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	209 'relpath:<path>' - a path relative to cwd
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	210 'relre:<regexp>' - a regexp that doesn't have to match the start of a name
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	211 '<something>' - one of the cases above, selected by the dflt_pat argument
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	212
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	213 returns:
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	214 a 3-tuple containing
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	215 - list of roots (places where one should start a recursive walk of the fs);
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	216 this often matches the explicit non-pattern names passed in, but also
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	217 includes the initial part of glob: patterns that has no glob characters
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	218 - a bool match(filename) function
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	219 - a bool indicating if any patterns were passed in
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	220 """
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	221
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	222 # a common case: no patterns at all
7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	223 if not names and not inc and not exc:
8571 9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	224 return [], lambda f: True, False
8568 4fa1618bf495 match: refactor patkind Matt Mackall <mpm@selenic.com> parents: 8567 diff changeset	225
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	226 roots, pats, anypats = _normalizepats(names, dflt_pat, root, cwd)
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	227
8572 dd46948a07fa match: kill test in matchfn Matt Mackall <mpm@selenic.com> parents: 8571 diff changeset	228 if names:
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	229 patmatch = _matchfn(pats, '$')
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	230 if inc:
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	231 dummy, inckinds, dummy = _normalizepats(inc, 'glob', root, cwd)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	232 incmatch = _matchfn(inckinds, '(?:/\|$)')
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	233 if exc:
8574 63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	234 dummy, exckinds, dummy = _normalizepats(exc, 'glob', root, cwd)
63a7ed2128d5 match: unnest functions in _matcher Matt Mackall <mpm@selenic.com> parents: 8573 diff changeset	235 excmatch = _matchfn(exckinds, '(?:/\|$)')
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	236
8571 9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	237 if names:
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	238 if inc:
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	239 if exc:
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	240 m = lambda f: incmatch(f) and not excmatch(f) and patmatch(f)
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	241 else:
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	242 m = lambda f: incmatch(f) and patmatch(f)
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	243 else:
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	244 if exc:
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	245 m = lambda f: not excmatch(f) and patmatch(f)
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	246 else:
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	247 m = patmatch
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	248 else:
8571 9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	249 if inc:
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	250 if exc:
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	251 m = lambda f: incmatch(f) and not excmatch(f)
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	252 else:
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	253 m = incmatch
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	254 else:
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	255 if exc:
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	256 m = lambda f: not excmatch(f)
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	257 else:
9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	258 m = lambda f: True
8570 7fe2012b3bd0 match: move util match functions over Matt Mackall <mpm@selenic.com> parents: 8568 diff changeset	259
8571 9f12e1a27a1b match: refactor matchfn generation Matt Mackall <mpm@selenic.com> parents: 8570 diff changeset	260 return (roots, m, (inc or exc or anypats) and True)

Mercurial > public > mercurial-scm > hg

annotate mercurial/match.py @ 8575:aa4fcb5c46f1