mercurial/match.py
author Denis Laxalde <denis@laxalde.org>
Sat, 06 Apr 2019 17:54:13 +0200
changeset 42082 413a75da98ce
parent 42081 bf777c1e78dd
child 42083 bee1647578b7
permissions -rw-r--r--
match: add doctest examples for patkind()
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
8761
0289f384e1e5 Generally replace "file name" with "filename" in help and comments.
timeless <timeless@gmail.com>
parents: 8682
diff changeset
     1
# match.py - filename matching
8231
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
10263
25e572394f5c Update license to GPLv2+
Matt Mackall <mpm@selenic.com>
parents: 9036
diff changeset
     6
# GNU General Public License version 2 or any later version.
8231
5d4d88a4f5e6 match: add copyright and license header
Martin Geisler <mg@lazybytes.net>
parents: 8152
diff changeset
     7
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
     8
from __future__ import absolute_import, print_function
25958
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
     9
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
    10
import copy
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
    11
import itertools
25958
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
    12
import os
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
    13
import re
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
    14
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
    15
from .i18n import _
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
    16
from . import (
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
    17
    encoding,
26587
56b2bcea2529 error: get Abort from 'error' instead of 'util'
Pierre-Yves David <pierre-yves.david@fb.com>
parents: 26014
diff changeset
    18
    error,
25958
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
    19
    pathutil,
36572
9adfa48792a7 match: some minimal pycompat fixes guided by test-hgignore.t
Augie Fackler <augie@google.com>
parents: 36050
diff changeset
    20
    pycompat,
25958
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
    21
    util,
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
    22
)
37084
f0b6fbea00cf stringutil: bulk-replace call sites to point to new module
Yuya Nishihara <yuya@tcha.org>
parents: 36572
diff changeset
    23
from .utils import (
f0b6fbea00cf stringutil: bulk-replace call sites to point to new module
Yuya Nishihara <yuya@tcha.org>
parents: 36572
diff changeset
    24
    stringutil,
f0b6fbea00cf stringutil: bulk-replace call sites to point to new module
Yuya Nishihara <yuya@tcha.org>
parents: 36572
diff changeset
    25
)
6576
69f3e9ac7c56 walk: introduce match objects
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
    26
33684
2be0bf186950 match: expose some data and functionality to other modules
Kostia Balytskyi <ikostia@fb.com>
parents: 33582
diff changeset
    27
allpatternkinds = ('re', 'glob', 'path', 'relglob', 'relpath', 'relre',
41282
4fab8a7d2d72 match: support rooted globs in hgignore
Valentin Gatien-Baron <vgatien-baron@janestreet.com>
parents: 41129
diff changeset
    28
                   'rootglob',
33684
2be0bf186950 match: expose some data and functionality to other modules
Kostia Balytskyi <ikostia@fb.com>
parents: 33582
diff changeset
    29
                   'listfile', 'listfile0', 'set', 'include', 'subinclude',
2be0bf186950 match: expose some data and functionality to other modules
Kostia Balytskyi <ikostia@fb.com>
parents: 33582
diff changeset
    30
                   'rootfilesin')
2be0bf186950 match: expose some data and functionality to other modules
Kostia Balytskyi <ikostia@fb.com>
parents: 33582
diff changeset
    31
cwdrelativepatternkinds = ('relpath', 'glob')
2be0bf186950 match: expose some data and functionality to other modules
Kostia Balytskyi <ikostia@fb.com>
parents: 33582
diff changeset
    32
24636
36872036169b treemanifest: further optimize treemanifest.matches()
Drew Gottlieb <drgott@google.com>
parents: 24452
diff changeset
    33
propertycache = util.propertycache
36872036169b treemanifest: further optimize treemanifest.matches()
Drew Gottlieb <drgott@google.com>
parents: 24452
diff changeset
    34
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
    35
def _rematcher(regex):
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
    36
    '''compile the regexp with the best available regexp engine and return a
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
    37
    matcher function'''
21909
335bb8b80443 match: use util.re.compile instead of util.compilere
Siddharth Agarwal <sid0@fb.com>
parents: 21815
diff changeset
    38
    m = util.re.compile(regex)
16943
8d08a28aa63e matcher: use re2 bindings if available
Bryan O'Sullivan <bryano@fb.com>
parents: 16791
diff changeset
    39
    try:
8d08a28aa63e matcher: use re2 bindings if available
Bryan O'Sullivan <bryano@fb.com>
parents: 16791
diff changeset
    40
        # slightly faster, provided by facebook's re2 bindings
8d08a28aa63e matcher: use re2 bindings if available
Bryan O'Sullivan <bryano@fb.com>
parents: 16791
diff changeset
    41
        return m.test_match
8d08a28aa63e matcher: use re2 bindings if available
Bryan O'Sullivan <bryano@fb.com>
parents: 16791
diff changeset
    42
    except AttributeError:
8d08a28aa63e matcher: use re2 bindings if available
Bryan O'Sullivan <bryano@fb.com>
parents: 16791
diff changeset
    43
        return m.match
8d08a28aa63e matcher: use re2 bindings if available
Bryan O'Sullivan <bryano@fb.com>
parents: 16791
diff changeset
    44
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
    45
def _expandsets(kindpats, ctx, listsubrepos, badfn):
38612
760cc5dc01e8 fileset: restrict getfileset() to not return a computed set (API)
Yuya Nishihara <yuya@tcha.org>
parents: 38611
diff changeset
    46
    '''Returns the kindpats list with the 'set' patterns expanded to matchers'''
760cc5dc01e8 fileset: restrict getfileset() to not return a computed set (API)
Yuya Nishihara <yuya@tcha.org>
parents: 38611
diff changeset
    47
    matchers = []
14675
cfc89398f710 match: introduce basic fileset support
Matt Mackall <mpm@selenic.com>
parents: 14674
diff changeset
    48
    other = []
cfc89398f710 match: introduce basic fileset support
Matt Mackall <mpm@selenic.com>
parents: 14674
diff changeset
    49
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
    50
    for kind, pat, source in kindpats:
14675
cfc89398f710 match: introduce basic fileset support
Matt Mackall <mpm@selenic.com>
parents: 14674
diff changeset
    51
        if kind == 'set':
41022
481249481392 match: fix assertion for fileset with no context (issue6046)
Yuya Nishihara <yuya@tcha.org>
parents: 40345
diff changeset
    52
            if ctx is None:
32444
57d6c0c74b1b match: use ProgrammingError where appropriate
Martin von Zweigbergk <martinvonz@google.com>
parents: 32415
diff changeset
    53
                raise error.ProgrammingError("fileset expression with no "
57d6c0c74b1b match: use ProgrammingError where appropriate
Martin von Zweigbergk <martinvonz@google.com>
parents: 32415
diff changeset
    54
                                             "context")
38612
760cc5dc01e8 fileset: restrict getfileset() to not return a computed set (API)
Yuya Nishihara <yuya@tcha.org>
parents: 38611
diff changeset
    55
            matchers.append(ctx.matchfileset(pat, badfn=badfn))
25122
755d23a49170 match: resolve filesets in subrepos for commands given the '-S' argument
Matt Harbison <matt_harbison@yahoo.com>
parents: 25114
diff changeset
    56
755d23a49170 match: resolve filesets in subrepos for commands given the '-S' argument
Matt Harbison <matt_harbison@yahoo.com>
parents: 25114
diff changeset
    57
            if listsubrepos:
755d23a49170 match: resolve filesets in subrepos for commands given the '-S' argument
Matt Harbison <matt_harbison@yahoo.com>
parents: 25114
diff changeset
    58
                for subpath in ctx.substate:
38612
760cc5dc01e8 fileset: restrict getfileset() to not return a computed set (API)
Yuya Nishihara <yuya@tcha.org>
parents: 38611
diff changeset
    59
                    sm = ctx.sub(subpath).matchfileset(pat, badfn=badfn)
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
    60
                    pm = prefixdirmatcher(subpath, sm, badfn=badfn)
38612
760cc5dc01e8 fileset: restrict getfileset() to not return a computed set (API)
Yuya Nishihara <yuya@tcha.org>
parents: 38611
diff changeset
    61
                    matchers.append(pm)
25122
755d23a49170 match: resolve filesets in subrepos for commands given the '-S' argument
Matt Harbison <matt_harbison@yahoo.com>
parents: 25114
diff changeset
    62
14675
cfc89398f710 match: introduce basic fileset support
Matt Mackall <mpm@selenic.com>
parents: 14674
diff changeset
    63
            continue
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
    64
        other.append((kind, pat, source))
38612
760cc5dc01e8 fileset: restrict getfileset() to not return a computed set (API)
Yuya Nishihara <yuya@tcha.org>
parents: 38611
diff changeset
    65
    return matchers, other
14675
cfc89398f710 match: introduce basic fileset support
Matt Mackall <mpm@selenic.com>
parents: 14674
diff changeset
    66
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    67
def _expandsubinclude(kindpats, root):
32132
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
    68
    '''Returns the list of subinclude matcher args and the kindpats without the
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    69
    subincludes in it.'''
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    70
    relmatchers = []
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    71
    other = []
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    72
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    73
    for kind, pat, source in kindpats:
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    74
        if kind == 'subinclude':
25301
caaf4045eca8 match: normpath the ignore source when expanding the 'subinclude' kind
Matt Harbison <matt_harbison@yahoo.com>
parents: 25283
diff changeset
    75
            sourceroot = pathutil.dirname(util.normpath(source))
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    76
            pat = util.pconvert(pat)
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    77
            path = pathutil.join(sourceroot, pat)
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    78
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    79
            newroot = pathutil.dirname(path)
32132
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
    80
            matcherargs = (newroot, '', [], ['include:%s' % path])
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    81
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    82
            prefix = pathutil.canonpath(root, root, newroot)
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    83
            if prefix:
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    84
                prefix += '/'
32132
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
    85
            relmatchers.append((prefix, matcherargs))
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    86
        else:
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    87
            other.append((kind, pat, source))
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    88
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    89
    return relmatchers, other
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    90
24447
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
    91
def _kindpatsalwaysmatch(kindpats):
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
    92
    """"Checks whether the kindspats match everything, as e.g.
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
    93
    'relpath:.' does.
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
    94
    """
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
    95
    for kind, pat, source in kindpats:
24447
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
    96
        if pat != '' or kind not in ['relpath', 'glob']:
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
    97
            return False
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
    98
    return True
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
    99
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   100
def _buildkindpatsmatcher(matchercls, root, kindpats, ctx=None,
38580
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   101
                          listsubrepos=False, badfn=None):
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   102
    matchers = []
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   103
    fms, kindpats = _expandsets(kindpats, ctx=ctx,
38612
760cc5dc01e8 fileset: restrict getfileset() to not return a computed set (API)
Yuya Nishihara <yuya@tcha.org>
parents: 38611
diff changeset
   104
                                listsubrepos=listsubrepos, badfn=badfn)
38580
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   105
    if kindpats:
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   106
        m = matchercls(root, kindpats, badfn=badfn)
38580
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   107
        matchers.append(m)
38612
760cc5dc01e8 fileset: restrict getfileset() to not return a computed set (API)
Yuya Nishihara <yuya@tcha.org>
parents: 38611
diff changeset
   108
    if fms:
760cc5dc01e8 fileset: restrict getfileset() to not return a computed set (API)
Yuya Nishihara <yuya@tcha.org>
parents: 38611
diff changeset
   109
        matchers.extend(fms)
38580
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   110
    if not matchers:
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   111
        return nevermatcher(badfn=badfn)
38580
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   112
    if len(matchers) == 1:
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   113
        return matchers[0]
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   114
    return unionmatcher(matchers)
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   115
32728
3e8eb6d84a5c match: allow pats to be None
Martin von Zweigbergk <martinvonz@google.com>
parents: 32650
diff changeset
   116
def match(root, cwd, patterns=None, include=None, exclude=None, default='glob',
41622
635a12c53ea6 match: remove unused "exact" argument (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41528
diff changeset
   117
          auditor=None, ctx=None, listsubrepos=False, warn=None,
32401
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   118
          badfn=None, icasefs=False):
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   119
    """build an object to match a set of file patterns
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   120
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   121
    arguments:
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   122
    root - the canonical root of the tree you're matching against
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   123
    cwd - the current working directory, if relevant
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   124
    patterns - patterns to find
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   125
    include - patterns to include (unless they are excluded)
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   126
    exclude - patterns to exclude (even if they are included)
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   127
    default - if a pattern in patterns has no explicit type, assume this one
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   128
    warn - optional function used for printing warnings
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   129
    badfn - optional bad() callback for this matcher instead of the default
32401
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   130
    icasefs - make a matcher for wdir on case insensitive filesystems, which
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   131
        normalizes the given patterns to the case in the filesystem
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   132
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   133
    a pattern is one of:
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   134
    'glob:<glob>' - a glob relative to cwd
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   135
    're:<regexp>' - a regular expression
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   136
    'path:<path>' - a path relative to repository root, which is matched
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   137
                    recursively
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   138
    'rootfilesin:<path>' - a path relative to repository root, which is
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   139
                    matched non-recursively (will not match subdirectories)
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   140
    'relglob:<glob>' - an unrooted glob (*.c matches C files in all dirs)
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   141
    'relpath:<path>' - a path relative to cwd
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   142
    'relre:<regexp>' - a regexp that needn't match the start of a name
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   143
    'set:<fileset>' - a fileset expression
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   144
    'include:<path>' - a file of patterns to read and include
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   145
    'subinclude:<path>' - a file of patterns to match against files under
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   146
                          the same directory
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   147
    '<something>' - a pattern of the specified default type
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   148
    """
32401
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   149
    normalize = _donormalize
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   150
    if icasefs:
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   151
        dirstate = ctx.repo().dirstate
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   152
        dsnormalize = dirstate.normalize
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   153
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   154
        def normalize(patterns, default, root, cwd, auditor, warn):
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   155
            kp = _donormalize(patterns, default, root, cwd, auditor, warn)
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   156
            kindpats = []
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   157
            for kind, pats, source in kp:
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   158
                if kind not in ('re', 'relre'):  # regex can't be normalized
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   159
                    p = pats
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   160
                    pats = dsnormalize(pats)
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   161
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   162
                    # Preserve the original to handle a case only rename.
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   163
                    if p != pats and p in dirstate:
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   164
                        kindpats.append((kind, p, source))
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   165
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   166
                kindpats.append((kind, pats, source))
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   167
            return kindpats
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   168
41622
635a12c53ea6 match: remove unused "exact" argument (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41528
diff changeset
   169
    if patterns:
32556
5f08eca8f8d3 match: move normalize() call out of matcher constructors
Martin von Zweigbergk <martinvonz@google.com>
parents: 32555
diff changeset
   170
        kindpats = normalize(patterns, default, root, cwd, auditor, warn)
32557
3fdcc34c0aba match: remove special-casing of always-matching patterns in patternmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32556
diff changeset
   171
        if _kindpatsalwaysmatch(kindpats):
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   172
            m = alwaysmatcher(badfn)
32557
3fdcc34c0aba match: remove special-casing of always-matching patterns in patternmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32556
diff changeset
   173
        else:
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   174
            m = _buildkindpatsmatcher(patternmatcher, root, kindpats, ctx=ctx,
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   175
                                      listsubrepos=listsubrepos, badfn=badfn)
32553
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   176
    else:
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   177
        # It's a little strange that no patterns means to match everything.
32650
783394c0c978 match: simplify nevermatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32605
diff changeset
   178
        # Consider changing this to match nothing (probably using nevermatcher).
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   179
        m = alwaysmatcher(badfn)
32553
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   180
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   181
    if include:
32556
5f08eca8f8d3 match: move normalize() call out of matcher constructors
Martin von Zweigbergk <martinvonz@google.com>
parents: 32555
diff changeset
   182
        kindpats = normalize(include, 'glob', root, cwd, auditor, warn)
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   183
        im = _buildkindpatsmatcher(includematcher, root, kindpats, ctx=ctx,
38580
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   184
                                   listsubrepos=listsubrepos, badfn=None)
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   185
        m = intersectmatchers(m, im)
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   186
    if exclude:
32556
5f08eca8f8d3 match: move normalize() call out of matcher constructors
Martin von Zweigbergk <martinvonz@google.com>
parents: 32555
diff changeset
   187
        kindpats = normalize(exclude, 'glob', root, cwd, auditor, warn)
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   188
        em = _buildkindpatsmatcher(includematcher, root, kindpats, ctx=ctx,
38580
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   189
                                   listsubrepos=listsubrepos, badfn=None)
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   190
        m = differencematcher(m, em)
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   191
    return m
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   192
41676
0531dff73d0b match: delete unused root and cwd arguments from {always,never,exact}() (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41675
diff changeset
   193
def exact(files, badfn=None):
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   194
    return exactmatcher(files, badfn=badfn)
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   195
41676
0531dff73d0b match: delete unused root and cwd arguments from {always,never,exact}() (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41675
diff changeset
   196
def always(badfn=None):
0531dff73d0b match: delete unused root and cwd arguments from {always,never,exact}() (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41675
diff changeset
   197
    return alwaysmatcher(badfn)
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   198
41676
0531dff73d0b match: delete unused root and cwd arguments from {always,never,exact}() (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41675
diff changeset
   199
def never(badfn=None):
0531dff73d0b match: delete unused root and cwd arguments from {always,never,exact}() (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41675
diff changeset
   200
    return nevermatcher(badfn)
32605
e6ff007e107e match: introduce nevermatcher for when no ignore files are present
Siddharth Agarwal <sid0@fb.com>
parents: 32557
diff changeset
   201
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   202
def badmatch(match, badfn):
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   203
    """Make a copy of the given matcher, replacing its bad method with the given
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   204
    one.
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   205
    """
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   206
    m = copy.copy(match)
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   207
    m.bad = badfn
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   208
    return m
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   209
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   210
def _donormalize(patterns, default, root, cwd, auditor, warn):
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   211
    '''Convert 'kind:pat' from the patterns list to tuples with kind and
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   212
    normalized and rooted patterns and with listfiles expanded.'''
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   213
    kindpats = []
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   214
    for kind, pat in [_patsplit(p, default) for p in patterns]:
33684
2be0bf186950 match: expose some data and functionality to other modules
Kostia Balytskyi <ikostia@fb.com>
parents: 33582
diff changeset
   215
        if kind in cwdrelativepatternkinds:
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   216
            pat = pathutil.canonpath(root, cwd, pat, auditor)
41282
4fab8a7d2d72 match: support rooted globs in hgignore
Valentin Gatien-Baron <vgatien-baron@janestreet.com>
parents: 41129
diff changeset
   217
        elif kind in ('relglob', 'path', 'rootfilesin', 'rootglob'):
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   218
            pat = util.normpath(pat)
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   219
        elif kind in ('listfile', 'listfile0'):
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   220
            try:
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   221
                files = util.readfile(pat)
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   222
                if kind == 'listfile0':
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   223
                    files = files.split('\0')
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   224
                else:
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   225
                    files = files.splitlines()
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   226
                files = [f for f in files if f]
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   227
            except EnvironmentError:
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   228
                raise error.Abort(_("unable to read file list (%s)") % pat)
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   229
            for k, p, source in _donormalize(files, default, root, cwd,
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   230
                                             auditor, warn):
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   231
                kindpats.append((k, p, pat))
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   232
            continue
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   233
        elif kind == 'include':
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   234
            try:
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   235
                fullpath = os.path.join(root, util.localpath(pat))
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   236
                includepats = readpatternfile(fullpath, warn)
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   237
                for k, p, source in _donormalize(includepats, default,
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   238
                                                 root, cwd, auditor, warn):
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   239
                    kindpats.append((k, p, source or pat))
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   240
            except error.Abort as inst:
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   241
                raise error.Abort('%s: %s' % (pat, inst[0]))
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   242
            except IOError as inst:
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   243
                if warn:
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   244
                    warn(_("skipping unreadable pattern file '%s': %s\n") %
37084
f0b6fbea00cf stringutil: bulk-replace call sites to point to new module
Yuya Nishihara <yuya@tcha.org>
parents: 36572
diff changeset
   245
                         (pat, stringutil.forcebytestr(inst.strerror)))
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   246
            continue
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   247
        # else: re or relre - which cannot be normalized
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   248
        kindpats.append((kind, pat, ''))
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   249
    return kindpats
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   250
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   251
class basematcher(object):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   252
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   253
    def __init__(self, badfn=None):
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   254
        if badfn is not None:
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   255
            self.bad = badfn
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   256
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   257
    def __call__(self, fn):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   258
        return self.matchfn(fn)
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   259
    def __iter__(self):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   260
        for f in self._files:
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   261
            yield f
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   262
    # Callbacks related to how the matcher is used by dirstate.walk.
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   263
    # Subscribers to these events must monkeypatch the matcher object.
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   264
    def bad(self, f, msg):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   265
        '''Callback from dirstate.walk for each explicit file that can't be
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   266
        found/accessed, with an error message.'''
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   267
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   268
    # If an explicitdir is set, it will be called when an explicitly listed
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   269
    # directory is visited.
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   270
    explicitdir = None
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   271
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   272
    # If an traversedir is set, it will be called when a directory discovered
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   273
    # by recursive traversal is visited.
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   274
    traversedir = None
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   275
32459
9f781f43f2ce match: make basematcher._files a @propertycache
Martin von Zweigbergk <martinvonz@google.com>
parents: 32458
diff changeset
   276
    @propertycache
9f781f43f2ce match: make basematcher._files a @propertycache
Martin von Zweigbergk <martinvonz@google.com>
parents: 32458
diff changeset
   277
    def _files(self):
9f781f43f2ce match: make basematcher._files a @propertycache
Martin von Zweigbergk <martinvonz@google.com>
parents: 32458
diff changeset
   278
        return []
9f781f43f2ce match: make basematcher._files a @propertycache
Martin von Zweigbergk <martinvonz@google.com>
parents: 32458
diff changeset
   279
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   280
    def files(self):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   281
        '''Explicitly listed files or patterns or roots:
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   282
        if no patterns or .always(): empty list,
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   283
        if exact: list exact files,
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   284
        if not .anypats(): list all files and dirs,
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   285
        else: optimal roots'''
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   286
        return self._files
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   287
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   288
    @propertycache
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   289
    def _fileset(self):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   290
        return set(self._files)
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   291
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   292
    def exact(self, f):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   293
        '''Returns True if f is in .files().'''
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   294
        return f in self._fileset
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   295
32463
43e091847c4d match: make matchfn a method on the class
Martin von Zweigbergk <martinvonz@google.com>
parents: 32461
diff changeset
   296
    def matchfn(self, f):
43e091847c4d match: make matchfn a method on the class
Martin von Zweigbergk <martinvonz@google.com>
parents: 32461
diff changeset
   297
        return False
43e091847c4d match: make matchfn a method on the class
Martin von Zweigbergk <martinvonz@google.com>
parents: 32461
diff changeset
   298
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   299
    def visitdir(self, dir):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   300
        '''Decides whether a directory should be visited based on whether it
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   301
        has potential matches in it or one of its subdirectories. This is
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   302
        based on the match's primary, included, and excluded patterns.
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   303
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   304
        Returns the string 'all' if the given directory and all subdirectories
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   305
        should be visited. Otherwise returns True or False indicating whether
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   306
        the given directory should be visited.
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   307
        '''
33478
cf15c3cc304c match: make base matcher return True for visitdir
Durham Goode <durham@fb.com>
parents: 33448
diff changeset
   308
        return True
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   309
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   310
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   311
        '''Decides whether a directory should be visited based on whether it
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   312
        has potential matches in it or one of its subdirectories, and
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   313
        potentially lists which subdirectories of that directory should be
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   314
        visited. This is based on the match's primary, included, and excluded
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   315
        patterns.
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   316
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   317
        This function is very similar to 'visitdir', and the following mapping
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   318
        can be applied:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   319
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   320
             visitdir | visitchildrenlist
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   321
            ----------+-------------------
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   322
             False    | set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   323
             'all'    | 'all'
39260
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   324
             True     | 'this' OR non-empty set of subdirs -or files- to visit
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   325
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   326
        Example:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   327
          Assume matchers ['path:foo/bar', 'rootfilesin:qux'], we would return
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   328
          the following values (assuming the implementation of visitchildrenset
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   329
          is capable of recognizing this; some implementations are not).
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   330
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   331
          '.' -> {'foo', 'qux'}
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   332
          'baz' -> set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   333
          'foo' -> {'bar'}
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   334
          # Ideally this would be 'all', but since the prefix nature of matchers
39260
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   335
          # is applied to the entire matcher, we have to downgrade this to
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   336
          # 'this' due to the non-prefix 'rootfilesin'-kind matcher being mixed
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   337
          # in.
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   338
          'foo/bar' -> 'this'
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   339
          'qux' -> 'this'
39260
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   340
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   341
        Important:
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   342
          Most matchers do not know if they're representing files or
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   343
          directories. They see ['path:dir/f'] and don't know whether 'f' is a
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   344
          file or a directory, so visitchildrenset('dir') for most matchers will
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   345
          return {'f'}, but if the matcher knows it's a file (like exactmatcher
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   346
          does), it may return 'this'. Do not rely on the return being a set
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   347
          indicating that there are no files in this dir to investigate (or
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   348
          equivalently that if there are files to investigate in 'dir' that it
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   349
          will always return 'this').
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   350
        '''
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   351
        return 'this'
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   352
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   353
    def always(self):
33379
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   354
        '''Matcher will match everything and .files() will be empty --
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   355
        optimization might be possible.'''
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   356
        return False
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   357
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   358
    def isexact(self):
33379
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   359
        '''Matcher will match exactly the list of files in .files() --
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   360
        optimization might be possible.'''
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   361
        return False
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   362
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   363
    def prefix(self):
33379
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   364
        '''Matcher will match the paths in .files() recursively --
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   365
        optimization might be possible.'''
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   366
        return False
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   367
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   368
    def anypats(self):
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   369
        '''None of .always(), .isexact(), and .prefix() is true --
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   370
        optimizations will be difficult.'''
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   371
        return not self.always() and not self.isexact() and not self.prefix()
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   372
32553
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   373
class alwaysmatcher(basematcher):
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   374
    '''Matches everything.'''
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   375
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   376
    def __init__(self, badfn=None):
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   377
        super(alwaysmatcher, self).__init__(badfn)
32553
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   378
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   379
    def always(self):
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   380
        return True
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   381
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   382
    def matchfn(self, f):
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   383
        return True
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   384
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   385
    def visitdir(self, dir):
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   386
        return 'all'
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   387
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   388
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   389
        return 'all'
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   390
32553
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   391
    def __repr__(self):
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   392
        return r'<alwaysmatcher>'
32553
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   393
32605
e6ff007e107e match: introduce nevermatcher for when no ignore files are present
Siddharth Agarwal <sid0@fb.com>
parents: 32557
diff changeset
   394
class nevermatcher(basematcher):
e6ff007e107e match: introduce nevermatcher for when no ignore files are present
Siddharth Agarwal <sid0@fb.com>
parents: 32557
diff changeset
   395
    '''Matches nothing.'''
e6ff007e107e match: introduce nevermatcher for when no ignore files are present
Siddharth Agarwal <sid0@fb.com>
parents: 32557
diff changeset
   396
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   397
    def __init__(self, badfn=None):
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   398
        super(nevermatcher, self).__init__(badfn)
32605
e6ff007e107e match: introduce nevermatcher for when no ignore files are present
Siddharth Agarwal <sid0@fb.com>
parents: 32557
diff changeset
   399
33378
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   400
    # It's a little weird to say that the nevermatcher is an exact matcher
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   401
    # or a prefix matcher, but it seems to make sense to let callers take
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   402
    # fast paths based on either. There will be no exact matches, nor any
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   403
    # prefixes (files() returns []), so fast paths iterating over them should
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   404
    # be efficient (and correct).
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   405
    def isexact(self):
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   406
        return True
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   407
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   408
    def prefix(self):
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   409
        return True
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   410
33582
44bc181b9835 match: override visitdir() in nevermatcher to return False
Martin von Zweigbergk <martinvonz@google.com>
parents: 33497
diff changeset
   411
    def visitdir(self, dir):
44bc181b9835 match: override visitdir() in nevermatcher to return False
Martin von Zweigbergk <martinvonz@google.com>
parents: 33497
diff changeset
   412
        return False
44bc181b9835 match: override visitdir() in nevermatcher to return False
Martin von Zweigbergk <martinvonz@google.com>
parents: 33497
diff changeset
   413
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   414
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   415
        return set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   416
32605
e6ff007e107e match: introduce nevermatcher for when no ignore files are present
Siddharth Agarwal <sid0@fb.com>
parents: 32557
diff changeset
   417
    def __repr__(self):
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   418
        return r'<nevermatcher>'
32605
e6ff007e107e match: introduce nevermatcher for when no ignore files are present
Siddharth Agarwal <sid0@fb.com>
parents: 32557
diff changeset
   419
38577
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   420
class predicatematcher(basematcher):
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   421
    """A matcher adapter for a simple boolean function"""
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   422
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   423
    def __init__(self, predfn, predrepr=None, badfn=None):
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   424
        super(predicatematcher, self).__init__(badfn)
38577
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   425
        self.matchfn = predfn
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   426
        self._predrepr = predrepr
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   427
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   428
    @encoding.strmethod
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   429
    def __repr__(self):
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   430
        s = (stringutil.buildrepr(self._predrepr)
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   431
             or pycompat.byterepr(self.matchfn))
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   432
        return '<predicatenmatcher pred=%s>' % s
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   433
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   434
class patternmatcher(basematcher):
42081
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   435
    """Matches a set of (kind, pat, source) against a 'root' directory.
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   436
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   437
    >>> kindpats = [
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   438
    ...     ('re', '.*\.c$', ''),
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   439
    ...     ('path', 'foo/a', ''),
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   440
    ...     ('relpath', 'b', ''),
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   441
    ...     ('glob', '*.h', ''),
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   442
    ... ]
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   443
    >>> m = patternmatcher('foo', kindpats)
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   444
    >>> bool(m('main.c'))  # matches re:.*\.c$
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   445
    True
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   446
    >>> bool(m('b.txt'))
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   447
    False
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   448
    >>> bool(m('foo/a'))  # matches path:foo/a
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   449
    True
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   450
    >>> bool(m('a'))  # does not match path:b, since 'root' is 'foo'
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   451
    False
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   452
    >>> bool(m('b'))  # matches relpath:b, since 'root' is 'foo'
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   453
    True
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   454
    >>> bool(m('lib.h'))  # matches glob:*.h
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   455
    True
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   456
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   457
    >>> m.files()
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   458
    ['.', 'foo/a', 'b', '.']
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   459
    >>> m.exact('foo/a')
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   460
    True
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   461
    >>> m.exact('b')
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   462
    True
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   463
    >>> m.exact('lib.h')  # exact matches are for (rel)path kinds
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   464
    False
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   465
    """
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   466
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   467
    def __init__(self, root, kindpats, badfn=None):
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   468
        super(patternmatcher, self).__init__(badfn)
8581
101d305c1d0b match: fold _matcher into match.__init__
Matt Mackall <mpm@selenic.com>
parents: 8580
diff changeset
   469
32557
3fdcc34c0aba match: remove special-casing of always-matching patterns in patternmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32556
diff changeset
   470
        self._files = _explicitfiles(kindpats)
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
   471
        self._prefix = _prefix(kindpats)
41669
a13268524c25 match: delete unused argument "listsubrepos" from _buildmatch()
Martin von Zweigbergk <martinvonz@google.com>
parents: 41668
diff changeset
   472
        self._pats, self.matchfn = _buildmatch(kindpats, '$', root)
8587
8f15d54437b9 match: fold match into _match base class
Matt Mackall <mpm@selenic.com>
parents: 8586
diff changeset
   473
32322
23c9a2a71c6e match: make _fileroots a @propertycache and rename it to _fileset
Martin von Zweigbergk <martinvonz@google.com>
parents: 32312
diff changeset
   474
    @propertycache
24636
36872036169b treemanifest: further optimize treemanifest.matches()
Drew Gottlieb <drgott@google.com>
parents: 24452
diff changeset
   475
    def _dirs(self):
32322
23c9a2a71c6e match: make _fileroots a @propertycache and rename it to _fileset
Martin von Zweigbergk <martinvonz@google.com>
parents: 32312
diff changeset
   476
        return set(util.dirs(self._fileset)) | {'.'}
24636
36872036169b treemanifest: further optimize treemanifest.matches()
Drew Gottlieb <drgott@google.com>
parents: 24452
diff changeset
   477
36872036169b treemanifest: further optimize treemanifest.matches()
Drew Gottlieb <drgott@google.com>
parents: 24452
diff changeset
   478
    def visitdir(self, dir):
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
   479
        if self._prefix and dir in self._fileset:
27343
c59647c6694d treemanifest: don't iterate entire matching submanifests on match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 27327
diff changeset
   480
            return 'all'
32554
f44ea253ffe2 match: optimize visitdir() for when no explicit files are listed
Martin von Zweigbergk <martinvonz@google.com>
parents: 32553
diff changeset
   481
        return ('.' in self._fileset or
32322
23c9a2a71c6e match: make _fileroots a @propertycache and rename it to _fileset
Martin von Zweigbergk <martinvonz@google.com>
parents: 32312
diff changeset
   482
                dir in self._fileset or
25576
d02f4b3e71f5 match: break boolean expressions into one operand per line
Martin von Zweigbergk <martinvonz@google.com>
parents: 25575
diff changeset
   483
                dir in self._dirs or
32322
23c9a2a71c6e match: make _fileroots a @propertycache and rename it to _fileset
Martin von Zweigbergk <martinvonz@google.com>
parents: 32312
diff changeset
   484
                any(parentdir in self._fileset
25577
a410479c7ee7 match: drop optimization (?) of 'parentdirs' calculation
Martin von Zweigbergk <martinvonz@google.com>
parents: 25576
diff changeset
   485
                    for parentdir in util.finddirs(dir)))
24636
36872036169b treemanifest: further optimize treemanifest.matches()
Drew Gottlieb <drgott@google.com>
parents: 24452
diff changeset
   486
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   487
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   488
        ret = self.visitdir(dir)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   489
        if ret is True:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   490
            return 'this'
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   491
        elif not ret:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   492
            return set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   493
        assert ret == 'all'
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   494
        return 'all'
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   495
33379
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   496
    def prefix(self):
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
   497
        return self._prefix
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
   498
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   499
    @encoding.strmethod
32406
952017471f93 match: implement __repr__() and update users (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32401
diff changeset
   500
    def __repr__(self):
38019
2f406142d7b4 py3: use pycompat.bytestr() on bytes before %r-ing it
Pulkit Goyal <7895pulkit@gmail.com>
parents: 37084
diff changeset
   501
        return ('<patternmatcher patterns=%r>' % pycompat.bytestr(self._pats))
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   502
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   503
# This is basically a reimplementation of util.dirs that stores the children
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   504
# instead of just a count of them, plus a small optional optimization to avoid
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   505
# some directories we don't need.
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   506
class _dirchildren(object):
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   507
    def __init__(self, paths, onlyinclude=None):
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   508
        self._dirs = {}
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   509
        self._onlyinclude = onlyinclude or []
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   510
        addpath = self.addpath
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   511
        for f in paths:
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   512
            addpath(f)
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   513
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   514
    def addpath(self, path):
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   515
        if path == '.':
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   516
            return
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   517
        dirs = self._dirs
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   518
        findsplitdirs = _dirchildren._findsplitdirs
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   519
        for d, b in findsplitdirs(path):
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   520
            if d not in self._onlyinclude:
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   521
                continue
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   522
            dirs.setdefault(d, set()).add(b)
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   523
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   524
    @staticmethod
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   525
    def _findsplitdirs(path):
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   526
        # yields (dirname, basename) tuples, walking back to the root.  This is
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   527
        # very similar to util.finddirs, except:
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   528
        #  - produces a (dirname, basename) tuple, not just 'dirname'
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   529
        #  - includes root dir
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   530
        # Unlike manifest._splittopdir, this does not suffix `dirname` with a
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   531
        # slash, and produces '.' for the root instead of ''.
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   532
        oldpos = len(path)
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   533
        pos = path.rfind('/')
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   534
        while pos != -1:
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   535
            yield path[:pos], path[pos + 1:oldpos]
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   536
            oldpos = pos
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   537
            pos = path.rfind('/', 0, pos)
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   538
        yield '.', path[:oldpos]
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   539
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   540
    def get(self, path):
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   541
        return self._dirs.get(path, set())
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   542
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   543
class includematcher(basematcher):
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   544
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   545
    def __init__(self, root, kindpats, badfn=None):
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   546
        super(includematcher, self).__init__(badfn)
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   547
41669
a13268524c25 match: delete unused argument "listsubrepos" from _buildmatch()
Martin von Zweigbergk <martinvonz@google.com>
parents: 41668
diff changeset
   548
        self._pats, self.matchfn = _buildmatch(kindpats, '(?:/|$)', root)
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
   549
        self._prefix = _prefix(kindpats)
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
   550
        roots, dirs, parents = _rootsdirsandparents(kindpats)
32502
3026f19b4b01 match: remove support for non-include patterns from includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32501
diff changeset
   551
        # roots are directories which are recursively included.
32503
361808a2b0b8 match: simplify includematcher a bit
Martin von Zweigbergk <martinvonz@google.com>
parents: 32502
diff changeset
   552
        self._roots = set(roots)
32502
3026f19b4b01 match: remove support for non-include patterns from includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32501
diff changeset
   553
        # dirs are directories which are non-recursively included.
32503
361808a2b0b8 match: simplify includematcher a bit
Martin von Zweigbergk <martinvonz@google.com>
parents: 32502
diff changeset
   554
        self._dirs = set(dirs)
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
   555
        # parents are directories which are non-recursively included because
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
   556
        # they are needed to get to items in _dirs or _roots.
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
   557
        self._parents = set(parents)
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   558
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   559
    def visitdir(self, dir):
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
   560
        if self._prefix and dir in self._roots:
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   561
            return 'all'
32503
361808a2b0b8 match: simplify includematcher a bit
Martin von Zweigbergk <martinvonz@google.com>
parents: 32502
diff changeset
   562
        return ('.' in self._roots or
361808a2b0b8 match: simplify includematcher a bit
Martin von Zweigbergk <martinvonz@google.com>
parents: 32502
diff changeset
   563
                dir in self._roots or
361808a2b0b8 match: simplify includematcher a bit
Martin von Zweigbergk <martinvonz@google.com>
parents: 32502
diff changeset
   564
                dir in self._dirs or
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
   565
                dir in self._parents or
32503
361808a2b0b8 match: simplify includematcher a bit
Martin von Zweigbergk <martinvonz@google.com>
parents: 32502
diff changeset
   566
                any(parentdir in self._roots
361808a2b0b8 match: simplify includematcher a bit
Martin von Zweigbergk <martinvonz@google.com>
parents: 32502
diff changeset
   567
                    for parentdir in util.finddirs(dir)))
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   568
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   569
    @propertycache
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   570
    def _allparentschildren(self):
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   571
        # It may seem odd that we add dirs, roots, and parents, and then
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   572
        # restrict to only parents. This is to catch the case of:
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   573
        #   dirs = ['foo/bar']
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   574
        #   parents = ['foo']
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   575
        # if we asked for the children of 'foo', but had only added
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   576
        # self._parents, we wouldn't be able to respond ['bar'].
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   577
        return _dirchildren(
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   578
                itertools.chain(self._dirs, self._roots, self._parents),
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   579
                onlyinclude=self._parents)
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   580
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   581
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   582
        if self._prefix and dir in self._roots:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   583
            return 'all'
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   584
        # Note: this does *not* include the 'dir in self._parents' case from
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   585
        # visitdir, that's handled below.
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   586
        if ('.' in self._roots or
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   587
            dir in self._roots or
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   588
            dir in self._dirs or
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   589
            any(parentdir in self._roots
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   590
                for parentdir in util.finddirs(dir))):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   591
            return 'this'
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   592
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   593
        if dir in self._parents:
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   594
            return self._allparentschildren.get(dir) or set()
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   595
        return set()
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   596
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   597
    @encoding.strmethod
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   598
    def __repr__(self):
36572
9adfa48792a7 match: some minimal pycompat fixes guided by test-hgignore.t
Augie Fackler <augie@google.com>
parents: 36050
diff changeset
   599
        return ('<includematcher includes=%r>' % pycompat.bytestr(self._pats))
32406
952017471f93 match: implement __repr__() and update users (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32401
diff changeset
   600
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   601
class exactmatcher(basematcher):
42080
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   602
    r'''Matches the input files exactly. They are interpreted as paths, not
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   603
    patterns (so no kind-prefixes).
42080
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   604
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   605
    >>> m = exactmatcher(['a.txt', 're:.*\.c$'])
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   606
    >>> m('a.txt')
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   607
    True
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   608
    >>> m('b.txt')
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   609
    False
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   610
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   611
    Input files that would be matched are exactly those returned by .files()
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   612
    >>> m.files()
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   613
    ['a.txt', 're:.*\\.c$']
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   614
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   615
    So pattern 're:.*\.c$' is not considered as a regex, but as a file name
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   616
    >>> m('main.c')
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   617
    False
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   618
    >>> m('re:.*\.c$')
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   619
    True
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   620
    '''
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   621
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   622
    def __init__(self, files, badfn=None):
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   623
        super(exactmatcher, self).__init__(badfn)
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   624
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   625
        if isinstance(files, list):
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   626
            self._files = files
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   627
        else:
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   628
            self._files = list(files)
32543
cf7c88986e9f match: define exactmatcher.matchfn statically
Yuya Nishihara <yuya@tcha.org>
parents: 32542
diff changeset
   629
cf7c88986e9f match: define exactmatcher.matchfn statically
Yuya Nishihara <yuya@tcha.org>
parents: 32542
diff changeset
   630
    matchfn = basematcher.exact
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   631
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   632
    @propertycache
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   633
    def _dirs(self):
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   634
        return set(util.dirs(self._fileset)) | {'.'}
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   635
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   636
    def visitdir(self, dir):
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   637
        return dir in self._dirs
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   638
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   639
    def visitchildrenset(self, dir):
39261
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   640
        if not self._fileset or dir not in self._dirs:
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   641
            return set()
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   642
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   643
        candidates = self._fileset | self._dirs - {'.'}
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   644
        if dir != '.':
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   645
            d = dir + '/'
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   646
            candidates = set(c[len(d):] for c in candidates if
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   647
                             c.startswith(d))
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   648
        # self._dirs includes all of the directories, recursively, so if
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   649
        # we're attempting to match foo/bar/baz.txt, it'll have '.', 'foo',
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   650
        # 'foo/bar' in it. Thus we can safely ignore a candidate that has a
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   651
        # '/' in it, indicating a it's for a subdir-of-a-subdir; the
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   652
        # immediate subdir will be in there without a slash.
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   653
        ret = {c for c in candidates if '/' not in c}
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   654
        # We really do not expect ret to be empty, since that would imply that
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   655
        # there's something in _dirs that didn't have a file in _fileset.
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   656
        assert ret
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   657
        return ret
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   658
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   659
    def isexact(self):
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   660
        return True
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   661
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   662
    @encoding.strmethod
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   663
    def __repr__(self):
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   664
        return ('<exactmatcher files=%r>' % self._files)
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   665
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   666
class differencematcher(basematcher):
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   667
    '''Composes two matchers by matching if the first matches and the second
35659
821d8a5ab4ff match: do not weirdly include explicit files excluded by -X option
Yuya Nishihara <yuya@tcha.org>
parents: 35164
diff changeset
   668
    does not.
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   669
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   670
    The second matcher's non-matching-attributes (bad, explicitdir,
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   671
    traversedir) are ignored.
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   672
    '''
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   673
    def __init__(self, m1, m2):
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   674
        super(differencematcher, self).__init__()
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   675
        self._m1 = m1
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   676
        self._m2 = m2
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   677
        self.bad = m1.bad
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   678
        self.explicitdir = m1.explicitdir
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   679
        self.traversedir = m1.traversedir
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   680
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   681
    def matchfn(self, f):
35659
821d8a5ab4ff match: do not weirdly include explicit files excluded by -X option
Yuya Nishihara <yuya@tcha.org>
parents: 35164
diff changeset
   682
        return self._m1(f) and not self._m2(f)
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   683
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   684
    @propertycache
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   685
    def _files(self):
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   686
        if self.isexact():
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   687
            return [f for f in self._m1.files() if self(f)]
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   688
        # If m1 is not an exact matcher, we can't easily figure out the set of
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   689
        # files, because its files() are not always files. For example, if
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   690
        # m1 is "path:dir" and m2 is "rootfileins:.", we don't
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   691
        # want to remove "dir" from the set even though it would match m2,
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   692
        # because the "dir" in m1 may not be a file.
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   693
        return self._m1.files()
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   694
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   695
    def visitdir(self, dir):
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   696
        if self._m2.visitdir(dir) == 'all':
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   697
            return False
41528
b7a0efb3c370 match: teach diffmatcher.visitdir() to return 'all' if possible
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 41282
diff changeset
   698
        elif not self._m2.visitdir(dir):
b7a0efb3c370 match: teach diffmatcher.visitdir() to return 'all' if possible
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 41282
diff changeset
   699
            # m2 does not match dir, we can return 'all' here if possible
b7a0efb3c370 match: teach diffmatcher.visitdir() to return 'all' if possible
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 41282
diff changeset
   700
            return self._m1.visitdir(dir)
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   701
        return bool(self._m1.visitdir(dir))
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   702
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   703
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   704
        m2_set = self._m2.visitchildrenset(dir)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   705
        if m2_set == 'all':
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   706
            return set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   707
        m1_set = self._m1.visitchildrenset(dir)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   708
        # Possible values for m1: 'all', 'this', set(...), set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   709
        # Possible values for m2:        'this', set(...), set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   710
        # If m2 has nothing under here that we care about, return m1, even if
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   711
        # it's 'all'. This is a change in behavior from visitdir, which would
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   712
        # return True, not 'all', for some reason.
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   713
        if not m2_set:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   714
            return m1_set
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   715
        if m1_set in ['all', 'this']:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   716
            # Never return 'all' here if m2_set is any kind of non-empty (either
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   717
            # 'this' or set(foo)), since m2 might return set() for a
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   718
            # subdirectory.
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   719
            return 'this'
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   720
        # Possible values for m1:         set(...), set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   721
        # Possible values for m2: 'this', set(...)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   722
        # We ignore m2's set results. They're possibly incorrect:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   723
        #  m1 = path:dir/subdir, m2=rootfilesin:dir, visitchildrenset('.'):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   724
        #    m1 returns {'dir'}, m2 returns {'dir'}, if we subtracted we'd
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   725
        #    return set(), which is *not* correct, we still need to visit 'dir'!
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   726
        return m1_set
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   727
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   728
    def isexact(self):
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   729
        return self._m1.isexact()
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   730
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   731
    @encoding.strmethod
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   732
    def __repr__(self):
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   733
        return ('<differencematcher m1=%r, m2=%r>' % (self._m1, self._m2))
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   734
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   735
def intersectmatchers(m1, m2):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   736
    '''Composes two matchers by matching if both of them match.
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   737
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   738
    The second matcher's non-matching-attributes (bad, explicitdir,
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   739
    traversedir) are ignored.
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   740
    '''
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   741
    if m1 is None or m2 is None:
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   742
        return m1 or m2
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   743
    if m1.always():
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   744
        m = copy.copy(m2)
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   745
        # TODO: Consider encapsulating these things in a class so there's only
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   746
        # one thing to copy from m1.
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   747
        m.bad = m1.bad
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   748
        m.explicitdir = m1.explicitdir
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   749
        m.traversedir = m1.traversedir
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   750
        return m
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   751
    if m2.always():
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   752
        m = copy.copy(m1)
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   753
        return m
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   754
    return intersectionmatcher(m1, m2)
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   755
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   756
class intersectionmatcher(basematcher):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   757
    def __init__(self, m1, m2):
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   758
        super(intersectionmatcher, self).__init__()
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   759
        self._m1 = m1
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   760
        self._m2 = m2
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   761
        self.bad = m1.bad
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   762
        self.explicitdir = m1.explicitdir
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   763
        self.traversedir = m1.traversedir
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   764
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   765
    @propertycache
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   766
    def _files(self):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   767
        if self.isexact():
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   768
            m1, m2 = self._m1, self._m2
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   769
            if not m1.isexact():
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   770
                m1, m2 = m2, m1
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   771
            return [f for f in m1.files() if m2(f)]
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   772
        # It neither m1 nor m2 is an exact matcher, we can't easily intersect
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   773
        # the set of files, because their files() are not always files. For
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   774
        # example, if intersecting a matcher "-I glob:foo.txt" with matcher of
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   775
        # "path:dir2", we don't want to remove "dir2" from the set.
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   776
        return self._m1.files() + self._m2.files()
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   777
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   778
    def matchfn(self, f):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   779
        return self._m1(f) and self._m2(f)
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   780
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   781
    def visitdir(self, dir):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   782
        visit1 = self._m1.visitdir(dir)
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   783
        if visit1 == 'all':
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   784
            return self._m2.visitdir(dir)
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   785
        # bool() because visit1=True + visit2='all' should not be 'all'
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   786
        return bool(visit1 and self._m2.visitdir(dir))
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   787
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   788
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   789
        m1_set = self._m1.visitchildrenset(dir)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   790
        if not m1_set:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   791
            return set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   792
        m2_set = self._m2.visitchildrenset(dir)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   793
        if not m2_set:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   794
            return set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   795
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   796
        if m1_set == 'all':
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   797
            return m2_set
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   798
        elif m2_set == 'all':
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   799
            return m1_set
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   800
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   801
        if m1_set == 'this' or m2_set == 'this':
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   802
            return 'this'
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   803
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   804
        assert isinstance(m1_set, set) and isinstance(m2_set, set)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   805
        return m1_set.intersection(m2_set)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   806
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   807
    def always(self):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   808
        return self._m1.always() and self._m2.always()
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   809
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   810
    def isexact(self):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   811
        return self._m1.isexact() or self._m2.isexact()
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   812
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   813
    @encoding.strmethod
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   814
    def __repr__(self):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   815
        return ('<intersectionmatcher m1=%r, m2=%r>' % (self._m1, self._m2))
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   816
32460
f9445b528687 match: make subdirmatcher extend basematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32459
diff changeset
   817
class subdirmatcher(basematcher):
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   818
    """Adapt a matcher to work on a subdirectory only.
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   819
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   820
    The paths are remapped to remove/insert the path as needed:
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   821
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
   822
    >>> from . import pycompat
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
   823
    >>> m1 = match(b'root', b'', [b'a.txt', b'sub/b.txt'])
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
   824
    >>> m2 = subdirmatcher(b'sub', m1)
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
   825
    >>> bool(m2(b'a.txt'))
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   826
    False
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
   827
    >>> bool(m2(b'b.txt'))
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   828
    True
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
   829
    >>> bool(m2.matchfn(b'a.txt'))
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   830
    False
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
   831
    >>> bool(m2.matchfn(b'b.txt'))
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   832
    True
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   833
    >>> m2.files()
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   834
    ['b.txt']
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
   835
    >>> m2.exact(b'b.txt')
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   836
    True
12268
83aaeba32b88 narrowmatcher: propagate bad method
Martin Geisler <mg@lazybytes.net>
parents: 12267
diff changeset
   837
    >>> def bad(f, msg):
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
   838
    ...     print(pycompat.sysstr(b"%s: %s" % (f, msg)))
12268
83aaeba32b88 narrowmatcher: propagate bad method
Martin Geisler <mg@lazybytes.net>
parents: 12267
diff changeset
   839
    >>> m1.bad = bad
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
   840
    >>> m2.bad(b'x.txt', b'No such file')
12268
83aaeba32b88 narrowmatcher: propagate bad method
Martin Geisler <mg@lazybytes.net>
parents: 12267
diff changeset
   841
    sub/x.txt: No such file
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   842
    """
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   843
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   844
    def __init__(self, path, matcher):
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   845
        super(subdirmatcher, self).__init__()
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   846
        self._path = path
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   847
        self._matcher = matcher
32460
f9445b528687 match: make subdirmatcher extend basematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32459
diff changeset
   848
        self._always = matcher.always()
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   849
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   850
        self._files = [f[len(path) + 1:] for f in matcher._files
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   851
                       if f.startswith(path + "/")]
25194
ef4538ba67ef match: explicitly naming a subrepo implies always() for the submatcher
Matt Harbison <matt_harbison@yahoo.com>
parents: 24790
diff changeset
   852
32325
763d72925691 match: use match.prefix() in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32324
diff changeset
   853
        # If the parent repo had a path to this subrepo and the matcher is
763d72925691 match: use match.prefix() in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32324
diff changeset
   854
        # a prefix matcher, this submatcher always matches.
763d72925691 match: use match.prefix() in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32324
diff changeset
   855
        if matcher.prefix():
25195
472a685a4961 merge with stable
Matt Mackall <mpm@selenic.com>
parents: 25189 25194
diff changeset
   856
            self._always = any(f == path for f in matcher._files)
25194
ef4538ba67ef match: explicitly naming a subrepo implies always() for the submatcher
Matt Harbison <matt_harbison@yahoo.com>
parents: 24790
diff changeset
   857
32324
77dac8fd30ee match: avoid accessing match._pathrestricted from subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32323
diff changeset
   858
    def bad(self, f, msg):
77dac8fd30ee match: avoid accessing match._pathrestricted from subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32323
diff changeset
   859
        self._matcher.bad(self._path + "/" + f, msg)
77dac8fd30ee match: avoid accessing match._pathrestricted from subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32323
diff changeset
   860
32464
2e80a691e575 match: override matchfn() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32463
diff changeset
   861
    def matchfn(self, f):
2e80a691e575 match: override matchfn() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32463
diff changeset
   862
        # Some information is lost in the superclass's constructor, so we
2e80a691e575 match: override matchfn() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32463
diff changeset
   863
        # can not accurately create the matching function for the subdirectory
2e80a691e575 match: override matchfn() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32463
diff changeset
   864
        # from the inputs. Instead, we override matchfn() and visitdir() to
2e80a691e575 match: override matchfn() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32463
diff changeset
   865
        # call the original matcher with the subdirectory path prepended.
2e80a691e575 match: override matchfn() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32463
diff changeset
   866
        return self._matcher.matchfn(self._path + "/" + f)
2e80a691e575 match: override matchfn() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32463
diff changeset
   867
32323
0aa4032a97e1 match: override visitdir() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32322
diff changeset
   868
    def visitdir(self, dir):
0aa4032a97e1 match: override visitdir() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32322
diff changeset
   869
        if dir == '.':
0aa4032a97e1 match: override visitdir() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32322
diff changeset
   870
            dir = self._path
0aa4032a97e1 match: override visitdir() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32322
diff changeset
   871
        else:
0aa4032a97e1 match: override visitdir() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32322
diff changeset
   872
            dir = self._path + "/" + dir
0aa4032a97e1 match: override visitdir() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32322
diff changeset
   873
        return self._matcher.visitdir(dir)
0aa4032a97e1 match: override visitdir() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32322
diff changeset
   874
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   875
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   876
        if dir == '.':
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   877
            dir = self._path
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   878
        else:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   879
            dir = self._path + "/" + dir
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   880
        return self._matcher.visitchildrenset(dir)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   881
32460
f9445b528687 match: make subdirmatcher extend basematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32459
diff changeset
   882
    def always(self):
f9445b528687 match: make subdirmatcher extend basematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32459
diff changeset
   883
        return self._always
f9445b528687 match: make subdirmatcher extend basematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32459
diff changeset
   884
33379
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   885
    def prefix(self):
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   886
        return self._matcher.prefix() and not self._always
32460
f9445b528687 match: make subdirmatcher extend basematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32459
diff changeset
   887
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   888
    @encoding.strmethod
32552
e7aa11f3abcd match: add __repr__ for subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32543
diff changeset
   889
    def __repr__(self):
e7aa11f3abcd match: add __repr__ for subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32543
diff changeset
   890
        return ('<subdirmatcher path=%r, matcher=%r>' %
e7aa11f3abcd match: add __repr__ for subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32543
diff changeset
   891
                (self._path, self._matcher))
e7aa11f3abcd match: add __repr__ for subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32543
diff changeset
   892
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   893
class prefixdirmatcher(basematcher):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   894
    """Adapt a matcher to work on a parent directory.
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   895
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   896
    The matcher's non-matching-attributes (bad, explicitdir, traversedir) are
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   897
    ignored.
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   898
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   899
    The prefix path should usually be the relative path from the root of
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   900
    this matcher to the root of the wrapped matcher.
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   901
38750
a8bfaf592033 doctest: convert matcher root to native path
Yuya Nishihara <yuya@tcha.org>
parents: 38612
diff changeset
   902
    >>> m1 = match(util.localpath(b'root/d/e'), b'f', [b'../a.txt', b'b.txt'])
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   903
    >>> m2 = prefixdirmatcher(b'd/e', m1)
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   904
    >>> bool(m2(b'a.txt'),)
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   905
    False
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   906
    >>> bool(m2(b'd/e/a.txt'))
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   907
    True
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   908
    >>> bool(m2(b'd/e/b.txt'))
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   909
    False
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   910
    >>> m2.files()
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   911
    ['d/e/a.txt', 'd/e/f/b.txt']
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   912
    >>> m2.exact(b'd/e/a.txt')
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   913
    True
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   914
    >>> m2.visitdir(b'd')
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   915
    True
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   916
    >>> m2.visitdir(b'd/e')
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   917
    True
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   918
    >>> m2.visitdir(b'd/e/f')
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   919
    True
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   920
    >>> m2.visitdir(b'd/e/g')
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   921
    False
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   922
    >>> m2.visitdir(b'd/ef')
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   923
    False
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   924
    """
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   925
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   926
    def __init__(self, path, matcher, badfn=None):
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   927
        super(prefixdirmatcher, self).__init__(badfn)
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   928
        if not path:
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   929
            raise error.ProgrammingError('prefix path must not be empty')
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   930
        self._path = path
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   931
        self._pathprefix = path + '/'
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   932
        self._matcher = matcher
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   933
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   934
    @propertycache
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   935
    def _files(self):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   936
        return [self._pathprefix + f for f in self._matcher._files]
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   937
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   938
    def matchfn(self, f):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   939
        if not f.startswith(self._pathprefix):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   940
            return False
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   941
        return self._matcher.matchfn(f[len(self._pathprefix):])
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   942
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   943
    @propertycache
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   944
    def _pathdirs(self):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   945
        return set(util.finddirs(self._path)) | {'.'}
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   946
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   947
    def visitdir(self, dir):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   948
        if dir == self._path:
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   949
            return self._matcher.visitdir('.')
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   950
        if dir.startswith(self._pathprefix):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   951
            return self._matcher.visitdir(dir[len(self._pathprefix):])
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   952
        return dir in self._pathdirs
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   953
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   954
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   955
        if dir == self._path:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   956
            return self._matcher.visitchildrenset('.')
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   957
        if dir.startswith(self._pathprefix):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   958
            return self._matcher.visitchildrenset(dir[len(self._pathprefix):])
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   959
        if dir in self._pathdirs:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   960
            return 'this'
38958
b9f94d67ea73 match: add missing "return set()", add FIXME to test to doc a bug
Kyle Lippincott <spectral@google.com>
parents: 38957
diff changeset
   961
        return set()
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   962
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   963
    def isexact(self):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   964
        return self._matcher.isexact()
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   965
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   966
    def prefix(self):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   967
        return self._matcher.prefix()
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   968
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   969
    @encoding.strmethod
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   970
    def __repr__(self):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   971
        return ('<prefixdirmatcher path=%r, matcher=%r>'
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   972
                % (pycompat.bytestr(self._path), self._matcher))
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
   973
33319
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
   974
class unionmatcher(basematcher):
33448
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
   975
    """A matcher that is the union of several matchers.
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
   976
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   977
    The non-matching-attributes (bad, explicitdir, traversedir) are taken from
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   978
    the first matcher.
33448
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
   979
    """
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
   980
33319
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
   981
    def __init__(self, matchers):
33448
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
   982
        m1 = matchers[0]
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   983
        super(unionmatcher, self).__init__()
33448
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
   984
        self.explicitdir = m1.explicitdir
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
   985
        self.traversedir = m1.traversedir
33319
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
   986
        self._matchers = matchers
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
   987
33380
892d255ec2a1 match: override matchfn instead of __call__ for consistency
Martin von Zweigbergk <martinvonz@google.com>
parents: 33379
diff changeset
   988
    def matchfn(self, f):
33319
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
   989
        for match in self._matchers:
33380
892d255ec2a1 match: override matchfn instead of __call__ for consistency
Martin von Zweigbergk <martinvonz@google.com>
parents: 33379
diff changeset
   990
            if match(f):
33319
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
   991
                return True
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
   992
        return False
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
   993
33448
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
   994
    def visitdir(self, dir):
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
   995
        r = False
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
   996
        for m in self._matchers:
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
   997
            v = m.visitdir(dir)
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
   998
            if v == 'all':
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
   999
                return v
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1000
            r |= v
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1001
        return r
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1002
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1003
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1004
        r = set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1005
        this = False
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1006
        for m in self._matchers:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1007
            v = m.visitchildrenset(dir)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1008
            if not v:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1009
                continue
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1010
            if v == 'all':
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1011
                return v
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1012
            if this or v == 'this':
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1013
                this = True
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1014
                # don't break, we might have an 'all' in here.
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1015
                continue
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1016
            assert isinstance(v, set)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1017
            r = r.union(v)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1018
        if this:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1019
            return 'this'
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1020
        return r
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1021
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
  1022
    @encoding.strmethod
33319
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
  1023
    def __repr__(self):
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
  1024
        return ('<unionmatcher matchers=%r>' % self._matchers)
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
  1025
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1026
def patkind(pattern, default=None):
42082
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1027
    '''If pattern is 'kind:pat' with a known kind, return kind.
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1028
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1029
    >>> patkind('re:.*\.c$')
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1030
    're'
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1031
    >>> patkind('glob:*.c')
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1032
    'glob'
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1033
    >>> patkind('relpath:test.py')
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1034
    'relpath'
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1035
    >>> patkind('main.py')
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1036
    >>> patkind('main.py', default='re')
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1037
    're'
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1038
    '''
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1039
    return _patsplit(pattern, default)[0]
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1040
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1041
def _patsplit(pattern, default):
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1042
    """Split a string into the optional pattern kind prefix and the actual
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1043
    pattern."""
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1044
    if ':' in pattern:
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1045
        kind, pat = pattern.split(':', 1)
33684
2be0bf186950 match: expose some data and functionality to other modules
Kostia Balytskyi <ikostia@fb.com>
parents: 33582
diff changeset
  1046
        if kind in allpatternkinds:
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1047
            return kind, pat
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1048
    return default, pattern
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1049
8582
a4c199e12b5a match: remove head and tail args from _globre
Matt Mackall <mpm@selenic.com>
parents: 8581
diff changeset
  1050
def _globre(pat):
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1051
    r'''Convert an extended glob string to a regexp string.
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1052
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1053
    >>> from . import pycompat
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1054
    >>> def bprint(s):
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1055
    ...     print(pycompat.sysstr(s))
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1056
    >>> bprint(_globre(br'?'))
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1057
    .
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1058
    >>> bprint(_globre(br'*'))
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1059
    [^/]*
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1060
    >>> bprint(_globre(br'**'))
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1061
    .*
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1062
    >>> bprint(_globre(br'**/a'))
21815
a4b67bf1f0a5 match: make glob '**/' match the empty string
Siddharth Agarwal <sid0@fb.com>
parents: 21191
diff changeset
  1063
    (?:.*/)?a
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1064
    >>> bprint(_globre(br'a/**/b'))
38475
67dc32d4e790 cleanup: migrate from re.escape to stringutil.reescape
Augie Fackler <augie@google.com>
parents: 38019
diff changeset
  1065
    a/(?:.*/)?b
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1066
    >>> bprint(_globre(br'[a*?!^][^b][!c]'))
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1067
    [a*?!^][\^b][^c]
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1068
    >>> bprint(_globre(br'{a,b}'))
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1069
    (?:a|b)
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1070
    >>> bprint(_globre(br'.\*\?'))
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1071
    \.\*\?
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1072
    '''
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1073
    i, n = 0, len(pat)
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1074
    res = ''
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1075
    group = 0
40684
e6c9ef5e11a0 match: provide and use a quick way to escape a single byte
Boris Feld <boris.feld@octobus.net>
parents: 40345
diff changeset
  1076
    escape = util.stringutil.regexbytesescapemap.get
10282
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10263
diff changeset
  1077
    def peek():
31421
5c9cda37d7f6 match: slice over bytes to get the byteschr instead of ascii value
Pulkit Goyal <7895pulkit@gmail.com>
parents: 31420
diff changeset
  1078
        return i < n and pat[i:i + 1]
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1079
    while i < n:
31421
5c9cda37d7f6 match: slice over bytes to get the byteschr instead of ascii value
Pulkit Goyal <7895pulkit@gmail.com>
parents: 31420
diff changeset
  1080
        c = pat[i:i + 1]
10282
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10263
diff changeset
  1081
        i += 1
8583
19d1b2aec562 match: optimize escaping in _globre
Matt Mackall <mpm@selenic.com>
parents: 8582
diff changeset
  1082
        if c not in '*?[{},\\':
40684
e6c9ef5e11a0 match: provide and use a quick way to escape a single byte
Boris Feld <boris.feld@octobus.net>
parents: 40345
diff changeset
  1083
            res += escape(c, c)
8583
19d1b2aec562 match: optimize escaping in _globre
Matt Mackall <mpm@selenic.com>
parents: 8582
diff changeset
  1084
        elif c == '*':
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1085
            if peek() == '*':
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1086
                i += 1
21815
a4b67bf1f0a5 match: make glob '**/' match the empty string
Siddharth Agarwal <sid0@fb.com>
parents: 21191
diff changeset
  1087
                if peek() == '/':
a4b67bf1f0a5 match: make glob '**/' match the empty string
Siddharth Agarwal <sid0@fb.com>
parents: 21191
diff changeset
  1088
                    i += 1
a4b67bf1f0a5 match: make glob '**/' match the empty string
Siddharth Agarwal <sid0@fb.com>
parents: 21191
diff changeset
  1089
                    res += '(?:.*/)?'
a4b67bf1f0a5 match: make glob '**/' match the empty string
Siddharth Agarwal <sid0@fb.com>
parents: 21191
diff changeset
  1090
                else:
a4b67bf1f0a5 match: make glob '**/' match the empty string
Siddharth Agarwal <sid0@fb.com>
parents: 21191
diff changeset
  1091
                    res += '.*'
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1092
            else:
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1093
                res += '[^/]*'
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1094
        elif c == '?':
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1095
            res += '.'
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1096
        elif c == '[':
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1097
            j = i
31421
5c9cda37d7f6 match: slice over bytes to get the byteschr instead of ascii value
Pulkit Goyal <7895pulkit@gmail.com>
parents: 31420
diff changeset
  1098
            if j < n and pat[j:j + 1] in '!]':
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1099
                j += 1
31421
5c9cda37d7f6 match: slice over bytes to get the byteschr instead of ascii value
Pulkit Goyal <7895pulkit@gmail.com>
parents: 31420
diff changeset
  1100
            while j < n and pat[j:j + 1] != ']':
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1101
                j += 1
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1102
            if j >= n:
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1103
                res += '\\['
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1104
            else:
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1105
                stuff = pat[i:j].replace('\\','\\\\')
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1106
                i = j + 1
31421
5c9cda37d7f6 match: slice over bytes to get the byteschr instead of ascii value
Pulkit Goyal <7895pulkit@gmail.com>
parents: 31420
diff changeset
  1107
                if stuff[0:1] == '!':
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1108
                    stuff = '^' + stuff[1:]
31421
5c9cda37d7f6 match: slice over bytes to get the byteschr instead of ascii value
Pulkit Goyal <7895pulkit@gmail.com>
parents: 31420
diff changeset
  1109
                elif stuff[0:1] == '^':
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1110
                    stuff = '\\' + stuff
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1111
                res = '%s[%s]' % (res, stuff)
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1112
        elif c == '{':
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1113
            group += 1
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1114
            res += '(?:'
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1115
        elif c == '}' and group:
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1116
            res += ')'
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1117
            group -= 1
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1118
        elif c == ',' and group:
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1119
            res += '|'
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1120
        elif c == '\\':
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1121
            p = peek()
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1122
            if p:
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1123
                i += 1
40684
e6c9ef5e11a0 match: provide and use a quick way to escape a single byte
Boris Feld <boris.feld@octobus.net>
parents: 40345
diff changeset
  1124
                res += escape(p, p)
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1125
            else:
40684
e6c9ef5e11a0 match: provide and use a quick way to escape a single byte
Boris Feld <boris.feld@octobus.net>
parents: 40345
diff changeset
  1126
                res += escape(c, c)
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1127
        else:
40684
e6c9ef5e11a0 match: provide and use a quick way to escape a single byte
Boris Feld <boris.feld@octobus.net>
parents: 40345
diff changeset
  1128
            res += escape(c, c)
8582
a4c199e12b5a match: remove head and tail args from _globre
Matt Mackall <mpm@selenic.com>
parents: 8581
diff changeset
  1129
    return res
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1130
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1131
def _regex(kind, pat, globsuffix):
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1132
    '''Convert a (normalized) pattern of any kind into a regular expression.
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1133
    globsuffix is appended to the regexp of globs.'''
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1134
    if not pat:
8574
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1135
        return ''
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1136
    if kind == 're':
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1137
        return pat
33358
38b6122df5c7 match: combine regex code for path: and relpath:
Martin von Zweigbergk <martinvonz@google.com>
parents: 33357
diff changeset
  1138
    if kind in ('path', 'relpath'):
25636
bfe9ed85f27c match: let 'path:.' and 'path:' match everything (issue4687)
Matt Harbison <matt_harbison@yahoo.com>
parents: 25194
diff changeset
  1139
        if pat == '.':
bfe9ed85f27c match: let 'path:.' and 'path:' match everything (issue4687)
Matt Harbison <matt_harbison@yahoo.com>
parents: 25194
diff changeset
  1140
            return ''
38475
67dc32d4e790 cleanup: migrate from re.escape to stringutil.reescape
Augie Fackler <augie@google.com>
parents: 38019
diff changeset
  1141
        return util.stringutil.reescape(pat) + '(?:/|$)'
31012
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1142
    if kind == 'rootfilesin':
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1143
        if pat == '.':
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1144
            escaped = ''
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1145
        else:
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1146
            # Pattern is a directory name.
38475
67dc32d4e790 cleanup: migrate from re.escape to stringutil.reescape
Augie Fackler <augie@google.com>
parents: 38019
diff changeset
  1147
            escaped = util.stringutil.reescape(pat) + '/'
31012
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1148
        # Anything after the pattern must be a non-directory.
33357
a21819f439fe match: remove unnecessary '^' from regexes
Martin von Zweigbergk <martinvonz@google.com>
parents: 33319
diff changeset
  1149
        return escaped + '[^/]+$'
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1150
    if kind == 'relglob':
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1151
        return '(?:|.*/)' + _globre(pat) + globsuffix
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1152
    if kind == 'relre':
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1153
        if pat.startswith('^'):
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1154
            return pat
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1155
        return '.*' + pat
41282
4fab8a7d2d72 match: support rooted globs in hgignore
Valentin Gatien-Baron <vgatien-baron@janestreet.com>
parents: 41129
diff changeset
  1156
    if kind in ('glob', 'rootglob'):
38578
2d487b9cac07 match: explode if unsupported pattern passed down to _regex() builder
Yuya Nishihara <yuya@tcha.org>
parents: 38577
diff changeset
  1157
        return _globre(pat) + globsuffix
2d487b9cac07 match: explode if unsupported pattern passed down to _regex() builder
Yuya Nishihara <yuya@tcha.org>
parents: 38577
diff changeset
  1158
    raise error.ProgrammingError('not a regex pattern: %s:%s' % (kind, pat))
8574
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1159
41669
a13268524c25 match: delete unused argument "listsubrepos" from _buildmatch()
Martin von Zweigbergk <martinvonz@google.com>
parents: 41668
diff changeset
  1160
def _buildmatch(kindpats, globsuffix, root):
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1161
    '''Return regexp string and a matcher function for kindpats.
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1162
    globsuffix is appended to the regexp of globs.'''
25239
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1163
    matchfuncs = []
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1164
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
  1165
    subincludes, kindpats = _expandsubinclude(kindpats, root)
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
  1166
    if subincludes:
32132
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1167
        submatchers = {}
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
  1168
        def matchsubinclude(f):
32132
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1169
            for prefix, matcherargs in subincludes:
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1170
                if f.startswith(prefix):
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1171
                    mf = submatchers.get(prefix)
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1172
                    if mf is None:
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1173
                        mf = match(*matcherargs)
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1174
                        submatchers[prefix] = mf
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1175
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1176
                    if mf(f[len(prefix):]):
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1177
                        return True
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
  1178
            return False
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
  1179
        matchfuncs.append(matchsubinclude)
14675
cfc89398f710 match: introduce basic fileset support
Matt Mackall <mpm@selenic.com>
parents: 14674
diff changeset
  1180
25239
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1181
    regex = ''
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1182
    if kindpats:
40242
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1183
        if all(k == 'rootfilesin' for k, p, s in kindpats):
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1184
            dirs = {p for k, p, s in kindpats}
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1185
            def mf(f):
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1186
                i = f.rfind('/')
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1187
                if i >= 0:
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1188
                    dir = f[:i]
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1189
                else:
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1190
                    dir = '.'
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1191
                return dir in dirs
40345
d30a19d10441 match: fix up a repr to not crash on Python 3
Augie Fackler <augie@google.com>
parents: 40242
diff changeset
  1192
            regex = b'rootfilesin: %s' % stringutil.pprint(list(sorted(dirs)))
40242
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1193
            matchfuncs.append(mf)
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1194
        else:
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1195
            regex, mf = _buildregexmatch(kindpats, globsuffix)
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1196
            matchfuncs.append(mf)
25239
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1197
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1198
    if len(matchfuncs) == 1:
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1199
        return regex, matchfuncs[0]
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1200
    else:
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1201
        return regex, lambda f: any(mf(f) for mf in matchfuncs)
14675
cfc89398f710 match: introduce basic fileset support
Matt Mackall <mpm@selenic.com>
parents: 14674
diff changeset
  1202
40774
8306dac48061 match: extract a literal constant into a symbolic one
Boris Feld <boris.feld@octobus.net>
parents: 40757
diff changeset
  1203
MAX_RE_SIZE = 20000
8306dac48061 match: extract a literal constant into a symbolic one
Boris Feld <boris.feld@octobus.net>
parents: 40757
diff changeset
  1204
40776
ce401300f981 match: extract function that group regexps
Boris Feld <boris.feld@octobus.net>
parents: 40775
diff changeset
  1205
def _joinregexes(regexps):
ce401300f981 match: extract function that group regexps
Boris Feld <boris.feld@octobus.net>
parents: 40775
diff changeset
  1206
    """gather multiple regular expressions into a single one"""
40782
3984409e144b match: drop unnecessary wrapping of regex in group
Martin von Zweigbergk <martinvonz@google.com>
parents: 40781
diff changeset
  1207
    return '|'.join(regexps)
40776
ce401300f981 match: extract function that group regexps
Boris Feld <boris.feld@octobus.net>
parents: 40775
diff changeset
  1208
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1209
def _buildregexmatch(kindpats, globsuffix):
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1210
    """Build a match function from a list of kinds and kindpats,
40775
4e02f25f31c6 match: test for overflow error in pattern
Boris Feld <boris.feld@octobus.net>
parents: 40774
diff changeset
  1211
    return regexp string and a matcher function.
4e02f25f31c6 match: test for overflow error in pattern
Boris Feld <boris.feld@octobus.net>
parents: 40774
diff changeset
  1212
4e02f25f31c6 match: test for overflow error in pattern
Boris Feld <boris.feld@octobus.net>
parents: 40774
diff changeset
  1213
    Test too large input
4e02f25f31c6 match: test for overflow error in pattern
Boris Feld <boris.feld@octobus.net>
parents: 40774
diff changeset
  1214
    >>> _buildregexmatch([
40947
9e462fb88f79 match: fix doctest to use bytes instead of str
Augie Fackler <augie@google.com>
parents: 40782
diff changeset
  1215
    ...     (b'relglob', b'?' * MAX_RE_SIZE, b'')
9e462fb88f79 match: fix doctest to use bytes instead of str
Augie Fackler <augie@google.com>
parents: 40782
diff changeset
  1216
    ... ], b'$')
40775
4e02f25f31c6 match: test for overflow error in pattern
Boris Feld <boris.feld@octobus.net>
parents: 40774
diff changeset
  1217
    Traceback (most recent call last):
4e02f25f31c6 match: test for overflow error in pattern
Boris Feld <boris.feld@octobus.net>
parents: 40774
diff changeset
  1218
    ...
40778
69bd3176da7c match: raise an Abort error instead of OverflowError
Boris Feld <boris.feld@octobus.net>
parents: 40777
diff changeset
  1219
    Abort: matcher pattern is too long (20009 bytes)
40775
4e02f25f31c6 match: test for overflow error in pattern
Boris Feld <boris.feld@octobus.net>
parents: 40774
diff changeset
  1220
    """
8574
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1221
    try:
40777
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1222
        allgroups = []
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1223
        regexps = [_regex(k, p, globsuffix) for (k, p, s) in kindpats]
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1224
        fullregexp = _joinregexes(regexps)
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1225
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1226
        startidx = 0
40782
3984409e144b match: drop unnecessary wrapping of regex in group
Martin von Zweigbergk <martinvonz@google.com>
parents: 40781
diff changeset
  1227
        groupsize = 0
40777
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1228
        for idx, r in enumerate(regexps):
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1229
            piecesize = len(r)
40782
3984409e144b match: drop unnecessary wrapping of regex in group
Martin von Zweigbergk <martinvonz@google.com>
parents: 40781
diff changeset
  1230
            if piecesize > MAX_RE_SIZE:
40778
69bd3176da7c match: raise an Abort error instead of OverflowError
Boris Feld <boris.feld@octobus.net>
parents: 40777
diff changeset
  1231
                msg = _("matcher pattern is too long (%d bytes)") % piecesize
69bd3176da7c match: raise an Abort error instead of OverflowError
Boris Feld <boris.feld@octobus.net>
parents: 40777
diff changeset
  1232
                raise error.Abort(msg)
40780
1e019f45fa88 match: make "groupsize" include the trailing "|"
Martin von Zweigbergk <martinvonz@google.com>
parents: 40779
diff changeset
  1233
            elif (groupsize + piecesize) > MAX_RE_SIZE:
40777
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1234
                group = regexps[startidx:idx]
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1235
                allgroups.append(_joinregexes(group))
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1236
                startidx = idx
40782
3984409e144b match: drop unnecessary wrapping of regex in group
Martin von Zweigbergk <martinvonz@google.com>
parents: 40781
diff changeset
  1237
                groupsize = 0
40777
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1238
            groupsize += piecesize + 1
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1239
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1240
        if startidx == 0:
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1241
            func = _rematcher(fullregexp)
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1242
        else:
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1243
            group = regexps[startidx:]
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1244
            allgroups.append(_joinregexes(group))
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1245
            allmatchers = [_rematcher(g) for g in allgroups]
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1246
            func = lambda s: any(m(s) for m in allmatchers)
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1247
        return fullregexp, func
8574
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1248
    except re.error:
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
  1249
        for k, p, s in kindpats:
8574
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1250
            try:
40782
3984409e144b match: drop unnecessary wrapping of regex in group
Martin von Zweigbergk <martinvonz@google.com>
parents: 40781
diff changeset
  1251
                _rematcher(_regex(k, p, globsuffix))
8574
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1252
            except re.error:
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
  1253
                if s:
26587
56b2bcea2529 error: get Abort from 'error' instead of 'util'
Pierre-Yves David <pierre-yves.david@fb.com>
parents: 26014
diff changeset
  1254
                    raise error.Abort(_("%s: invalid pattern (%s): %s") %
40779
e8c992d56465 match: fix an unaligned (but harmless) indent
Martin von Zweigbergk <martinvonz@google.com>
parents: 40778
diff changeset
  1255
                                      (s, k, p))
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
  1256
                else:
26587
56b2bcea2529 error: get Abort from 'error' instead of 'util'
Pierre-Yves David <pierre-yves.david@fb.com>
parents: 26014
diff changeset
  1257
                    raise error.Abort(_("invalid pattern (%s): %s") % (k, p))
56b2bcea2529 error: get Abort from 'error' instead of 'util'
Pierre-Yves David <pierre-yves.david@fb.com>
parents: 26014
diff changeset
  1258
        raise error.Abort(_("invalid pattern"))
8574
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1259
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1260
def _patternrootsanddirs(kindpats):
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1261
    '''Returns roots and directories corresponding to each pattern.
21079
b02ab6486a78 match: make it more clear what _roots do and that it ends up in match()._files
Mads Kiilerich <madski@unity3d.com>
parents: 20401
diff changeset
  1262
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1263
    This calculates the roots and directories exactly matching the patterns and
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1264
    returns a tuple of (roots, dirs) for each. It does not return other
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1265
    directories which may also need to be considered, like the parent
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1266
    directories.
21079
b02ab6486a78 match: make it more clear what _roots do and that it ends up in match()._files
Mads Kiilerich <madski@unity3d.com>
parents: 20401
diff changeset
  1267
    '''
8576
ec4ed21db4b2 match: split up _normalizepats
Matt Mackall <mpm@selenic.com>
parents: 8575
diff changeset
  1268
    r = []
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1269
    d = []
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
  1270
    for kind, pat, source in kindpats:
41282
4fab8a7d2d72 match: support rooted globs in hgignore
Valentin Gatien-Baron <vgatien-baron@janestreet.com>
parents: 41129
diff changeset
  1271
        if kind in ('glob', 'rootglob'): # find the non-glob prefix
8584
0f06e72abfdc match: fold _globprefix into _roots
Matt Mackall <mpm@selenic.com>
parents: 8583
diff changeset
  1272
            root = []
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1273
            for p in pat.split('/'):
8584
0f06e72abfdc match: fold _globprefix into _roots
Matt Mackall <mpm@selenic.com>
parents: 8583
diff changeset
  1274
                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
  1275
                    break
0f06e72abfdc match: fold _globprefix into _roots
Matt Mackall <mpm@selenic.com>
parents: 8583
diff changeset
  1276
                root.append(p)
0f06e72abfdc match: fold _globprefix into _roots
Matt Mackall <mpm@selenic.com>
parents: 8583
diff changeset
  1277
            r.append('/'.join(root) or '.')
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1278
        elif kind in ('relpath', 'path'):
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1279
            r.append(pat or '.')
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1280
        elif kind in ('rootfilesin',):
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1281
            d.append(pat or '.')
19107
fcf08023c011 match: fix root calculation for combining regexps with simple paths
Mads Kiilerich <madski@unity3d.com>
parents: 18713
diff changeset
  1282
        else: # relglob, re, relre
8576
ec4ed21db4b2 match: split up _normalizepats
Matt Mackall <mpm@selenic.com>
parents: 8575
diff changeset
  1283
            r.append('.')
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1284
    return r, d
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1285
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1286
def _roots(kindpats):
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1287
    '''Returns root directories to match recursively from the given patterns.'''
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1288
    roots, dirs = _patternrootsanddirs(kindpats)
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1289
    return roots
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1290
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1291
def _rootsdirsandparents(kindpats):
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1292
    '''Returns roots and exact directories from patterns.
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1293
38957
f356be1a7ba3 match: correct doc for _rootsdirsandparents after 5a7df82de142
Kyle Lippincott <spectral@google.com>
parents: 38955
diff changeset
  1294
    `roots` are directories to match recursively, `dirs` should
f356be1a7ba3 match: correct doc for _rootsdirsandparents after 5a7df82de142
Kyle Lippincott <spectral@google.com>
parents: 38955
diff changeset
  1295
    be matched non-recursively, and `parents` are the implicitly required
f356be1a7ba3 match: correct doc for _rootsdirsandparents after 5a7df82de142
Kyle Lippincott <spectral@google.com>
parents: 38955
diff changeset
  1296
    directories to walk to items in either roots or dirs.
f356be1a7ba3 match: correct doc for _rootsdirsandparents after 5a7df82de142
Kyle Lippincott <spectral@google.com>
parents: 38955
diff changeset
  1297
f356be1a7ba3 match: correct doc for _rootsdirsandparents after 5a7df82de142
Kyle Lippincott <spectral@google.com>
parents: 38955
diff changeset
  1298
    Returns a tuple of (roots, dirs, parents).
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1299
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1300
    >>> _rootsdirsandparents(
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1301
    ...     [(b'glob', b'g/h/*', b''), (b'glob', b'g/h', b''),
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1302
    ...      (b'glob', b'g*', b'')])
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1303
    (['g/h', 'g/h', '.'], [], ['g', '.'])
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1304
    >>> _rootsdirsandparents(
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1305
    ...     [(b'rootfilesin', b'g/h', b''), (b'rootfilesin', b'', b'')])
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1306
    ([], ['g/h', '.'], ['g', '.'])
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1307
    >>> _rootsdirsandparents(
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1308
    ...     [(b'relpath', b'r', b''), (b'path', b'p/p', b''),
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1309
    ...      (b'path', b'', b'')])
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1310
    (['r', 'p/p', '.'], [], ['p', '.'])
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1311
    >>> _rootsdirsandparents(
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1312
    ...     [(b'relglob', b'rg*', b''), (b're', b're/', b''),
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1313
    ...      (b'relre', b'rr', b'')])
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1314
    (['.', '.', '.'], [], ['.'])
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1315
    '''
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1316
    r, d = _patternrootsanddirs(kindpats)
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1317
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1318
    p = []
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1319
    # Append the parents as non-recursive/exact directories, since they must be
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1320
    # scanned to get to either the roots or the other exact directories.
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1321
    p.extend(util.dirs(d))
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1322
    p.extend(util.dirs(r))
32176
cf042543afa2 match: optimize visitdir() for patterns matching only root directory
Martin von Zweigbergk <martinvonz@google.com>
parents: 32132
diff changeset
  1323
    # util.dirs() does not include the root directory, so add it manually
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1324
    p.append('.')
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1325
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
  1326
    # FIXME: all uses of this function convert these to sets, do so before
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
  1327
    # returning.
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
  1328
    # FIXME: all uses of this function do not need anything in 'roots' and
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
  1329
    # 'dirs' to also be in 'parents', consider removing them before returning.
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1330
    return r, d, p
8576
ec4ed21db4b2 match: split up _normalizepats
Matt Mackall <mpm@selenic.com>
parents: 8575
diff changeset
  1331
31012
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1332
def _explicitfiles(kindpats):
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1333
    '''Returns the potential explicit filenames from the patterns.
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1334
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1335
    >>> _explicitfiles([(b'path', b'foo/bar', b'')])
31012
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1336
    ['foo/bar']
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1337
    >>> _explicitfiles([(b'rootfilesin', b'foo/bar', b'')])
31012
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1338
    []
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1339
    '''
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1340
    # Keep only the pattern kinds where one can specify filenames (vs only
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1341
    # directory names).
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1342
    filable = [kp for kp in kindpats if kp[0] not in ('rootfilesin',)]
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1343
    return _roots(filable)
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1344
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
  1345
def _prefix(kindpats):
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
  1346
    '''Whether all the patterns match a prefix (i.e. recursively)'''
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
  1347
    for kind, pat, source in kindpats:
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
  1348
        if kind not in ('path', 'relpath'):
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
  1349
            return False
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
  1350
    return True
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1351
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1352
_commentre = None
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1353
27595
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1354
def readpatternfile(filepath, warn, sourceinfo=False):
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1355
    '''parse a pattern file, returning a list of
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1356
    patterns. These patterns should be given to compile()
25216
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1357
    to be validated and converted into a match function.
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1358
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1359
    trailing white space is dropped.
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1360
    the escape character is backslash.
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1361
    comments start with #.
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1362
    empty lines are skipped.
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1363
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1364
    lines can be of the following formats:
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1365
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1366
    syntax: regexp # defaults following lines to non-rooted regexps
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1367
    syntax: glob   # defaults following lines to non-rooted globs
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1368
    re:pattern     # non-rooted regular expression
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1369
    glob:pattern   # non-rooted glob
41282
4fab8a7d2d72 match: support rooted globs in hgignore
Valentin Gatien-Baron <vgatien-baron@janestreet.com>
parents: 41129
diff changeset
  1370
    rootglob:pat   # rooted glob (same root as ^ in regexps)
27595
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1371
    pattern        # pattern of the current default type
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1372
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1373
    if sourceinfo is set, returns a list of tuples:
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1374
    (pattern, lineno, originalline). This is useful to debug ignore patterns.
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1375
    '''
25216
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1376
40685
e41f6c2e69c4 match: reformat `syntaxes` dictionary for better maintainability
Boris Feld <boris.feld@octobus.net>
parents: 40684
diff changeset
  1377
    syntaxes = {
e41f6c2e69c4 match: reformat `syntaxes` dictionary for better maintainability
Boris Feld <boris.feld@octobus.net>
parents: 40684
diff changeset
  1378
        're': 'relre:',
e41f6c2e69c4 match: reformat `syntaxes` dictionary for better maintainability
Boris Feld <boris.feld@octobus.net>
parents: 40684
diff changeset
  1379
        'regexp': 'relre:',
e41f6c2e69c4 match: reformat `syntaxes` dictionary for better maintainability
Boris Feld <boris.feld@octobus.net>
parents: 40684
diff changeset
  1380
        'glob': 'relglob:',
41282
4fab8a7d2d72 match: support rooted globs in hgignore
Valentin Gatien-Baron <vgatien-baron@janestreet.com>
parents: 41129
diff changeset
  1381
        'rootglob': 'rootglob:',
40685
e41f6c2e69c4 match: reformat `syntaxes` dictionary for better maintainability
Boris Feld <boris.feld@octobus.net>
parents: 40684
diff changeset
  1382
        'include': 'include',
e41f6c2e69c4 match: reformat `syntaxes` dictionary for better maintainability
Boris Feld <boris.feld@octobus.net>
parents: 40684
diff changeset
  1383
        'subinclude': 'subinclude',
e41f6c2e69c4 match: reformat `syntaxes` dictionary for better maintainability
Boris Feld <boris.feld@octobus.net>
parents: 40684
diff changeset
  1384
    }
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1385
    syntax = 'relre:'
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1386
    patterns = []
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1387
31403
10c17f8bfcf3 py3: open file in rb mode
Rishabh Madan <rishabhmadan96@gmail.com>
parents: 31392
diff changeset
  1388
    fp = open(filepath, 'rb')
30399
7f3593c29473 match: migrate to util.iterfile
Jun Wu <quark@fb.com>
parents: 29803
diff changeset
  1389
    for lineno, line in enumerate(util.iterfile(fp), start=1):
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1390
        if "#" in line:
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1391
            global _commentre
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1392
            if not _commentre:
31420
40704098853f match: make regular expression bytes to prevent TypeError
Pulkit Goyal <7895pulkit@gmail.com>
parents: 31403
diff changeset
  1393
                _commentre = util.re.compile(br'((?:^|[^\\])(?:\\\\)*)#.*')
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1394
            # remove comments prefixed by an even number of escapes
27327
d500341e4f55 match: use re2 in readpatternfile if possible
Bryan O'Sullivan <bos@serpentine.com>
parents: 26781
diff changeset
  1395
            m = _commentre.search(line)
d500341e4f55 match: use re2 in readpatternfile if possible
Bryan O'Sullivan <bos@serpentine.com>
parents: 26781
diff changeset
  1396
            if m:
d500341e4f55 match: use re2 in readpatternfile if possible
Bryan O'Sullivan <bos@serpentine.com>
parents: 26781
diff changeset
  1397
                line = line[:m.end(1)]
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1398
            # fixup properly escaped comments that survived the above
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1399
            line = line.replace("\\#", "#")
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1400
        line = line.rstrip()
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1401
        if not line:
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1402
            continue
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1403
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1404
        if line.startswith('syntax:'):
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1405
            s = line[7:].strip()
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1406
            try:
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1407
                syntax = syntaxes[s]
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1408
            except KeyError:
25214
08703b10c3ae match: add optional warn argument
Durham Goode <durham@fb.com>
parents: 25213
diff changeset
  1409
                if warn:
08703b10c3ae match: add optional warn argument
Durham Goode <durham@fb.com>
parents: 25213
diff changeset
  1410
                    warn(_("%s: ignoring invalid syntax '%s'\n") %
08703b10c3ae match: add optional warn argument
Durham Goode <durham@fb.com>
parents: 25213
diff changeset
  1411
                         (filepath, s))
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1412
            continue
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1413
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1414
        linesyntax = syntax
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1415
        for s, rels in syntaxes.iteritems():
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1416
            if line.startswith(rels):
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1417
                linesyntax = rels
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1418
                line = line[len(rels):]
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1419
                break
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1420
            elif line.startswith(s+':'):
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1421
                linesyntax = rels
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1422
                line = line[len(s) + 1:]
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1423
                break
27595
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1424
        if sourceinfo:
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1425
            patterns.append((linesyntax + line, lineno, line))
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1426
        else:
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1427
            patterns.append(linesyntax + line)
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1428
    fp.close()
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1429
    return patterns