mercurial/match.py
author Raphaël Gomès <rgomes@octobus.net>
Fri, 24 Jan 2020 11:06:01 +0100
changeset 44530 4d1634e59f13
parent 44452 9d2b2df2c2ba
child 45682 d2e1dcd4490d
permissions -rw-r--r--
rust-status: use bare hg status fastpath from Python This change also adds a test case for subrepos. Repeating the benchmark information from the `hg-core` commit: On the Netbeans repository: C: 840ms Rust+C: 556ms Mozilla Central with the one pattern that causes a fallback removed: C: 2.315s Rust+C: 1.700s Differential Revision: https://phab.mercurial-scm.org/D7931
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 _
43085
eef9a2d67051 py3: manually import pycompat.open into files that need it
Gregory Szorc <gregory.szorc@gmail.com>
parents: 43077
diff changeset
    16
from .pycompat import open
25958
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
    17
from . import (
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
    18
    encoding,
26587
56b2bcea2529 error: get Abort from 'error' instead of 'util'
Pierre-Yves David <pierre-yves.david@fb.com>
parents: 26014
diff changeset
    19
    error,
25958
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
    20
    pathutil,
42452
a3a8887e4426 rust: using policy.importrust from Python callers
Georges Racinet <georges.racinet@octobus.net>
parents: 42367
diff changeset
    21
    policy,
36572
9adfa48792a7 match: some minimal pycompat fixes guided by test-hgignore.t
Augie Fackler <augie@google.com>
parents: 36050
diff changeset
    22
    pycompat,
25958
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
    23
    util,
c4ccf2d394a7 match: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 25875
diff changeset
    24
)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
    25
from .utils import stringutil
6576
69f3e9ac7c56 walk: introduce match objects
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
    26
44137
3bd77c64bc74 rust-filepatterns: remove bridge code for filepatterns-related functions
Raphaël Gomès <rgomes@octobus.net>
parents: 44009
diff changeset
    27
rustmod = policy.importrust('dirstate')
42329
c7652f7440d9 rust-filepatterns: call new Rust implementations from Python
Raphaël Gomès <rgomes@octobus.net>
parents: 42195
diff changeset
    28
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
    29
allpatternkinds = (
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    30
    b're',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    31
    b'glob',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    32
    b'path',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    33
    b'relglob',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    34
    b'relpath',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    35
    b'relre',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    36
    b'rootglob',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    37
    b'listfile',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    38
    b'listfile0',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    39
    b'set',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    40
    b'include',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    41
    b'subinclude',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    42
    b'rootfilesin',
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
    43
)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    44
cwdrelativepatternkinds = (b'relpath', b'glob')
33684
2be0bf186950 match: expose some data and functionality to other modules
Kostia Balytskyi <ikostia@fb.com>
parents: 33582
diff changeset
    45
24636
36872036169b treemanifest: further optimize treemanifest.matches()
Drew Gottlieb <drgott@google.com>
parents: 24452
diff changeset
    46
propertycache = util.propertycache
36872036169b treemanifest: further optimize treemanifest.matches()
Drew Gottlieb <drgott@google.com>
parents: 24452
diff changeset
    47
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
    48
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
    49
def _rematcher(regex):
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
    50
    '''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
    51
    matcher function'''
21909
335bb8b80443 match: use util.re.compile instead of util.compilere
Siddharth Agarwal <sid0@fb.com>
parents: 21815
diff changeset
    52
    m = util.re.compile(regex)
16943
8d08a28aa63e matcher: use re2 bindings if available
Bryan O'Sullivan <bryano@fb.com>
parents: 16791
diff changeset
    53
    try:
8d08a28aa63e matcher: use re2 bindings if available
Bryan O'Sullivan <bryano@fb.com>
parents: 16791
diff changeset
    54
        # 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
    55
        return m.test_match
8d08a28aa63e matcher: use re2 bindings if available
Bryan O'Sullivan <bryano@fb.com>
parents: 16791
diff changeset
    56
    except AttributeError:
8d08a28aa63e matcher: use re2 bindings if available
Bryan O'Sullivan <bryano@fb.com>
parents: 16791
diff changeset
    57
        return m.match
8d08a28aa63e matcher: use re2 bindings if available
Bryan O'Sullivan <bryano@fb.com>
parents: 16791
diff changeset
    58
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
    59
44009
e685fac56693 match: resolve filesets against the passed `cwd`, not the current one
Matt Harbison <matt_harbison@yahoo.com>
parents: 44002
diff changeset
    60
def _expandsets(cwd, kindpats, ctx=None, listsubrepos=False, badfn=None):
38612
760cc5dc01e8 fileset: restrict getfileset() to not return a computed set (API)
Yuya Nishihara <yuya@tcha.org>
parents: 38611
diff changeset
    61
    '''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
    62
    matchers = []
14675
cfc89398f710 match: introduce basic fileset support
Matt Mackall <mpm@selenic.com>
parents: 14674
diff changeset
    63
    other = []
cfc89398f710 match: introduce basic fileset support
Matt Mackall <mpm@selenic.com>
parents: 14674
diff changeset
    64
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
    65
    for kind, pat, source in kindpats:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    66
        if kind == b'set':
41022
481249481392 match: fix assertion for fileset with no context (issue6046)
Yuya Nishihara <yuya@tcha.org>
parents: 40345
diff changeset
    67
            if ctx is None:
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
    68
                raise error.ProgrammingError(
43117
8ff1ecfadcd1 cleanup: join string literals that are already on one line
Martin von Zweigbergk <martinvonz@google.com>
parents: 43106
diff changeset
    69
                    b"fileset expression with no context"
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
    70
                )
44009
e685fac56693 match: resolve filesets against the passed `cwd`, not the current one
Matt Harbison <matt_harbison@yahoo.com>
parents: 44002
diff changeset
    71
            matchers.append(ctx.matchfileset(cwd, 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
    72
755d23a49170 match: resolve filesets in subrepos for commands given the '-S' argument
Matt Harbison <matt_harbison@yahoo.com>
parents: 25114
diff changeset
    73
            if listsubrepos:
755d23a49170 match: resolve filesets in subrepos for commands given the '-S' argument
Matt Harbison <matt_harbison@yahoo.com>
parents: 25114
diff changeset
    74
                for subpath in ctx.substate:
44009
e685fac56693 match: resolve filesets against the passed `cwd`, not the current one
Matt Harbison <matt_harbison@yahoo.com>
parents: 44002
diff changeset
    75
                    sm = ctx.sub(subpath).matchfileset(cwd, 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
    76
                    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
    77
                    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
    78
14675
cfc89398f710 match: introduce basic fileset support
Matt Mackall <mpm@selenic.com>
parents: 14674
diff changeset
    79
            continue
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
    80
        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
    81
    return matchers, other
14675
cfc89398f710 match: introduce basic fileset support
Matt Mackall <mpm@selenic.com>
parents: 14674
diff changeset
    82
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
    83
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    84
def _expandsubinclude(kindpats, root):
32132
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
    85
    '''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
    86
    subincludes in it.'''
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    87
    relmatchers = []
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    88
    other = []
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    89
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    90
    for kind, pat, source in kindpats:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    91
        if kind == b'subinclude':
25301
caaf4045eca8 match: normpath the ignore source when expanding the 'subinclude' kind
Matt Harbison <matt_harbison@yahoo.com>
parents: 25283
diff changeset
    92
            sourceroot = pathutil.dirname(util.normpath(source))
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    93
            pat = util.pconvert(pat)
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    94
            path = pathutil.join(sourceroot, pat)
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    95
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    96
            newroot = pathutil.dirname(path)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
    97
            matcherargs = (newroot, b'', [], [b'include:%s' % path])
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    98
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
    99
            prefix = pathutil.canonpath(root, root, newroot)
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
   100
            if prefix:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   101
                prefix += b'/'
32132
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
   102
            relmatchers.append((prefix, matcherargs))
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
   103
        else:
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
   104
            other.append((kind, pat, source))
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
   105
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
   106
    return relmatchers, other
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
   107
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   108
24447
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
   109
def _kindpatsalwaysmatch(kindpats):
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
   110
    """"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
   111
    'relpath:.' does.
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
   112
    """
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
   113
    for kind, pat, source in kindpats:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   114
        if pat != b'' or kind not in [b'relpath', b'glob']:
24447
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
   115
            return False
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
   116
    return True
d44d53bc9a1e matcher: make e.g. 'relpath:.' lead to fast paths
Martin von Zweigbergk <martinvonz@google.com>
parents: 23686
diff changeset
   117
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   118
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   119
def _buildkindpatsmatcher(
44009
e685fac56693 match: resolve filesets against the passed `cwd`, not the current one
Matt Harbison <matt_harbison@yahoo.com>
parents: 44002
diff changeset
   120
    matchercls, root, cwd, kindpats, ctx=None, listsubrepos=False, badfn=None,
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   121
):
38580
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   122
    matchers = []
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   123
    fms, kindpats = _expandsets(
44009
e685fac56693 match: resolve filesets against the passed `cwd`, not the current one
Matt Harbison <matt_harbison@yahoo.com>
parents: 44002
diff changeset
   124
        cwd, kindpats, ctx=ctx, listsubrepos=listsubrepos, badfn=badfn,
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   125
    )
38580
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   126
    if kindpats:
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   127
        m = matchercls(root, kindpats, badfn=badfn)
38580
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   128
        matchers.append(m)
38612
760cc5dc01e8 fileset: restrict getfileset() to not return a computed set (API)
Yuya Nishihara <yuya@tcha.org>
parents: 38611
diff changeset
   129
    if fms:
760cc5dc01e8 fileset: restrict getfileset() to not return a computed set (API)
Yuya Nishihara <yuya@tcha.org>
parents: 38611
diff changeset
   130
        matchers.extend(fms)
38580
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   131
    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
   132
        return nevermatcher(badfn=badfn)
38580
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   133
    if len(matchers) == 1:
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   134
        return matchers[0]
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   135
    return unionmatcher(matchers)
9f9ffe5f687c match: compose 'set:' pattern as matcher
Yuya Nishihara <yuya@tcha.org>
parents: 38579
diff changeset
   136
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   137
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   138
def match(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   139
    root,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   140
    cwd,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   141
    patterns=None,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   142
    include=None,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   143
    exclude=None,
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   144
    default=b'glob',
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   145
    auditor=None,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   146
    ctx=None,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   147
    listsubrepos=False,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   148
    warn=None,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   149
    badfn=None,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   150
    icasefs=False,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   151
):
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   152
    r"""build an object to match a set of file patterns
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   153
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   154
    arguments:
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   155
    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
   156
    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
   157
    patterns - patterns to find
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   158
    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
   159
    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
   160
    default - if a pattern in patterns has no explicit type, assume this one
42083
bee1647578b7 match: complete documentation of match() parameters
Denis Laxalde <denis@laxalde.org>
parents: 42082
diff changeset
   161
    auditor - optional path auditor
bee1647578b7 match: complete documentation of match() parameters
Denis Laxalde <denis@laxalde.org>
parents: 42082
diff changeset
   162
    ctx - optional changecontext
bee1647578b7 match: complete documentation of match() parameters
Denis Laxalde <denis@laxalde.org>
parents: 42082
diff changeset
   163
    listsubrepos - if True, recurse into subrepositories
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   164
    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
   165
    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
   166
    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
   167
        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
   168
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   169
    a pattern is one of:
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   170
    '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
   171
    're:<regexp>' - a regular expression
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   172
    '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
   173
                    recursively
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   174
    '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
   175
                    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
   176
    '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
   177
    '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
   178
    '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
   179
    'set:<fileset>' - a fileset expression
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   180
    '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
   181
    '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
   182
                          the same directory
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   183
    '<something>' - a pattern of the specified default type
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   184
43964
8f67735344ae tests: convert the `root` arg of matchmod.match() to local path separators
Matt Harbison <matt_harbison@yahoo.com>
parents: 43950
diff changeset
   185
    >>> def _match(root, *args, **kwargs):
8f67735344ae tests: convert the `root` arg of matchmod.match() to local path separators
Matt Harbison <matt_harbison@yahoo.com>
parents: 43950
diff changeset
   186
    ...     return match(util.localpath(root), *args, **kwargs)
8f67735344ae tests: convert the `root` arg of matchmod.match() to local path separators
Matt Harbison <matt_harbison@yahoo.com>
parents: 43950
diff changeset
   187
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   188
    Usually a patternmatcher is returned:
43964
8f67735344ae tests: convert the `root` arg of matchmod.match() to local path separators
Matt Harbison <matt_harbison@yahoo.com>
parents: 43950
diff changeset
   189
    >>> _match(b'/foo', b'.', [b're:.*\.c$', b'path:foo/a', b'*.py'])
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   190
    <patternmatcher patterns='.*\\.c$|foo/a(?:/|$)|[^/]*\\.py$'>
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   191
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   192
    Combining 'patterns' with 'include' (resp. 'exclude') gives an
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   193
    intersectionmatcher (resp. a differencematcher):
43964
8f67735344ae tests: convert the `root` arg of matchmod.match() to local path separators
Matt Harbison <matt_harbison@yahoo.com>
parents: 43950
diff changeset
   194
    >>> type(_match(b'/foo', b'.', [b're:.*\.c$'], include=[b'path:lib']))
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   195
    <class 'mercurial.match.intersectionmatcher'>
43964
8f67735344ae tests: convert the `root` arg of matchmod.match() to local path separators
Matt Harbison <matt_harbison@yahoo.com>
parents: 43950
diff changeset
   196
    >>> type(_match(b'/foo', b'.', [b're:.*\.c$'], exclude=[b'path:build']))
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   197
    <class 'mercurial.match.differencematcher'>
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   198
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   199
    Notice that, if 'patterns' is empty, an alwaysmatcher is returned:
43964
8f67735344ae tests: convert the `root` arg of matchmod.match() to local path separators
Matt Harbison <matt_harbison@yahoo.com>
parents: 43950
diff changeset
   200
    >>> _match(b'/foo', b'.', [])
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   201
    <alwaysmatcher>
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   202
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   203
    The 'default' argument determines which kind of pattern is assumed if a
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   204
    pattern has no prefix:
43964
8f67735344ae tests: convert the `root` arg of matchmod.match() to local path separators
Matt Harbison <matt_harbison@yahoo.com>
parents: 43950
diff changeset
   205
    >>> _match(b'/foo', b'.', [b'.*\.c$'], default=b're')
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   206
    <patternmatcher patterns='.*\\.c$'>
43964
8f67735344ae tests: convert the `root` arg of matchmod.match() to local path separators
Matt Harbison <matt_harbison@yahoo.com>
parents: 43950
diff changeset
   207
    >>> _match(b'/foo', b'.', [b'main.py'], default=b'relpath')
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   208
    <patternmatcher patterns='main\\.py(?:/|$)'>
43964
8f67735344ae tests: convert the `root` arg of matchmod.match() to local path separators
Matt Harbison <matt_harbison@yahoo.com>
parents: 43950
diff changeset
   209
    >>> _match(b'/foo', b'.', [b'main.py'], default=b're')
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   210
    <patternmatcher patterns='main.py'>
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   211
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   212
    The primary use of matchers is to check whether a value (usually a file
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   213
    name) matches againset one of the patterns given at initialization. There
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   214
    are two ways of doing this check.
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   215
43964
8f67735344ae tests: convert the `root` arg of matchmod.match() to local path separators
Matt Harbison <matt_harbison@yahoo.com>
parents: 43950
diff changeset
   216
    >>> m = _match(b'/foo', b'', [b're:.*\.c$', b'relpath:a'])
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   217
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   218
    1. Calling the matcher with a file name returns True if any pattern
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   219
    matches that file name:
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   220
    >>> m(b'a')
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   221
    True
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   222
    >>> m(b'main.c')
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   223
    True
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   224
    >>> m(b'test.py')
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   225
    False
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   226
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   227
    2. Using the exact() method only returns True if the file name matches one
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   228
    of the exact patterns (i.e. not re: or glob: patterns):
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   229
    >>> m.exact(b'a')
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   230
    True
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   231
    >>> m.exact(b'main.c')
42084
42537dfc7a7c match: add doctest examples in match()
Denis Laxalde <denis@laxalde.org>
parents: 42083
diff changeset
   232
    False
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   233
    """
43949
8b1a9ba375e5 match: make sure `root` argument is always an absolute path (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 43735
diff changeset
   234
    assert os.path.isabs(root)
43965
8a81fa44f7bb match: don't util.normpath() cwd
Matt Harbison <matt_harbison@yahoo.com>
parents: 43964
diff changeset
   235
    cwd = os.path.join(root, util.localpath(cwd))
32401
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   236
    normalize = _donormalize
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   237
    if icasefs:
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   238
        dirstate = ctx.repo().dirstate
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   239
        dsnormalize = dirstate.normalize
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   240
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   241
        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
   242
            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
   243
            kindpats = []
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   244
            for kind, pats, source in kp:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   245
                if kind not in (b're', b'relre'):  # regex can't be normalized
32401
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   246
                    p = pats
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   247
                    pats = dsnormalize(pats)
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   248
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   249
                    # 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
   250
                    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
   251
                        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
   252
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   253
                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
   254
            return kindpats
284b18303f61 match: replace icasefsmatch() function by flag to regular match()
Martin von Zweigbergk <martinvonz@google.com>
parents: 32400
diff changeset
   255
41622
635a12c53ea6 match: remove unused "exact" argument (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41528
diff changeset
   256
    if patterns:
32556
5f08eca8f8d3 match: move normalize() call out of matcher constructors
Martin von Zweigbergk <martinvonz@google.com>
parents: 32555
diff changeset
   257
        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
   258
        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
   259
            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
   260
        else:
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   261
            m = _buildkindpatsmatcher(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   262
                patternmatcher,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   263
                root,
44009
e685fac56693 match: resolve filesets against the passed `cwd`, not the current one
Matt Harbison <matt_harbison@yahoo.com>
parents: 44002
diff changeset
   264
                cwd,
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   265
                kindpats,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   266
                ctx=ctx,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   267
                listsubrepos=listsubrepos,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   268
                badfn=badfn,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   269
            )
32553
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   270
    else:
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   271
        # 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
   272
        # 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
   273
        m = alwaysmatcher(badfn)
32553
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   274
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   275
    if include:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   276
        kindpats = normalize(include, b'glob', root, cwd, auditor, warn)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   277
        im = _buildkindpatsmatcher(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   278
            includematcher,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   279
            root,
44009
e685fac56693 match: resolve filesets against the passed `cwd`, not the current one
Matt Harbison <matt_harbison@yahoo.com>
parents: 44002
diff changeset
   280
            cwd,
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   281
            kindpats,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   282
            ctx=ctx,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   283
            listsubrepos=listsubrepos,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   284
            badfn=None,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   285
        )
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   286
        m = intersectmatchers(m, im)
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   287
    if exclude:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   288
        kindpats = normalize(exclude, b'glob', root, cwd, auditor, warn)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   289
        em = _buildkindpatsmatcher(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   290
            includematcher,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   291
            root,
44009
e685fac56693 match: resolve filesets against the passed `cwd`, not the current one
Matt Harbison <matt_harbison@yahoo.com>
parents: 44002
diff changeset
   292
            cwd,
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   293
            kindpats,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   294
            ctx=ctx,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   295
            listsubrepos=listsubrepos,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   296
            badfn=None,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   297
        )
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   298
        m = differencematcher(m, em)
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   299
    return m
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   300
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   301
41676
0531dff73d0b match: delete unused root and cwd arguments from {always,never,exact}() (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41675
diff changeset
   302
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
   303
    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
   304
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   305
41676
0531dff73d0b match: delete unused root and cwd arguments from {always,never,exact}() (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41675
diff changeset
   306
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
   307
    return alwaysmatcher(badfn)
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   308
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   309
41676
0531dff73d0b match: delete unused root and cwd arguments from {always,never,exact}() (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41675
diff changeset
   310
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
   311
    return nevermatcher(badfn)
32605
e6ff007e107e match: introduce nevermatcher for when no ignore files are present
Siddharth Agarwal <sid0@fb.com>
parents: 32557
diff changeset
   312
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   313
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   314
def badmatch(match, badfn):
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   315
    """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
   316
    one.
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   317
    """
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   318
    m = copy.copy(match)
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   319
    m.bad = badfn
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   320
    return m
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   321
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   322
42085
54e6d7ef5ca5 match: make _donormalize's auditor and warn arguments optional
Denis Laxalde <denis@laxalde.org>
parents: 42084
diff changeset
   323
def _donormalize(patterns, default, root, cwd, auditor=None, warn=None):
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   324
    '''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
   325
    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
   326
    kindpats = []
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   327
    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
   328
        if kind in cwdrelativepatternkinds:
42085
54e6d7ef5ca5 match: make _donormalize's auditor and warn arguments optional
Denis Laxalde <denis@laxalde.org>
parents: 42084
diff changeset
   329
            pat = pathutil.canonpath(root, cwd, pat, auditor=auditor)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   330
        elif kind in (b'relglob', b'path', b'rootfilesin', b'rootglob'):
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   331
            pat = util.normpath(pat)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   332
        elif kind in (b'listfile', b'listfile0'):
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   333
            try:
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   334
                files = util.readfile(pat)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   335
                if kind == b'listfile0':
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   336
                    files = files.split(b'\0')
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   337
                else:
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   338
                    files = files.splitlines()
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   339
                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
   340
            except EnvironmentError:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   341
                raise error.Abort(_(b"unable to read file list (%s)") % pat)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   342
            for k, p, source in _donormalize(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   343
                files, default, root, cwd, auditor, warn
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   344
            ):
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   345
                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
   346
            continue
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   347
        elif kind == b'include':
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   348
            try:
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   349
                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
   350
                includepats = readpatternfile(fullpath, warn)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   351
                for k, p, source in _donormalize(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   352
                    includepats, default, root, cwd, auditor, warn
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   353
                ):
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   354
                    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
   355
            except error.Abort as inst:
43485
6ff196d2553b match: suppress error about subscripting an exception
Augie Fackler <augie@google.com>
parents: 43117
diff changeset
   356
                raise error.Abort(
6ff196d2553b match: suppress error about subscripting an exception
Augie Fackler <augie@google.com>
parents: 43117
diff changeset
   357
                    b'%s: %s'
6ff196d2553b match: suppress error about subscripting an exception
Augie Fackler <augie@google.com>
parents: 43117
diff changeset
   358
                    % (pat, inst[0])  # pytype: disable=unsupported-operands
6ff196d2553b match: suppress error about subscripting an exception
Augie Fackler <augie@google.com>
parents: 43117
diff changeset
   359
                )
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   360
            except IOError as inst:
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   361
                if warn:
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   362
                    warn(
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   363
                        _(b"skipping unreadable pattern file '%s': %s\n")
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   364
                        % (pat, stringutil.forcebytestr(inst.strerror))
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   365
                    )
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   366
            continue
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   367
        # else: re or relre - which cannot be normalized
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   368
        kindpats.append((kind, pat, b''))
32397
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   369
    return kindpats
0ec4cd6fe051 match: move body of _normalize() to a static function
Martin von Zweigbergk <martinvonz@google.com>
parents: 32396
diff changeset
   370
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   371
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   372
class basematcher(object):
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   373
    def __init__(self, badfn=None):
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   374
        if badfn is not None:
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   375
            self.bad = badfn
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   376
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   377
    def __call__(self, fn):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   378
        return self.matchfn(fn)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   379
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   380
    # 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
   381
    # 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
   382
    def bad(self, f, msg):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   383
        '''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
   384
        found/accessed, with an error message.'''
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   385
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   386
    # 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
   387
    # by recursive traversal is visited.
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   388
    traversedir = None
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   389
32459
9f781f43f2ce match: make basematcher._files a @propertycache
Martin von Zweigbergk <martinvonz@google.com>
parents: 32458
diff changeset
   390
    @propertycache
9f781f43f2ce match: make basematcher._files a @propertycache
Martin von Zweigbergk <martinvonz@google.com>
parents: 32458
diff changeset
   391
    def _files(self):
9f781f43f2ce match: make basematcher._files a @propertycache
Martin von Zweigbergk <martinvonz@google.com>
parents: 32458
diff changeset
   392
        return []
9f781f43f2ce match: make basematcher._files a @propertycache
Martin von Zweigbergk <martinvonz@google.com>
parents: 32458
diff changeset
   393
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   394
    def files(self):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   395
        '''Explicitly listed files or patterns or roots:
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   396
        if no patterns or .always(): empty list,
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   397
        if exact: list exact files,
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   398
        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
   399
        else: optimal roots'''
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   400
        return self._files
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   401
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   402
    @propertycache
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   403
    def _fileset(self):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   404
        return set(self._files)
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   405
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   406
    def exact(self, f):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   407
        '''Returns True if f is in .files().'''
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   408
        return f in self._fileset
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   409
32463
43e091847c4d match: make matchfn a method on the class
Martin von Zweigbergk <martinvonz@google.com>
parents: 32461
diff changeset
   410
    def matchfn(self, f):
43e091847c4d match: make matchfn a method on the class
Martin von Zweigbergk <martinvonz@google.com>
parents: 32461
diff changeset
   411
        return False
43e091847c4d match: make matchfn a method on the class
Martin von Zweigbergk <martinvonz@google.com>
parents: 32461
diff changeset
   412
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   413
    def visitdir(self, dir):
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   414
        '''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
   415
        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
   416
        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
   417
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   418
        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
   419
        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
   420
        the given directory should be visited.
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   421
        '''
33478
cf15c3cc304c match: make base matcher return True for visitdir
Durham Goode <durham@fb.com>
parents: 33448
diff changeset
   422
        return True
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   423
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   424
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   425
        '''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
   426
        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
   427
        potentially lists which subdirectories of that directory should be
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   428
        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
   429
        patterns.
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   430
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   431
        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
   432
        can be applied:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   433
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   434
             visitdir | visitchildrenlist
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   435
            ----------+-------------------
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   436
             False    | set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   437
             'all'    | 'all'
39260
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   438
             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
   439
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   440
        Example:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   441
          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
   442
          the following values (assuming the implementation of visitchildrenset
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   443
          is capable of recognizing this; some implementations are not).
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   444
42341
27d6956d386b match: use '' instead of '.' for root directory (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 42329
diff changeset
   445
          '' -> {'foo', 'qux'}
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   446
          'baz' -> set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   447
          'foo' -> {'bar'}
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   448
          # 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
   449
          # 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
   450
          # '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
   451
          # in.
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   452
          'foo/bar' -> 'this'
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   453
          'qux' -> 'this'
39260
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   454
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   455
        Important:
27946fca8a05 match: document that visitchildrenset might return files
Kyle Lippincott <spectral@google.com>
parents: 38958
diff changeset
   456
          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
   457
          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
   458
          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
   459
          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
   460
          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
   461
          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
   462
          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
   463
          will always return 'this').
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   464
        '''
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   465
        return b'this'
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   466
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   467
    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
   468
        '''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
   469
        optimization might be possible.'''
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   470
        return False
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   471
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   472
    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
   473
        '''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
   474
        optimization might be possible.'''
32458
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   475
        return False
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   476
a04bc55201c3 match: extract base class for matchers
Martin von Zweigbergk <martinvonz@google.com>
parents: 32444
diff changeset
   477
    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
   478
        '''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
   479
        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
   480
        return False
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   481
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   482
    def anypats(self):
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   483
        '''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
   484
        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
   485
        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
   486
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   487
32553
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   488
class alwaysmatcher(basematcher):
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   489
    '''Matches everything.'''
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   490
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   491
    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
   492
        super(alwaysmatcher, self).__init__(badfn)
32553
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   493
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   494
    def always(self):
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   495
        return True
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   496
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   497
    def matchfn(self, f):
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   498
        return True
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   499
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   500
    def visitdir(self, dir):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   501
        return b'all'
32553
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   502
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   503
    def visitchildrenset(self, dir):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   504
        return b'all'
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   505
32553
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   506
    def __repr__(self):
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   507
        return r'<alwaysmatcher>'
32553
20c9f3ecc192 match: handle everything-matching using new alwaysmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32552
diff changeset
   508
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   509
32605
e6ff007e107e match: introduce nevermatcher for when no ignore files are present
Siddharth Agarwal <sid0@fb.com>
parents: 32557
diff changeset
   510
class nevermatcher(basematcher):
e6ff007e107e match: introduce nevermatcher for when no ignore files are present
Siddharth Agarwal <sid0@fb.com>
parents: 32557
diff changeset
   511
    '''Matches nothing.'''
e6ff007e107e match: introduce nevermatcher for when no ignore files are present
Siddharth Agarwal <sid0@fb.com>
parents: 32557
diff changeset
   512
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   513
    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
   514
        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
   515
33378
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   516
    # 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
   517
    # 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
   518
    # 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
   519
    # 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
   520
    # 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
   521
    def isexact(self):
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   522
        return True
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   523
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   524
    def prefix(self):
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   525
        return True
adf95bfb423a match: make nevermatcher an exact matcher and a prefix matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33358
diff changeset
   526
33582
44bc181b9835 match: override visitdir() in nevermatcher to return False
Martin von Zweigbergk <martinvonz@google.com>
parents: 33497
diff changeset
   527
    def visitdir(self, dir):
44bc181b9835 match: override visitdir() in nevermatcher to return False
Martin von Zweigbergk <martinvonz@google.com>
parents: 33497
diff changeset
   528
        return False
44bc181b9835 match: override visitdir() in nevermatcher to return False
Martin von Zweigbergk <martinvonz@google.com>
parents: 33497
diff changeset
   529
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   530
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   531
        return set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   532
32605
e6ff007e107e match: introduce nevermatcher for when no ignore files are present
Siddharth Agarwal <sid0@fb.com>
parents: 32557
diff changeset
   533
    def __repr__(self):
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   534
        return r'<nevermatcher>'
32605
e6ff007e107e match: introduce nevermatcher for when no ignore files are present
Siddharth Agarwal <sid0@fb.com>
parents: 32557
diff changeset
   535
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   536
38577
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   537
class predicatematcher(basematcher):
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   538
    """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
   539
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   540
    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
   541
        super(predicatematcher, self).__init__(badfn)
38577
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   542
        self.matchfn = predfn
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   543
        self._predrepr = predrepr
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   544
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   545
    @encoding.strmethod
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   546
    def __repr__(self):
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   547
        s = stringutil.buildrepr(self._predrepr) or pycompat.byterepr(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   548
            self.matchfn
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   549
        )
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   550
        return b'<predicatenmatcher pred=%s>' % s
38577
76838305b9dd match: add basic wrapper for boolean function
Yuya Nishihara <yuya@tcha.org>
parents: 38475
diff changeset
   551
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   552
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   553
class patternmatcher(basematcher):
43721
bfbbf48d51e8 py3: make doc strings containing deprecated '\.' escape sequence raw strings
Daniel Ploch <dploch@google.com>
parents: 43117
diff changeset
   554
    r"""Matches a set of (kind, pat, source) against a 'root' directory.
42081
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   555
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   556
    >>> kindpats = [
42186
fd384911f51b match: use raw strings to avoid illegal baskslash escape
Gregory Szorc <gregory.szorc@gmail.com>
parents: 42099
diff changeset
   557
    ...     (b're', br'.*\.c$', b''),
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   558
    ...     (b'path', b'foo/a', b''),
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   559
    ...     (b'relpath', b'b', b''),
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   560
    ...     (b'glob', b'*.h', b''),
42081
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   561
    ... ]
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   562
    >>> m = patternmatcher(b'foo', kindpats)
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   563
    >>> m(b'main.c')  # matches re:.*\.c$
42081
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   564
    True
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   565
    >>> m(b'b.txt')
42081
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   566
    False
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   567
    >>> m(b'foo/a')  # matches path:foo/a
42081
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   568
    True
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   569
    >>> m(b'a')  # does not match path:b, since 'root' is 'foo'
42081
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   570
    False
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   571
    >>> m(b'b')  # matches relpath:b, since 'root' is 'foo'
42081
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   572
    True
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   573
    >>> m(b'lib.h')  # matches glob:*.h
42081
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   574
    True
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   575
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   576
    >>> m.files()
42341
27d6956d386b match: use '' instead of '.' for root directory (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 42329
diff changeset
   577
    ['', 'foo/a', 'b', '']
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   578
    >>> m.exact(b'foo/a')
42081
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   579
    True
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   580
    >>> m.exact(b'b')
42081
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   581
    True
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   582
    >>> m.exact(b'lib.h')  # exact matches are for (rel)path kinds
42081
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   583
    False
bf777c1e78dd match: add a docstring with doctest examples to patternmatcher
Denis Laxalde <denis@laxalde.org>
parents: 42080
diff changeset
   584
    """
32395
24245b54aa8a match: replace match class by match function (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32325
diff changeset
   585
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   586
    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
   587
        super(patternmatcher, self).__init__(badfn)
8581
101d305c1d0b match: fold _matcher into match.__init__
Matt Mackall <mpm@selenic.com>
parents: 8580
diff changeset
   588
32557
3fdcc34c0aba match: remove special-casing of always-matching patterns in patternmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32556
diff changeset
   589
        self._files = _explicitfiles(kindpats)
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
   590
        self._prefix = _prefix(kindpats)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   591
        self._pats, self.matchfn = _buildmatch(kindpats, b'$', root)
8587
8f15d54437b9 match: fold match into _match base class
Matt Mackall <mpm@selenic.com>
parents: 8586
diff changeset
   592
32322
23c9a2a71c6e match: make _fileroots a @propertycache and rename it to _fileset
Martin von Zweigbergk <martinvonz@google.com>
parents: 32312
diff changeset
   593
    @propertycache
24636
36872036169b treemanifest: further optimize treemanifest.matches()
Drew Gottlieb <drgott@google.com>
parents: 24452
diff changeset
   594
    def _dirs(self):
43523
c21aca51b392 utils: move the `dirs` definition in pathutil (API)
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 43506
diff changeset
   595
        return set(pathutil.dirs(self._fileset))
24636
36872036169b treemanifest: further optimize treemanifest.matches()
Drew Gottlieb <drgott@google.com>
parents: 24452
diff changeset
   596
36872036169b treemanifest: further optimize treemanifest.matches()
Drew Gottlieb <drgott@google.com>
parents: 24452
diff changeset
   597
    def visitdir(self, dir):
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
   598
        if self._prefix and dir in self._fileset:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   599
            return b'all'
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   600
        return (
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   601
            dir in self._fileset
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   602
            or dir in self._dirs
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   603
            or any(
43633
0b7733719d21 utils: move finddirs() to pathutil
Martin von Zweigbergk <martinvonz@google.com>
parents: 43595
diff changeset
   604
                parentdir in self._fileset
0b7733719d21 utils: move finddirs() to pathutil
Martin von Zweigbergk <martinvonz@google.com>
parents: 43595
diff changeset
   605
                for parentdir in pathutil.finddirs(dir)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   606
            )
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   607
        )
24636
36872036169b treemanifest: further optimize treemanifest.matches()
Drew Gottlieb <drgott@google.com>
parents: 24452
diff changeset
   608
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   609
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   610
        ret = self.visitdir(dir)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   611
        if ret is True:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   612
            return b'this'
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   613
        elif not ret:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   614
            return set()
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   615
        assert ret == b'all'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   616
        return b'all'
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   617
33379
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
   618
    def prefix(self):
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
   619
        return self._prefix
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
   620
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   621
    @encoding.strmethod
32406
952017471f93 match: implement __repr__() and update users (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32401
diff changeset
   622
    def __repr__(self):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   623
        return b'<patternmatcher patterns=%r>' % pycompat.bytestr(self._pats)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   624
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   625
43523
c21aca51b392 utils: move the `dirs` definition in pathutil (API)
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 43506
diff changeset
   626
# This is basically a reimplementation of pathutil.dirs that stores the
c21aca51b392 utils: move the `dirs` definition in pathutil (API)
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 43506
diff changeset
   627
# children instead of just a count of them, plus a small optional optimization
c21aca51b392 utils: move the `dirs` definition in pathutil (API)
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 43506
diff changeset
   628
# to avoid some directories we don't need.
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   629
class _dirchildren(object):
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   630
    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
   631
        self._dirs = {}
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   632
        self._onlyinclude = onlyinclude or []
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   633
        addpath = self.addpath
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   634
        for f in paths:
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   635
            addpath(f)
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   636
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   637
    def addpath(self, path):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   638
        if path == b'':
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   639
            return
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   640
        dirs = self._dirs
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   641
        findsplitdirs = _dirchildren._findsplitdirs
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   642
        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
   643
            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
   644
                continue
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   645
            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
   646
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   647
    @staticmethod
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   648
    def _findsplitdirs(path):
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   649
        # yields (dirname, basename) tuples, walking back to the root.  This is
43633
0b7733719d21 utils: move finddirs() to pathutil
Martin von Zweigbergk <martinvonz@google.com>
parents: 43595
diff changeset
   650
        # very similar to pathutil.finddirs, except:
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   651
        #  - 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
   652
        # Unlike manifest._splittopdir, this does not suffix `dirname` with a
42341
27d6956d386b match: use '' instead of '.' for root directory (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 42329
diff changeset
   653
        # slash.
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   654
        oldpos = len(path)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   655
        pos = path.rfind(b'/')
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   656
        while pos != -1:
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   657
            yield path[:pos], path[pos + 1 : oldpos]
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   658
            oldpos = pos
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   659
            pos = path.rfind(b'/', 0, pos)
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   660
        yield b'', path[:oldpos]
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   661
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   662
    def get(self, path):
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   663
        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
   664
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   665
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   666
class includematcher(basematcher):
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   667
    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
   668
        super(includematcher, self).__init__(badfn)
44530
4d1634e59f13 rust-status: use bare hg status fastpath from Python
Raphaël Gomès <rgomes@octobus.net>
parents: 44452
diff changeset
   669
        if rustmod is not None:
4d1634e59f13 rust-status: use bare hg status fastpath from Python
Raphaël Gomès <rgomes@octobus.net>
parents: 44452
diff changeset
   670
            # We need to pass the patterns to Rust because they can contain
4d1634e59f13 rust-status: use bare hg status fastpath from Python
Raphaël Gomès <rgomes@octobus.net>
parents: 44452
diff changeset
   671
            # patterns from the user interface
4d1634e59f13 rust-status: use bare hg status fastpath from Python
Raphaël Gomès <rgomes@octobus.net>
parents: 44452
diff changeset
   672
            self._kindpats = kindpats
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   673
        self._pats, self.matchfn = _buildmatch(kindpats, b'(?:/|$)', root)
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
   674
        self._prefix = _prefix(kindpats)
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
   675
        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
   676
        # roots are directories which are recursively included.
32503
361808a2b0b8 match: simplify includematcher a bit
Martin von Zweigbergk <martinvonz@google.com>
parents: 32502
diff changeset
   677
        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
   678
        # 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
   679
        self._dirs = set(dirs)
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
   680
        # parents are directories which are non-recursively included because
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
   681
        # they are needed to get to items in _dirs or _roots.
42361
c4b8f8637d7a match: de-flake test-doctest.py by not depending on util.dirs() order
Martin von Zweigbergk <martinvonz@google.com>
parents: 42346
diff changeset
   682
        self._parents = parents
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   683
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   684
    def visitdir(self, dir):
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
   685
        if self._prefix and dir in self._roots:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   686
            return b'all'
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   687
        return (
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   688
            dir in self._roots
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   689
            or dir in self._dirs
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   690
            or dir in self._parents
43633
0b7733719d21 utils: move finddirs() to pathutil
Martin von Zweigbergk <martinvonz@google.com>
parents: 43595
diff changeset
   691
            or any(
0b7733719d21 utils: move finddirs() to pathutil
Martin von Zweigbergk <martinvonz@google.com>
parents: 43595
diff changeset
   692
                parentdir in self._roots for parentdir in pathutil.finddirs(dir)
0b7733719d21 utils: move finddirs() to pathutil
Martin von Zweigbergk <martinvonz@google.com>
parents: 43595
diff changeset
   693
            )
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   694
        )
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   695
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   696
    @propertycache
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   697
    def _allparentschildren(self):
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   698
        # 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
   699
        # 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
   700
        #   dirs = ['foo/bar']
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   701
        #   parents = ['foo']
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   702
        # 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
   703
        # 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
   704
        return _dirchildren(
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   705
            itertools.chain(self._dirs, self._roots, self._parents),
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   706
            onlyinclude=self._parents,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   707
        )
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
   708
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   709
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   710
        if self._prefix and dir in self._roots:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   711
            return b'all'
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   712
        # 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
   713
        # visitdir, that's handled below.
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   714
        if (
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   715
            b'' in self._roots
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   716
            or dir in self._roots
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   717
            or dir in self._dirs
43633
0b7733719d21 utils: move finddirs() to pathutil
Martin von Zweigbergk <martinvonz@google.com>
parents: 43595
diff changeset
   718
            or any(
0b7733719d21 utils: move finddirs() to pathutil
Martin von Zweigbergk <martinvonz@google.com>
parents: 43595
diff changeset
   719
                parentdir in self._roots for parentdir in pathutil.finddirs(dir)
0b7733719d21 utils: move finddirs() to pathutil
Martin von Zweigbergk <martinvonz@google.com>
parents: 43595
diff changeset
   720
            )
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   721
        ):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   722
            return b'this'
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   723
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   724
        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
   725
            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
   726
        return set()
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   727
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   728
    @encoding.strmethod
32501
7095dbc266e3 match: split up main matcher into patternmatcher and includematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32500
diff changeset
   729
    def __repr__(self):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   730
        return b'<includematcher includes=%r>' % pycompat.bytestr(self._pats)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   731
32406
952017471f93 match: implement __repr__() and update users (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 32401
diff changeset
   732
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   733
class exactmatcher(basematcher):
42080
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   734
    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
   735
    patterns (so no kind-prefixes).
42080
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   736
42186
fd384911f51b match: use raw strings to avoid illegal baskslash escape
Gregory Szorc <gregory.szorc@gmail.com>
parents: 42099
diff changeset
   737
    >>> m = exactmatcher([b'a.txt', br're:.*\.c$'])
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   738
    >>> m(b'a.txt')
42080
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   739
    True
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   740
    >>> m(b'b.txt')
42080
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   741
    False
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   742
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   743
    Input files that would be matched are exactly those returned by .files()
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   744
    >>> m.files()
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   745
    ['a.txt', 're:.*\\.c$']
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   746
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   747
    So pattern 're:.*\.c$' is not considered as a regex, but as a file name
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
   748
    >>> m(b'main.c')
42080
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   749
    False
42186
fd384911f51b match: use raw strings to avoid illegal baskslash escape
Gregory Szorc <gregory.szorc@gmail.com>
parents: 42099
diff changeset
   750
    >>> m(br're:.*\.c$')
42080
f3db5c805a67 match: add doctest examples for exactmatcher
Denis Laxalde <denis@laxalde.org>
parents: 41676
diff changeset
   751
    True
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   752
    '''
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   753
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
   754
    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
   755
        super(exactmatcher, self).__init__(badfn)
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   756
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   757
        if isinstance(files, list):
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   758
            self._files = files
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   759
        else:
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   760
            self._files = list(files)
32543
cf7c88986e9f match: define exactmatcher.matchfn statically
Yuya Nishihara <yuya@tcha.org>
parents: 32542
diff changeset
   761
cf7c88986e9f match: define exactmatcher.matchfn statically
Yuya Nishihara <yuya@tcha.org>
parents: 32542
diff changeset
   762
    matchfn = basematcher.exact
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   763
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   764
    @propertycache
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   765
    def _dirs(self):
43523
c21aca51b392 utils: move the `dirs` definition in pathutil (API)
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 43506
diff changeset
   766
        return set(pathutil.dirs(self._fileset))
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   767
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   768
    def visitdir(self, dir):
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   769
        return dir in self._dirs
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   770
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   771
    def visitchildrenset(self, dir):
39261
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   772
        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
   773
            return set()
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   774
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   775
        candidates = self._fileset | self._dirs - {b''}
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   776
        if dir != b'':
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   777
            d = dir + b'/'
44452
9d2b2df2c2ba cleanup: run pyupgrade on our source tree to clean up varying things
Augie Fackler <augie@google.com>
parents: 44137
diff changeset
   778
            candidates = {c[len(d) :] for c in candidates if c.startswith(d)}
39261
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   779
        # self._dirs includes all of the directories, recursively, so if
42341
27d6956d386b match: use '' instead of '.' for root directory (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 42329
diff changeset
   780
        # we're attempting to match foo/bar/baz.txt, it'll have '', 'foo',
39261
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   781
        # '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
   782
        # '/' 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
   783
        # immediate subdir will be in there without a slash.
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   784
        ret = {c for c in candidates if b'/' not in c}
39261
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   785
        # 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
   786
        # 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
   787
        assert ret
c9a3f7f5c023 match: make exactmatcher.visitchildrenset return file children as well
Kyle Lippincott <spectral@google.com>
parents: 39260
diff changeset
   788
        return ret
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   789
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   790
    def isexact(self):
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   791
        return True
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   792
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   793
    @encoding.strmethod
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   794
    def __repr__(self):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   795
        return b'<exactmatcher files=%r>' % self._files
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   796
32499
a3583852861a match: handle exact matching using new exactmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32497
diff changeset
   797
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   798
class differencematcher(basematcher):
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   799
    '''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
   800
    does not.
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   801
43715
5e1b0470cee7 match: remove explicitdir attribute
Martin von Zweigbergk <martinvonz@google.com>
parents: 43633
diff changeset
   802
    The second matcher's non-matching-attributes (bad, traversedir) are ignored.
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   803
    '''
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   804
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   805
    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
   806
        super(differencematcher, self).__init__()
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   807
        self._m1 = m1
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   808
        self._m2 = m2
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   809
        self.bad = m1.bad
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   810
        self.traversedir = m1.traversedir
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   811
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   812
    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
   813
        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
   814
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   815
    @propertycache
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   816
    def _files(self):
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   817
        if self.isexact():
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   818
            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
   819
        # 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
   820
        # 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
   821
        # 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
   822
        # 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
   823
        # 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
   824
        return self._m1.files()
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   825
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   826
    def visitdir(self, dir):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   827
        if self._m2.visitdir(dir) == b'all':
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   828
            return False
41528
b7a0efb3c370 match: teach diffmatcher.visitdir() to return 'all' if possible
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 41282
diff changeset
   829
        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
   830
            # 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
   831
            return self._m1.visitdir(dir)
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   832
        return bool(self._m1.visitdir(dir))
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   833
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   834
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   835
        m2_set = self._m2.visitchildrenset(dir)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   836
        if m2_set == b'all':
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   837
            return set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   838
        m1_set = self._m1.visitchildrenset(dir)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   839
        # Possible values for m1: 'all', 'this', set(...), set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   840
        # Possible values for m2:        'this', set(...), set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   841
        # 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
   842
        # 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
   843
        # return True, not 'all', for some reason.
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   844
        if not m2_set:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   845
            return m1_set
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   846
        if m1_set in [b'all', b'this']:
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   847
            # 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
   848
            # '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
   849
            # subdirectory.
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   850
            return b'this'
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   851
        # Possible values for m1:         set(...), set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   852
        # Possible values for m2: 'this', set(...)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   853
        # We ignore m2's set results. They're possibly incorrect:
42341
27d6956d386b match: use '' instead of '.' for root directory (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 42329
diff changeset
   854
        #  m1 = path:dir/subdir, m2=rootfilesin:dir, visitchildrenset(''):
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   855
        #    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
   856
        #    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
   857
        return m1_set
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   858
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   859
    def isexact(self):
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   860
        return self._m1.isexact()
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   861
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   862
    @encoding.strmethod
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   863
    def __repr__(self):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   864
        return b'<differencematcher m1=%r, m2=%r>' % (self._m1, self._m2)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   865
32465
a83a7d27911e match: handle excludes using new differencematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32464
diff changeset
   866
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   867
def intersectmatchers(m1, m2):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   868
    '''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
   869
43715
5e1b0470cee7 match: remove explicitdir attribute
Martin von Zweigbergk <martinvonz@google.com>
parents: 43633
diff changeset
   870
    The second matcher's non-matching-attributes (bad, traversedir) are ignored.
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   871
    '''
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   872
    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
   873
        return m1 or m2
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   874
    if m1.always():
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   875
        m = copy.copy(m2)
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   876
        # 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
   877
        # one thing to copy from m1.
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   878
        m.bad = m1.bad
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   879
        m.traversedir = m1.traversedir
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   880
        return m
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   881
    if m2.always():
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   882
        m = copy.copy(m1)
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   883
        return m
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   884
    return intersectionmatcher(m1, m2)
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   885
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   886
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   887
class intersectionmatcher(basematcher):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   888
    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
   889
        super(intersectionmatcher, self).__init__()
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   890
        self._m1 = m1
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   891
        self._m2 = m2
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   892
        self.bad = m1.bad
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   893
        self.traversedir = m1.traversedir
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   894
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   895
    @propertycache
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   896
    def _files(self):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   897
        if self.isexact():
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   898
            m1, m2 = self._m1, self._m2
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   899
            if not m1.isexact():
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   900
                m1, m2 = m2, m1
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   901
            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
   902
        # 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
   903
        # 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
   904
        # 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
   905
        # "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
   906
        return self._m1.files() + self._m2.files()
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   907
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   908
    def matchfn(self, f):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   909
        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
   910
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   911
    def visitdir(self, dir):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   912
        visit1 = self._m1.visitdir(dir)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   913
        if visit1 == b'all':
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   914
            return self._m2.visitdir(dir)
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   915
        # 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
   916
        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
   917
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   918
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   919
        m1_set = self._m1.visitchildrenset(dir)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   920
        if not m1_set:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   921
            return set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   922
        m2_set = self._m2.visitchildrenset(dir)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   923
        if not m2_set:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   924
            return set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   925
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   926
        if m1_set == b'all':
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   927
            return m2_set
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   928
        elif m2_set == b'all':
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   929
            return m1_set
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   930
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   931
        if m1_set == b'this' or m2_set == b'this':
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   932
            return b'this'
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   933
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   934
        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
   935
        return m1_set.intersection(m2_set)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
   936
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   937
    def always(self):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   938
        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
   939
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   940
    def isexact(self):
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   941
        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
   942
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
   943
    @encoding.strmethod
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   944
    def __repr__(self):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   945
        return b'<intersectionmatcher m1=%r, m2=%r>' % (self._m1, self._m2)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   946
32497
9eccd559c592 match: handle includes using new intersectionmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32496
diff changeset
   947
32460
f9445b528687 match: make subdirmatcher extend basematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32459
diff changeset
   948
class subdirmatcher(basematcher):
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   949
    """Adapt a matcher to work on a subdirectory only.
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   950
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   951
    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
   952
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
   953
    >>> from . import pycompat
44002
d9d78e70149a tests: fix failing doctest in match.py by adding dummy auditor
Martin von Zweigbergk <martinvonz@google.com>
parents: 43965
diff changeset
   954
    >>> m1 = match(util.localpath(b'/root'), b'', [b'a.txt', b'sub/b.txt'], auditor=lambda name: None)
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
   955
    >>> m2 = subdirmatcher(b'sub', m1)
42087
2e2699af5649 match: let regex match function return a boolean
Denis Laxalde <denis@laxalde.org>
parents: 42086
diff changeset
   956
    >>> m2(b'a.txt')
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   957
    False
42087
2e2699af5649 match: let regex match function return a boolean
Denis Laxalde <denis@laxalde.org>
parents: 42086
diff changeset
   958
    >>> m2(b'b.txt')
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   959
    True
42087
2e2699af5649 match: let regex match function return a boolean
Denis Laxalde <denis@laxalde.org>
parents: 42086
diff changeset
   960
    >>> m2.matchfn(b'a.txt')
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   961
    False
42087
2e2699af5649 match: let regex match function return a boolean
Denis Laxalde <denis@laxalde.org>
parents: 42086
diff changeset
   962
    >>> m2.matchfn(b'b.txt')
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   963
    True
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   964
    >>> m2.files()
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   965
    ['b.txt']
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
   966
    >>> m2.exact(b'b.txt')
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   967
    True
12268
83aaeba32b88 narrowmatcher: propagate bad method
Martin Geisler <mg@lazybytes.net>
parents: 12267
diff changeset
   968
    >>> 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
   969
    ...     print(pycompat.sysstr(b"%s: %s" % (f, msg)))
12268
83aaeba32b88 narrowmatcher: propagate bad method
Martin Geisler <mg@lazybytes.net>
parents: 12267
diff changeset
   970
    >>> m1.bad = bad
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
   971
    >>> m2.bad(b'x.txt', b'No such file')
12268
83aaeba32b88 narrowmatcher: propagate bad method
Martin Geisler <mg@lazybytes.net>
parents: 12267
diff changeset
   972
    sub/x.txt: No such file
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   973
    """
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   974
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   975
    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
   976
        super(subdirmatcher, self).__init__()
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   977
        self._path = path
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   978
        self._matcher = matcher
32460
f9445b528687 match: make subdirmatcher extend basematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32459
diff changeset
   979
        self._always = matcher.always()
12165
b7fbf24c8a93 match: add narrowmatcher class
Martin Geisler <mg@lazybytes.net>
parents: 12163
diff changeset
   980
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   981
        self._files = [
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   982
            f[len(path) + 1 :]
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   983
            for f in matcher._files
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   984
            if f.startswith(path + b"/")
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
   985
        ]
25194
ef4538ba67ef match: explicitly naming a subrepo implies always() for the submatcher
Matt Harbison <matt_harbison@yahoo.com>
parents: 24790
diff changeset
   986
32325
763d72925691 match: use match.prefix() in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32324
diff changeset
   987
        # 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
   988
        # a prefix matcher, this submatcher always matches.
763d72925691 match: use match.prefix() in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32324
diff changeset
   989
        if matcher.prefix():
25195
472a685a4961 merge with stable
Matt Mackall <mpm@selenic.com>
parents: 25189 25194
diff changeset
   990
            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
   991
32324
77dac8fd30ee match: avoid accessing match._pathrestricted from subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32323
diff changeset
   992
    def bad(self, f, msg):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
   993
        self._matcher.bad(self._path + b"/" + f, msg)
32324
77dac8fd30ee match: avoid accessing match._pathrestricted from subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32323
diff changeset
   994
32464
2e80a691e575 match: override matchfn() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32463
diff changeset
   995
    def matchfn(self, f):
2e80a691e575 match: override matchfn() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32463
diff changeset
   996
        # 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
   997
        # 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
   998
        # 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
   999
        # call the original matcher with the subdirectory path prepended.
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1000
        return self._matcher.matchfn(self._path + b"/" + f)
32464
2e80a691e575 match: override matchfn() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32463
diff changeset
  1001
32323
0aa4032a97e1 match: override visitdir() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32322
diff changeset
  1002
    def visitdir(self, dir):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1003
        if dir == b'':
32323
0aa4032a97e1 match: override visitdir() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32322
diff changeset
  1004
            dir = self._path
0aa4032a97e1 match: override visitdir() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32322
diff changeset
  1005
        else:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1006
            dir = self._path + b"/" + dir
32323
0aa4032a97e1 match: override visitdir() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32322
diff changeset
  1007
        return self._matcher.visitdir(dir)
0aa4032a97e1 match: override visitdir() the usual way in subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32322
diff changeset
  1008
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1009
    def visitchildrenset(self, dir):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1010
        if dir == b'':
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1011
            dir = self._path
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1012
        else:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1013
            dir = self._path + b"/" + dir
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1014
        return self._matcher.visitchildrenset(dir)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1015
32460
f9445b528687 match: make subdirmatcher extend basematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32459
diff changeset
  1016
    def always(self):
f9445b528687 match: make subdirmatcher extend basematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32459
diff changeset
  1017
        return self._always
f9445b528687 match: make subdirmatcher extend basematcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32459
diff changeset
  1018
33379
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
  1019
    def prefix(self):
7ddb2aa2b7af match: express anypats(), not prefix(), in terms of the others
Martin von Zweigbergk <martinvonz@google.com>
parents: 33378
diff changeset
  1020
        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
  1021
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
  1022
    @encoding.strmethod
32552
e7aa11f3abcd match: add __repr__ for subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32543
diff changeset
  1023
    def __repr__(self):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1024
        return b'<subdirmatcher path=%r, matcher=%r>' % (
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1025
            self._path,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1026
            self._matcher,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1027
        )
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1028
32552
e7aa11f3abcd match: add __repr__ for subdirmatcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 32543
diff changeset
  1029
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1030
class prefixdirmatcher(basematcher):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1031
    """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
  1032
43715
5e1b0470cee7 match: remove explicitdir attribute
Martin von Zweigbergk <martinvonz@google.com>
parents: 43633
diff changeset
  1033
    The matcher's non-matching-attributes (bad, traversedir) are ignored.
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1034
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1035
    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
  1036
    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
  1037
43949
8b1a9ba375e5 match: make sure `root` argument is always an absolute path (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 43735
diff changeset
  1038
    >>> m1 = match(util.localpath(b'/root/d/e'), b'f', [b'../a.txt', b'b.txt'], auditor=lambda name: None)
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
  1039
    >>> m2 = prefixdirmatcher(b'd/e', m1)
42087
2e2699af5649 match: let regex match function return a boolean
Denis Laxalde <denis@laxalde.org>
parents: 42086
diff changeset
  1040
    >>> m2(b'a.txt')
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1041
    False
42087
2e2699af5649 match: let regex match function return a boolean
Denis Laxalde <denis@laxalde.org>
parents: 42086
diff changeset
  1042
    >>> m2(b'd/e/a.txt')
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1043
    True
42087
2e2699af5649 match: let regex match function return a boolean
Denis Laxalde <denis@laxalde.org>
parents: 42086
diff changeset
  1044
    >>> m2(b'd/e/b.txt')
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1045
    False
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1046
    >>> m2.files()
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1047
    ['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
  1048
    >>> 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
  1049
    True
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1050
    >>> m2.visitdir(b'd')
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1051
    True
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1052
    >>> m2.visitdir(b'd/e')
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1053
    True
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1054
    >>> m2.visitdir(b'd/e/f')
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1055
    True
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1056
    >>> m2.visitdir(b'd/e/g')
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1057
    False
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1058
    >>> m2.visitdir(b'd/ef')
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1059
    False
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1060
    """
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1061
41675
ddbebce94665 match: delete unused root and cwd arguments to constructors (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 41672
diff changeset
  1062
    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
  1063
        super(prefixdirmatcher, self).__init__(badfn)
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1064
        if not path:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1065
            raise error.ProgrammingError(b'prefix path must not be empty')
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1066
        self._path = path
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1067
        self._pathprefix = path + b'/'
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1068
        self._matcher = matcher
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1069
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1070
    @propertycache
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1071
    def _files(self):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1072
        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
  1073
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1074
    def matchfn(self, f):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1075
        if not f.startswith(self._pathprefix):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1076
            return False
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1077
        return self._matcher.matchfn(f[len(self._pathprefix) :])
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1078
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1079
    @propertycache
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1080
    def _pathdirs(self):
43633
0b7733719d21 utils: move finddirs() to pathutil
Martin von Zweigbergk <martinvonz@google.com>
parents: 43595
diff changeset
  1081
        return set(pathutil.finddirs(self._path))
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1082
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1083
    def visitdir(self, dir):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1084
        if dir == self._path:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1085
            return self._matcher.visitdir(b'')
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1086
        if dir.startswith(self._pathprefix):
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1087
            return self._matcher.visitdir(dir[len(self._pathprefix) :])
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1088
        return dir in self._pathdirs
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1089
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1090
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1091
        if dir == self._path:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1092
            return self._matcher.visitchildrenset(b'')
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1093
        if dir.startswith(self._pathprefix):
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1094
            return self._matcher.visitchildrenset(dir[len(self._pathprefix) :])
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1095
        if dir in self._pathdirs:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1096
            return b'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
  1097
        return set()
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1098
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1099
    def isexact(self):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1100
        return self._matcher.isexact()
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1101
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1102
    def prefix(self):
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1103
        return self._matcher.prefix()
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1104
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1105
    @encoding.strmethod
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1106
    def __repr__(self):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1107
        return b'<prefixdirmatcher path=%r, matcher=%r>' % (
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1108
            pycompat.bytestr(self._path),
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1109
            self._matcher,
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1110
        )
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1111
38611
0ba4cf3f088f match: add prefixdirmatcher to adapt subrepo matcher back
Yuya Nishihara <yuya@tcha.org>
parents: 38581
diff changeset
  1112
33319
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
  1113
class unionmatcher(basematcher):
33448
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1114
    """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
  1115
43715
5e1b0470cee7 match: remove explicitdir attribute
Martin von Zweigbergk <martinvonz@google.com>
parents: 43633
diff changeset
  1116
    The non-matching-attributes (bad, traversedir) are taken from the first
5e1b0470cee7 match: remove explicitdir attribute
Martin von Zweigbergk <martinvonz@google.com>
parents: 43633
diff changeset
  1117
    matcher.
33448
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1118
    """
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1119
33319
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
  1120
    def __init__(self, matchers):
33448
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1121
        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
  1122
        super(unionmatcher, self).__init__()
33448
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1123
        self.traversedir = m1.traversedir
33319
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
  1124
        self._matchers = matchers
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
  1125
33380
892d255ec2a1 match: override matchfn instead of __call__ for consistency
Martin von Zweigbergk <martinvonz@google.com>
parents: 33379
diff changeset
  1126
    def matchfn(self, f):
33319
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
  1127
        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
  1128
            if match(f):
33319
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
  1129
                return True
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
  1130
        return False
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
  1131
33448
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1132
    def visitdir(self, dir):
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1133
        r = False
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1134
        for m in self._matchers:
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1135
            v = m.visitdir(dir)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1136
            if v == b'all':
33448
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1137
                return v
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1138
            r |= v
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1139
        return r
04be8aec44a8 match: make unionmatcher a proper matcher
Martin von Zweigbergk <martinvonz@google.com>
parents: 33447
diff changeset
  1140
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1141
    def visitchildrenset(self, dir):
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1142
        r = set()
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1143
        this = False
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1144
        for m in self._matchers:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1145
            v = m.visitchildrenset(dir)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1146
            if not v:
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1147
                continue
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1148
            if v == b'all':
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1149
                return v
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1150
            if this or v == b'this':
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1151
                this = True
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1152
                # 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
  1153
                continue
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1154
            assert isinstance(v, set)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1155
            r = r.union(v)
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1156
        if this:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1157
            return b'this'
38955
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1158
        return r
081cc9a95b65 match: add visitchildrenset complement to visitdir
spectral <spectral@google.com>
parents: 38954
diff changeset
  1159
36050
c4fa47f880d3 py3: make sure we return str from __repr__
Pulkit Goyal <7895pulkit@gmail.com>
parents: 35659
diff changeset
  1160
    @encoding.strmethod
33319
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
  1161
    def __repr__(self):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1162
        return b'<unionmatcher matchers=%r>' % self._matchers
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1163
33319
3c84591e7321 match: move matchers from sparse into core
Gregory Szorc <gregory.szorc@gmail.com>
parents: 33306
diff changeset
  1164
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1165
def patkind(pattern, default=None):
43721
bfbbf48d51e8 py3: make doc strings containing deprecated '\.' escape sequence raw strings
Daniel Ploch <dploch@google.com>
parents: 43117
diff changeset
  1166
    r'''If pattern is 'kind:pat' with a known kind, return kind.
42082
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1167
42186
fd384911f51b match: use raw strings to avoid illegal baskslash escape
Gregory Szorc <gregory.szorc@gmail.com>
parents: 42099
diff changeset
  1168
    >>> patkind(br're:.*\.c$')
42082
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1169
    're'
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
  1170
    >>> patkind(b'glob:*.c')
42082
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1171
    'glob'
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
  1172
    >>> patkind(b'relpath:test.py')
42082
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1173
    'relpath'
42098
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
  1174
    >>> patkind(b'main.py')
5753e5949b51 py3: add b'' prefixes to new doctests in match.py
Pulkit Goyal <pulkit@yandex-team.ru>
parents: 42087
diff changeset
  1175
    >>> patkind(b'main.py', default=b're')
42082
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1176
    're'
413a75da98ce match: add doctest examples for patkind()
Denis Laxalde <denis@laxalde.org>
parents: 42081
diff changeset
  1177
    '''
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1178
    return _patsplit(pattern, default)[0]
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1179
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1180
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1181
def _patsplit(pattern, default):
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1182
    """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
  1183
    pattern."""
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1184
    if b':' in pattern:
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1185
        kind, pat = pattern.split(b':', 1)
33684
2be0bf186950 match: expose some data and functionality to other modules
Kostia Balytskyi <ikostia@fb.com>
parents: 33582
diff changeset
  1186
        if kind in allpatternkinds:
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1187
            return kind, pat
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1188
    return default, pattern
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1189
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1190
8582
a4c199e12b5a match: remove head and tail args from _globre
Matt Mackall <mpm@selenic.com>
parents: 8581
diff changeset
  1191
def _globre(pat):
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1192
    r'''Convert an extended glob string to a regexp string.
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1193
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1194
    >>> from . import pycompat
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1195
    >>> def bprint(s):
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1196
    ...     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
  1197
    >>> bprint(_globre(br'?'))
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1198
    .
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1199
    >>> bprint(_globre(br'*'))
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1200
    [^/]*
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1201
    >>> bprint(_globre(br'**'))
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1202
    .*
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1203
    >>> bprint(_globre(br'**/a'))
21815
a4b67bf1f0a5 match: make glob '**/' match the empty string
Siddharth Agarwal <sid0@fb.com>
parents: 21191
diff changeset
  1204
    (?:.*/)?a
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1205
    >>> bprint(_globre(br'a/**/b'))
38475
67dc32d4e790 cleanup: migrate from re.escape to stringutil.reescape
Augie Fackler <augie@google.com>
parents: 38019
diff changeset
  1206
    a/(?:.*/)?b
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1207
    >>> bprint(_globre(br'[a*?!^][^b][!c]'))
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1208
    [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
  1209
    >>> bprint(_globre(br'{a,b}'))
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1210
    (?:a|b)
34137
a8994d08e4a2 doctest: use print_function and convert bytes to unicode where needed
Yuya Nishihara <yuya@tcha.org>
parents: 34131
diff changeset
  1211
    >>> bprint(_globre(br'.\*\?'))
21112
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1212
    \.\*\?
03782d2fc776 match: _globre doctests
Mads Kiilerich <madski@unity3d.com>
parents: 21111
diff changeset
  1213
    '''
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1214
    i, n = 0, len(pat)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1215
    res = b''
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1216
    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
  1217
    escape = util.stringutil.regexbytesescapemap.get
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1218
10282
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10263
diff changeset
  1219
    def peek():
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1220
        return i < n and pat[i : i + 1]
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1221
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1222
    while i < n:
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1223
        c = pat[i : i + 1]
10282
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10263
diff changeset
  1224
        i += 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1225
        if c not in b'*?[{},\\':
40684
e6c9ef5e11a0 match: provide and use a quick way to escape a single byte
Boris Feld <boris.feld@octobus.net>
parents: 40345
diff changeset
  1226
            res += escape(c, c)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1227
        elif c == b'*':
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1228
            if peek() == b'*':
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1229
                i += 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1230
                if peek() == b'/':
21815
a4b67bf1f0a5 match: make glob '**/' match the empty string
Siddharth Agarwal <sid0@fb.com>
parents: 21191
diff changeset
  1231
                    i += 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1232
                    res += b'(?:.*/)?'
21815
a4b67bf1f0a5 match: make glob '**/' match the empty string
Siddharth Agarwal <sid0@fb.com>
parents: 21191
diff changeset
  1233
                else:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1234
                    res += b'.*'
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1235
            else:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1236
                res += b'[^/]*'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1237
        elif c == b'?':
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1238
            res += b'.'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1239
        elif c == b'[':
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1240
            j = i
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1241
            if j < n and pat[j : j + 1] in b'!]':
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1242
                j += 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1243
            while j < n and pat[j : j + 1] != b']':
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1244
                j += 1
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1245
            if j >= n:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1246
                res += b'\\['
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1247
            else:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1248
                stuff = pat[i:j].replace(b'\\', b'\\\\')
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1249
                i = j + 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1250
                if stuff[0:1] == b'!':
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1251
                    stuff = b'^' + stuff[1:]
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1252
                elif stuff[0:1] == b'^':
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1253
                    stuff = b'\\' + stuff
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1254
                res = b'%s[%s]' % (res, stuff)
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1255
        elif c == b'{':
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1256
            group += 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1257
            res += b'(?:'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1258
        elif c == b'}' and group:
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1259
            res += b')'
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1260
            group -= 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1261
        elif c == b',' and group:
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1262
            res += b'|'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1263
        elif c == b'\\':
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1264
            p = peek()
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1265
            if p:
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1266
                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
  1267
                res += escape(p, p)
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1268
            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
  1269
                res += escape(c, c)
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1270
        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
  1271
            res += escape(c, c)
8582
a4c199e12b5a match: remove head and tail args from _globre
Matt Mackall <mpm@selenic.com>
parents: 8581
diff changeset
  1272
    return res
8570
7fe2012b3bd0 match: move util match functions over
Matt Mackall <mpm@selenic.com>
parents: 8568
diff changeset
  1273
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1274
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1275
def _regex(kind, pat, globsuffix):
42329
c7652f7440d9 rust-filepatterns: call new Rust implementations from Python
Raphaël Gomès <rgomes@octobus.net>
parents: 42195
diff changeset
  1276
    '''Convert a (normalized) pattern of any kind into a
c7652f7440d9 rust-filepatterns: call new Rust implementations from Python
Raphaël Gomès <rgomes@octobus.net>
parents: 42195
diff changeset
  1277
    regular expression.
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1278
    globsuffix is appended to the regexp of globs.'''
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1279
    if not pat and kind in (b'glob', b'relpath'):
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1280
        return b''
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1281
    if kind == b're':
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1282
        return pat
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1283
    if kind in (b'path', b'relpath'):
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1284
        if pat == b'.':
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1285
            return b''
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1286
        return util.stringutil.reescape(pat) + b'(?:/|$)'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1287
    if kind == b'rootfilesin':
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1288
        if pat == b'.':
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1289
            escaped = b''
31012
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1290
        else:
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1291
            # Pattern is a directory name.
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1292
            escaped = util.stringutil.reescape(pat) + b'/'
31012
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1293
        # Anything after the pattern must be a non-directory.
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1294
        return escaped + b'[^/]+$'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1295
    if kind == b'relglob':
42864
72890d8f9860 match: simplify the regexps created for glob patterns
Valentin Gatien-Baron <valentin.gatienbaron@gmail.com>
parents: 42766
diff changeset
  1296
        globre = _globre(pat)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1297
        if globre.startswith(b'[^/]*'):
42864
72890d8f9860 match: simplify the regexps created for glob patterns
Valentin Gatien-Baron <valentin.gatienbaron@gmail.com>
parents: 42766
diff changeset
  1298
            # When pat has the form *XYZ (common), make the returned regex more
72890d8f9860 match: simplify the regexps created for glob patterns
Valentin Gatien-Baron <valentin.gatienbaron@gmail.com>
parents: 42766
diff changeset
  1299
            # legible by returning the regex for **XYZ instead of **/*XYZ.
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1300
            return b'.*' + globre[len(b'[^/]*') :] + globsuffix
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1301
        return b'(?:|.*/)' + globre + globsuffix
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1302
    if kind == b'relre':
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1303
        if pat.startswith(b'^'):
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1304
            return pat
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1305
        return b'.*' + pat
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1306
    if kind in (b'glob', b'rootglob'):
38578
2d487b9cac07 match: explode if unsupported pattern passed down to _regex() builder
Yuya Nishihara <yuya@tcha.org>
parents: 38577
diff changeset
  1307
        return _globre(pat) + globsuffix
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1308
    raise error.ProgrammingError(b'not a regex pattern: %s:%s' % (kind, pat))
8574
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1309
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1310
41669
a13268524c25 match: delete unused argument "listsubrepos" from _buildmatch()
Martin von Zweigbergk <martinvonz@google.com>
parents: 41668
diff changeset
  1311
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
  1312
    '''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
  1313
    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
  1314
    matchfuncs = []
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1315
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
  1316
    subincludes, kindpats = _expandsubinclude(kindpats, root)
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
  1317
    if subincludes:
32132
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1318
        submatchers = {}
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1319
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
  1320
        def matchsubinclude(f):
32132
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1321
            for prefix, matcherargs in subincludes:
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1322
                if f.startswith(prefix):
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1323
                    mf = submatchers.get(prefix)
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1324
                    if mf is None:
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1325
                        mf = match(*matcherargs)
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1326
                        submatchers[prefix] = mf
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1327
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1328
                    if mf(f[len(prefix) :]):
32132
6dea1701f170 match: make subinclude construction lazy
Durham Goode <durham@fb.com>
parents: 31433
diff changeset
  1329
                        return True
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
  1330
            return False
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1331
25283
19d0e5efa6ca match: enable 'subinclude:' syntax
Durham Goode <durham@fb.com>
parents: 25250
diff changeset
  1332
        matchfuncs.append(matchsubinclude)
14675
cfc89398f710 match: introduce basic fileset support
Matt Mackall <mpm@selenic.com>
parents: 14674
diff changeset
  1333
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1334
    regex = b''
25239
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1335
    if kindpats:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1336
        if all(k == b'rootfilesin' for k, p, s in kindpats):
40242
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1337
            dirs = {p for k, p, s in kindpats}
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1338
40242
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1339
            def mf(f):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1340
                i = f.rfind(b'/')
40242
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1341
                if i >= 0:
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1342
                    dir = f[:i]
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1343
                else:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1344
                    dir = b'.'
40242
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1345
                return dir in dirs
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1346
40345
d30a19d10441 match: fix up a repr to not crash on Python 3
Augie Fackler <augie@google.com>
parents: 40242
diff changeset
  1347
            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
  1348
            matchfuncs.append(mf)
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1349
        else:
19ed212de2d1 match: optimize matcher when all patterns are of rootfilesin kind
Martin von Zweigbergk <martinvonz@google.com>
parents: 39460
diff changeset
  1350
            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
  1351
            matchfuncs.append(mf)
25239
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1352
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1353
    if len(matchfuncs) == 1:
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1354
        return regex, matchfuncs[0]
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1355
    else:
714f612f2afc match: allow unioning arbitrary match functions
Durham Goode <durham@fb.com>
parents: 25238
diff changeset
  1356
        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
  1357
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1358
40774
8306dac48061 match: extract a literal constant into a symbolic one
Boris Feld <boris.feld@octobus.net>
parents: 40757
diff changeset
  1359
MAX_RE_SIZE = 20000
8306dac48061 match: extract a literal constant into a symbolic one
Boris Feld <boris.feld@octobus.net>
parents: 40757
diff changeset
  1360
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1361
40776
ce401300f981 match: extract function that group regexps
Boris Feld <boris.feld@octobus.net>
parents: 40775
diff changeset
  1362
def _joinregexes(regexps):
ce401300f981 match: extract function that group regexps
Boris Feld <boris.feld@octobus.net>
parents: 40775
diff changeset
  1363
    """gather multiple regular expressions into a single one"""
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1364
    return b'|'.join(regexps)
40776
ce401300f981 match: extract function that group regexps
Boris Feld <boris.feld@octobus.net>
parents: 40775
diff changeset
  1365
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1366
21111
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1367
def _buildregexmatch(kindpats, globsuffix):
9d28fd795215 match: improve documentation - docstrings and more descriptive variable naming
Mads Kiilerich <madski@unity3d.com>
parents: 21079
diff changeset
  1368
    """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
  1369
    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
  1370
4e02f25f31c6 match: test for overflow error in pattern
Boris Feld <boris.feld@octobus.net>
parents: 40774
diff changeset
  1371
    Test too large input
4e02f25f31c6 match: test for overflow error in pattern
Boris Feld <boris.feld@octobus.net>
parents: 40774
diff changeset
  1372
    >>> _buildregexmatch([
40947
9e462fb88f79 match: fix doctest to use bytes instead of str
Augie Fackler <augie@google.com>
parents: 40782
diff changeset
  1373
    ...     (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
  1374
    ... ], b'$')
40775
4e02f25f31c6 match: test for overflow error in pattern
Boris Feld <boris.feld@octobus.net>
parents: 40774
diff changeset
  1375
    Traceback (most recent call last):
4e02f25f31c6 match: test for overflow error in pattern
Boris Feld <boris.feld@octobus.net>
parents: 40774
diff changeset
  1376
    ...
40778
69bd3176da7c match: raise an Abort error instead of OverflowError
Boris Feld <boris.feld@octobus.net>
parents: 40777
diff changeset
  1377
    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
  1378
    """
8574
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1379
    try:
40777
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1380
        allgroups = []
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1381
        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
  1382
        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
  1383
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1384
        startidx = 0
40782
3984409e144b match: drop unnecessary wrapping of regex in group
Martin von Zweigbergk <martinvonz@google.com>
parents: 40781
diff changeset
  1385
        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
  1386
        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
  1387
            piecesize = len(r)
40782
3984409e144b match: drop unnecessary wrapping of regex in group
Martin von Zweigbergk <martinvonz@google.com>
parents: 40781
diff changeset
  1388
            if piecesize > MAX_RE_SIZE:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1389
                msg = _(b"matcher pattern is too long (%d bytes)") % piecesize
40778
69bd3176da7c match: raise an Abort error instead of OverflowError
Boris Feld <boris.feld@octobus.net>
parents: 40777
diff changeset
  1390
                raise error.Abort(msg)
40780
1e019f45fa88 match: make "groupsize" include the trailing "|"
Martin von Zweigbergk <martinvonz@google.com>
parents: 40779
diff changeset
  1391
            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
  1392
                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
  1393
                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
  1394
                startidx = idx
40782
3984409e144b match: drop unnecessary wrapping of regex in group
Martin von Zweigbergk <martinvonz@google.com>
parents: 40781
diff changeset
  1395
                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
  1396
            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
  1397
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1398
        if startidx == 0:
42087
2e2699af5649 match: let regex match function return a boolean
Denis Laxalde <denis@laxalde.org>
parents: 42086
diff changeset
  1399
            matcher = _rematcher(fullregexp)
42099
bccb322f1496 match: fix re2 compability broken in 2e2699af5649
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 42098
diff changeset
  1400
            func = lambda s: bool(matcher(s))
40777
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1401
        else:
3c842749debc match: avoid translating glob to matcher multiple times for large sets
Boris Feld <boris.feld@octobus.net>
parents: 40776
diff changeset
  1402
            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
  1403
            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
  1404
            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
  1405
            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
  1406
        return fullregexp, func
8574
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1407
    except re.error:
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
  1408
        for k, p, s in kindpats:
8574
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1409
            try:
40782
3984409e144b match: drop unnecessary wrapping of regex in group
Martin von Zweigbergk <martinvonz@google.com>
parents: 40781
diff changeset
  1410
                _rematcher(_regex(k, p, globsuffix))
8574
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1411
            except re.error:
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
  1412
                if s:
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1413
                    raise error.Abort(
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1414
                        _(b"%s: invalid pattern (%s): %s") % (s, k, p)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1415
                    )
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
  1416
                else:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1417
                    raise error.Abort(_(b"invalid pattern (%s): %s") % (k, p))
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1418
        raise error.Abort(_(b"invalid pattern"))
8574
63a7ed2128d5 match: unnest functions in _matcher
Matt Mackall <mpm@selenic.com>
parents: 8573
diff changeset
  1419
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1420
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1421
def _patternrootsanddirs(kindpats):
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1422
    '''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
  1423
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1424
    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
  1425
    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
  1426
    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
  1427
    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
  1428
    '''
8576
ec4ed21db4b2 match: split up _normalizepats
Matt Mackall <mpm@selenic.com>
parents: 8575
diff changeset
  1429
    r = []
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1430
    d = []
25213
08a8e9da0ae7 match: add source to kindpats list
Durham Goode <durham@fb.com>
parents: 25195
diff changeset
  1431
    for kind, pat, source in kindpats:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1432
        if kind in (b'glob', b'rootglob'):  # find the non-glob prefix
8584
0f06e72abfdc match: fold _globprefix into _roots
Matt Mackall <mpm@selenic.com>
parents: 8583
diff changeset
  1433
            root = []
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1434
            for p in pat.split(b'/'):
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1435
                if b'[' in p or b'{' in p or b'*' in p or b'?' in p:
8584
0f06e72abfdc match: fold _globprefix into _roots
Matt Mackall <mpm@selenic.com>
parents: 8583
diff changeset
  1436
                    break
0f06e72abfdc match: fold _globprefix into _roots
Matt Mackall <mpm@selenic.com>
parents: 8583
diff changeset
  1437
                root.append(p)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1438
            r.append(b'/'.join(root))
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1439
        elif kind in (b'relpath', b'path'):
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1440
            if pat == b'.':
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1441
                pat = b''
42341
27d6956d386b match: use '' instead of '.' for root directory (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 42329
diff changeset
  1442
            r.append(pat)
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1443
        elif kind in (b'rootfilesin',):
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1444
            if pat == b'.':
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1445
                pat = b''
42341
27d6956d386b match: use '' instead of '.' for root directory (API)
Martin von Zweigbergk <martinvonz@google.com>
parents: 42329
diff changeset
  1446
            d.append(pat)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1447
        else:  # relglob, re, relre
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1448
            r.append(b'')
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1449
    return r, d
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1450
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1451
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1452
def _roots(kindpats):
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1453
    '''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
  1454
    roots, dirs = _patternrootsanddirs(kindpats)
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1455
    return roots
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1456
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1457
38954
5a7df82de142 includematcher: separate "parents" from "dirs"
spectral <spectral@google.com>
parents: 38750
diff changeset
  1458
def _rootsdirsandparents(kindpats):
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1459
    '''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
  1460
38957
f356be1a7ba3 match: correct doc for _rootsdirsandparents after 5a7df82de142
Kyle Lippincott <spectral@google.com>
parents: 38955
diff changeset
  1461
    `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
  1462
    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
  1463
    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
  1464
f356be1a7ba3 match: correct doc for _rootsdirsandparents after 5a7df82de142
Kyle Lippincott <spectral@google.com>
parents: 38955
diff changeset
  1465
    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
  1466
42367
96fc696a9cb2 match: stabilize _rootsdirsandparents doctest
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 42362
diff changeset
  1467
    >>> r = _rootsdirsandparents(
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1468
    ...     [(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
  1469
    ...      (b'glob', b'g*', b'')])
42367
96fc696a9cb2 match: stabilize _rootsdirsandparents doctest
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 42362
diff changeset
  1470
    >>> print(r[0:2], sorted(r[2])) # the set has an unstable output
96fc696a9cb2 match: stabilize _rootsdirsandparents doctest
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 42362
diff changeset
  1471
    (['g/h', 'g/h', ''], []) ['', 'g']
96fc696a9cb2 match: stabilize _rootsdirsandparents doctest
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 42362
diff changeset
  1472
    >>> r = _rootsdirsandparents(
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1473
    ...     [(b'rootfilesin', b'g/h', b''), (b'rootfilesin', b'', b'')])
42367
96fc696a9cb2 match: stabilize _rootsdirsandparents doctest
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 42362
diff changeset
  1474
    >>> print(r[0:2], sorted(r[2])) # the set has an unstable output
96fc696a9cb2 match: stabilize _rootsdirsandparents doctest
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 42362
diff changeset
  1475
    ([], ['g/h', '']) ['', 'g']
96fc696a9cb2 match: stabilize _rootsdirsandparents doctest
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 42362
diff changeset
  1476
    >>> r = _rootsdirsandparents(
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1477
    ...     [(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
  1478
    ...      (b'path', b'', b'')])
42367
96fc696a9cb2 match: stabilize _rootsdirsandparents doctest
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 42362
diff changeset
  1479
    >>> print(r[0:2], sorted(r[2])) # the set has an unstable output
96fc696a9cb2 match: stabilize _rootsdirsandparents doctest
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 42362
diff changeset
  1480
    (['r', 'p/p', ''], []) ['', 'p']
96fc696a9cb2 match: stabilize _rootsdirsandparents doctest
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 42362
diff changeset
  1481
    >>> r = _rootsdirsandparents(
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1482
    ...     [(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
  1483
    ...      (b'relre', b'rr', b'')])
42367
96fc696a9cb2 match: stabilize _rootsdirsandparents doctest
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 42362
diff changeset
  1484
    >>> print(r[0:2], sorted(r[2])) # the set has an unstable output
96fc696a9cb2 match: stabilize _rootsdirsandparents doctest
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 42362
diff changeset
  1485
    (['', '', ''], []) ['']
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1486
    '''
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1487
    r, d = _patternrootsanddirs(kindpats)
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1488
42361
c4b8f8637d7a match: de-flake test-doctest.py by not depending on util.dirs() order
Martin von Zweigbergk <martinvonz@google.com>
parents: 42346
diff changeset
  1489
    p = set()
c4b8f8637d7a match: de-flake test-doctest.py by not depending on util.dirs() order
Martin von Zweigbergk <martinvonz@google.com>
parents: 42346
diff changeset
  1490
    # Add the parents as non-recursive/exact directories, since they must be
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1491
    # scanned to get to either the roots or the other exact directories.
43523
c21aca51b392 utils: move the `dirs` definition in pathutil (API)
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 43506
diff changeset
  1492
    p.update(pathutil.dirs(d))
c21aca51b392 utils: move the `dirs` definition in pathutil (API)
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 43506
diff changeset
  1493
    p.update(pathutil.dirs(r))
31013
693a5bb47854 match: making visitdir() deal with non-recursive entries
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 31012
diff changeset
  1494
39460
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
  1495
    # 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
  1496
    # returning.
35ecaa999a12 match: improve includematcher.visitchildrenset to be much faster and cached
Kyle Lippincott <spectral@google.com>
parents: 39261
diff changeset
  1497
    # 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
  1498
    # '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
  1499
    return r, d, p
8576
ec4ed21db4b2 match: split up _normalizepats
Matt Mackall <mpm@selenic.com>
parents: 8575
diff changeset
  1500
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1501
31012
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1502
def _explicitfiles(kindpats):
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1503
    '''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
  1504
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1505
    >>> _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
  1506
    ['foo/bar']
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 33684
diff changeset
  1507
    >>> _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
  1508
    []
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1509
    '''
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1510
    # 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
  1511
    # directory names).
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1512
    filable = [kp for kp in kindpats if kp[0] not in (b'rootfilesin',)]
31012
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1513
    return _roots(filable)
88358446da16 match: adding support for matching files inside a directory
Rodrigo Damazio Bovendorp <rdamazio@google.com>
parents: 30399
diff changeset
  1514
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1515
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
  1516
def _prefix(kindpats):
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
  1517
    '''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
  1518
    for kind, pat, source in kindpats:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1519
        if kind not in (b'path', b'relpath'):
33405
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
  1520
            return False
6aa643762641 match: inverse _anypats(), making it _prefix()
Martin von Zweigbergk <martinvonz@google.com>
parents: 33380
diff changeset
  1521
    return True
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1522
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1523
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1524
_commentre = None
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1525
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1526
27595
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1527
def readpatternfile(filepath, warn, sourceinfo=False):
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1528
    '''parse a pattern file, returning a list of
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1529
    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
  1530
    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
  1531
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1532
    trailing white space is dropped.
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1533
    the escape character is backslash.
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1534
    comments start with #.
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1535
    empty lines are skipped.
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1536
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1537
    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
  1538
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1539
    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
  1540
    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
  1541
    re:pattern     # non-rooted regular expression
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1542
    glob:pattern   # non-rooted glob
41282
4fab8a7d2d72 match: support rooted globs in hgignore
Valentin Gatien-Baron <vgatien-baron@janestreet.com>
parents: 41129
diff changeset
  1543
    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
  1544
    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
  1545
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1546
    if sourceinfo is set, returns a list of tuples:
42329
c7652f7440d9 rust-filepatterns: call new Rust implementations from Python
Raphaël Gomès <rgomes@octobus.net>
parents: 42195
diff changeset
  1547
    (pattern, lineno, originalline).
c7652f7440d9 rust-filepatterns: call new Rust implementations from Python
Raphaël Gomès <rgomes@octobus.net>
parents: 42195
diff changeset
  1548
    This is useful to debug ignore patterns.
27595
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1549
    '''
25216
dc562165044a ignore: use 'include:' rules instead of custom syntax
Durham Goode <durham@fb.com>
parents: 25215
diff changeset
  1550
40685
e41f6c2e69c4 match: reformat `syntaxes` dictionary for better maintainability
Boris Feld <boris.feld@octobus.net>
parents: 40684
diff changeset
  1551
    syntaxes = {
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1552
        b're': b'relre:',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1553
        b'regexp': b'relre:',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1554
        b'glob': b'relglob:',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1555
        b'rootglob': b'rootglob:',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1556
        b'include': b'include',
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1557
        b'subinclude': b'subinclude',
40685
e41f6c2e69c4 match: reformat `syntaxes` dictionary for better maintainability
Boris Feld <boris.feld@octobus.net>
parents: 40684
diff changeset
  1558
    }
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1559
    syntax = b'relre:'
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1560
    patterns = []
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1561
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1562
    fp = open(filepath, b'rb')
30399
7f3593c29473 match: migrate to util.iterfile
Jun Wu <quark@fb.com>
parents: 29803
diff changeset
  1563
    for lineno, line in enumerate(util.iterfile(fp), start=1):
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1564
        if b"#" in line:
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1565
            global _commentre
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1566
            if not _commentre:
31420
40704098853f match: make regular expression bytes to prevent TypeError
Pulkit Goyal <7895pulkit@gmail.com>
parents: 31403
diff changeset
  1567
                _commentre = util.re.compile(br'((?:^|[^\\])(?:\\\\)*)#.*')
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1568
            # 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
  1569
            m = _commentre.search(line)
d500341e4f55 match: use re2 in readpatternfile if possible
Bryan O'Sullivan <bos@serpentine.com>
parents: 26781
diff changeset
  1570
            if m:
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1571
                line = line[: m.end(1)]
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1572
            # fixup properly escaped comments that survived the above
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1573
            line = line.replace(b"\\#", b"#")
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1574
        line = line.rstrip()
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1575
        if not line:
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1576
            continue
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1577
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1578
        if line.startswith(b'syntax:'):
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1579
            s = line[7:].strip()
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1580
            try:
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1581
                syntax = syntaxes[s]
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1582
            except KeyError:
25214
08703b10c3ae match: add optional warn argument
Durham Goode <durham@fb.com>
parents: 25213
diff changeset
  1583
                if warn:
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1584
                    warn(
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1585
                        _(b"%s: ignoring invalid syntax '%s'\n") % (filepath, s)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1586
                    )
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1587
            continue
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1588
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1589
        linesyntax = syntax
43106
d783f945a701 py3: finish porting iteritems() to pycompat and remove source transformer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 43085
diff changeset
  1590
        for s, rels in pycompat.iteritems(syntaxes):
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1591
            if line.startswith(rels):
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1592
                linesyntax = rels
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1593
                line = line[len(rels) :]
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1594
                break
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
  1595
            elif line.startswith(s + b':'):
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1596
                linesyntax = rels
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 42864
diff changeset
  1597
                line = line[len(s) + 1 :]
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1598
                break
27595
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1599
        if sourceinfo:
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1600
            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
  1601
        else:
9e2d01707e71 match: add option to return line and lineno from readpattern
Laurent Charignon <lcharignon@fb.com>
parents: 27343
diff changeset
  1602
            patterns.append(linesyntax + line)
25167
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1603
    fp.close()
6f7048cc2419 ignore: move readpatternfile to match.py
Durham Goode <durham@fb.com>
parents: 25122
diff changeset
  1604
    return patterns