view mercurial/match.py @ 23785:cb99bacb9b4e

branchcache: introduce revbranchcache for caching of revision branch names

It is expensive to retrieve the branch name of a revision. Very expensive when
creating a changectx and calling .branch() every time - slightly less when
using changelog.branchinfo().

Now, to speed things up, provide a way to cache the results on disk in an
efficient format. Each branchname is assigned a number, and for each revision
we store the number of the corresponding branch name. The branch names are
stored in a dedicated file which is strictly append only.

Branch names are usually reused across several revisions, and the total list of
branch names will thus be so small that it is feasible to read the whole set of
names before using the cache. It will however do that it might be more
efficient to use the changelog for retrieving the branch info for a single
revision.

The revision entries are stored in another file. This file is usually append
only, but if the repository has been modified, the file will be truncated and
the relevant parts rewritten on demand.

The entries for each revision are 8 bytes each, and the whole revision file
will thus be 1/8 of 00changelog.i.

Each revision entry contains the first 4 bytes of the corresponding node hash.
This is used as a check sum that always is verified before the entry is used.
That check is relatively expensive but it makes sure history modification is
detected and handled correctly. It will also detect and handle most revision
file corruptions.

This is just a cache. A new format can always be introduced if other
requirements or ideas make that seem like a good idea. Rebuilding the cache is
not really more expensive than it was to run for example 'hg log -b branchname'
before this cache was introduced.

This new method is still unused but promise to make some operations several
times faster once it actually is used.

Abandoning Python 2.4 would make it possible to implement this more efficiently
by using struct classes and pack_into. The Python code could probably also be
micro optimized or it could be implemented very efficiently in C where it would
be easy to control the data access.

# match.py - filename matching
#
#  Copyright 2008, 2009 Matt Mackall <[e-mail]> and others
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.

import re
import util, pathutil
from i18n import _

def _rematcher(regex):
    '''compile the regexp with the best available regexp engine and return a
    matcher function'''
    m = util.re.compile(regex)
    try:
        # slightly faster, provided by facebook's re2 bindings
        return m.test_match
    except AttributeError:
        return m.match

def _expandsets(kindpats, ctx):
    '''Returns the kindpats list with the 'set' patterns expanded.'''
    fset = set()
    other = []

    for kind, pat in kindpats:
        if kind == 'set':
            if not ctx:
                raise util.Abort("fileset expression with no context")
            s = ctx.getfileset(pat)
            fset.update(s)
            continue
        other.append((kind, pat))
    return fset, other

class match(object):
    def __init__(self, root, cwd, patterns, include=[], exclude=[],
                 default='glob', exact=False, auditor=None, ctx=None):
        """build an object to match a set of file patterns

        arguments:
        root - the canonical root of the tree you're matching against
        cwd - the current working directory, if relevant
        patterns - patterns to find
        include - patterns to include (unless they are excluded)
        exclude - patterns to exclude (even if they are included)
        default - if a pattern in patterns has no explicit type, assume this one
        exact - patterns are actually filenames (include/exclude still apply)

        a pattern is one of:
        'glob:<glob>' - a glob relative to cwd
        're:<regexp>' - a regular expression
        'path:<path>' - a path relative to repository root
        'relglob:<glob>' - an unrooted glob (*.c matches C files in all dirs)
        'relpath:<path>' - a path relative to cwd
        'relre:<regexp>' - a regexp that needn't match the start of a name
        'set:<fileset>' - a fileset expression
        '<something>' - a pattern of the specified default type
        """

        self._root = root
        self._cwd = cwd
        self._files = [] # exact files and roots of patterns
        self._anypats = bool(include or exclude)
        self._ctx = ctx
        self._always = False
        self._pathrestricted = bool(include or exclude or patterns)

        matchfns = []
        if include:
            kindpats = _normalize(include, 'glob', root, cwd, auditor)
            self.includepat, im = _buildmatch(ctx, kindpats, '(?:/|$)')
            matchfns.append(im)
        if exclude:
            kindpats = _normalize(exclude, 'glob', root, cwd, auditor)
            self.excludepat, em = _buildmatch(ctx, kindpats, '(?:/|$)')
            matchfns.append(lambda f: not em(f))
        if exact:
            if isinstance(patterns, list):
                self._files = patterns
            else:
                self._files = list(patterns)
            matchfns.append(self.exact)
        elif patterns:
            kindpats = _normalize(patterns, default, root, cwd, auditor)
            self._files = _roots(kindpats)
            self._anypats = self._anypats or _anypats(kindpats)
            self.patternspat, pm = _buildmatch(ctx, kindpats, '$')
            matchfns.append(pm)

        if not matchfns:
            m = util.always
            self._always = True
        elif len(matchfns) == 1:
            m = matchfns[0]
        else:
            def m(f):
                for matchfn in matchfns:
                    if not matchfn(f):
                        return False
                return True

        self.matchfn = m
        self._fmap = set(self._files)

    def __call__(self, fn):
        return self.matchfn(fn)
    def __iter__(self):
        for f in self._files:
            yield f

    # Callbacks related to how the matcher is used by dirstate.walk.
    # Subscribers to these events must monkeypatch the matcher object.
    def bad(self, f, msg):
        '''Callback from dirstate.walk for each explicit file that can't be
        found/accessed, with an error message.'''
        pass

    # If an explicitdir is set, it will be called when an explicitly listed
    # directory is visited.
    explicitdir = None

    # If an traversedir is set, it will be called when a directory discovered
    # by recursive traversal is visited.
    traversedir = None

    def abs(self, f):
        '''Convert a repo path back to path that is relative to the root of the
        matcher.'''
        return f

    def rel(self, f):
        '''Convert repo path back to path that is relative to cwd of matcher.'''
        return util.pathto(self._root, self._cwd, f)

    def uipath(self, f):
        '''Convert repo path to a display path.  If patterns or -I/-X were used
        to create this matcher, the display path will be relative to cwd.
        Otherwise it is relative to the root of the repo.'''
        return (self._pathrestricted and self.rel(f)) or self.abs(f)

    def files(self):
        '''Explicitly listed files or patterns or roots:
        if no patterns or .always(): empty list,
        if exact: list exact files,
        if not .anypats(): list all files and dirs,
        else: optimal roots'''
        return self._files

    def exact(self, f):
        '''Returns True if f is in .files().'''
        return f in self._fmap

    def anypats(self):
        '''Matcher uses patterns or include/exclude.'''
        return self._anypats

    def always(self):
        '''Matcher will match everything and .files() will be empty
        - optimization might be possible and necessary.'''
        return self._always

def exact(root, cwd, files):
    return match(root, cwd, files, exact=True)

def always(root, cwd):
    return match(root, cwd, [])

class narrowmatcher(match):
    """Adapt a matcher to work on a subdirectory only.

    The paths are remapped to remove/insert the path as needed:

    >>> m1 = match('root', '', ['a.txt', 'sub/b.txt'])
    >>> m2 = narrowmatcher('sub', m1)
    >>> bool(m2('a.txt'))
    False
    >>> bool(m2('b.txt'))
    True
    >>> bool(m2.matchfn('a.txt'))
    False
    >>> bool(m2.matchfn('b.txt'))
    True
    >>> m2.files()
    ['b.txt']
    >>> m2.exact('b.txt')
    True
    >>> util.pconvert(m2.rel('b.txt'))
    'sub/b.txt'
    >>> def bad(f, msg):
    ...     print "%s: %s" % (f, msg)
    >>> m1.bad = bad
    >>> m2.bad('x.txt', 'No such file')
    sub/x.txt: No such file
    >>> m2.abs('c.txt')
    'sub/c.txt'
    """

    def __init__(self, path, matcher):
        self._root = matcher._root
        self._cwd = matcher._cwd
        self._path = path
        self._matcher = matcher
        self._always = matcher._always
        self._pathrestricted = matcher._pathrestricted

        self._files = [f[len(path) + 1:] for f in matcher._files
                       if f.startswith(path + "/")]
        self._anypats = matcher._anypats
        self.matchfn = lambda fn: matcher.matchfn(self._path + "/" + fn)
        self._fmap = set(self._files)

    def abs(self, f):
        return self._matcher.abs(self._path + "/" + f)

    def bad(self, f, msg):
        self._matcher.bad(self._path + "/" + f, msg)

    def rel(self, f):
        return self._matcher.rel(self._path + "/" + f)

def patkind(pattern, default=None):
    '''If pattern is 'kind:pat' with a known kind, return kind.'''
    return _patsplit(pattern, default)[0]

def _patsplit(pattern, default):
    """Split a string into the optional pattern kind prefix and the actual
    pattern."""
    if ':' in pattern:
        kind, pat = pattern.split(':', 1)
        if kind in ('re', 'glob', 'path', 'relglob', 'relpath', 'relre',
                    'listfile', 'listfile0', 'set'):
            return kind, pat
    return default, pattern

def _globre(pat):
    r'''Convert an extended glob string to a regexp string.

    >>> print _globre(r'?')
    .
    >>> print _globre(r'*')
    [^/]*
    >>> print _globre(r'**')
    .*
    >>> print _globre(r'**/a')
    (?:.*/)?a
    >>> print _globre(r'a/**/b')
    a\/(?:.*/)?b
    >>> print _globre(r'[a*?!^][^b][!c]')
    [a*?!^][\^b][^c]
    >>> print _globre(r'{a,b}')
    (?:a|b)
    >>> print _globre(r'.\*\?')
    \.\*\?
    '''
    i, n = 0, len(pat)
    res = ''
    group = 0
    escape = util.re.escape
    def peek():
        return i < n and pat[i]
    while i < n:
        c = pat[i]
        i += 1
        if c not in '*?[{},\\':
            res += escape(c)
        elif c == '*':
            if peek() == '*':
                i += 1
                if peek() == '/':
                    i += 1
                    res += '(?:.*/)?'
                else:
                    res += '.*'
            else:
                res += '[^/]*'
        elif c == '?':
            res += '.'
        elif c == '[':
            j = i
            if j < n and pat[j] in '!]':
                j += 1
            while j < n and pat[j] != ']':
                j += 1
            if j >= n:
                res += '\\['
            else:
                stuff = pat[i:j].replace('\\','\\\\')
                i = j + 1
                if stuff[0] == '!':
                    stuff = '^' + stuff[1:]
                elif stuff[0] == '^':
                    stuff = '\\' + stuff
                res = '%s[%s]' % (res, stuff)
        elif c == '{':
            group += 1
            res += '(?:'
        elif c == '}' and group:
            res += ')'
            group -= 1
        elif c == ',' and group:
            res += '|'
        elif c == '\\':
            p = peek()
            if p:
                i += 1
                res += escape(p)
            else:
                res += escape(c)
        else:
            res += escape(c)
    return res

def _regex(kind, pat, globsuffix):
    '''Convert a (normalized) pattern of any kind into a regular expression.
    globsuffix is appended to the regexp of globs.'''
    if not pat:
        return ''
    if kind == 're':
        return pat
    if kind == 'path':
        return '^' + util.re.escape(pat) + '(?:/|$)'
    if kind == 'relglob':
        return '(?:|.*/)' + _globre(pat) + globsuffix
    if kind == 'relpath':
        return util.re.escape(pat) + '(?:/|$)'
    if kind == 'relre':
        if pat.startswith('^'):
            return pat
        return '.*' + pat
    return _globre(pat) + globsuffix

def _buildmatch(ctx, kindpats, globsuffix):
    '''Return regexp string and a matcher function for kindpats.
    globsuffix is appended to the regexp of globs.'''
    fset, kindpats = _expandsets(kindpats, ctx)
    if not kindpats:
        return "", fset.__contains__

    regex, mf = _buildregexmatch(kindpats, globsuffix)
    if fset:
        return regex, lambda f: f in fset or mf(f)
    return regex, mf

def _buildregexmatch(kindpats, globsuffix):
    """Build a match function from a list of kinds and kindpats,
    return regexp string and a matcher function."""
    try:
        regex = '(?:%s)' % '|'.join([_regex(k, p, globsuffix)
                                     for (k, p) in kindpats])
        if len(regex) > 20000:
            raise OverflowError
        return regex, _rematcher(regex)
    except OverflowError:
        # We're using a Python with a tiny regex engine and we
        # made it explode, so we'll divide the pattern list in two
        # until it works
        l = len(kindpats)
        if l < 2:
            raise
        regexa, a = _buildregexmatch(kindpats[:l//2], globsuffix)
        regexb, b = _buildregexmatch(kindpats[l//2:], globsuffix)
        return regex, lambda s: a(s) or b(s)
    except re.error:
        for k, p in kindpats:
            try:
                _rematcher('(?:%s)' % _regex(k, p, globsuffix))
            except re.error:
                raise util.Abort(_("invalid pattern (%s): %s") % (k, p))
        raise util.Abort(_("invalid pattern"))

def _normalize(patterns, default, root, cwd, auditor):
    '''Convert 'kind:pat' from the patterns list to tuples with kind and
    normalized and rooted patterns and with listfiles expanded.'''
    kindpats = []
    for kind, pat in [_patsplit(p, default) for p in patterns]:
        if kind in ('glob', 'relpath'):
            pat = pathutil.canonpath(root, cwd, pat, auditor)
        elif kind in ('relglob', 'path'):
            pat = util.normpath(pat)
        elif kind in ('listfile', 'listfile0'):
            try:
                files = util.readfile(pat)
                if kind == 'listfile0':
                    files = files.split('\0')
                else:
                    files = files.splitlines()
                files = [f for f in files if f]
            except EnvironmentError:
                raise util.Abort(_("unable to read file list (%s)") % pat)
            kindpats += _normalize(files, default, root, cwd, auditor)
            continue
        # else: re or relre - which cannot be normalized
        kindpats.append((kind, pat))
    return kindpats

def _roots(kindpats):
    '''return roots and exact explicitly listed files from patterns

    >>> _roots([('glob', 'g/*'), ('glob', 'g'), ('glob', 'g*')])
    ['g', 'g', '.']
    >>> _roots([('relpath', 'r'), ('path', 'p/p'), ('path', '')])
    ['r', 'p/p', '.']
    >>> _roots([('relglob', 'rg*'), ('re', 're/'), ('relre', 'rr')])
    ['.', '.', '.']
    '''
    r = []
    for kind, pat in kindpats:
        if kind == 'glob': # find the non-glob prefix
            root = []
            for p in pat.split('/'):
                if '[' in p or '{' in p or '*' in p or '?' in p:
                    break
                root.append(p)
            r.append('/'.join(root) or '.')
        elif kind in ('relpath', 'path'):
            r.append(pat or '.')
        else: # relglob, re, relre
            r.append('.')
    return r

def _anypats(kindpats):
    for kind, pat in kindpats:
        if kind in ('glob', 're', 'relglob', 'relre', 'set'):
            return True