py3: add utility to forward __str__() to __bytes__()

It calls unifromlocal() instead of sysstr() because __bytes__() may contain
locale-dependent values such as paths.

mercurial/context.py

@@ -1,2312 +1,2303 b''
 # context.py - changeset and file context objects for mercurial
 #
 # Copyright 2006, 2007 Matt Mackall <[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.
 
 from __future__ import absolute_import
 
 import errno
 import os
 import re
 import stat
 
 from .i18n import _
 from .node import (
     addednodeid,
     bin,
     hex,
     modifiednodeid,
     nullid,
     nullrev,
     short,
     wdirid,
     wdirnodes,
     wdirrev,
 )
 from . import (
     encoding,
     error,
     fileset,
     match as matchmod,
     mdiff,
     obsolete as obsmod,
     patch,
     phases,
     pycompat,
     repoview,
     revlog,
     scmutil,
     subrepo,
     util,
 )
 
 propertycache = util.propertycache
 
 nonascii = re.compile(r'[^\x21-\x7f]').search
 
 class basectx(object):
     """A basectx object represents the common logic for its children:
     changectx: read-only context that is already present in the repo,
     workingctx: a context that represents the working directory and can
                 be committed,
     memctx: a context that represents changes in-memory and can also
             be committed."""
     def __new__(cls, repo, changeid='', *args, **kwargs):
         if isinstance(changeid, basectx):
             return changeid
 
         o = super(basectx, cls).__new__(cls)
 
         o._repo = repo
         o._rev = nullrev
         o._node = nullid
 
         return o
 
-    def __str__(self):
-        r = short(self.node())
-        if pycompat.ispy3:
-            return r.decode('ascii')
-        return r
-
     def __bytes__(self):
         return short(self.node())
 
+    __str__ = encoding.strmethod(__bytes__)
+
     def __int__(self):
         return self.rev()
 
     def __repr__(self):
         return r"<%s %s>" % (type(self).__name__, str(self))
 
     def __eq__(self, other):
         try:
             return type(self) == type(other) and self._rev == other._rev
         except AttributeError:
             return False
 
     def __ne__(self, other):
         return not (self == other)
 
     def __contains__(self, key):
         return key in self._manifest
 
     def __getitem__(self, key):
         return self.filectx(key)
 
     def __iter__(self):
         return iter(self._manifest)
 
     def _buildstatusmanifest(self, status):
         """Builds a manifest that includes the given status results, if this is
         a working copy context. For non-working copy contexts, it just returns
         the normal manifest."""
         return self.manifest()
 
     def _matchstatus(self, other, match):
         """return match.always if match is none
 
         This internal method provides a way for child objects to override the
         match operator.
         """
         return match or matchmod.always(self._repo.root, self._repo.getcwd())
 
     def _buildstatus(self, other, s, match, listignored, listclean,
                      listunknown):
         """build a status with respect to another context"""
         # Load earliest manifest first for caching reasons. More specifically,
         # if you have revisions 1000 and 1001, 1001 is probably stored as a
         # delta against 1000. Thus, if you read 1000 first, we'll reconstruct
         # 1000 and cache it so that when you read 1001, we just need to apply a
         # delta to what's in the cache. So that's one full reconstruction + one
         # delta application.
         mf2 = None
         if self.rev() is not None and self.rev() < other.rev():
             mf2 = self._buildstatusmanifest(s)
         mf1 = other._buildstatusmanifest(s)
         if mf2 is None:
             mf2 = self._buildstatusmanifest(s)
 
         modified, added = [], []
         removed = []
         clean = []
         deleted, unknown, ignored = s.deleted, s.unknown, s.ignored
         deletedset = set(deleted)
         d = mf1.diff(mf2, match=match, clean=listclean)
         for fn, value in d.iteritems():
             if fn in deletedset:
                 continue
             if value is None:
                 clean.append(fn)
                 continue
             (node1, flag1), (node2, flag2) = value
             if node1 is None:
                 added.append(fn)
             elif node2 is None:
                 removed.append(fn)
             elif flag1 != flag2:
                 modified.append(fn)
             elif node2 not in wdirnodes:
                 # When comparing files between two commits, we save time by
                 # not comparing the file contents when the nodeids differ.
                 # Note that this means we incorrectly report a reverted change
                 # to a file as a modification.
                 modified.append(fn)
             elif self[fn].cmp(other[fn]):
                 modified.append(fn)
             else:
                 clean.append(fn)
 
         if removed:
             # need to filter files if they are already reported as removed
             unknown = [fn for fn in unknown if fn not in mf1 and
                                                (not match or match(fn))]
             ignored = [fn for fn in ignored if fn not in mf1 and
                                                (not match or match(fn))]
             # if they're deleted, don't report them as removed
             removed = [fn for fn in removed if fn not in deletedset]
 
         return scmutil.status(modified, added, removed, deleted, unknown,
                               ignored, clean)
 
     @propertycache
     def substate(self):
         return subrepo.state(self, self._repo.ui)
 
     def subrev(self, subpath):
         return self.substate[subpath][1]
 
     def rev(self):
         return self._rev
     def node(self):
         return self._node
     def hex(self):
         return hex(self.node())
     def manifest(self):
         return self._manifest
     def manifestctx(self):
         return self._manifestctx
     def repo(self):
         return self._repo
     def phasestr(self):
         return phases.phasenames[self.phase()]
     def mutable(self):
         return self.phase() > phases.public
 
     def getfileset(self, expr):
         return fileset.getfileset(self, expr)
 
     def obsolete(self):
         """True if the changeset is obsolete"""
         return self.rev() in obsmod.getrevs(self._repo, 'obsolete')
 
     def extinct(self):
         """True if the changeset is extinct"""
         return self.rev() in obsmod.getrevs(self._repo, 'extinct')
 
     def unstable(self):
         """True if the changeset is not obsolete but it's ancestor are"""
         return self.rev() in obsmod.getrevs(self._repo, 'unstable')
 
     def bumped(self):
         """True if the changeset try to be a successor of a public changeset
 
         Only non-public and non-obsolete changesets may be bumped.
         """
         return self.rev() in obsmod.getrevs(self._repo, 'bumped')
 
     def divergent(self):
         """Is a successors of a changeset with multiple possible successors set
 
         Only non-public and non-obsolete changesets may be divergent.
         """
         return self.rev() in obsmod.getrevs(self._repo, 'divergent')
 
     def troubled(self):
         """True if the changeset is either unstable, bumped or divergent"""
         return self.unstable() or self.bumped() or self.divergent()
 
     def troubles(self):
         """return the list of troubles affecting this changesets.
 
         Troubles are returned as strings. possible values are:
         - unstable,
         - bumped,
         - divergent.
         """
         troubles = []
         if self.unstable():
             troubles.append('unstable')
         if self.bumped():
             troubles.append('bumped')
         if self.divergent():
             troubles.append('divergent')
         return troubles
 
     def parents(self):
         """return contexts for each parent changeset"""
         return self._parents
 
     def p1(self):
         return self._parents[0]
 
     def p2(self):
         parents = self._parents
         if len(parents) == 2:
             return parents[1]
         return changectx(self._repo, nullrev)
 
     def _fileinfo(self, path):
         if r'_manifest' in self.__dict__:
             try:
                 return self._manifest[path], self._manifest.flags(path)
             except KeyError:
                 raise error.ManifestLookupError(self._node, path,
                                                 _('not found in manifest'))
         if r'_manifestdelta' in self.__dict__ or path in self.files():
             if path in self._manifestdelta:
                 return (self._manifestdelta[path],
                         self._manifestdelta.flags(path))
         mfl = self._repo.manifestlog
         try:
             node, flag = mfl[self._changeset.manifest].find(path)
         except KeyError:
             raise error.ManifestLookupError(self._node, path,
                                             _('not found in manifest'))
 
         return node, flag
 
     def filenode(self, path):
         return self._fileinfo(path)[0]
 
     def flags(self, path):
         try:
             return self._fileinfo(path)[1]
         except error.LookupError:
             return ''
 
     def sub(self, path, allowcreate=True):
         '''return a subrepo for the stored revision of path, never wdir()'''
         return subrepo.subrepo(self, path, allowcreate=allowcreate)
 
     def nullsub(self, path, pctx):
         return subrepo.nullsubrepo(self, path, pctx)
 
     def workingsub(self, path):
         '''return a subrepo for the stored revision, or wdir if this is a wdir
         context.
         '''
         return subrepo.subrepo(self, path, allowwdir=True)
 
     def match(self, pats=None, include=None, exclude=None, default='glob',
               listsubrepos=False, badfn=None):
         r = self._repo
         return matchmod.match(r.root, r.getcwd(), pats,
                               include, exclude, default,
                               auditor=r.nofsauditor, ctx=self,
                               listsubrepos=listsubrepos, badfn=badfn)
 
     def diff(self, ctx2=None, match=None, **opts):
         """Returns a diff generator for the given contexts and matcher"""
         if ctx2 is None:
             ctx2 = self.p1()
         if ctx2 is not None:
             ctx2 = self._repo[ctx2]
         diffopts = patch.diffopts(self._repo.ui, opts)
         return patch.diff(self._repo, ctx2, self, match=match, opts=diffopts)
 
     def dirs(self):
         return self._manifest.dirs()
 
     def hasdir(self, dir):
         return self._manifest.hasdir(dir)
 
     def status(self, other=None, match=None, listignored=False,
                listclean=False, listunknown=False, listsubrepos=False):
         """return status of files between two nodes or node and working
         directory.
 
         If other is None, compare this node with working directory.
 
         returns (modified, added, removed, deleted, unknown, ignored, clean)
         """
 
         ctx1 = self
         ctx2 = self._repo[other]
 
         # This next code block is, admittedly, fragile logic that tests for
         # reversing the contexts and wouldn't need to exist if it weren't for
         # the fast (and common) code path of comparing the working directory
         # with its first parent.
         #
         # What we're aiming for here is the ability to call:
         #
         # workingctx.status(parentctx)
         #
         # If we always built the manifest for each context and compared those,
         # then we'd be done. But the special case of the above call means we
         # just copy the manifest of the parent.
         reversed = False
         if (not isinstance(ctx1, changectx)
             and isinstance(ctx2, changectx)):
             reversed = True
             ctx1, ctx2 = ctx2, ctx1
 
         match = ctx2._matchstatus(ctx1, match)
         r = scmutil.status([], [], [], [], [], [], [])
         r = ctx2._buildstatus(ctx1, r, match, listignored, listclean,
                               listunknown)
 
         if reversed:
             # Reverse added and removed. Clear deleted, unknown and ignored as
             # these make no sense to reverse.
             r = scmutil.status(r.modified, r.removed, r.added, [], [], [],
                                r.clean)
 
         if listsubrepos:
             for subpath, sub in scmutil.itersubrepos(ctx1, ctx2):
                 try:
                     rev2 = ctx2.subrev(subpath)
                 except KeyError:
                     # A subrepo that existed in node1 was deleted between
                     # node1 and node2 (inclusive). Thus, ctx2's substate
                     # won't contain that subpath. The best we can do ignore it.
                     rev2 = None
                 submatch = matchmod.subdirmatcher(subpath, match)
                 s = sub.status(rev2, match=submatch, ignored=listignored,
                                clean=listclean, unknown=listunknown,
                                listsubrepos=True)
                 for rfiles, sfiles in zip(r, s):
                     rfiles.extend("%s/%s" % (subpath, f) for f in sfiles)
 
         for l in r:
             l.sort()
 
         return r
 
 def _filterederror(repo, changeid):
     """build an exception to be raised about a filtered changeid
 
     This is extracted in a function to help extensions (eg: evolve) to
     experiment with various message variants."""
     if repo.filtername.startswith('visible'):
         msg = _("hidden revision '%s'") % changeid
         hint = _('use --hidden to access hidden revisions')
         return error.FilteredRepoLookupError(msg, hint=hint)
     msg = _("filtered revision '%s' (not in '%s' subset)")
     msg %= (changeid, repo.filtername)
     return error.FilteredRepoLookupError(msg)
 
 class changectx(basectx):
     """A changecontext object makes access to data related to a particular
     changeset convenient. It represents a read-only context already present in
     the repo."""
     def __init__(self, repo, changeid=''):
         """changeid is a revision number, node, or tag"""
 
         # since basectx.__new__ already took care of copying the object, we
         # don't need to do anything in __init__, so we just exit here
         if isinstance(changeid, basectx):
             return
 
         if changeid == '':
             changeid = '.'
         self._repo = repo
 
         try:
             if isinstance(changeid, int):
                 self._node = repo.changelog.node(changeid)
                 self._rev = changeid
                 return
             if not pycompat.ispy3 and isinstance(changeid, long):
                 changeid = str(changeid)
             if changeid == 'null':
                 self._node = nullid
                 self._rev = nullrev
                 return
             if changeid == 'tip':
                 self._node = repo.changelog.tip()
                 self._rev = repo.changelog.rev(self._node)
                 return
             if changeid == '.' or changeid == repo.dirstate.p1():
                 # this is a hack to delay/avoid loading obsmarkers
                 # when we know that '.' won't be hidden
                 self._node = repo.dirstate.p1()
                 self._rev = repo.unfiltered().changelog.rev(self._node)
                 return
             if len(changeid) == 20:
                 try:
                     self._node = changeid
                     self._rev = repo.changelog.rev(changeid)
                     return
                 except error.FilteredRepoLookupError:
                     raise
                 except LookupError:
                     pass
 
             try:
                 r = int(changeid)
                 if '%d' % r != changeid:
                     raise ValueError
                 l = len(repo.changelog)
                 if r < 0:
                     r += l
                 if r < 0 or r >= l and r != wdirrev:
                     raise ValueError
                 self._rev = r
                 self._node = repo.changelog.node(r)
                 return
             except error.FilteredIndexError:
                 raise
             except (ValueError, OverflowError, IndexError):
                 pass
 
             if len(changeid) == 40:
                 try:
                     self._node = bin(changeid)
                     self._rev = repo.changelog.rev(self._node)
                     return
                 except error.FilteredLookupError:
                     raise
                 except (TypeError, LookupError):
                     pass
 
             # lookup bookmarks through the name interface
             try:
                 self._node = repo.names.singlenode(repo, changeid)
                 self._rev = repo.changelog.rev(self._node)
                 return
             except KeyError:
                 pass
             except error.FilteredRepoLookupError:
                 raise
             except error.RepoLookupError:
                 pass
 
             self._node = repo.unfiltered().changelog._partialmatch(changeid)
             if self._node is not None:
                 self._rev = repo.changelog.rev(self._node)
                 return
 
             # lookup failed
             # check if it might have come from damaged dirstate
             #
             # XXX we could avoid the unfiltered if we had a recognizable
             # exception for filtered changeset access
             if changeid in repo.unfiltered().dirstate.parents():
                 msg = _("working directory has unknown parent '%s'!")
                 raise error.Abort(msg % short(changeid))
             try:
                 if len(changeid) == 20 and nonascii(changeid):
                     changeid = hex(changeid)
             except TypeError:
                 pass
         except (error.FilteredIndexError, error.FilteredLookupError,
                 error.FilteredRepoLookupError):
             raise _filterederror(repo, changeid)
         except IndexError:
             pass
         raise error.RepoLookupError(
             _("unknown revision '%s'") % changeid)
 
     def __hash__(self):
         try:
             return hash(self._rev)
         except AttributeError:
             return id(self)
 
     def __nonzero__(self):
         return self._rev != nullrev
 
     __bool__ = __nonzero__
 
     @propertycache
     def _changeset(self):
         return self._repo.changelog.changelogrevision(self.rev())
 
     @propertycache
     def _manifest(self):
         return self._manifestctx.read()
 
     @property
     def _manifestctx(self):
         return self._repo.manifestlog[self._changeset.manifest]
 
     @propertycache
     def _manifestdelta(self):
         return self._manifestctx.readdelta()
 
     @propertycache
     def _parents(self):
         repo = self._repo
         p1, p2 = repo.changelog.parentrevs(self._rev)
         if p2 == nullrev:
             return [changectx(repo, p1)]
         return [changectx(repo, p1), changectx(repo, p2)]
 
     def changeset(self):
         c = self._changeset
         return (
             c.manifest,
             c.user,
             c.date,
             c.files,
             c.description,
             c.extra,
         )
     def manifestnode(self):
         return self._changeset.manifest
 
     def user(self):
         return self._changeset.user
     def date(self):
         return self._changeset.date
     def files(self):
         return self._changeset.files
     def description(self):
         return self._changeset.description
     def branch(self):
         return encoding.tolocal(self._changeset.extra.get("branch"))
     def closesbranch(self):
         return 'close' in self._changeset.extra
     def extra(self):
         return self._changeset.extra
     def tags(self):
         return self._repo.nodetags(self._node)
     def bookmarks(self):
         return self._repo.nodebookmarks(self._node)
     def phase(self):
         return self._repo._phasecache.phase(self._repo, self._rev)
     def hidden(self):
         return self._rev in repoview.filterrevs(self._repo, 'visible')
 
     def children(self):
         """return contexts for each child changeset"""
         c = self._repo.changelog.children(self._node)
         return [changectx(self._repo, x) for x in c]
 
     def ancestors(self):
         for a in self._repo.changelog.ancestors([self._rev]):
             yield changectx(self._repo, a)
 
     def descendants(self):
         for d in self._repo.changelog.descendants([self._rev]):
             yield changectx(self._repo, d)
 
     def filectx(self, path, fileid=None, filelog=None):
         """get a file context from this changeset"""
         if fileid is None:
             fileid = self.filenode(path)
         return filectx(self._repo, path, fileid=fileid,
                        changectx=self, filelog=filelog)
 
     def ancestor(self, c2, warn=False):
         """return the "best" ancestor context of self and c2
 
         If there are multiple candidates, it will show a message and check
         merge.preferancestor configuration before falling back to the
         revlog ancestor."""
         # deal with workingctxs
         n2 = c2._node
         if n2 is None:
             n2 = c2._parents[0]._node
         cahs = self._repo.changelog.commonancestorsheads(self._node, n2)
         if not cahs:
             anc = nullid
         elif len(cahs) == 1:
             anc = cahs[0]
         else:
             # experimental config: merge.preferancestor
             for r in self._repo.ui.configlist('merge', 'preferancestor', ['*']):
                 try:
                     ctx = changectx(self._repo, r)
                 except error.RepoLookupError:
                     continue
                 anc = ctx.node()
                 if anc in cahs:
                     break
             else:
                 anc = self._repo.changelog.ancestor(self._node, n2)
             if warn:
                 self._repo.ui.status(
                     (_("note: using %s as ancestor of %s and %s\n") %
                      (short(anc), short(self._node), short(n2))) +
                     ''.join(_("      alternatively, use --config "
                               "merge.preferancestor=%s\n") %
                             short(n) for n in sorted(cahs) if n != anc))
         return changectx(self._repo, anc)
 
     def descendant(self, other):
         """True if other is descendant of this changeset"""
         return self._repo.changelog.descendant(self._rev, other._rev)
 
     def walk(self, match):
         '''Generates matching file names.'''
 
         # Wrap match.bad method to have message with nodeid
         def bad(fn, msg):
             # The manifest doesn't know about subrepos, so don't complain about
             # paths into valid subrepos.
             if any(fn == s or fn.startswith(s + '/')
                    for s in self.substate):
                 return
             match.bad(fn, _('no such file in rev %s') % self)
 
         m = matchmod.badmatch(match, bad)
         return self._manifest.walk(m)
 
     def matches(self, match):
         return self.walk(match)
 
 class basefilectx(object):
     """A filecontext object represents the common logic for its children:
     filectx: read-only access to a filerevision that is already present
              in the repo,
     workingfilectx: a filecontext that represents files from the working
                     directory,
     memfilectx: a filecontext that represents files in-memory,
     overlayfilectx: duplicate another filecontext with some fields overridden.
     """
     @propertycache
     def _filelog(self):
         return self._repo.file(self._path)
 
     @propertycache
     def _changeid(self):
         if r'_changeid' in self.__dict__:
             return self._changeid
         elif r'_changectx' in self.__dict__:
             return self._changectx.rev()
         elif r'_descendantrev' in self.__dict__:
             # this file context was created from a revision with a known
             # descendant, we can (lazily) correct for linkrev aliases
             return self._adjustlinkrev(self._descendantrev)
         else:
             return self._filelog.linkrev(self._filerev)
 
     @propertycache
     def _filenode(self):
         if r'_fileid' in self.__dict__:
             return self._filelog.lookup(self._fileid)
         else:
             return self._changectx.filenode(self._path)
 
     @propertycache
     def _filerev(self):
         return self._filelog.rev(self._filenode)
 
     @propertycache
     def _repopath(self):
         return self._path
 
     def __nonzero__(self):
         try:
             self._filenode
             return True
         except error.LookupError:
             # file is missing
             return False
 
     __bool__ = __nonzero__
 
-    def __str__(self):
+    def __bytes__(self):
         try:
             return "%s@%s" % (self.path(), self._changectx)
         except error.LookupError:
             return "%s@???" % self.path()
 
-    def __bytes__(self):
-        try:
-            return "%s@%s" % (self.path(), self._changectx)
-        except error.LookupError:
-            return "%s@???" % self.path()
+    __str__ = encoding.strmethod(__bytes__)
 
     def __repr__(self):
         return "<%s %s>" % (type(self).__name__, str(self))
 
     def __hash__(self):
         try:
             return hash((self._path, self._filenode))
         except AttributeError:
             return id(self)
 
     def __eq__(self, other):
         try:
             return (type(self) == type(other) and self._path == other._path
                     and self._filenode == other._filenode)
         except AttributeError:
             return False
 
     def __ne__(self, other):
         return not (self == other)
 
     def filerev(self):
         return self._filerev
     def filenode(self):
         return self._filenode
     @propertycache
     def _flags(self):
         return self._changectx.flags(self._path)
     def flags(self):
         return self._flags
     def filelog(self):
         return self._filelog
     def rev(self):
         return self._changeid
     def linkrev(self):
         return self._filelog.linkrev(self._filerev)
     def node(self):
         return self._changectx.node()
     def hex(self):
         return self._changectx.hex()
     def user(self):
         return self._changectx.user()
     def date(self):
         return self._changectx.date()
     def files(self):
         return self._changectx.files()
     def description(self):
         return self._changectx.description()
     def branch(self):
         return self._changectx.branch()
     def extra(self):
         return self._changectx.extra()
     def phase(self):
         return self._changectx.phase()
     def phasestr(self):
         return self._changectx.phasestr()
     def manifest(self):
         return self._changectx.manifest()
     def changectx(self):
         return self._changectx
     def renamed(self):
         return self._copied
     def repo(self):
         return self._repo
     def size(self):
         return len(self.data())
 
     def path(self):
         return self._path
 
     def isbinary(self):
         try:
             return util.binary(self.data())
         except IOError:
             return False
     def isexec(self):
         return 'x' in self.flags()
     def islink(self):
         return 'l' in self.flags()
 
     def isabsent(self):
         """whether this filectx represents a file not in self._changectx
 
         This is mainly for merge code to detect change/delete conflicts. This is
         expected to be True for all subclasses of basectx."""
         return False
 
     _customcmp = False
     def cmp(self, fctx):
         """compare with other file context
 
         returns True if different than fctx.
         """
         if fctx._customcmp:
             return fctx.cmp(self)
 
         if (fctx._filenode is None
             and (self._repo._encodefilterpats
                  # if file data starts with '\1\n', empty metadata block is
                  # prepended, which adds 4 bytes to filelog.size().
                  or self.size() - 4 == fctx.size())
             or self.size() == fctx.size()):
             return self._filelog.cmp(self._filenode, fctx.data())
 
         return True
 
     def _adjustlinkrev(self, srcrev, inclusive=False):
         """return the first ancestor of <srcrev> introducing <fnode>
 
         If the linkrev of the file revision does not point to an ancestor of
         srcrev, we'll walk down the ancestors until we find one introducing
         this file revision.
 
         :srcrev: the changeset revision we search ancestors from
         :inclusive: if true, the src revision will also be checked
         """
         repo = self._repo
         cl = repo.unfiltered().changelog
         mfl = repo.manifestlog
         # fetch the linkrev
         lkr = self.linkrev()
         # hack to reuse ancestor computation when searching for renames
         memberanc = getattr(self, '_ancestrycontext', None)
         iteranc = None
         if srcrev is None:
             # wctx case, used by workingfilectx during mergecopy
             revs = [p.rev() for p in self._repo[None].parents()]
             inclusive = True # we skipped the real (revless) source
         else:
             revs = [srcrev]
         if memberanc is None:
             memberanc = iteranc = cl.ancestors(revs, lkr,
                                                inclusive=inclusive)
         # check if this linkrev is an ancestor of srcrev
         if lkr not in memberanc:
             if iteranc is None:
                 iteranc = cl.ancestors(revs, lkr, inclusive=inclusive)
             fnode = self._filenode
             path = self._path
             for a in iteranc:
                 ac = cl.read(a) # get changeset data (we avoid object creation)
                 if path in ac[3]: # checking the 'files' field.
                     # The file has been touched, check if the content is
                     # similar to the one we search for.
                     if fnode == mfl[ac[0]].readfast().get(path):
                         return a
             # In theory, we should never get out of that loop without a result.
             # But if manifest uses a buggy file revision (not children of the
             # one it replaces) we could. Such a buggy situation will likely
             # result is crash somewhere else at to some point.
         return lkr
 
     def introrev(self):
         """return the rev of the changeset which introduced this file revision
 
         This method is different from linkrev because it take into account the
         changeset the filectx was created from. It ensures the returned
         revision is one of its ancestors. This prevents bugs from
         'linkrev-shadowing' when a file revision is used by multiple
         changesets.
         """
         lkr = self.linkrev()
         attrs = vars(self)
         noctx = not ('_changeid' in attrs or '_changectx' in attrs)
         if noctx or self.rev() == lkr:
             return self.linkrev()
         return self._adjustlinkrev(self.rev(), inclusive=True)
 
     def _parentfilectx(self, path, fileid, filelog):
         """create parent filectx keeping ancestry info for _adjustlinkrev()"""
         fctx = filectx(self._repo, path, fileid=fileid, filelog=filelog)
         if '_changeid' in vars(self) or '_changectx' in vars(self):
             # If self is associated with a changeset (probably explicitly
             # fed), ensure the created filectx is associated with a
             # changeset that is an ancestor of self.changectx.
             # This lets us later use _adjustlinkrev to get a correct link.
             fctx._descendantrev = self.rev()
             fctx._ancestrycontext = getattr(self, '_ancestrycontext', None)
         elif '_descendantrev' in vars(self):
             # Otherwise propagate _descendantrev if we have one associated.
             fctx._descendantrev = self._descendantrev
             fctx._ancestrycontext = getattr(self, '_ancestrycontext', None)
         return fctx
 
     def parents(self):
         _path = self._path
         fl = self._filelog
         parents = self._filelog.parents(self._filenode)
         pl = [(_path, node, fl) for node in parents if node != nullid]
 
         r = fl.renamed(self._filenode)
         if r:
             # - In the simple rename case, both parent are nullid, pl is empty.
             # - In case of merge, only one of the parent is null id and should
             # be replaced with the rename information. This parent is -always-
             # the first one.
             #
             # As null id have always been filtered out in the previous list
             # comprehension, inserting to 0 will always result in "replacing
             # first nullid parent with rename information.
             pl.insert(0, (r[0], r[1], self._repo.file(r[0])))
 
         return [self._parentfilectx(path, fnode, l) for path, fnode, l in pl]
 
     def p1(self):
         return self.parents()[0]
 
     def p2(self):
         p = self.parents()
         if len(p) == 2:
             return p[1]
         return filectx(self._repo, self._path, fileid=-1, filelog=self._filelog)
 
     def annotate(self, follow=False, linenumber=False, skiprevs=None,
                  diffopts=None):
         '''returns a list of tuples of ((ctx, number), line) for each line
         in the file, where ctx is the filectx of the node where
         that line was last changed; if linenumber parameter is true, number is
         the line number at the first appearance in the managed file, otherwise,
         number has a fixed value of False.
         '''
 
         def lines(text):
             if text.endswith("\n"):
                 return text.count("\n")
             return text.count("\n") + int(bool(text))
 
         if linenumber:
             def decorate(text, rev):
                 return ([(rev, i) for i in xrange(1, lines(text) + 1)], text)
         else:
             def decorate(text, rev):
                 return ([(rev, False)] * lines(text), text)
 
         getlog = util.lrucachefunc(lambda x: self._repo.file(x))
 
         def parents(f):
             # Cut _descendantrev here to mitigate the penalty of lazy linkrev
             # adjustment. Otherwise, p._adjustlinkrev() would walk changelog
             # from the topmost introrev (= srcrev) down to p.linkrev() if it
             # isn't an ancestor of the srcrev.
             f._changeid
             pl = f.parents()
 
             # Don't return renamed parents if we aren't following.
             if not follow:
                 pl = [p for p in pl if p.path() == f.path()]
 
             # renamed filectx won't have a filelog yet, so set it
             # from the cache to save time
             for p in pl:
                 if not '_filelog' in p.__dict__:
                     p._filelog = getlog(p.path())
 
             return pl
 
         # use linkrev to find the first changeset where self appeared
         base = self
         introrev = self.introrev()
         if self.rev() != introrev:
             base = self.filectx(self.filenode(), changeid=introrev)
         if getattr(base, '_ancestrycontext', None) is None:
             cl = self._repo.changelog
             if introrev is None:
                 # wctx is not inclusive, but works because _ancestrycontext
                 # is used to test filelog revisions
                 ac = cl.ancestors([p.rev() for p in base.parents()],
                                   inclusive=True)
             else:
                 ac = cl.ancestors([introrev], inclusive=True)
             base._ancestrycontext = ac
 
         # This algorithm would prefer to be recursive, but Python is a
         # bit recursion-hostile. Instead we do an iterative
         # depth-first search.
 
         # 1st DFS pre-calculates pcache and needed
         visit = [base]
         pcache = {}
         needed = {base: 1}
         while visit:
             f = visit.pop()
             if f in pcache:
                 continue
             pl = parents(f)
             pcache[f] = pl
             for p in pl:
                 needed[p] = needed.get(p, 0) + 1
                 if p not in pcache:
                     visit.append(p)
 
         # 2nd DFS does the actual annotate
         visit[:] = [base]
         hist = {}
         while visit:
             f = visit[-1]
             if f in hist:
                 visit.pop()
                 continue
 
             ready = True
             pl = pcache[f]
             for p in pl:
                 if p not in hist:
                     ready = False
                     visit.append(p)
             if ready:
                 visit.pop()
                 curr = decorate(f.data(), f)
                 skipchild = False
                 if skiprevs is not None:
                     skipchild = f._changeid in skiprevs
                 curr = _annotatepair([hist[p] for p in pl], f, curr, skipchild,
                                      diffopts)
                 for p in pl:
                     if needed[p] == 1:
                         del hist[p]
                         del needed[p]
                     else:
                         needed[p] -= 1
 
                 hist[f] = curr
                 del pcache[f]
 
         return zip(hist[base][0], hist[base][1].splitlines(True))
 
     def ancestors(self, followfirst=False):
         visit = {}
         c = self
         if followfirst:
             cut = 1
         else:
             cut = None
 
         while True:
             for parent in c.parents()[:cut]:
                 visit[(parent.linkrev(), parent.filenode())] = parent
             if not visit:
                 break
             c = visit.pop(max(visit))
             yield c
 
 def _annotatepair(parents, childfctx, child, skipchild, diffopts):
     r'''
     Given parent and child fctxes and annotate data for parents, for all lines
     in either parent that match the child, annotate the child with the parent's
     data.
 
     Additionally, if `skipchild` is True, replace all other lines with parent
     annotate data as well such that child is never blamed for any lines.
 
     >>> oldfctx = 'old'
     >>> p1fctx, p2fctx, childfctx = 'p1', 'p2', 'c'
     >>> olddata = 'a\nb\n'
     >>> p1data = 'a\nb\nc\n'
     >>> p2data = 'a\nc\nd\n'
     >>> childdata = 'a\nb2\nc\nc2\nd\n'
     >>> diffopts = mdiff.diffopts()
 
     >>> def decorate(text, rev):
     ...     return ([(rev, i) for i in xrange(1, text.count('\n') + 1)], text)
 
     Basic usage:
 
     >>> oldann = decorate(olddata, oldfctx)
     >>> p1ann = decorate(p1data, p1fctx)
     >>> p1ann = _annotatepair([oldann], p1fctx, p1ann, False, diffopts)
     >>> p1ann[0]
     [('old', 1), ('old', 2), ('p1', 3)]
     >>> p2ann = decorate(p2data, p2fctx)
     >>> p2ann = _annotatepair([oldann], p2fctx, p2ann, False, diffopts)
     >>> p2ann[0]
     [('old', 1), ('p2', 2), ('p2', 3)]
 
     Test with multiple parents (note the difference caused by ordering):
 
     >>> childann = decorate(childdata, childfctx)
     >>> childann = _annotatepair([p1ann, p2ann], childfctx, childann, False,
     ...                          diffopts)
     >>> childann[0]
     [('old', 1), ('c', 2), ('p2', 2), ('c', 4), ('p2', 3)]
 
     >>> childann = decorate(childdata, childfctx)
     >>> childann = _annotatepair([p2ann, p1ann], childfctx, childann, False,
     ...                          diffopts)
     >>> childann[0]
     [('old', 1), ('c', 2), ('p1', 3), ('c', 4), ('p2', 3)]
 
     Test with skipchild (note the difference caused by ordering):
 
     >>> childann = decorate(childdata, childfctx)
     >>> childann = _annotatepair([p1ann, p2ann], childfctx, childann, True,
     ...                          diffopts)
     >>> childann[0]
     [('old', 1), ('old', 2), ('p2', 2), ('p2', 2), ('p2', 3)]
 
     >>> childann = decorate(childdata, childfctx)
     >>> childann = _annotatepair([p2ann, p1ann], childfctx, childann, True,
     ...                          diffopts)
     >>> childann[0]
     [('old', 1), ('old', 2), ('p1', 3), ('p1', 3), ('p2', 3)]
     '''
     pblocks = [(parent, mdiff.allblocks(parent[1], child[1], opts=diffopts))
                for parent in parents]
 
     if skipchild:
         # Need to iterate over the blocks twice -- make it a list
         pblocks = [(p, list(blocks)) for (p, blocks) in pblocks]
     # Mercurial currently prefers p2 over p1 for annotate.
     # TODO: change this?
     for parent, blocks in pblocks:
         for (a1, a2, b1, b2), t in blocks:
             # Changed blocks ('!') or blocks made only of blank lines ('~')
             # belong to the child.
             if t == '=':
                 child[0][b1:b2] = parent[0][a1:a2]
 
     if skipchild:
         # Now try and match up anything that couldn't be matched,
         # Reversing pblocks maintains bias towards p2, matching above
         # behavior.
         pblocks.reverse()
 
         # The heuristics are:
         # * Work on blocks of changed lines (effectively diff hunks with -U0).
         # This could potentially be smarter but works well enough.
         # * For a non-matching section, do a best-effort fit. Match lines in
         #   diff hunks 1:1, dropping lines as necessary.
         # * Repeat the last line as a last resort.
 
         # First, replace as much as possible without repeating the last line.
         remaining = [(parent, []) for parent, _blocks in pblocks]
         for idx, (parent, blocks) in enumerate(pblocks):
             for (a1, a2, b1, b2), _t in blocks:
                 if a2 - a1 >= b2 - b1:
                     for bk in xrange(b1, b2):
                         if child[0][bk][0] == childfctx:
                             ak = min(a1 + (bk - b1), a2 - 1)
                             child[0][bk] = parent[0][ak]
                 else:
                     remaining[idx][1].append((a1, a2, b1, b2))
 
         # Then, look at anything left, which might involve repeating the last
         # line.
         for parent, blocks in remaining:
             for a1, a2, b1, b2 in blocks:
                 for bk in xrange(b1, b2):
                     if child[0][bk][0] == childfctx:
                         ak = min(a1 + (bk - b1), a2 - 1)
                         child[0][bk] = parent[0][ak]
     return child
 
 class filectx(basefilectx):
     """A filecontext object makes access to data related to a particular
        filerevision convenient."""
     def __init__(self, repo, path, changeid=None, fileid=None,
                  filelog=None, changectx=None):
         """changeid can be a changeset revision, node, or tag.
            fileid can be a file revision or node."""
         self._repo = repo
         self._path = path
 
         assert (changeid is not None
                 or fileid is not None
                 or changectx is not None), \
                 ("bad args: changeid=%r, fileid=%r, changectx=%r"
                  % (changeid, fileid, changectx))
 
         if filelog is not None:
             self._filelog = filelog
 
         if changeid is not None:
             self._changeid = changeid
         if changectx is not None:
             self._changectx = changectx
         if fileid is not None:
             self._fileid = fileid
 
     @propertycache
     def _changectx(self):
         try:
             return changectx(self._repo, self._changeid)
         except error.FilteredRepoLookupError:
             # Linkrev may point to any revision in the repository.  When the
             # repository is filtered this may lead to `filectx` trying to build
             # `changectx` for filtered revision. In such case we fallback to
             # creating `changectx` on the unfiltered version of the reposition.
             # This fallback should not be an issue because `changectx` from
             # `filectx` are not used in complex operations that care about
             # filtering.
             #
             # This fallback is a cheap and dirty fix that prevent several
             # crashes. It does not ensure the behavior is correct. However the
             # behavior was not correct before filtering either and "incorrect
             # behavior" is seen as better as "crash"
             #
             # Linkrevs have several serious troubles with filtering that are
             # complicated to solve. Proper handling of the issue here should be
             # considered when solving linkrev issue are on the table.
             return changectx(self._repo.unfiltered(), self._changeid)
 
     def filectx(self, fileid, changeid=None):
         '''opens an arbitrary revision of the file without
         opening a new filelog'''
         return filectx(self._repo, self._path, fileid=fileid,
                        filelog=self._filelog, changeid=changeid)
 
     def rawdata(self):
         return self._filelog.revision(self._filenode, raw=True)
 
     def rawflags(self):
         """low-level revlog flags"""
         return self._filelog.flags(self._filerev)
 
     def data(self):
         try:
             return self._filelog.read(self._filenode)
         except error.CensoredNodeError:
             if self._repo.ui.config("censor", "policy", "abort") == "ignore":
                 return ""
             raise error.Abort(_("censored node: %s") % short(self._filenode),
                              hint=_("set censor.policy to ignore errors"))
 
     def size(self):
         return self._filelog.size(self._filerev)
 
     @propertycache
     def _copied(self):
         """check if file was actually renamed in this changeset revision
 
         If rename logged in file revision, we report copy for changeset only
         if file revisions linkrev points back to the changeset in question
         or both changeset parents contain different file revisions.
         """
 
         renamed = self._filelog.renamed(self._filenode)
         if not renamed:
             return renamed
 
         if self.rev() == self.linkrev():
             return renamed
 
         name = self.path()
         fnode = self._filenode
         for p in self._changectx.parents():
             try:
                 if fnode == p.filenode(name):
                     return None
             except error.LookupError:
                 pass
         return renamed
 
     def children(self):
         # hard for renames
         c = self._filelog.children(self._filenode)
         return [filectx(self._repo, self._path, fileid=x,
                         filelog=self._filelog) for x in c]
 
 class committablectx(basectx):
     """A committablectx object provides common functionality for a context that
     wants the ability to commit, e.g. workingctx or memctx."""
     def __init__(self, repo, text="", user=None, date=None, extra=None,
                  changes=None):
         self._repo = repo
         self._rev = None
         self._node = None
         self._text = text
         if date:
             self._date = util.parsedate(date)
         if user:
             self._user = user
         if changes:
             self._status = changes
 
         self._extra = {}
         if extra:
             self._extra = extra.copy()
         if 'branch' not in self._extra:
             try:
                 branch = encoding.fromlocal(self._repo.dirstate.branch())
             except UnicodeDecodeError:
                 raise error.Abort(_('branch name not in UTF-8!'))
             self._extra['branch'] = branch
         if self._extra['branch'] == '':
             self._extra['branch'] = 'default'
 
-    def __str__(self):
-        return str(self._parents[0]) + r"+"
-
     def __bytes__(self):
         return bytes(self._parents[0]) + "+"
 
+    __str__ = encoding.strmethod(__bytes__)
+
     def __nonzero__(self):
         return True
 
     __bool__ = __nonzero__
 
     def _buildflagfunc(self):
         # Create a fallback function for getting file flags when the
         # filesystem doesn't support them
 
         copiesget = self._repo.dirstate.copies().get
         parents = self.parents()
         if len(parents) < 2:
             # when we have one parent, it's easy: copy from parent
             man = parents[0].manifest()
             def func(f):
                 f = copiesget(f, f)
                 return man.flags(f)
         else:
             # merges are tricky: we try to reconstruct the unstored
             # result from the merge (issue1802)
             p1, p2 = parents
             pa = p1.ancestor(p2)
             m1, m2, ma = p1.manifest(), p2.manifest(), pa.manifest()
 
             def func(f):
                 f = copiesget(f, f) # may be wrong for merges with copies
                 fl1, fl2, fla = m1.flags(f), m2.flags(f), ma.flags(f)
                 if fl1 == fl2:
                     return fl1
                 if fl1 == fla:
                     return fl2
                 if fl2 == fla:
                     return fl1
                 return '' # punt for conflicts
 
         return func
 
     @propertycache
     def _flagfunc(self):
         return self._repo.dirstate.flagfunc(self._buildflagfunc)
 
     @propertycache
     def _status(self):
         return self._repo.status()
 
     @propertycache
     def _user(self):
         return self._repo.ui.username()
 
     @propertycache
     def _date(self):
         ui = self._repo.ui
         date = ui.configdate('devel', 'default-date')
         if date is None:
             date = util.makedate()
         return date
 
     def subrev(self, subpath):
         return None
 
     def manifestnode(self):
         return None
     def user(self):
         return self._user or self._repo.ui.username()
     def date(self):
         return self._date
     def description(self):
         return self._text
     def files(self):
         return sorted(self._status.modified + self._status.added +
                       self._status.removed)
 
     def modified(self):
         return self._status.modified
     def added(self):
         return self._status.added
     def removed(self):
         return self._status.removed
     def deleted(self):
         return self._status.deleted
     def branch(self):
         return encoding.tolocal(self._extra['branch'])
     def closesbranch(self):
         return 'close' in self._extra
     def extra(self):
         return self._extra
 
     def tags(self):
         return []
 
     def bookmarks(self):
         b = []
         for p in self.parents():
             b.extend(p.bookmarks())
         return b
 
     def phase(self):
         phase = phases.draft # default phase to draft
         for p in self.parents():
             phase = max(phase, p.phase())
         return phase
 
     def hidden(self):
         return False
 
     def children(self):
         return []
 
     def flags(self, path):
         if r'_manifest' in self.__dict__:
             try:
                 return self._manifest.flags(path)
             except KeyError:
                 return ''
 
         try:
             return self._flagfunc(path)
         except OSError:
             return ''
 
     def ancestor(self, c2):
         """return the "best" ancestor context of self and c2"""
         return self._parents[0].ancestor(c2) # punt on two parents for now
 
     def walk(self, match):
         '''Generates matching file names.'''
         return sorted(self._repo.dirstate.walk(match, sorted(self.substate),
                                                True, False))
 
     def matches(self, match):
         return sorted(self._repo.dirstate.matches(match))
 
     def ancestors(self):
         for p in self._parents:
             yield p
         for a in self._repo.changelog.ancestors(
             [p.rev() for p in self._parents]):
             yield changectx(self._repo, a)
 
     def markcommitted(self, node):
         """Perform post-commit cleanup necessary after committing this ctx
 
         Specifically, this updates backing stores this working context
         wraps to reflect the fact that the changes reflected by this
         workingctx have been committed.  For example, it marks
         modified and added files as normal in the dirstate.
 
         """
 
         with self._repo.dirstate.parentchange():
             for f in self.modified() + self.added():
                 self._repo.dirstate.normal(f)
             for f in self.removed():
                 self._repo.dirstate.drop(f)
             self._repo.dirstate.setparents(node)
 
         # write changes out explicitly, because nesting wlock at
         # runtime may prevent 'wlock.release()' in 'repo.commit()'
         # from immediately doing so for subsequent changing files
         self._repo.dirstate.write(self._repo.currenttransaction())
 
     def dirty(self, missing=False, merge=True, branch=True):
         return False
 
 class workingctx(committablectx):
     """A workingctx object makes access to data related to
     the current working directory convenient.
     date - any valid date string or (unixtime, offset), or None.
     user - username string, or None.
     extra - a dictionary of extra values, or None.
     changes - a list of file lists as returned by localrepo.status()
                or None to use the repository status.
     """
     def __init__(self, repo, text="", user=None, date=None, extra=None,
                  changes=None):
         super(workingctx, self).__init__(repo, text, user, date, extra, changes)
 
     def __iter__(self):
         d = self._repo.dirstate
         for f in d:
             if d[f] != 'r':
                 yield f
 
     def __contains__(self, key):
         return self._repo.dirstate[key] not in "?r"
 
     def hex(self):
         return hex(wdirid)
 
     @propertycache
     def _parents(self):
         p = self._repo.dirstate.parents()
         if p[1] == nullid:
             p = p[:-1]
         return [changectx(self._repo, x) for x in p]
 
     def filectx(self, path, filelog=None):
         """get a file context from the working directory"""
         return workingfilectx(self._repo, path, workingctx=self,
                               filelog=filelog)
 
     def dirty(self, missing=False, merge=True, branch=True):
         "check whether a working directory is modified"
         # check subrepos first
         for s in sorted(self.substate):
             if self.sub(s).dirty():
                 return True
         # check current working dir
         return ((merge and self.p2()) or
                 (branch and self.branch() != self.p1().branch()) or
                 self.modified() or self.added() or self.removed() or
                 (missing and self.deleted()))
 
     def add(self, list, prefix=""):
         join = lambda f: os.path.join(prefix, f)
         with self._repo.wlock():
             ui, ds = self._repo.ui, self._repo.dirstate
             rejected = []
             lstat = self._repo.wvfs.lstat
             for f in list:
                 scmutil.checkportable(ui, join(f))
                 try:
                     st = lstat(f)
                 except OSError:
                     ui.warn(_("%s does not exist!\n") % join(f))
                     rejected.append(f)
                     continue
                 if st.st_size > 10000000:
                     ui.warn(_("%s: up to %d MB of RAM may be required "
                               "to manage this file\n"
                               "(use 'hg revert %s' to cancel the "
                               "pending addition)\n")
                               % (f, 3 * st.st_size // 1000000, join(f)))
                 if not (stat.S_ISREG(st.st_mode) or stat.S_ISLNK(st.st_mode)):
                     ui.warn(_("%s not added: only files and symlinks "
                               "supported currently\n") % join(f))
                     rejected.append(f)
                 elif ds[f] in 'amn':
                     ui.warn(_("%s already tracked!\n") % join(f))
                 elif ds[f] == 'r':
                     ds.normallookup(f)
                 else:
                     ds.add(f)
             return rejected
 
     def forget(self, files, prefix=""):
         join = lambda f: os.path.join(prefix, f)
         with self._repo.wlock():
             rejected = []
             for f in files:
                 if f not in self._repo.dirstate:
                     self._repo.ui.warn(_("%s not tracked!\n") % join(f))
                     rejected.append(f)
                 elif self._repo.dirstate[f] != 'a':
                     self._repo.dirstate.remove(f)
                 else:
                     self._repo.dirstate.drop(f)
             return rejected
 
     def undelete(self, list):
         pctxs = self.parents()
         with self._repo.wlock():
             for f in list:
                 if self._repo.dirstate[f] != 'r':
                     self._repo.ui.warn(_("%s not removed!\n") % f)
                 else:
                     fctx = f in pctxs[0] and pctxs[0][f] or pctxs[1][f]
                     t = fctx.data()
                     self._repo.wwrite(f, t, fctx.flags())
                     self._repo.dirstate.normal(f)
 
     def copy(self, source, dest):
         try:
             st = self._repo.wvfs.lstat(dest)
         except OSError as err:
             if err.errno != errno.ENOENT:
                 raise
             self._repo.ui.warn(_("%s does not exist!\n") % dest)
             return
         if not (stat.S_ISREG(st.st_mode) or stat.S_ISLNK(st.st_mode)):
             self._repo.ui.warn(_("copy failed: %s is not a file or a "
                                  "symbolic link\n") % dest)
         else:
             with self._repo.wlock():
                 if self._repo.dirstate[dest] in '?':
                     self._repo.dirstate.add(dest)
                 elif self._repo.dirstate[dest] in 'r':
                     self._repo.dirstate.normallookup(dest)
                 self._repo.dirstate.copy(source, dest)
 
     def match(self, pats=None, include=None, exclude=None, default='glob',
               listsubrepos=False, badfn=None):
         r = self._repo
 
         # Only a case insensitive filesystem needs magic to translate user input
         # to actual case in the filesystem.
         icasefs = not util.fscasesensitive(r.root)
         return matchmod.match(r.root, r.getcwd(), pats, include, exclude,
                               default, auditor=r.auditor, ctx=self,
                               listsubrepos=listsubrepos, badfn=badfn,
                               icasefs=icasefs)
 
     def _filtersuspectsymlink(self, files):
         if not files or self._repo.dirstate._checklink:
             return files
 
         # Symlink placeholders may get non-symlink-like contents
         # via user error or dereferencing by NFS or Samba servers,
         # so we filter out any placeholders that don't look like a
         # symlink
         sane = []
         for f in files:
             if self.flags(f) == 'l':
                 d = self[f].data()
                 if d == '' or len(d) >= 1024 or '\n' in d or util.binary(d):
                     self._repo.ui.debug('ignoring suspect symlink placeholder'
                                         ' "%s"\n' % f)
                     continue
             sane.append(f)
         return sane
 
     def _checklookup(self, files):
         # check for any possibly clean files
         if not files:
             return [], [], []
 
         modified = []
         deleted = []
         fixup = []
         pctx = self._parents[0]
         # do a full compare of any files that might have changed
         for f in sorted(files):
             try:
                 # This will return True for a file that got replaced by a
                 # directory in the interim, but fixing that is pretty hard.
                 if (f not in pctx or self.flags(f) != pctx.flags(f)
                     or pctx[f].cmp(self[f])):
                     modified.append(f)
                 else:
                     fixup.append(f)
             except (IOError, OSError):
                 # A file become inaccessible in between? Mark it as deleted,
                 # matching dirstate behavior (issue5584).
                 # The dirstate has more complex behavior around whether a
                 # missing file matches a directory, etc, but we don't need to
                 # bother with that: if f has made it to this point, we're sure
                 # it's in the dirstate.
                 deleted.append(f)
 
         return modified, deleted, fixup
 
     def _poststatusfixup(self, status, fixup):
         """update dirstate for files that are actually clean"""
         poststatus = self._repo.postdsstatus()
         if fixup or poststatus:
             try:
                 oldid = self._repo.dirstate.identity()
 
                 # updating the dirstate is optional
                 # so we don't wait on the lock
                 # wlock can invalidate the dirstate, so cache normal _after_
                 # taking the lock
                 with self._repo.wlock(False):
                     if self._repo.dirstate.identity() == oldid:
                         if fixup:
                             normal = self._repo.dirstate.normal
                             for f in fixup:
                                 normal(f)
                             # write changes out explicitly, because nesting
                             # wlock at runtime may prevent 'wlock.release()'
                             # after this block from doing so for subsequent
                             # changing files
                             tr = self._repo.currenttransaction()
                             self._repo.dirstate.write(tr)
 
                         if poststatus:
                             for ps in poststatus:
                                 ps(self, status)
                     else:
                         # in this case, writing changes out breaks
                         # consistency, because .hg/dirstate was
                         # already changed simultaneously after last
                         # caching (see also issue5584 for detail)
                         self._repo.ui.debug('skip updating dirstate: '
                                             'identity mismatch\n')
             except error.LockError:
                 pass
             finally:
                 # Even if the wlock couldn't be grabbed, clear out the list.
                 self._repo.clearpostdsstatus()
 
     def _dirstatestatus(self, match=None, ignored=False, clean=False,
                         unknown=False):
         '''Gets the status from the dirstate -- internal use only.'''
         listignored, listclean, listunknown = ignored, clean, unknown
         match = match or matchmod.always(self._repo.root, self._repo.getcwd())
         subrepos = []
         if '.hgsub' in self:
             subrepos = sorted(self.substate)
         cmp, s = self._repo.dirstate.status(match, subrepos, listignored,
                                             listclean, listunknown)
 
         # check for any possibly clean files
         fixup = []
         if cmp:
             modified2, deleted2, fixup = self._checklookup(cmp)
             s.modified.extend(modified2)
             s.deleted.extend(deleted2)
 
             if fixup and listclean:
                 s.clean.extend(fixup)
 
         self._poststatusfixup(s, fixup)
 
         if match.always():
             # cache for performance
             if s.unknown or s.ignored or s.clean:
                 # "_status" is cached with list*=False in the normal route
                 self._status = scmutil.status(s.modified, s.added, s.removed,
                                               s.deleted, [], [], [])
             else:
                 self._status = s
 
         return s
 
     @propertycache
     def _manifest(self):
         """generate a manifest corresponding to the values in self._status
 
         This reuse the file nodeid from parent, but we use special node
         identifiers for added and modified files. This is used by manifests
         merge to see that files are different and by update logic to avoid
         deleting newly added files.
         """
         return self._buildstatusmanifest(self._status)
 
     def _buildstatusmanifest(self, status):
         """Builds a manifest that includes the given status results."""
         parents = self.parents()
 
         man = parents[0].manifest().copy()
 
         ff = self._flagfunc
         for i, l in ((addednodeid, status.added),
                      (modifiednodeid, status.modified)):
             for f in l:
                 man[f] = i
                 try:
                     man.setflag(f, ff(f))
                 except OSError:
                     pass
 
         for f in status.deleted + status.removed:
             if f in man:
                 del man[f]
 
         return man
 
     def _buildstatus(self, other, s, match, listignored, listclean,
                      listunknown):
         """build a status with respect to another context
 
         This includes logic for maintaining the fast path of status when
         comparing the working directory against its parent, which is to skip
         building a new manifest if self (working directory) is not comparing
         against its parent (repo['.']).
         """
         s = self._dirstatestatus(match, listignored, listclean, listunknown)
         # Filter out symlinks that, in the case of FAT32 and NTFS filesystems,
         # might have accidentally ended up with the entire contents of the file
         # they are supposed to be linking to.
         s.modified[:] = self._filtersuspectsymlink(s.modified)
         if other != self._repo['.']:
             s = super(workingctx, self)._buildstatus(other, s, match,
                                                      listignored, listclean,
                                                      listunknown)
         return s
 
     def _matchstatus(self, other, match):
         """override the match method with a filter for directory patterns
 
         We use inheritance to customize the match.bad method only in cases of
         workingctx since it belongs only to the working directory when
         comparing against the parent changeset.
 
         If we aren't comparing against the working directory's parent, then we
         just use the default match object sent to us.
         """
         superself = super(workingctx, self)
         match = superself._matchstatus(other, match)
         if other != self._repo['.']:
             def bad(f, msg):
                 # 'f' may be a directory pattern from 'match.files()',
                 # so 'f not in ctx1' is not enough
                 if f not in other and not other.hasdir(f):
                     self._repo.ui.warn('%s: %s\n' %
                                        (self._repo.dirstate.pathto(f), msg))
             match.bad = bad
         return match
 
 class committablefilectx(basefilectx):
     """A committablefilectx provides common functionality for a file context
     that wants the ability to commit, e.g. workingfilectx or memfilectx."""
     def __init__(self, repo, path, filelog=None, ctx=None):
         self._repo = repo
         self._path = path
         self._changeid = None
         self._filerev = self._filenode = None
 
         if filelog is not None:
             self._filelog = filelog
         if ctx:
             self._changectx = ctx
 
     def __nonzero__(self):
         return True
 
     __bool__ = __nonzero__
 
     def linkrev(self):
         # linked to self._changectx no matter if file is modified or not
         return self.rev()
 
     def parents(self):
         '''return parent filectxs, following copies if necessary'''
         def filenode(ctx, path):
             return ctx._manifest.get(path, nullid)
 
         path = self._path
         fl = self._filelog
         pcl = self._changectx._parents
         renamed = self.renamed()
 
         if renamed:
             pl = [renamed + (None,)]
         else:
             pl = [(path, filenode(pcl[0], path), fl)]
 
         for pc in pcl[1:]:
             pl.append((path, filenode(pc, path), fl))
 
         return [self._parentfilectx(p, fileid=n, filelog=l)
                 for p, n, l in pl if n != nullid]
 
     def children(self):
         return []
 
 class workingfilectx(committablefilectx):
     """A workingfilectx object makes access to data related to a particular
        file in the working directory convenient."""
     def __init__(self, repo, path, filelog=None, workingctx=None):
         super(workingfilectx, self).__init__(repo, path, filelog, workingctx)
 
     @propertycache
     def _changectx(self):
         return workingctx(self._repo)
 
     def data(self):
         return self._repo.wread(self._path)
     def renamed(self):
         rp = self._repo.dirstate.copied(self._path)
         if not rp:
             return None
         return rp, self._changectx._parents[0]._manifest.get(rp, nullid)
 
     def size(self):
         return self._repo.wvfs.lstat(self._path).st_size
     def date(self):
         t, tz = self._changectx.date()
         try:
             return (self._repo.wvfs.lstat(self._path).st_mtime, tz)
         except OSError as err:
             if err.errno != errno.ENOENT:
                 raise
             return (t, tz)
 
     def cmp(self, fctx):
         """compare with other file context
 
         returns True if different than fctx.
         """
         # fctx should be a filectx (not a workingfilectx)
         # invert comparison to reuse the same code path
         return fctx.cmp(self)
 
     def remove(self, ignoremissing=False):
         """wraps unlink for a repo's working directory"""
         self._repo.wvfs.unlinkpath(self._path, ignoremissing=ignoremissing)
 
     def write(self, data, flags):
         """wraps repo.wwrite"""
         self._repo.wwrite(self._path, data, flags)
 
 class workingcommitctx(workingctx):
     """A workingcommitctx object makes access to data related to
     the revision being committed convenient.
 
     This hides changes in the working directory, if they aren't
     committed in this context.
     """
     def __init__(self, repo, changes,
                  text="", user=None, date=None, extra=None):
         super(workingctx, self).__init__(repo, text, user, date, extra,
                                          changes)
 
     def _dirstatestatus(self, match=None, ignored=False, clean=False,
                         unknown=False):
         """Return matched files only in ``self._status``
 
         Uncommitted files appear "clean" via this context, even if
         they aren't actually so in the working directory.
         """
         match = match or matchmod.always(self._repo.root, self._repo.getcwd())
         if clean:
             clean = [f for f in self._manifest if f not in self._changedset]
         else:
             clean = []
         return scmutil.status([f for f in self._status.modified if match(f)],
                               [f for f in self._status.added if match(f)],
                               [f for f in self._status.removed if match(f)],
                               [], [], [], clean)
 
     @propertycache
     def _changedset(self):
         """Return the set of files changed in this context
         """
         changed = set(self._status.modified)
         changed.update(self._status.added)
         changed.update(self._status.removed)
         return changed
 
 def makecachingfilectxfn(func):
     """Create a filectxfn that caches based on the path.
 
     We can't use util.cachefunc because it uses all arguments as the cache
     key and this creates a cycle since the arguments include the repo and
     memctx.
     """
     cache = {}
 
     def getfilectx(repo, memctx, path):
         if path not in cache:
             cache[path] = func(repo, memctx, path)
         return cache[path]
 
     return getfilectx
 
 def memfilefromctx(ctx):
     """Given a context return a memfilectx for ctx[path]
 
     This is a convenience method for building a memctx based on another
     context.
     """
     def getfilectx(repo, memctx, path):
         fctx = ctx[path]
         # this is weird but apparently we only keep track of one parent
         # (why not only store that instead of a tuple?)
         copied = fctx.renamed()
         if copied:
             copied = copied[0]
         return memfilectx(repo, path, fctx.data(),
                           islink=fctx.islink(), isexec=fctx.isexec(),
                           copied=copied, memctx=memctx)
 
     return getfilectx
 
 def memfilefrompatch(patchstore):
     """Given a patch (e.g. patchstore object) return a memfilectx
 
     This is a convenience method for building a memctx based on a patchstore.
     """
     def getfilectx(repo, memctx, path):
         data, mode, copied = patchstore.getfile(path)
         if data is None:
             return None
         islink, isexec = mode
         return memfilectx(repo, path, data, islink=islink,
                           isexec=isexec, copied=copied,
                           memctx=memctx)
 
     return getfilectx
 
 class memctx(committablectx):
     """Use memctx to perform in-memory commits via localrepo.commitctx().
 
     Revision information is supplied at initialization time while
     related files data and is made available through a callback
     mechanism.  'repo' is the current localrepo, 'parents' is a
     sequence of two parent revisions identifiers (pass None for every
     missing parent), 'text' is the commit message and 'files' lists
     names of files touched by the revision (normalized and relative to
     repository root).
 
     filectxfn(repo, memctx, path) is a callable receiving the
     repository, the current memctx object and the normalized path of
     requested file, relative to repository root. It is fired by the
     commit function for every file in 'files', but calls order is
     undefined. If the file is available in the revision being
     committed (updated or added), filectxfn returns a memfilectx
     object. If the file was removed, filectxfn return None for recent
     Mercurial. Moved files are represented by marking the source file
     removed and the new file added with copy information (see
     memfilectx).
 
     user receives the committer name and defaults to current
     repository username, date is the commit date in any format
     supported by util.parsedate() and defaults to current date, extra
     is a dictionary of metadata or is left empty.
     """
 
     # Mercurial <= 3.1 expects the filectxfn to raise IOError for missing files.
     # Extensions that need to retain compatibility across Mercurial 3.1 can use
     # this field to determine what to do in filectxfn.
     _returnnoneformissingfiles = True
 
     def __init__(self, repo, parents, text, files, filectxfn, user=None,
                  date=None, extra=None, branch=None, editor=False):
         super(memctx, self).__init__(repo, text, user, date, extra)
         self._rev = None
         self._node = None
         parents = [(p or nullid) for p in parents]
         p1, p2 = parents
         self._parents = [changectx(self._repo, p) for p in (p1, p2)]
         files = sorted(set(files))
         self._files = files
         if branch is not None:
             self._extra['branch'] = encoding.fromlocal(branch)
         self.substate = {}
 
         if isinstance(filectxfn, patch.filestore):
             filectxfn = memfilefrompatch(filectxfn)
         elif not callable(filectxfn):
             # if store is not callable, wrap it in a function
             filectxfn = memfilefromctx(filectxfn)
 
         # memoizing increases performance for e.g. vcs convert scenarios.
         self._filectxfn = makecachingfilectxfn(filectxfn)
 
         if editor:
             self._text = editor(self._repo, self, [])
             self._repo.savecommitmessage(self._text)
 
     def filectx(self, path, filelog=None):
         """get a file context from the working directory
 
         Returns None if file doesn't exist and should be removed."""
         return self._filectxfn(self._repo, self, path)
 
     def commit(self):
         """commit context to the repo"""
         return self._repo.commitctx(self)
 
     @propertycache
     def _manifest(self):
         """generate a manifest based on the return values of filectxfn"""
 
         # keep this simple for now; just worry about p1
         pctx = self._parents[0]
         man = pctx.manifest().copy()
 
         for f in self._status.modified:
             p1node = nullid
             p2node = nullid
             p = pctx[f].parents() # if file isn't in pctx, check p2?
             if len(p) > 0:
                 p1node = p[0].filenode()
                 if len(p) > 1:
                     p2node = p[1].filenode()
             man[f] = revlog.hash(self[f].data(), p1node, p2node)
 
         for f in self._status.added:
             man[f] = revlog.hash(self[f].data(), nullid, nullid)
 
         for f in self._status.removed:
             if f in man:
                 del man[f]
 
         return man
 
     @propertycache
     def _status(self):
         """Calculate exact status from ``files`` specified at construction
         """
         man1 = self.p1().manifest()
         p2 = self._parents[1]
         # "1 < len(self._parents)" can't be used for checking
         # existence of the 2nd parent, because "memctx._parents" is
         # explicitly initialized by the list, of which length is 2.
         if p2.node() != nullid:
             man2 = p2.manifest()
             managing = lambda f: f in man1 or f in man2
         else:
             managing = lambda f: f in man1
 
         modified, added, removed = [], [], []
         for f in self._files:
             if not managing(f):
                 added.append(f)
             elif self[f]:
                 modified.append(f)
             else:
                 removed.append(f)
 
         return scmutil.status(modified, added, removed, [], [], [], [])
 
 class memfilectx(committablefilectx):
     """memfilectx represents an in-memory file to commit.
 
     See memctx and committablefilectx for more details.
     """
     def __init__(self, repo, path, data, islink=False,
                  isexec=False, copied=None, memctx=None):
         """
         path is the normalized file path relative to repository root.
         data is the file content as a string.
         islink is True if the file is a symbolic link.
         isexec is True if the file is executable.
         copied is the source file path if current file was copied in the
         revision being committed, or None."""
         super(memfilectx, self).__init__(repo, path, None, memctx)
         self._data = data
         self._flags = (islink and 'l' or '') + (isexec and 'x' or '')
         self._copied = None
         if copied:
             self._copied = (copied, nullid)
 
     def data(self):
         return self._data
 
     def remove(self, ignoremissing=False):
         """wraps unlink for a repo's working directory"""
         # need to figure out what to do here
         del self._changectx[self._path]
 
     def write(self, data, flags):
         """wraps repo.wwrite"""
         self._data = data
 
 class overlayfilectx(committablefilectx):
     """Like memfilectx but take an original filectx and optional parameters to
     override parts of it. This is useful when fctx.data() is expensive (i.e.
     flag processor is expensive) and raw data, flags, and filenode could be
     reused (ex. rebase or mode-only amend a REVIDX_EXTSTORED file).
     """
 
     def __init__(self, originalfctx, datafunc=None, path=None, flags=None,
                  copied=None, ctx=None):
         """originalfctx: filecontext to duplicate
 
         datafunc: None or a function to override data (file content). It is a
         function to be lazy. path, flags, copied, ctx: None or overridden value
 
         copied could be (path, rev), or False. copied could also be just path,
         and will be converted to (path, nullid). This simplifies some callers.
         """
 
         if path is None:
             path = originalfctx.path()
         if ctx is None:
             ctx = originalfctx.changectx()
             ctxmatch = lambda: True
         else:
             ctxmatch = lambda: ctx == originalfctx.changectx()
 
         repo = originalfctx.repo()
         flog = originalfctx.filelog()
         super(overlayfilectx, self).__init__(repo, path, flog, ctx)
 
         if copied is None:
             copied = originalfctx.renamed()
             copiedmatch = lambda: True
         else:
             if copied and not isinstance(copied, tuple):
                 # repo._filecommit will recalculate copyrev so nullid is okay
                 copied = (copied, nullid)
             copiedmatch = lambda: copied == originalfctx.renamed()
 
         # When data, copied (could affect data), ctx (could affect filelog
         # parents) are not overridden, rawdata, rawflags, and filenode may be
         # reused (repo._filecommit should double check filelog parents).
         #
         # path, flags are not hashed in filelog (but in manifestlog) so they do
         # not affect reusable here.
         #
         # If ctx or copied is overridden to a same value with originalfctx,
         # still consider it's reusable. originalfctx.renamed() may be a bit
         # expensive so it's not called unless necessary. Assuming datafunc is
         # always expensive, do not call it for this "reusable" test.
         reusable = datafunc is None and ctxmatch() and copiedmatch()
 
         if datafunc is None:
             datafunc = originalfctx.data
         if flags is None:
             flags = originalfctx.flags()
 
         self._datafunc = datafunc
         self._flags = flags
         self._copied = copied
 
         if reusable:
             # copy extra fields from originalfctx
             attrs = ['rawdata', 'rawflags', '_filenode', '_filerev']
             for attr in attrs:
                 if util.safehasattr(originalfctx, attr):
                     setattr(self, attr, getattr(originalfctx, attr))
 
     def data(self):
         return self._datafunc()
 
 class metadataonlyctx(committablectx):
     """Like memctx but it's reusing the manifest of different commit.
     Intended to be used by lightweight operations that are creating
     metadata-only changes.
 
     Revision information is supplied at initialization time.  'repo' is the
     current localrepo, 'ctx' is original revision which manifest we're reuisng
     'parents' is a sequence of two parent revisions identifiers (pass None for
     every missing parent), 'text' is the commit.
 
     user receives the committer name and defaults to current repository
     username, date is the commit date in any format supported by
     util.parsedate() and defaults to current date, extra is a dictionary of
     metadata or is left empty.
     """
     def __new__(cls, repo, originalctx, *args, **kwargs):
         return super(metadataonlyctx, cls).__new__(cls, repo)
 
     def __init__(self, repo, originalctx, parents, text, user=None, date=None,
                  extra=None, editor=False):
         super(metadataonlyctx, self).__init__(repo, text, user, date, extra)
         self._rev = None
         self._node = None
         self._originalctx = originalctx
         self._manifestnode = originalctx.manifestnode()
         parents = [(p or nullid) for p in parents]
         p1, p2 = self._parents = [changectx(self._repo, p) for p in parents]
 
         # sanity check to ensure that the reused manifest parents are
         # manifests of our commit parents
         mp1, mp2 = self.manifestctx().parents
         if p1 != nullid and p1.manifestnode() != mp1:
             raise RuntimeError('can\'t reuse the manifest: '
                                'its p1 doesn\'t match the new ctx p1')
         if p2 != nullid and p2.manifestnode() != mp2:
             raise RuntimeError('can\'t reuse the manifest: '
                                'its p2 doesn\'t match the new ctx p2')
 
         self._files = originalctx.files()
         self.substate = {}
 
         if editor:
             self._text = editor(self._repo, self, [])
             self._repo.savecommitmessage(self._text)
 
     def manifestnode(self):
         return self._manifestnode
 
     @property
     def _manifestctx(self):
         return self._repo.manifestlog[self._manifestnode]
 
     def filectx(self, path, filelog=None):
         return self._originalctx.filectx(path, filelog=filelog)
 
     def commit(self):
         """commit context to the repo"""
         return self._repo.commitctx(self)
 
     @property
     def _manifest(self):
         return self._originalctx.manifest()
 
     @propertycache
     def _status(self):
         """Calculate exact status from ``files`` specified in the ``origctx``
         and parents manifests.
         """
         man1 = self.p1().manifest()
         p2 = self._parents[1]
         # "1 < len(self._parents)" can't be used for checking
         # existence of the 2nd parent, because "metadataonlyctx._parents" is
         # explicitly initialized by the list, of which length is 2.
         if p2.node() != nullid:
             man2 = p2.manifest()
             managing = lambda f: f in man1 or f in man2
         else:
             managing = lambda f: f in man1
 
         modified, added, removed = [], [], []
         for f in self._files:
             if not managing(f):
                 added.append(f)
             elif self[f]:
                 modified.append(f)
             else:
                 removed.append(f)
 
         return scmutil.status(modified, added, removed, [], [], [], [])

mercurial/encoding.py

@@ -1,593 +1,602 b''
 # encoding.py - character transcoding support for Mercurial
 #
 #  Copyright 2005-2009 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
 
 import array
 import locale
 import os
 import unicodedata
 
 from . import (
     error,
     policy,
     pycompat,
 )
 
 _sysstr = pycompat.sysstr
 
 if pycompat.ispy3:
     unichr = chr
 
 # These unicode characters are ignored by HFS+ (Apple Technote 1150,
 # "Unicode Subtleties"), so we need to ignore them in some places for
 # sanity.
 _ignore = [unichr(int(x, 16)).encode("utf-8") for x in
            "200c 200d 200e 200f 202a 202b 202c 202d 202e "
            "206a 206b 206c 206d 206e 206f feff".split()]
 # verify the next function will work
 assert all(i.startswith(("\xe2", "\xef")) for i in _ignore)
 
 def hfsignoreclean(s):
     """Remove codepoints ignored by HFS+ from s.
 
     >>> hfsignoreclean(u'.h\u200cg'.encode('utf-8'))
     '.hg'
     >>> hfsignoreclean(u'.h\ufeffg'.encode('utf-8'))
     '.hg'
     """
     if "\xe2" in s or "\xef" in s:
         for c in _ignore:
             s = s.replace(c, '')
     return s
 
 # encoding.environ is provided read-only, which may not be used to modify
 # the process environment
 _nativeenviron = (not pycompat.ispy3 or os.supports_bytes_environ)
 if not pycompat.ispy3:
     environ = os.environ  # re-exports
 elif _nativeenviron:
     environ = os.environb  # re-exports
 else:
     # preferred encoding isn't known yet; use utf-8 to avoid unicode error
     # and recreate it once encoding is settled
     environ = dict((k.encode(u'utf-8'), v.encode(u'utf-8'))
                    for k, v in os.environ.items())  # re-exports
 
 _encodingfixers = {
     '646': lambda: 'ascii',
     'ANSI_X3.4-1968': lambda: 'ascii',
 }
 
 try:
     encoding = environ.get("HGENCODING")
     if not encoding:
         encoding = locale.getpreferredencoding().encode('ascii') or 'ascii'
         encoding = _encodingfixers.get(encoding, lambda: encoding)()
 except locale.Error:
     encoding = 'ascii'
 encodingmode = environ.get("HGENCODINGMODE", "strict")
 fallbackencoding = 'ISO-8859-1'
 
 class localstr(str):
     '''This class allows strings that are unmodified to be
     round-tripped to the local encoding and back'''
     def __new__(cls, u, l):
         s = str.__new__(cls, l)
         s._utf8 = u
         return s
     def __hash__(self):
         return hash(self._utf8) # avoid collisions in local string space
 
 def tolocal(s):
     """
     Convert a string from internal UTF-8 to local encoding
 
     All internal strings should be UTF-8 but some repos before the
     implementation of locale support may contain latin1 or possibly
     other character sets. We attempt to decode everything strictly
     using UTF-8, then Latin-1, and failing that, we use UTF-8 and
     replace unknown characters.
 
     The localstr class is used to cache the known UTF-8 encoding of
     strings next to their local representation to allow lossless
     round-trip conversion back to UTF-8.
 
     >>> u = 'foo: \\xc3\\xa4' # utf-8
     >>> l = tolocal(u)
     >>> l
     'foo: ?'
     >>> fromlocal(l)
     'foo: \\xc3\\xa4'
     >>> u2 = 'foo: \\xc3\\xa1'
     >>> d = { l: 1, tolocal(u2): 2 }
     >>> len(d) # no collision
     2
     >>> 'foo: ?' in d
     False
     >>> l1 = 'foo: \\xe4' # historical latin1 fallback
     >>> l = tolocal(l1)
     >>> l
     'foo: ?'
     >>> fromlocal(l) # magically in utf-8
     'foo: \\xc3\\xa4'
     """
 
     try:
         try:
             # make sure string is actually stored in UTF-8
             u = s.decode('UTF-8')
             if encoding == 'UTF-8':
                 # fast path
                 return s
             r = u.encode(_sysstr(encoding), u"replace")
             if u == r.decode(_sysstr(encoding)):
                 # r is a safe, non-lossy encoding of s
                 return r
             return localstr(s, r)
         except UnicodeDecodeError:
             # we should only get here if we're looking at an ancient changeset
             try:
                 u = s.decode(_sysstr(fallbackencoding))
                 r = u.encode(_sysstr(encoding), u"replace")
                 if u == r.decode(_sysstr(encoding)):
                     # r is a safe, non-lossy encoding of s
                     return r
                 return localstr(u.encode('UTF-8'), r)
             except UnicodeDecodeError:
                 u = s.decode("utf-8", "replace") # last ditch
                 # can't round-trip
                 return u.encode(_sysstr(encoding), u"replace")
     except LookupError as k:
         raise error.Abort(k, hint="please check your locale settings")
 
 def fromlocal(s):
     """
     Convert a string from the local character encoding to UTF-8
 
     We attempt to decode strings using the encoding mode set by
     HGENCODINGMODE, which defaults to 'strict'. In this mode, unknown
     characters will cause an error message. Other modes include
     'replace', which replaces unknown characters with a special
     Unicode character, and 'ignore', which drops the character.
     """
 
     # can we do a lossless round-trip?
     if isinstance(s, localstr):
         return s._utf8
 
     try:
         u = s.decode(_sysstr(encoding), _sysstr(encodingmode))
         return u.encode("utf-8")
     except UnicodeDecodeError as inst:
         sub = s[max(0, inst.start - 10):inst.start + 10]
         raise error.Abort("decoding near '%s': %s!" % (sub, inst))
     except LookupError as k:
         raise error.Abort(k, hint="please check your locale settings")
 
 def unitolocal(u):
     """Convert a unicode string to a byte string of local encoding"""
     return tolocal(u.encode('utf-8'))
 
 def unifromlocal(s):
     """Convert a byte string of local encoding to a unicode string"""
     return fromlocal(s).decode('utf-8')
 
+def unimethod(bytesfunc):
+    """Create a proxy method that forwards __unicode__() and __str__() of
+    Python 3 to __bytes__()"""
+    def unifunc(obj):
+        return unifromlocal(bytesfunc(obj))
+    return unifunc
+
 # converter functions between native str and byte string. use these if the
 # character encoding is not aware (e.g. exception message) or is known to
 # be locale dependent (e.g. date formatting.)
 if pycompat.ispy3:
     strtolocal = unitolocal
     strfromlocal = unifromlocal
+    strmethod = unimethod
 else:
     strtolocal = pycompat.identity
     strfromlocal = pycompat.identity
+    strmethod = pycompat.identity
 
 if not _nativeenviron:
     # now encoding and helper functions are available, recreate the environ
     # dict to be exported to other modules
     environ = dict((tolocal(k.encode(u'utf-8')), tolocal(v.encode(u'utf-8')))
                    for k, v in os.environ.items())  # re-exports
 
 # How to treat ambiguous-width characters. Set to 'wide' to treat as wide.
 _wide = _sysstr(environ.get("HGENCODINGAMBIGUOUS", "narrow") == "wide"
                 and "WFA" or "WF")
 
 def colwidth(s):
     "Find the column width of a string for display in the local encoding"
     return ucolwidth(s.decode(_sysstr(encoding), u'replace'))
 
 def ucolwidth(d):
     "Find the column width of a Unicode string for display"
     eaw = getattr(unicodedata, 'east_asian_width', None)
     if eaw is not None:
         return sum([eaw(c) in _wide and 2 or 1 for c in d])
     return len(d)
 
 def getcols(s, start, c):
     '''Use colwidth to find a c-column substring of s starting at byte
     index start'''
     for x in xrange(start + c, len(s)):
         t = s[start:x]
         if colwidth(t) == c:
             return t
 
 def trim(s, width, ellipsis='', leftside=False):
     """Trim string 's' to at most 'width' columns (including 'ellipsis').
 
     If 'leftside' is True, left side of string 's' is trimmed.
     'ellipsis' is always placed at trimmed side.
 
     >>> ellipsis = '+++'
     >>> from . import encoding
     >>> encoding.encoding = 'utf-8'
     >>> t= '1234567890'
     >>> print trim(t, 12, ellipsis=ellipsis)
     1234567890
     >>> print trim(t, 10, ellipsis=ellipsis)
     1234567890
     >>> print trim(t, 8, ellipsis=ellipsis)
     12345+++
     >>> print trim(t, 8, ellipsis=ellipsis, leftside=True)
     +++67890
     >>> print trim(t, 8)
     12345678
     >>> print trim(t, 8, leftside=True)
     34567890
     >>> print trim(t, 3, ellipsis=ellipsis)
     +++
     >>> print trim(t, 1, ellipsis=ellipsis)
     +
     >>> u = u'\u3042\u3044\u3046\u3048\u304a' # 2 x 5 = 10 columns
     >>> t = u.encode(encoding.encoding)
     >>> print trim(t, 12, ellipsis=ellipsis)
     \xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88\xe3\x81\x8a
     >>> print trim(t, 10, ellipsis=ellipsis)
     \xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88\xe3\x81\x8a
     >>> print trim(t, 8, ellipsis=ellipsis)
     \xe3\x81\x82\xe3\x81\x84+++
     >>> print trim(t, 8, ellipsis=ellipsis, leftside=True)
     +++\xe3\x81\x88\xe3\x81\x8a
     >>> print trim(t, 5)
     \xe3\x81\x82\xe3\x81\x84
     >>> print trim(t, 5, leftside=True)
     \xe3\x81\x88\xe3\x81\x8a
     >>> print trim(t, 4, ellipsis=ellipsis)
     +++
     >>> print trim(t, 4, ellipsis=ellipsis, leftside=True)
     +++
     >>> t = '\x11\x22\x33\x44\x55\x66\x77\x88\x99\xaa' # invalid byte sequence
     >>> print trim(t, 12, ellipsis=ellipsis)
     \x11\x22\x33\x44\x55\x66\x77\x88\x99\xaa
     >>> print trim(t, 10, ellipsis=ellipsis)
     \x11\x22\x33\x44\x55\x66\x77\x88\x99\xaa
     >>> print trim(t, 8, ellipsis=ellipsis)
     \x11\x22\x33\x44\x55+++
     >>> print trim(t, 8, ellipsis=ellipsis, leftside=True)
     +++\x66\x77\x88\x99\xaa
     >>> print trim(t, 8)
     \x11\x22\x33\x44\x55\x66\x77\x88
     >>> print trim(t, 8, leftside=True)
     \x33\x44\x55\x66\x77\x88\x99\xaa
     >>> print trim(t, 3, ellipsis=ellipsis)
     +++
     >>> print trim(t, 1, ellipsis=ellipsis)
     +
     """
     try:
         u = s.decode(_sysstr(encoding))
     except UnicodeDecodeError:
         if len(s) <= width: # trimming is not needed
             return s
         width -= len(ellipsis)
         if width <= 0: # no enough room even for ellipsis
             return ellipsis[:width + len(ellipsis)]
         if leftside:
             return ellipsis + s[-width:]
         return s[:width] + ellipsis
 
     if ucolwidth(u) <= width: # trimming is not needed
         return s
 
     width -= len(ellipsis)
     if width <= 0: # no enough room even for ellipsis
         return ellipsis[:width + len(ellipsis)]
 
     if leftside:
         uslice = lambda i: u[i:]
         concat = lambda s: ellipsis + s
     else:
         uslice = lambda i: u[:-i]
         concat = lambda s: s + ellipsis
     for i in xrange(1, len(u)):
         usub = uslice(i)
         if ucolwidth(usub) <= width:
             return concat(usub.encode(_sysstr(encoding)))
     return ellipsis # no enough room for multi-column characters
 
 def _asciilower(s):
     '''convert a string to lowercase if ASCII
 
     Raises UnicodeDecodeError if non-ASCII characters are found.'''
     s.decode('ascii')
     return s.lower()
 
 def asciilower(s):
     # delay importing avoids cyclic dependency around "parsers" in
     # pure Python build (util => i18n => encoding => parsers => util)
     parsers = policy.importmod(r'parsers')
     impl = getattr(parsers, 'asciilower', _asciilower)
     global asciilower
     asciilower = impl
     return impl(s)
 
 def _asciiupper(s):
     '''convert a string to uppercase if ASCII
 
     Raises UnicodeDecodeError if non-ASCII characters are found.'''
     s.decode('ascii')
     return s.upper()
 
 def asciiupper(s):
     # delay importing avoids cyclic dependency around "parsers" in
     # pure Python build (util => i18n => encoding => parsers => util)
     parsers = policy.importmod(r'parsers')
     impl = getattr(parsers, 'asciiupper', _asciiupper)
     global asciiupper
     asciiupper = impl
     return impl(s)
 
 def lower(s):
     "best-effort encoding-aware case-folding of local string s"
     try:
         return asciilower(s)
     except UnicodeDecodeError:
         pass
     try:
         if isinstance(s, localstr):
             u = s._utf8.decode("utf-8")
         else:
             u = s.decode(_sysstr(encoding), _sysstr(encodingmode))
 
         lu = u.lower()
         if u == lu:
             return s # preserve localstring
         return lu.encode(_sysstr(encoding))
     except UnicodeError:
         return s.lower() # we don't know how to fold this except in ASCII
     except LookupError as k:
         raise error.Abort(k, hint="please check your locale settings")
 
 def upper(s):
     "best-effort encoding-aware case-folding of local string s"
     try:
         return asciiupper(s)
     except UnicodeDecodeError:
         return upperfallback(s)
 
 def upperfallback(s):
     try:
         if isinstance(s, localstr):
             u = s._utf8.decode("utf-8")
         else:
             u = s.decode(_sysstr(encoding), _sysstr(encodingmode))
 
         uu = u.upper()
         if u == uu:
             return s # preserve localstring
         return uu.encode(_sysstr(encoding))
     except UnicodeError:
         return s.upper() # we don't know how to fold this except in ASCII
     except LookupError as k:
         raise error.Abort(k, hint="please check your locale settings")
 
 class normcasespecs(object):
     '''what a platform's normcase does to ASCII strings
 
     This is specified per platform, and should be consistent with what normcase
     on that platform actually does.
 
     lower: normcase lowercases ASCII strings
     upper: normcase uppercases ASCII strings
     other: the fallback function should always be called
 
     This should be kept in sync with normcase_spec in util.h.'''
     lower = -1
     upper = 1
     other = 0
 
 _jsonmap = []
 _jsonmap.extend("\\u%04x" % x for x in range(32))
 _jsonmap.extend(pycompat.bytechr(x) for x in range(32, 127))
 _jsonmap.append('\\u007f')
 _jsonmap[0x09] = '\\t'
 _jsonmap[0x0a] = '\\n'
 _jsonmap[0x22] = '\\"'
 _jsonmap[0x5c] = '\\\\'
 _jsonmap[0x08] = '\\b'
 _jsonmap[0x0c] = '\\f'
 _jsonmap[0x0d] = '\\r'
 _paranoidjsonmap = _jsonmap[:]
 _paranoidjsonmap[0x3c] = '\\u003c'  # '<' (e.g. escape "</script>")
 _paranoidjsonmap[0x3e] = '\\u003e'  # '>'
 _jsonmap.extend(pycompat.bytechr(x) for x in range(128, 256))
 
 def jsonescape(s, paranoid=False):
     '''returns a string suitable for JSON
 
     JSON is problematic for us because it doesn't support non-Unicode
     bytes. To deal with this, we take the following approach:
 
     - localstr objects are converted back to UTF-8
     - valid UTF-8/ASCII strings are passed as-is
     - other strings are converted to UTF-8b surrogate encoding
     - apply JSON-specified string escaping
 
     (escapes are doubled in these tests)
 
     >>> jsonescape('this is a test')
     'this is a test'
     >>> jsonescape('escape characters: \\0 \\x0b \\x7f')
     'escape characters: \\\\u0000 \\\\u000b \\\\u007f'
     >>> jsonescape('escape characters: \\t \\n \\r \\" \\\\')
     'escape characters: \\\\t \\\\n \\\\r \\\\" \\\\\\\\'
     >>> jsonescape('a weird byte: \\xdd')
     'a weird byte: \\xed\\xb3\\x9d'
     >>> jsonescape('utf-8: caf\\xc3\\xa9')
     'utf-8: caf\\xc3\\xa9'
     >>> jsonescape('')
     ''
 
     If paranoid, non-ascii and common troublesome characters are also escaped.
     This is suitable for web output.
 
     >>> jsonescape('escape boundary: \\x7e \\x7f \\xc2\\x80', paranoid=True)
     'escape boundary: ~ \\\\u007f \\\\u0080'
     >>> jsonescape('a weird byte: \\xdd', paranoid=True)
     'a weird byte: \\\\udcdd'
     >>> jsonescape('utf-8: caf\\xc3\\xa9', paranoid=True)
     'utf-8: caf\\\\u00e9'
     >>> jsonescape('non-BMP: \\xf0\\x9d\\x84\\x9e', paranoid=True)
     'non-BMP: \\\\ud834\\\\udd1e'
     >>> jsonescape('<[e-mail]>', paranoid=True)
     '\\\\[e-mail]\\\\u003e'
     '''
 
     if paranoid:
         jm = _paranoidjsonmap
     else:
         jm = _jsonmap
 
     u8chars = toutf8b(s)
     try:
         return ''.join(jm[x] for x in bytearray(u8chars))  # fast path
     except IndexError:
         pass
     # non-BMP char is represented as UTF-16 surrogate pair
     u16codes = array.array('H', u8chars.decode('utf-8').encode('utf-16'))
     u16codes.pop(0)  # drop BOM
     return ''.join(jm[x] if x < 128 else '\\u%04x' % x for x in u16codes)
 
 _utf8len = [0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 3, 4]
 
 def getutf8char(s, pos):
     '''get the next full utf-8 character in the given string, starting at pos
 
     Raises a UnicodeError if the given location does not start a valid
     utf-8 character.
     '''
 
     # find how many bytes to attempt decoding from first nibble
     l = _utf8len[ord(s[pos]) >> 4]
     if not l: # ascii
         return s[pos]
 
     c = s[pos:pos + l]
     # validate with attempted decode
     c.decode("utf-8")
     return c
 
 def toutf8b(s):
     '''convert a local, possibly-binary string into UTF-8b
 
     This is intended as a generic method to preserve data when working
     with schemes like JSON and XML that have no provision for
     arbitrary byte strings. As Mercurial often doesn't know
     what encoding data is in, we use so-called UTF-8b.
 
     If a string is already valid UTF-8 (or ASCII), it passes unmodified.
     Otherwise, unsupported bytes are mapped to UTF-16 surrogate range,
     uDC00-uDCFF.
 
     Principles of operation:
 
     - ASCII and UTF-8 data successfully round-trips and is understood
       by Unicode-oriented clients
     - filenames and file contents in arbitrary other encodings can have
       be round-tripped or recovered by clueful clients
     - local strings that have a cached known UTF-8 encoding (aka
       localstr) get sent as UTF-8 so Unicode-oriented clients get the
       Unicode data they want
     - because we must preserve UTF-8 bytestring in places such as
       filenames, metadata can't be roundtripped without help
 
     (Note: "UTF-8b" often refers to decoding a mix of valid UTF-8 and
     arbitrary bytes into an internal Unicode format that can be
     re-encoded back into the original. Here we are exposing the
     internal surrogate encoding as a UTF-8 string.)
     '''
 
     if "\xed" not in s:
         if isinstance(s, localstr):
             return s._utf8
         try:
             s.decode('utf-8')
             return s
         except UnicodeDecodeError:
             pass
 
     r = ""
     pos = 0
     l = len(s)
     while pos < l:
         try:
             c = getutf8char(s, pos)
             if "\xed\xb0\x80" <= c <= "\xed\xb3\xbf":
                 # have to re-escape existing U+DCxx characters
                 c = unichr(0xdc00 + ord(s[pos])).encode('utf-8')
                 pos += 1
             else:
                 pos += len(c)
         except UnicodeDecodeError:
             c = unichr(0xdc00 + ord(s[pos])).encode('utf-8')
             pos += 1
         r += c
     return r
 
 def fromutf8b(s):
     '''Given a UTF-8b string, return a local, possibly-binary string.
 
     return the original binary string. This
     is a round-trip process for strings like filenames, but metadata
     that's was passed through tolocal will remain in UTF-8.
 
     >>> roundtrip = lambda x: fromutf8b(toutf8b(x)) == x
     >>> m = "\\xc3\\xa9\\x99abcd"
     >>> toutf8b(m)
     '\\xc3\\xa9\\xed\\xb2\\x99abcd'
     >>> roundtrip(m)
     True
     >>> roundtrip("\\xc2\\xc2\\x80")
     True
     >>> roundtrip("\\xef\\xbf\\xbd")
     True
     >>> roundtrip("\\xef\\xef\\xbf\\xbd")
     True
     >>> roundtrip("\\xf1\\x80\\x80\\x80\\x80")
     True
     '''
 
     # fast path - look for uDxxx prefixes in s
     if "\xed" not in s:
         return s
 
     # We could do this with the unicode type but some Python builds
     # use UTF-16 internally (issue5031) which causes non-BMP code
     # points to be escaped. Instead, we use our handy getutf8char
     # helper again to walk the string without "decoding" it.
 
     r = ""
     pos = 0
     l = len(s)
     while pos < l:
         c = getutf8char(s, pos)
         pos += len(c)
         # unescape U+DCxx characters
         if "\xed\xb0\x80" <= c <= "\xed\xb3\xbf":
             c = chr(ord(c.decode("utf-8")) & 0xff)
         r += c
     return r

mercurial/util.py

@@ -1,3744 +1,3743 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
 
 import bz2
 import calendar
 import codecs
 import collections
 import datetime
 import errno
 import gc
 import hashlib
 import imp
 import os
 import platform as pyplatform
 import re as remod
 import shutil
 import signal
 import socket
 import stat
 import string
 import subprocess
 import sys
 import tempfile
 import textwrap
 import time
 import traceback
 import warnings
 import zlib
 
 from . import (
     encoding,
     error,
     i18n,
     policy,
     pycompat,
 )
 
 base85 = policy.importmod(r'base85')
 osutil = policy.importmod(r'osutil')
 parsers = policy.importmod(r'parsers')
 
 b85decode = base85.b85decode
 b85encode = base85.b85encode
 
 cookielib = pycompat.cookielib
 empty = pycompat.empty
 httplib = pycompat.httplib
 httpserver = pycompat.httpserver
 pickle = pycompat.pickle
 queue = pycompat.queue
 socketserver = pycompat.socketserver
 stderr = pycompat.stderr
 stdin = pycompat.stdin
 stdout = pycompat.stdout
 stringio = pycompat.stringio
 urlerr = pycompat.urlerr
 urlreq = pycompat.urlreq
 xmlrpclib = pycompat.xmlrpclib
 
 # workaround for win32mbcs
 _filenamebytestr = pycompat.bytestr
 
 def isatty(fp):
     try:
         return fp.isatty()
     except AttributeError:
         return False
 
 # glibc determines buffering on first write to stdout - if we replace a TTY
 # destined stdout with a pipe destined stdout (e.g. pager), we want line
 # buffering
 if isatty(stdout):
     stdout = os.fdopen(stdout.fileno(), pycompat.sysstr('wb'), 1)
 
 if pycompat.osname == 'nt':
     from . import windows as platform
     stdout = platform.winstdout(stdout)
 else:
     from . import posix as platform
 
 _ = i18n._
 
 bindunixsocket = platform.bindunixsocket
 cachestat = platform.cachestat
 checkexec = platform.checkexec
 checklink = platform.checklink
 copymode = platform.copymode
 executablepath = platform.executablepath
 expandglobs = platform.expandglobs
 explainexit = platform.explainexit
 findexe = platform.findexe
 gethgcmd = platform.gethgcmd
 getuser = platform.getuser
 getpid = os.getpid
 groupmembers = platform.groupmembers
 groupname = platform.groupname
 hidewindow = platform.hidewindow
 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
 popen = platform.popen
 posixfile = platform.posixfile
 quotecommand = platform.quotecommand
 readpipe = platform.readpipe
 rename = platform.rename
 removedirs = platform.removedirs
 samedevice = platform.samedevice
 samefile = platform.samefile
 samestat = platform.samestat
 setbinary = platform.setbinary
 setflags = platform.setflags
 setsignalhandler = platform.setsignalhandler
 shellquote = platform.shellquote
 spawndetached = platform.spawndetached
 split = platform.split
 sshargs = platform.sshargs
 statfiles = getattr(osutil, 'statfiles', platform.statfiles)
 statisexec = platform.statisexec
 statislink = platform.statislink
 testpid = platform.testpid
 umask = platform.umask
 unlink = platform.unlink
 username = platform.username
 
 try:
     recvfds = osutil.recvfds
 except AttributeError:
     pass
 try:
     setprocname = osutil.setprocname
 except AttributeError:
     pass
 
 # Python compatibility
 
 _notset = object()
 
 # disable Python's problematic floating point timestamps (issue4836)
 # (Python hypocritically says you shouldn't change this behavior in
 # libraries, and sure enough Mercurial is not a library.)
 os.stat_float_times(False)
 
 def safehasattr(thing, attr):
     return getattr(thing, attr, _notset) is not _notset
 
 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')
 
 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(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(['md5', 'sha1'])
     >>> d.update('foo')
     >>> [k for k in sorted(d)]
     ['md5', 'sha1']
     >>> d['md5']
     'acbd18db4cc2f85cedef654fccc4a4d8'
     >>> d['sha1']
     '0beec7b5ea3f0fdbc95d0dd47f3c5bc275da8a33'
     >>> digester.preferred(['md5', 'sha1'])
     'sha1'
     """
 
     def __init__(self, digests, s=''):
         self._hashes = {}
         for k in digests:
             if k not in DIGESTS:
                 raise 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 Abort(_('unknown digest type: %s') % k)
         return self._hashes[key].hexdigest()
 
     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 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 Abort(_('%s mismatch: expected %s, got %s') %
                     (k, v, self._digester[k]))
 
 try:
     buffer = buffer
 except NameError:
     if not pycompat.ispy3:
         def buffer(sliceable, offset=0, length=None):
             if length is not None:
                 return sliceable[offset:offset + length]
             return sliceable[offset:]
     else:
         def buffer(sliceable, offset=0, length=None):
             if length is not None:
                 return memoryview(sliceable)[offset:offset + length]
             return memoryview(sliceable)[offset:]
 
 closefds = pycompat.osname == 'posix'
 
 _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 __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 readline(self, *args, **kwargs):
         if 1 < len(self._buffer):
             # 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 1 < len(self._buffer):
             # 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 1 < len(self._buffer):
             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):
         """read data to the buffer"""
         data = os.read(self._input.fileno(), _chunksize)
         if not data:
             self._eof = True
         else:
             self._lenbuf += len(data)
             self._buffer.append(data)
 
 def popen2(cmd, env=None, newlines=False):
     # Setting bufsize to -1 lets the system decide the buffer size.
     # The default for bufsize is 0, meaning unbuffered. This leads to
     # poor performance on Mac OS X: http://bugs.python.org/issue4194
     p = subprocess.Popen(cmd, shell=True, bufsize=-1,
                          close_fds=closefds,
                          stdin=subprocess.PIPE, stdout=subprocess.PIPE,
                          universal_newlines=newlines,
                          env=env)
     return p.stdin, p.stdout
 
 def popen3(cmd, env=None, newlines=False):
     stdin, stdout, stderr, p = popen4(cmd, env, newlines)
     return stdin, stdout, stderr
 
 def popen4(cmd, env=None, newlines=False, bufsize=-1):
     p = subprocess.Popen(cmd, shell=True, bufsize=bufsize,
                          close_fds=closefds,
                          stdin=subprocess.PIPE, stdout=subprocess.PIPE,
                          stderr=subprocess.PIPE,
                          universal_newlines=newlines,
                          env=env)
     return p.stdin, p.stdout, p.stderr, p
 
 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 = '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('3.6.1+190-df9b73d2d444+20151118')
     (3, 6, 1, '190-df9b73d2d444+20151118')
 
     >>> v = '3.6'
     >>> versiontuple(v, 2)
     (3, 6)
     >>> versiontuple(v, 3)
     (3, 6, None)
     >>> versiontuple(v, 4)
     (3, 6, None, None)
 
     >>> v = '3.9-rc'
     >>> versiontuple(v, 2)
     (3, 9)
     >>> versiontuple(v, 3)
     (3, 9, None)
     >>> versiontuple(v, 4)
     (3, 9, None, 'rc')
 
     >>> v = '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')
     """
     if not v:
         v = version()
     parts = remod.split('[\+-]', v, 1)
     if len(parts) == 1:
         vparts, extra = parts[0], None
     else:
         vparts, extra = parts
 
     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)
 
 # used by parsedate
 defaultdateformats = (
     '%Y-%m-%dT%H:%M:%S', # the 'real' ISO8601
     '%Y-%m-%dT%H:%M',    #   without seconds
     '%Y-%m-%dT%H%M%S',   # another awful but legal variant without :
     '%Y-%m-%dT%H%M',     #   without seconds
     '%Y-%m-%d %H:%M:%S', # our common legal variant
     '%Y-%m-%d %H:%M',    #   without seconds
     '%Y-%m-%d %H%M%S',   # without :
     '%Y-%m-%d %H%M',     #   without seconds
     '%Y-%m-%d %I:%M:%S%p',
     '%Y-%m-%d %H:%M',
     '%Y-%m-%d %I:%M%p',
     '%Y-%m-%d',
     '%m-%d',
     '%m/%d',
     '%m/%d/%y',
     '%m/%d/%Y',
     '%a %b %d %H:%M:%S %Y',
     '%a %b %d %I:%M:%S%p %Y',
     '%a, %d %b %Y %H:%M:%S',        #  GNU coreutils "/bin/date --rfc-2822"
     '%b %d %H:%M:%S %Y',
     '%b %d %I:%M:%S%p %Y',
     '%b %d %H:%M:%S',
     '%b %d %I:%M:%S%p',
     '%b %d %H:%M',
     '%b %d %I:%M%p',
     '%b %d %Y',
     '%b %d',
     '%H:%M:%S',
     '%I:%M:%S%p',
     '%H:%M',
     '%I:%M%p',
 )
 
 extendeddateformats = defaultdateformats + (
     "%Y",
     "%Y-%m",
     "%b",
     "%b %Y",
     )
 
 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 sortdict(collections.OrderedDict):
     '''a simple sorted dictionary
 
     >>> d1 = sortdict([('a', 0), ('b', 1)])
     >>> d2 = d1.copy()
     >>> d2
     sortdict([('a', 0), ('b', 1)])
     >>> d2.update([('a', 2)])
     >>> 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)
 
 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__ = (u'next', u'prev', u'key', u'value')
 
     def __init__(self):
         self.next = None
         self.prev = None
 
         self.key = _notset
         self.value = None
 
     def markempty(self):
         """Mark the node as emptied."""
         self.key = _notset
 
 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.
     """
     def __init__(self, max):
         self._cache = {}
 
         self._head = head = _lrucachenode()
         head.prev = head
         head.next = head
         self._size = 1
         self._capacity = max
 
     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 __setitem__(self, k, v):
         node = self._cache.get(k)
         # Replace existing value and mark as newest.
         if node is not None:
             node.value = v
             self._movetohead(node)
             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:
             del self._cache[node.key]
 
         node.key = k
         node.value = v
         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
 
     def __delitem__(self, k):
         node = self._cache.pop(k)
         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
 
     # Additional dict methods.
 
     def get(self, k, default=None):
         try:
             return self._cache[k].value
         except KeyError:
             return default
 
     def clear(self):
         n = self._head
         while n.key is not _notset:
             n.markempty()
             n = n.next
 
         self._cache.clear()
 
     def copy(self):
         result = lrucachedict(self._capacity)
         n = self._head.prev
         # Iterate in oldest-to-newest order, so the copy has the right ordering
         for i in range(len(self._cache)):
             result[n.key] = n.value
             n = n.prev
         return result
 
     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 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 pipefilter(s, cmd):
     '''filter string S through command CMD, returning its output'''
     p = subprocess.Popen(cmd, shell=True, close_fds=closefds,
                          stdin=subprocess.PIPE, stdout=subprocess.PIPE)
     pout, perr = p.communicate(s)
     return pout
 
 def tempfilter(s, cmd):
     '''filter string S through a pair of temporary files with CMD.
     CMD is used as a template to create the real command to be run,
     with the strings INFILE and OUTFILE replaced by the real names of
     the temporary files generated.'''
     inname, outname = None, None
     try:
         infd, inname = tempfile.mkstemp(prefix='hg-filter-in-')
         fp = os.fdopen(infd, pycompat.sysstr('wb'))
         fp.write(s)
         fp.close()
         outfd, outname = tempfile.mkstemp(prefix='hg-filter-out-')
         os.close(outfd)
         cmd = cmd.replace('INFILE', inname)
         cmd = cmd.replace('OUTFILE', outname)
         code = os.system(cmd)
         if pycompat.sysplatform == 'OpenVMS' and code & 1:
             code = 0
         if code:
             raise Abort(_("command '%s' failed: %s") %
                         (cmd, explainexit(code)))
         return readfile(outname)
     finally:
         try:
             if inname:
                 os.unlink(inname)
         except OSError:
             pass
         try:
             if outname:
                 os.unlink(outname)
         except OSError:
             pass
 
 filtertable = {
     'tempfile:': tempfilter,
     'pipe:': pipefilter,
     }
 
 def filter(s, cmd):
     "filter a string through a command that transforms its input to its output"
     for name, fn in filtertable.iteritems():
         if cmd.startswith(name):
             return fn(s, cmd[len(name):].lstrip())
     return pipefilter(s, cmd)
 
 def binary(s):
     """return true if a string is binary data"""
     return bool(s and '\0' in s)
 
 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)
 
 Abort = error.Abort
 
 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.
     """
     if sys.version_info >= (2, 7):
         return func
     def wrapper(*args, **kwargs):
         gcenabled = gc.isenabled()
         gc.disable()
         try:
             return func(*args, **kwargs)
         finally:
             if gcenabled:
                 gc.enable()
     return wrapper
 
 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 '.'
 
 def mainfrozen():
     """return True if we are a frozen executable.
 
     The code supports py2exe (most common, Windows only) and tools/freeze
     (portable, not much used).
     """
     return (safehasattr(sys, "frozen") or # new py2exe
             safehasattr(sys, "importers") or # old py2exe
             imp.is_frozen(u"__main__")) # tools/freeze
 
 # the location of data files matching the source code
 if 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)
 
 _hgexecutable = None
 
 def hgexecutable():
     """return location of the 'hg' executable.
 
     Defaults to $HG or 'hg' in the search path.
     """
     if _hgexecutable is None:
         hg = encoding.environ.get('HG')
         mainmod = sys.modules[pycompat.sysstr('__main__')]
         if hg:
             _sethgexecutable(hg)
         elif mainfrozen():
             if getattr(sys, 'frozen', None) == 'macosx_app':
                 # Env variable set by py2app
                 _sethgexecutable(encoding.environ['EXECUTABLEPATH'])
             else:
                 _sethgexecutable(pycompat.sysexecutable)
         elif (os.path.basename(
             pycompat.fsencode(getattr(mainmod, '__file__', ''))) == 'hg'):
             _sethgexecutable(pycompat.fsencode(mainmod.__file__))
         else:
             exe = findexe('hg') or os.path.basename(sys.argv[0])
             _sethgexecutable(exe)
     return _hgexecutable
 
 def _sethgexecutable(path):
     """set location of the 'hg' executable"""
     global _hgexecutable
     _hgexecutable = path
 
 def _isstdout(f):
     fileno = getattr(f, 'fileno', None)
     return fileno and fileno() == sys.__stdout__.fileno()
 
 def shellenviron(environ=None):
     """return environ with optional override, useful for shelling out"""
     def py2shell(val):
         'convert python object into string that is useful to shell'
         if val is None or val is False:
             return '0'
         if val is True:
             return '1'
         return str(val)
     env = dict(encoding.environ)
     if environ:
         env.update((k, py2shell(v)) for k, v in environ.iteritems())
     env['HG'] = hgexecutable()
     return env
 
 def system(cmd, environ=None, cwd=None, out=None):
     '''enhanced shell command execution.
     run with environment maybe modified, maybe in different dir.
 
     if out is specified, it is assumed to be a file-like object that has a
     write() method. stdout and stderr will be redirected to out.'''
     try:
         stdout.flush()
     except Exception:
         pass
     cmd = quotecommand(cmd)
     env = shellenviron(environ)
     if out is None or _isstdout(out):
         rc = subprocess.call(cmd, shell=True, close_fds=closefds,
                              env=env, cwd=cwd)
     else:
         proc = subprocess.Popen(cmd, shell=True, close_fds=closefds,
                                 env=env, cwd=cwd, stdout=subprocess.PIPE,
                                 stderr=subprocess.STDOUT)
         for line in iter(proc.stdout.readline, ''):
             out.write(line)
         proc.wait()
         rc = proc.returncode
     if pycompat.sysplatform == 'OpenVMS' and rc & 1:
         rc = 0
     return rc
 
 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 = {
     'btrfs',
     'ext2',
     'ext3',
     'ext4',
     'hfs',
     'jfs',
     '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.st_mtime + 1) & 0x7fffffff
                         os.utime(dest, (advanced, advanced))
         except shutil.Error as inst:
             raise Abort(str(inst))
 
 def copyfiles(src, dst, hardlink=None, progress=lambda t, pos: None):
     """Copy a directory tree using hardlinks if possible."""
     num = 0
 
     gettopic = lambda: hardlink and _('linking') or _('copying')
 
     if os.path.isdir(src):
         if hardlink is None:
             hardlink = (os.stat(src).st_dev ==
                         os.stat(os.path.dirname(dst)).st_dev)
         topic = gettopic()
         os.mkdir(dst)
         for name, kind in listdir(src):
             srcname = os.path.join(src, name)
             dstname = os.path.join(dst, name)
             def nprog(t, pos):
                 if pos is not None:
                     return progress(t, pos + num)
             hardlink, n = copyfiles(srcname, dstname, hardlink, progress=nprog)
             num += n
     else:
         if hardlink is None:
             hardlink = (os.stat(os.path.dirname(src)).st_dev ==
                         os.stat(os.path.dirname(dst)).st_dev)
         topic = gettopic()
 
         if hardlink:
             try:
                 oslink(src, dst)
             except (IOError, OSError):
                 hardlink = False
                 shutil.copy(src, dst)
         else:
             shutil.copy(src, dst)
         num += 1
         progress(topic, num)
     progress(topic, None)
 
     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'''.split()
 _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("just/a/normal/path")
     >>> checkwinfilename("foo/bar/con.xml")
     "filename contains 'con', which is reserved on Windows"
     >>> checkwinfilename("foo/con.xml/bar")
     "filename contains 'con', which is reserved on Windows"
     >>> checkwinfilename("foo/bar/xml.con")
     >>> checkwinfilename("foo/bar/AUX/bla.txt")
     "filename contains 'AUX', which is reserved on Windows"
     >>> checkwinfilename("foo/bar/bla:.txt")
     "filename contains ':', which is reserved on Windows"
     >>> checkwinfilename("foo/bar/b\07la.txt")
     "filename contains '\\x07', which is invalid on Windows"
     >>> checkwinfilename("foo/bar/bla ")
     "filename ends with ' ', which is not allowed on Windows"
     >>> checkwinfilename("../bar")
     >>> checkwinfilename("foo\\")
     "filename ends with '\\', which is invalid on Windows"
     >>> checkwinfilename("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 %r, which is invalid "
                          "on Windows") % 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.osname == 'nt':
     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):
     try:
         return os.symlink(info, pathname)
     except OSError as why:
         if why.errno == errno.EEXIST:
             raise
     except AttributeError: # no symlink in os
         pass
 
     ld = os.open(pathname, os.O_CREAT | os.O_WRONLY | os.O_EXCL)
     os.write(ld, info)
     os.close(ld)
 
 def readlock(pathname):
     try:
         return os.readlink(pathname)
     except OSError as why:
         if why.errno not in (errno.EINVAL, errno.ENOSYS):
             raise
     except AttributeError: # no symlink in os
         pass
     fp = posixfile(pathname)
     r = fp.read()
     fp.close()
     return r
 
 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 getfstype(dirpath):
     '''Get the filesystem type name from a directory (best-effort)
 
     Returns None if we are unsure. Raises OSError on ENOENT, EPERM, etc.
     '''
     return getattr(osutil, 'getfstype', lambda x: None)(dirpath)
 
 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 = testfile + ".hgtmp1"
     if os.path.lexists(f1):
         return False
     try:
         posixfile(f1, 'w').close()
     except IOError:
         try:
             os.unlink(f1)
         except OSError:
             pass
         return False
 
     f2 = testfile + ".hgtmp2"
     fd = None
     try:
         oslink(f1, f2)
         # nlinks() may behave differently for files on Windows shares if
         # the file is open.
         fd = posixfile(f2)
         return nlinks(f2) > 1
     except OSError:
         return False
     finally:
         if fd is not None:
             fd.close()
         for f in (f1, f2):
             try:
                 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 gui():
     '''Are we running in a GUI?'''
     if pycompat.sysplatform == 'darwin':
         if 'SSH_CONNECTION' in encoding.environ:
             # handle SSH access to a box where the user is logged in
             return False
         elif getattr(osutil, 'isgui', None):
             # check if a CoreGraphics session is available
             return osutil.isgui()
         else:
             # pure build; use a safe default
             return True
     else:
         return pycompat.osname == "nt" or encoding.environ.get("DISPLAY")
 
 def mktempcopy(name, emptyok=False, createmode=None):
     """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 = tempfile.mkstemp(prefix='.%s-' % fn, 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)
     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.st_ctime == old.stat.st_ctime and
                     self.stat.st_mtime == old.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.st_ctime == old.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.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)
         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.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):
     """unlink and remove the directory if it is empty"""
     if ignoremissing:
         tryunlink(f)
     else:
         unlink(f)
     # 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
 
 def makedate(timestamp=None):
     '''Return a unix timestamp (or the current time) as a (unixtime,
     offset) tuple based off the local timezone.'''
     if timestamp is None:
         timestamp = time.time()
     if timestamp < 0:
         hint = _("check your clock")
         raise Abort(_("negative timestamp: %d") % timestamp, hint=hint)
     delta = (datetime.datetime.utcfromtimestamp(timestamp) -
              datetime.datetime.fromtimestamp(timestamp))
     tz = delta.days * 86400 + delta.seconds
     return timestamp, tz
 
 def datestr(date=None, format='%a %b %d %H:%M:%S %Y %1%2'):
     """represent a (unixtime, offset) tuple as a localized time.
     unixtime is seconds since the epoch, and offset is the time zone's
     number of seconds away from UTC.
 
     >>> datestr((0, 0))
     'Thu Jan 01 00:00:00 1970 +0000'
     >>> datestr((42, 0))
     'Thu Jan 01 00:00:42 1970 +0000'
     >>> datestr((-42, 0))
     'Wed Dec 31 23:59:18 1969 +0000'
     >>> datestr((0x7fffffff, 0))
     'Tue Jan 19 03:14:07 2038 +0000'
     >>> datestr((-0x80000000, 0))
     'Fri Dec 13 20:45:52 1901 +0000'
     """
     t, tz = date or makedate()
     if "%1" in format or "%2" in format or "%z" in format:
         sign = (tz > 0) and "-" or "+"
         minutes = abs(tz) // 60
         q, r = divmod(minutes, 60)
         format = format.replace("%z", "%1%2")
         format = format.replace("%1", "%c%02d" % (sign, q))
         format = format.replace("%2", "%02d" % r)
     d = t - tz
     if d > 0x7fffffff:
         d = 0x7fffffff
     elif d < -0x80000000:
         d = -0x80000000
     # Never use time.gmtime() and datetime.datetime.fromtimestamp()
     # because they use the gmtime() system call which is buggy on Windows
     # for negative values.
     t = datetime.datetime(1970, 1, 1) + datetime.timedelta(seconds=d)
     s = encoding.strtolocal(t.strftime(encoding.strfromlocal(format)))
     return s
 
 def shortdate(date=None):
     """turn (timestamp, tzoff) tuple into iso 8631 date."""
     return datestr(date, format='%Y-%m-%d')
 
 def parsetimezone(s):
     """find a trailing timezone, if any, in string, and return a
        (offset, remainder) pair"""
 
     if s.endswith("GMT") or s.endswith("UTC"):
         return 0, s[:-3].rstrip()
 
     # Unix-style timezones [+-]hhmm
     if len(s) >= 5 and s[-5] in "+-" and s[-4:].isdigit():
         sign = (s[-5] == "+") and 1 or -1
         hours = int(s[-4:-2])
         minutes = int(s[-2:])
         return -sign * (hours * 60 + minutes) * 60, s[:-5].rstrip()
 
     # ISO8601 trailing Z
     if s.endswith("Z") and s[-2:-1].isdigit():
         return 0, s[:-1]
 
     # ISO8601-style [+-]hh:mm
     if (len(s) >= 6 and s[-6] in "+-" and s[-3] == ":" and
         s[-5:-3].isdigit() and s[-2:].isdigit()):
         sign = (s[-6] == "+") and 1 or -1
         hours = int(s[-5:-3])
         minutes = int(s[-2:])
         return -sign * (hours * 60 + minutes) * 60, s[:-6]
 
     return None, s
 
 def strdate(string, format, defaults=None):
     """parse a localized time string and return a (unixtime, offset) tuple.
     if the string cannot be parsed, ValueError is raised."""
     if defaults is None:
         defaults = {}
 
     # NOTE: unixtime = localunixtime + offset
     offset, date = parsetimezone(string)
 
     # add missing elements from defaults
     usenow = False # default to using biased defaults
     for part in ("S", "M", "HI", "d", "mb", "yY"): # decreasing specificity
         part = pycompat.bytestr(part)
         found = [True for p in part if ("%"+p) in format]
         if not found:
             date += "@" + defaults[part][usenow]
             format += "@%" + part[0]
         else:
             # We've found a specific time element, less specific time
             # elements are relative to today
             usenow = True
 
     timetuple = time.strptime(encoding.strfromlocal(date),
                               encoding.strfromlocal(format))
     localunixtime = int(calendar.timegm(timetuple))
     if offset is None:
         # local timezone
         unixtime = int(time.mktime(timetuple))
         offset = unixtime - localunixtime
     else:
         unixtime = localunixtime + offset
     return unixtime, offset
 
 def parsedate(date, formats=None, bias=None):
     """parse a localized date/time and return a (unixtime, offset) tuple.
 
     The date may be a "unixtime offset" string or in one of the specified
     formats. If the date already is a (unixtime, offset) tuple, it is returned.
 
     >>> parsedate(' today ') == parsedate(\
                                   datetime.date.today().strftime('%b %d'))
     True
     >>> parsedate( 'yesterday ') == parsedate((datetime.date.today() -\
                                                datetime.timedelta(days=1)\
                                               ).strftime('%b %d'))
     True
     >>> now, tz = makedate()
     >>> strnow, strtz = parsedate('now')
     >>> (strnow - now) < 1
     True
     >>> tz == strtz
     True
     """
     if bias is None:
         bias = {}
     if not date:
         return 0, 0
     if isinstance(date, tuple) and len(date) == 2:
         return date
     if not formats:
         formats = defaultdateformats
     date = date.strip()
 
     if date == 'now' or date == _('now'):
         return makedate()
     if date == 'today' or date == _('today'):
         date = datetime.date.today().strftime('%b %d')
     elif date == 'yesterday' or date == _('yesterday'):
         date = (datetime.date.today() -
                 datetime.timedelta(days=1)).strftime('%b %d')
 
     try:
         when, offset = map(int, date.split(' '))
     except ValueError:
         # fill out defaults
         now = makedate()
         defaults = {}
         for part in ("d", "mb", "yY", "HI", "M", "S"):
             # this piece is for rounding the specific end of unknowns
             b = bias.get(part)
             if b is None:
                 if part[0:1] in "HMS":
                     b = "00"
                 else:
                     b = "0"
 
             # this piece is for matching the generic end to today's date
             n = datestr(now, "%" + part[0:1])
 
             defaults[part] = (b, n)
 
         for format in formats:
             try:
                 when, offset = strdate(date, format, defaults)
             except (ValueError, OverflowError):
                 pass
             else:
                 break
         else:
             raise error.ParseError(_('invalid date: %r') % date)
     # validate explicit (probably user-specified) date and
     # time zone offset. values must fit in signed 32 bits for
     # current 32-bit linux runtimes. timezones go from UTC-12
     # to UTC+14
     if when < -0x80000000 or when > 0x7fffffff:
         raise error.ParseError(_('date exceeds 32 bits: %d') % when)
     if offset < -50400 or offset > 43200:
         raise error.ParseError(_('impossible time zone offset: %d') % offset)
     return when, offset
 
 def matchdate(date):
     """Return a function that matches a given date match specifier
 
     Formats include:
 
     '{date}' match a given date to the accuracy provided
 
     '<{date}' on or before a given date
 
     '>{date}' on or after a given date
 
     >>> p1 = parsedate("10:29:59")
     >>> p2 = parsedate("10:30:00")
     >>> p3 = parsedate("10:30:59")
     >>> p4 = parsedate("10:31:00")
     >>> p5 = parsedate("Sep 15 10:30:00 1999")
     >>> f = matchdate("10:30")
     >>> f(p1[0])
     False
     >>> f(p2[0])
     True
     >>> f(p3[0])
     True
     >>> f(p4[0])
     False
     >>> f(p5[0])
     False
     """
 
     def lower(date):
         d = {'mb': "1", 'd': "1"}
         return parsedate(date, extendeddateformats, d)[0]
 
     def upper(date):
         d = {'mb': "12", 'HI': "23", 'M': "59", 'S': "59"}
         for days in ("31", "30", "29"):
             try:
                 d["d"] = days
                 return parsedate(date, extendeddateformats, d)[0]
             except Abort:
                 pass
         d["d"] = "28"
         return parsedate(date, extendeddateformats, d)[0]
 
     date = date.strip()
 
     if not date:
         raise Abort(_("dates cannot consist entirely of whitespace"))
     elif date[0] == "<":
         if not date[1:]:
             raise Abort(_("invalid day spec, use '<DATE'"))
         when = upper(date[1:])
         return lambda x: x <= when
     elif date[0] == ">":
         if not date[1:]:
             raise Abort(_("invalid day spec, use '>DATE'"))
         when = lower(date[1:])
         return lambda x: x >= when
     elif date[0] == "-":
         try:
             days = int(date[1:])
         except ValueError:
             raise Abort(_("invalid day spec: %s") % date[1:])
         if days < 0:
             raise Abort(_("%s must be nonnegative (see 'hg help dates')")
                 % date[1:])
         when = makedate()[0] - days * 3600 * 24
         return lambda x: x >= when
     elif " to " in date:
         a, b = date.split(" to ")
         start, stop = lower(a), upper(b)
         return lambda x: x >= start and x <= stop
     else:
         start, stop = lower(date), upper(date)
         return lambda x: x >= start and x <= stop
 
 def stringmatcher(pattern, casesensitive=True):
     """
     accepts a string, possibly starting with 're:' or 'literal:' prefix.
     returns the matcher name, pattern, and matcher function.
     missing or unknown prefixes are treated as literal matches.
 
     helper for tests:
     >>> def test(pattern, *tests):
     ...     kind, pattern, matcher = stringmatcher(pattern)
     ...     return (kind, pattern, [bool(matcher(t)) for t in tests])
     >>> def itest(pattern, *tests):
     ...     kind, pattern, matcher = stringmatcher(pattern, casesensitive=False)
     ...     return (kind, pattern, [bool(matcher(t)) for t in tests])
 
     exact matching (no prefix):
     >>> test('abcdefg', 'abc', 'def', 'abcdefg')
     ('literal', 'abcdefg', [False, False, True])
 
     regex matching ('re:' prefix)
     >>> test('re:a.+b', 'nomatch', 'fooadef', 'fooadefbar')
     ('re', 'a.+b', [False, False, True])
 
     force exact matches ('literal:' prefix)
     >>> test('literal:re:foobar', 'foobar', 're:foobar')
     ('literal', 're:foobar', [False, True])
 
     unknown prefixes are ignored and treated as literals
     >>> test('foo:bar', 'foo', 'bar', 'foo:bar')
     ('literal', 'foo:bar', [False, False, True])
 
     case insensitive regex matches
     >>> itest('re:A.+b', 'nomatch', 'fooadef', 'fooadefBar')
     ('re', 'A.+b', [False, False, True])
 
     case insensitive literal matches
     >>> itest('ABCDEFG', 'abc', 'def', 'abcdefg')
     ('literal', 'ABCDEFG', [False, False, True])
     """
     if pattern.startswith('re:'):
         pattern = pattern[3:]
         try:
             flags = 0
             if not casesensitive:
                 flags = remod.I
             regex = remod.compile(pattern, flags)
         except remod.error as e:
             raise error.ParseError(_('invalid regular expression: %s')
                                    % e)
         return 're', pattern, regex.search
     elif pattern.startswith('literal:'):
         pattern = pattern[8:]
 
     match = pattern.__eq__
 
     if not casesensitive:
         ipat = encoding.lower(pattern)
         match = lambda s: ipat == encoding.lower(s)
     return 'literal', pattern, match
 
 def shortuser(user):
     """Return a short representation of a user name or email address."""
     f = user.find('@')
     if f >= 0:
         user = user[:f]
     f = user.find('<')
     if f >= 0:
         user = user[f + 1:]
     f = user.find(' ')
     if f >= 0:
         user = user[:f]
     f = user.find('.')
     if f >= 0:
         user = user[:f]
     return user
 
 def emailuser(user):
     """Return the user portion of an email address."""
     f = user.find('@')
     if f >= 0:
         user = user[:f]
     f = user.find('<')
     if f >= 0:
         user = user[f + 1:]
     return user
 
 def email(author):
     '''get email of author.'''
     r = author.find('>')
     if r == -1:
         r = None
     return author[author.find('<') + 1:r]
 
 def ellipsis(text, maxlength=400):
     """Trim string to at most maxlength (default: 400) columns in display."""
     return encoding.trim(text, maxlength, ellipsis='...')
 
 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')),
     )
 
 # 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)
 
 if pycompat.oslinesep == '\r\n':
     tonativeeol = tocrlf
     fromnativeeol = tolf
 else:
     tonativeeol = pycompat.identity
     fromnativeeol = pycompat.identity
 
 def escapestr(s):
     # call underlying function of s.encode('string_escape') directly for
     # Python 3 compatibility
     return codecs.escape_encode(s)[0]
 
 def unescapestr(s):
     return codecs.escape_decode(s)[0]
 
 def uirepr(s):
     # Avoid double backslash in Windows path repr()
     return repr(s).replace('\\\\', '\\')
 
 # delay import of textwrap
 def MBTextWrapper(**kwargs):
     class tw(textwrap.TextWrapper):
         """
         Extend TextWrapper for width-awareness.
 
         Neither number of 'bytes' in any encoding nor 'characters' is
         appropriate to calculate terminal columns for specified string.
 
         Original TextWrapper implementation uses built-in 'len()' directly,
         so overriding is needed to use width information of each characters.
 
         In addition, characters classified into 'ambiguous' width are
         treated as wide in East Asian area, but as narrow in other.
 
         This requires use decision to determine width of such characters.
         """
         def _cutdown(self, ucstr, space_left):
             l = 0
             colwidth = encoding.ucolwidth
             for i in xrange(len(ucstr)):
                 l += colwidth(ucstr[i])
                 if space_left < l:
                     return (ucstr[:i], ucstr[i:])
             return ucstr, ''
 
         # overriding of base class
         def _handle_long_word(self, reversed_chunks, cur_line, cur_len, width):
             space_left = max(width - cur_len, 1)
 
             if self.break_long_words:
                 cut, res = self._cutdown(reversed_chunks[-1], space_left)
                 cur_line.append(cut)
                 reversed_chunks[-1] = res
             elif not cur_line:
                 cur_line.append(reversed_chunks.pop())
 
         # this overriding code is imported from TextWrapper of Python 2.6
         # to calculate columns of string by 'encoding.ucolwidth()'
         def _wrap_chunks(self, chunks):
             colwidth = encoding.ucolwidth
 
             lines = []
             if self.width <= 0:
                 raise ValueError("invalid width %r (must be > 0)" % self.width)
 
             # Arrange in reverse order so items can be efficiently popped
             # from a stack of chucks.
             chunks.reverse()
 
             while chunks:
 
                 # Start the list of chunks that will make up the current line.
                 # cur_len is just the length of all the chunks in cur_line.
                 cur_line = []
                 cur_len = 0
 
                 # Figure out which static string will prefix this line.
                 if lines:
                     indent = self.subsequent_indent
                 else:
                     indent = self.initial_indent
 
                 # Maximum width for this line.
                 width = self.width - len(indent)
 
                 # First chunk on line is whitespace -- drop it, unless this
                 # is the very beginning of the text (i.e. no lines started yet).
                 if self.drop_whitespace and chunks[-1].strip() == r'' and lines:
                     del chunks[-1]
 
                 while chunks:
                     l = colwidth(chunks[-1])
 
                     # Can at least squeeze this chunk onto the current line.
                     if cur_len + l <= width:
                         cur_line.append(chunks.pop())
                         cur_len += l
 
                     # Nope, this line is full.
                     else:
                         break
 
                 # The current line is full, and the next chunk is too big to
                 # fit on *any* line (not just this one).
                 if chunks and colwidth(chunks[-1]) > width:
                     self._handle_long_word(chunks, cur_line, cur_len, width)
 
                 # If the last chunk on this line is all whitespace, drop it.
                 if (self.drop_whitespace and
                     cur_line and cur_line[-1].strip() == r''):
                     del cur_line[-1]
 
                 # Convert current line back to a string and store it in list
                 # of all lines (return value).
                 if cur_line:
                     lines.append(indent + r''.join(cur_line))
 
             return lines
 
     global MBTextWrapper
     MBTextWrapper = tw
     return tw(**kwargs)
 
 def wrap(line, width, initindent='', hangindent=''):
     maxindent = max(len(hangindent), len(initindent))
     if width <= maxindent:
         # adjust for weird terminal size
         width = max(78, maxindent + 1)
     line = line.decode(pycompat.sysstr(encoding.encoding),
                     pycompat.sysstr(encoding.encodingmode))
     initindent = initindent.decode(pycompat.sysstr(encoding.encoding),
                     pycompat.sysstr(encoding.encodingmode))
     hangindent = hangindent.decode(pycompat.sysstr(encoding.encoding),
                     pycompat.sysstr(encoding.encodingmode))
     wrapper = MBTextWrapper(width=width,
                             initial_indent=initindent,
                             subsequent_indent=hangindent)
     return wrapper.fill(line).encode(pycompat.sysstr(encoding.encoding))
 
 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 hgcmd():
     """Return the command used to execute current hg
 
     This is different from hgexecutable() because on Windows we want
     to avoid things opening new shell windows like batch files, so we
     get either the python call or current executable.
     """
     if mainfrozen():
         if getattr(sys, 'frozen', None) == 'macosx_app':
             # Env variable set by py2app
             return [encoding.environ['EXECUTABLEPATH']]
         else:
             return [pycompat.sysexecutable]
     return gethgcmd()
 
 def rundetached(args, condfn):
     """Execute the argument list in a detached process.
 
     condfn is a callable which is called repeatedly and should return
     True once the child process is known to have started successfully.
     At this point, the child process PID is returned. If the child
     process fails to start or finishes before condfn() evaluates to
     True, return -1.
     """
     # Windows case is easier because the child process is either
     # successfully starting and validating the condition or exiting
     # on failure. We just poll on its PID. On Unix, if the child
     # process fails to start, it will be left in a zombie state until
     # the parent wait on it, which we cannot do since we expect a long
     # running process on success. Instead we listen for SIGCHLD telling
     # us our child process terminated.
     terminated = set()
     def handler(signum, frame):
         terminated.add(os.wait())
     prevhandler = None
     SIGCHLD = getattr(signal, 'SIGCHLD', None)
     if SIGCHLD is not None:
         prevhandler = signal.signal(SIGCHLD, handler)
     try:
         pid = spawndetached(args)
         while not condfn():
             if ((pid in terminated or not testpid(pid))
                 and not condfn()):
                 return -1
             time.sleep(0.1)
         return pid
     finally:
         if prevhandler is not None:
             signal.signal(signal.SIGCHLD, prevhandler)
 
 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(r'%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(port)
     except socket.error:
         raise Abort(_("no port number associated with service '%s'") % port)
 
 _booleans = {'1': True, 'yes': True, 'true': True, 'on': True, 'always': True,
              '0': False, 'no': False, 'false': False, 'off': False,
              'never': False}
 
 def parsebool(s):
     """Parse s into a boolean.
 
     If s is not a valid boolean, returns None.
     """
     return _booleans.get(s.lower(), None)
 
 _hextochr = dict((a + b, chr(int(a + b, 16)))
                  for a in string.hexdigits for b in string.hexdigits)
 
 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('http://www.ietf.org/rfc/rfc2396.txt')
     <url scheme: 'http', host: 'www.ietf.org', path: 'rfc/rfc2396.txt'>
     >>> url('ssh://[::1]:2200//home/joe/repo')
     <url scheme: 'ssh', host: '[::1]', port: '2200', path: '/home/joe/repo'>
     >>> url('file:///home/joe/repo')
     <url scheme: 'file', path: '/home/joe/repo'>
     >>> url('file:///c:/temp/foo/')
     <url scheme: 'file', path: 'c:/temp/foo/'>
     >>> url('bundle:foo')
     <url scheme: 'bundle', path: 'foo'>
     >>> url('bundle://../foo')
     <url scheme: 'bundle', path: '../foo'>
     >>> url(r'c:\foo\bar')
     <url path: 'c:\\foo\\bar'>
     >>> url(r'\\blah\blah\blah')
     <url path: '\\\\blah\\blah\\blah'>
     >>> url(r'\\blah\blah\blah#baz')
     <url path: '\\\\blah\\blah\\blah', fragment: 'baz'>
     >>> url(r'file:///C:\users\me')
     <url scheme: 'file', path: 'C:\\users\\me'>
 
     Authentication credentials:
 
     >>> url('ssh://joe:xyz@x/repo')
     <url scheme: 'ssh', user: 'joe', passwd: 'xyz', host: 'x', path: 'repo'>
     >>> url('ssh://joe@x/repo')
     <url scheme: 'ssh', user: 'joe', host: 'x', path: 'repo'>
 
     Query strings and fragments:
 
     >>> url('http://host/a?b#c')
     <url scheme: 'http', host: 'host', path: 'a', query: 'b', fragment: 'c'>
     >>> url('http://host/a?b#c', parsequery=False, parsefragment=False)
     <url scheme: 'http', host: 'host', path: 'a?b#c'>
 
     Empty path:
 
     >>> url('')
     <url path: ''>
     >>> url('#a')
     <url path: '', fragment: 'a'>
     >>> url('http://host/')
     <url scheme: 'http', host: 'host', path: ''>
     >>> url('http://host/#a')
     <url scheme: 'http', host: 'host', path: '', fragment: 'a'>
 
     Only scheme:
 
     >>> url('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 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))
 
     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, v))
         return '<url %s>' % ', '.join(attrs)
 
-    def __str__(self):
+    def __bytes__(self):
         r"""Join the URL's components back into a URL string.
 
         Examples:
 
         >>> str(url('http://user:pw@host:80/c:/bob?fo:oo#ba:ar'))
         'http://user:pw@host:80/c:/bob?fo:oo#ba:ar'
         >>> str(url('http://user:pw@host:80/?foo=bar&baz=42'))
         'http://user:pw@host:80/?foo=bar&baz=42'
         >>> str(url('http://user:pw@host:80/?foo=bar%3dbaz'))
         'http://user:pw@host:80/?foo=bar%3dbaz'
         >>> str(url('ssh://user:pw@[::1]:2200//home/joe#'))
         'ssh://user:pw@[::1]:2200//home/joe#'
         >>> str(url('http://localhost:80//'))
         'http://localhost:80//'
         >>> str(url('http://localhost:80/'))
         'http://localhost:80/'
         >>> str(url('http://localhost:80'))
         'http://localhost:80/'
         >>> str(url('bundle:foo'))
         'bundle:foo'
         >>> str(url('bundle://../foo'))
         'bundle:../foo'
         >>> str(url('path'))
         'path'
         >>> str(url('file:///tmp/foo/bar'))
         'file:///tmp/foo/bar'
         >>> str(url('file:///c:/tmp/foo/bar'))
         'file:///c:/tmp/foo/bar'
         >>> print url(r'bundle:foo\bar')
         bundle:foo\bar
         >>> print url(r'file:///D:\data\hg')
         file:///D:\data\hg
         """
-        return encoding.strfromlocal(self.__bytes__())
-
-    def __bytes__(self):
         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(r'\\'):
             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 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 str(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')),
     )
 
 _timenesting = [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):
         start = timer()
         indent = 2
         _timenesting[0] += indent
         try:
             return func(*args, **kwargs)
         finally:
             elapsed = timer() - start
             _timenesting[0] -= indent
             stderr.write('%s%s: %s\n' %
                          (' ' * _timenesting[0], func.__name__,
                           timecount(elapsed)))
     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('30')
     30
     >>> sizetoint('2.2kb')
     2252
     >>> sizetoint('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:%s', 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 % (fn, ln), 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=stderr, otherf=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 safehasattr(map, 'iteritems') and skip is not None:
             for f, s in map.iteritems():
                 if s[0] != skip:
                     addpath(f)
         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
 
 def finddirs(path):
     pos = path.rfind('/')
     while pos != -1:
         yield path[:pos]
         pos = path.rfind('/', 0, pos)
 
 class ctxmanager(object):
     '''A context manager for use in 'with' blocks to allow multiple
     contexts to be entered at once.  This is both safer and more
     flexible than contextlib.nested.
 
     Once Mercurial supports Python 2.7+, this will become mostly
     unnecessary.
     '''
 
     def __init__(self, *args):
         '''Accepts a list of no-argument functions that return context
         managers.  These will be invoked at __call__ time.'''
         self._pending = args
         self._atexit = []
 
     def __enter__(self):
         return self
 
     def enter(self):
         '''Create and enter context managers in the order in which they were
         passed to the constructor.'''
         values = []
         for func in self._pending:
             obj = func()
             values.append(obj.__enter__())
             self._atexit.append(obj.__exit__)
         del self._pending
         return values
 
     def atexit(self, func, *args, **kwargs):
         '''Add a function to call when this context manager exits.  The
         ordering of multiple atexit calls is unspecified, save that
         they will happen before any __exit__ functions.'''
         def wrapper(exc_type, exc_val, exc_tb):
             func(*args, **kwargs)
         self._atexit.append(wrapper)
         return func
 
     def __exit__(self, exc_type, exc_val, exc_tb):
         '''Context managers are exited in the reverse order from which
         they were created.'''
         received = exc_type is not None
         suppressed = False
         pending = None
         self._atexit.reverse()
         for exitfunc in self._atexit:
             try:
                 if exitfunc(exc_type, exc_val, exc_tb):
                     suppressed = True
                     exc_type = None
                     exc_val = None
                     exc_tb = None
             except BaseException:
                 pending = sys.exc_info()
                 exc_type, exc_val, exc_tb = pending = sys.exc_info()
         del self._atexit
         if pending:
             raise exc_val
         return received and suppressed
 
 # compression code
 
 SERVERROLE = 'server'
 CLIENTROLE = 'client'
 
 compewireprotosupport = collections.namedtuple(u'compenginewireprotosupport',
                                                (u'name', u'serverpriority',
                                                 u'clientpriority'))
 
 class compressormanager(object):
     """Holds registrations of various compression engines.
 
     This class essentially abstracts the differences between compression
     engines to allow new compression formats to be added easily, possibly from
     extensions.
 
     Compressors are registered against the global instance by calling its
     ``register()`` method.
     """
     def __init__(self):
         self._engines = {}
         # Bundle spec human name to engine name.
         self._bundlenames = {}
         # Internal bundle identifier to engine name.
         self._bundletypes = {}
         # Revlog header to engine name.
         self._revlogheaders = {}
         # Wire proto identifier to engine name.
         self._wiretypes = {}
 
     def __getitem__(self, key):
         return self._engines[key]
 
     def __contains__(self, key):
         return key in self._engines
 
     def __iter__(self):
         return iter(self._engines.keys())
 
     def register(self, engine):
         """Register a compression engine with the manager.
 
         The argument must be a ``compressionengine`` instance.
         """
         if not isinstance(engine, compressionengine):
             raise ValueError(_('argument must be a compressionengine'))
 
         name = engine.name()
 
         if name in self._engines:
             raise error.Abort(_('compression engine %s already registered') %
                               name)
 
         bundleinfo = engine.bundletype()
         if bundleinfo:
             bundlename, bundletype = bundleinfo
 
             if bundlename in self._bundlenames:
                 raise error.Abort(_('bundle name %s already registered') %
                                   bundlename)
             if bundletype in self._bundletypes:
                 raise error.Abort(_('bundle type %s already registered by %s') %
                                   (bundletype, self._bundletypes[bundletype]))
 
             # No external facing name declared.
             if bundlename:
                 self._bundlenames[bundlename] = name
 
             self._bundletypes[bundletype] = name
 
         wiresupport = engine.wireprotosupport()
         if wiresupport:
             wiretype = wiresupport.name
             if wiretype in self._wiretypes:
                 raise error.Abort(_('wire protocol compression %s already '
                                     'registered by %s') %
                                   (wiretype, self._wiretypes[wiretype]))
 
             self._wiretypes[wiretype] = name
 
         revlogheader = engine.revlogheader()
         if revlogheader and revlogheader in self._revlogheaders:
             raise error.Abort(_('revlog header %s already registered by %s') %
                               (revlogheader, self._revlogheaders[revlogheader]))
 
         if revlogheader:
             self._revlogheaders[revlogheader] = name
 
         self._engines[name] = engine
 
     @property
     def supportedbundlenames(self):
         return set(self._bundlenames.keys())
 
     @property
     def supportedbundletypes(self):
         return set(self._bundletypes.keys())
 
     def forbundlename(self, bundlename):
         """Obtain a compression engine registered to a bundle name.
 
         Will raise KeyError if the bundle type isn't registered.
 
         Will abort if the engine is known but not available.
         """
         engine = self._engines[self._bundlenames[bundlename]]
         if not engine.available():
             raise error.Abort(_('compression engine %s could not be loaded') %
                               engine.name())
         return engine
 
     def forbundletype(self, bundletype):
         """Obtain a compression engine registered to a bundle type.
 
         Will raise KeyError if the bundle type isn't registered.
 
         Will abort if the engine is known but not available.
         """
         engine = self._engines[self._bundletypes[bundletype]]
         if not engine.available():
             raise error.Abort(_('compression engine %s could not be loaded') %
                               engine.name())
         return engine
 
     def supportedwireengines(self, role, onlyavailable=True):
         """Obtain compression engines that support the wire protocol.
 
         Returns a list of engines in prioritized order, most desired first.
 
         If ``onlyavailable`` is set, filter out engines that can't be
         loaded.
         """
         assert role in (SERVERROLE, CLIENTROLE)
 
         attr = 'serverpriority' if role == SERVERROLE else 'clientpriority'
 
         engines = [self._engines[e] for e in self._wiretypes.values()]
         if onlyavailable:
             engines = [e for e in engines if e.available()]
 
         def getkey(e):
             # Sort first by priority, highest first. In case of tie, sort
             # alphabetically. This is arbitrary, but ensures output is
             # stable.
             w = e.wireprotosupport()
             return -1 * getattr(w, attr), w.name
 
         return list(sorted(engines, key=getkey))
 
     def forwiretype(self, wiretype):
         engine = self._engines[self._wiretypes[wiretype]]
         if not engine.available():
             raise error.Abort(_('compression engine %s could not be loaded') %
                               engine.name())
         return engine
 
     def forrevlogheader(self, header):
         """Obtain a compression engine registered to a revlog header.
 
         Will raise KeyError if the revlog header value isn't registered.
         """
         return self._engines[self._revlogheaders[header]]
 
 compengines = compressormanager()
 
 class compressionengine(object):
     """Base class for compression engines.
 
     Compression engines must implement the interface defined by this class.
     """
     def name(self):
         """Returns the name of the compression engine.
 
         This is the key the engine is registered under.
 
         This method must be implemented.
         """
         raise NotImplementedError()
 
     def available(self):
         """Whether the compression engine is available.
 
         The intent of this method is to allow optional compression engines
         that may not be available in all installations (such as engines relying
         on C extensions that may not be present).
         """
         return True
 
     def bundletype(self):
         """Describes bundle identifiers for this engine.
 
         If this compression engine isn't supported for bundles, returns None.
 
         If this engine can be used for bundles, returns a 2-tuple of strings of
         the user-facing "bundle spec" compression name and an internal
         identifier used to denote the compression format within bundles. To
         exclude the name from external usage, set the first element to ``None``.
 
         If bundle compression is supported, the class must also implement
         ``compressstream`` and `decompressorreader``.
 
         The docstring of this method is used in the help system to tell users
         about this engine.
         """
         return None
 
     def wireprotosupport(self):
         """Declare support for this compression format on the wire protocol.
 
         If this compression engine isn't supported for compressing wire
         protocol payloads, returns None.
 
         Otherwise, returns ``compenginewireprotosupport`` with the following
         fields:
 
         * String format identifier
         * Integer priority for the server
         * Integer priority for the client
 
         The integer priorities are used to order the advertisement of format
         support by server and client. The highest integer is advertised
         first. Integers with non-positive values aren't advertised.
 
         The priority values are somewhat arbitrary and only used for default
         ordering. The relative order can be changed via config options.
 
         If wire protocol compression is supported, the class must also implement
         ``compressstream`` and ``decompressorreader``.
         """
         return None
 
     def revlogheader(self):
         """Header added to revlog chunks that identifies this engine.
 
         If this engine can be used to compress revlogs, this method should
         return the bytes used to identify chunks compressed with this engine.
         Else, the method should return ``None`` to indicate it does not
         participate in revlog compression.
         """
         return None
 
     def compressstream(self, it, opts=None):
         """Compress an iterator of chunks.
 
         The method receives an iterator (ideally a generator) of chunks of
         bytes to be compressed. It returns an iterator (ideally a generator)
         of bytes of chunks representing the compressed output.
 
         Optionally accepts an argument defining how to perform compression.
         Each engine treats this argument differently.
         """
         raise NotImplementedError()
 
     def decompressorreader(self, fh):
         """Perform decompression on a file object.
 
         Argument is an object with a ``read(size)`` method that returns
         compressed data. Return value is an object with a ``read(size)`` that
         returns uncompressed data.
         """
         raise NotImplementedError()
 
     def revlogcompressor(self, opts=None):
         """Obtain an object that can be used to compress revlog entries.
 
         The object has a ``compress(data)`` method that compresses binary
         data. This method returns compressed binary data or ``None`` if
         the data could not be compressed (too small, not compressible, etc).
         The returned data should have a header uniquely identifying this
         compression format so decompression can be routed to this engine.
         This header should be identified by the ``revlogheader()`` return
         value.
 
         The object has a ``decompress(data)`` method that decompresses
         data. The method will only be called if ``data`` begins with
         ``revlogheader()``. The method should return the raw, uncompressed
         data or raise a ``RevlogError``.
 
         The object is reusable but is not thread safe.
         """
         raise NotImplementedError()
 
 class _zlibengine(compressionengine):
     def name(self):
         return 'zlib'
 
     def bundletype(self):
         """zlib compression using the DEFLATE algorithm.
 
         All Mercurial clients should support this format. The compression
         algorithm strikes a reasonable balance between compression ratio
         and size.
         """
         return 'gzip', 'GZ'
 
     def wireprotosupport(self):
         return compewireprotosupport('zlib', 20, 20)
 
     def revlogheader(self):
         return 'x'
 
     def compressstream(self, it, opts=None):
         opts = opts or {}
 
         z = zlib.compressobj(opts.get('level', -1))
         for chunk in it:
             data = z.compress(chunk)
             # Not all calls to compress emit data. It is cheaper to inspect
             # here than to feed empty chunks through generator.
             if data:
                 yield data
 
         yield z.flush()
 
     def decompressorreader(self, fh):
         def gen():
             d = zlib.decompressobj()
             for chunk in filechunkiter(fh):
                 while chunk:
                     # Limit output size to limit memory.
                     yield d.decompress(chunk, 2 ** 18)
                     chunk = d.unconsumed_tail
 
         return chunkbuffer(gen())
 
     class zlibrevlogcompressor(object):
         def compress(self, data):
             insize = len(data)
             # Caller handles empty input case.
             assert insize > 0
 
             if insize < 44:
                 return None
 
             elif insize <= 1000000:
                 compressed = zlib.compress(data)
                 if len(compressed) < insize:
                     return compressed
                 return None
 
             # zlib makes an internal copy of the input buffer, doubling
             # memory usage for large inputs. So do streaming compression
             # on large inputs.
             else:
                 z = zlib.compressobj()
                 parts = []
                 pos = 0
                 while pos < insize:
                     pos2 = pos + 2**20
                     parts.append(z.compress(data[pos:pos2]))
                     pos = pos2
                 parts.append(z.flush())
 
                 if sum(map(len, parts)) < insize:
                     return ''.join(parts)
                 return None
 
         def decompress(self, data):
             try:
                 return zlib.decompress(data)
             except zlib.error as e:
                 raise error.RevlogError(_('revlog decompress error: %s') %
                                         str(e))
 
     def revlogcompressor(self, opts=None):
         return self.zlibrevlogcompressor()
 
 compengines.register(_zlibengine())
 
 class _bz2engine(compressionengine):
     def name(self):
         return 'bz2'
 
     def bundletype(self):
         """An algorithm that produces smaller bundles than ``gzip``.
 
         All Mercurial clients should support this format.
 
         This engine will likely produce smaller bundles than ``gzip`` but
         will be significantly slower, both during compression and
         decompression.
 
         If available, the ``zstd`` engine can yield similar or better
         compression at much higher speeds.
         """
         return 'bzip2', 'BZ'
 
     # We declare a protocol name but don't advertise by default because
     # it is slow.
     def wireprotosupport(self):
         return compewireprotosupport('bzip2', 0, 0)
 
     def compressstream(self, it, opts=None):
         opts = opts or {}
         z = bz2.BZ2Compressor(opts.get('level', 9))
         for chunk in it:
             data = z.compress(chunk)
             if data:
                 yield data
 
         yield z.flush()
 
     def decompressorreader(self, fh):
         def gen():
             d = bz2.BZ2Decompressor()
             for chunk in filechunkiter(fh):
                 yield d.decompress(chunk)
 
         return chunkbuffer(gen())
 
 compengines.register(_bz2engine())
 
 class _truncatedbz2engine(compressionengine):
     def name(self):
         return 'bz2truncated'
 
     def bundletype(self):
         return None, '_truncatedBZ'
 
     # We don't implement compressstream because it is hackily handled elsewhere.
 
     def decompressorreader(self, fh):
         def gen():
             # The input stream doesn't have the 'BZ' header. So add it back.
             d = bz2.BZ2Decompressor()
             d.decompress('BZ')
             for chunk in filechunkiter(fh):
                 yield d.decompress(chunk)
 
         return chunkbuffer(gen())
 
 compengines.register(_truncatedbz2engine())
 
 class _noopengine(compressionengine):
     def name(self):
         return 'none'
 
     def bundletype(self):
         """No compression is performed.
 
         Use this compression engine to explicitly disable compression.
         """
         return 'none', 'UN'
 
     # Clients always support uncompressed payloads. Servers don't because
     # unless you are on a fast network, uncompressed payloads can easily
     # saturate your network pipe.
     def wireprotosupport(self):
         return compewireprotosupport('none', 0, 10)
 
     # We don't implement revlogheader because it is handled specially
     # in the revlog class.
 
     def compressstream(self, it, opts=None):
         return it
 
     def decompressorreader(self, fh):
         return fh
 
     class nooprevlogcompressor(object):
         def compress(self, data):
             return None
 
     def revlogcompressor(self, opts=None):
         return self.nooprevlogcompressor()
 
 compengines.register(_noopengine())
 
 class _zstdengine(compressionengine):
     def name(self):
         return 'zstd'
 
     @propertycache
     def _module(self):
         # Not all installs have the zstd module available. So defer importing
         # until first access.
         try:
             from . import zstd
             # Force delayed import.
             zstd.__version__
             return zstd
         except ImportError:
             return None
 
     def available(self):
         return bool(self._module)
 
     def bundletype(self):
         """A modern compression algorithm that is fast and highly flexible.
 
         Only supported by Mercurial 4.1 and newer clients.
 
         With the default settings, zstd compression is both faster and yields
         better compression than ``gzip``. It also frequently yields better
         compression than ``bzip2`` while operating at much higher speeds.
 
         If this engine is available and backwards compatibility is not a
         concern, it is likely the best available engine.
         """
         return 'zstd', 'ZS'
 
     def wireprotosupport(self):
         return compewireprotosupport('zstd', 50, 50)
 
     def revlogheader(self):
         return '\x28'
 
     def compressstream(self, it, opts=None):
         opts = opts or {}
         # zstd level 3 is almost always significantly faster than zlib
         # while providing no worse compression. It strikes a good balance
         # between speed and compression.
         level = opts.get('level', 3)
 
         zstd = self._module
         z = zstd.ZstdCompressor(level=level).compressobj()
         for chunk in it:
             data = z.compress(chunk)
             if data:
                 yield data
 
         yield z.flush()
 
     def decompressorreader(self, fh):
         zstd = self._module
         dctx = zstd.ZstdDecompressor()
         return chunkbuffer(dctx.read_from(fh))
 
     class zstdrevlogcompressor(object):
         def __init__(self, zstd, level=3):
             # Writing the content size adds a few bytes to the output. However,
             # it allows decompression to be more optimal since we can
             # pre-allocate a buffer to hold the result.
             self._cctx = zstd.ZstdCompressor(level=level,
                                              write_content_size=True)
             self._dctx = zstd.ZstdDecompressor()
             self._compinsize = zstd.COMPRESSION_RECOMMENDED_INPUT_SIZE
             self._decompinsize = zstd.DECOMPRESSION_RECOMMENDED_INPUT_SIZE
 
         def compress(self, data):
             insize = len(data)
             # Caller handles empty input case.
             assert insize > 0
 
             if insize < 50:
                 return None
 
             elif insize <= 1000000:
                 compressed = self._cctx.compress(data)
                 if len(compressed) < insize:
                     return compressed
                 return None
             else:
                 z = self._cctx.compressobj()
                 chunks = []
                 pos = 0
                 while pos < insize:
                     pos2 = pos + self._compinsize
                     chunk = z.compress(data[pos:pos2])
                     if chunk:
                         chunks.append(chunk)
                     pos = pos2
                 chunks.append(z.flush())
 
                 if sum(map(len, chunks)) < insize:
                     return ''.join(chunks)
                 return None
 
         def decompress(self, data):
             insize = len(data)
 
             try:
                 # This was measured to be faster than other streaming
                 # decompressors.
                 dobj = self._dctx.decompressobj()
                 chunks = []
                 pos = 0
                 while pos < insize:
                     pos2 = pos + self._decompinsize
                     chunk = dobj.decompress(data[pos:pos2])
                     if chunk:
                         chunks.append(chunk)
                     pos = pos2
                 # Frame should be exhausted, so no finish() API.
 
                 return ''.join(chunks)
             except Exception as e:
                 raise error.RevlogError(_('revlog decompress error: %s') %
                                         str(e))
 
     def revlogcompressor(self, opts=None):
         opts = opts or {}
         return self.zstdrevlogcompressor(self._module,
                                          level=opts.get('level', 3))
 
 compengines.register(_zstdengine())
 
 def bundlecompressiontopics():
     """Obtains a list of available bundle compressions for use in help."""
     # help.makeitemsdocs() expects a dict of names to items with a .__doc__.
     items = {}
 
     # We need to format the docstring. So use a dummy object/type to hold it
     # rather than mutating the original.
     class docobject(object):
         pass
 
     for name in compengines:
         engine = compengines[name]
 
         if not engine.available():
             continue
 
         bt = engine.bundletype()
         if not bt or not bt[0]:
             continue
 
         doc = pycompat.sysstr('``%s``\n    %s') % (
             bt[0], engine.bundletype.__doc__)
 
         value = docobject()
         value.__doc__ = doc
 
         items[bt[0]] = value
 
     return items
 
 # convenient shortcut
 dst = debugstacktrace