revlog: flag processor

Add the ability for revlog objects to process revision flags and apply
registered transforms on read/write operations.

This patch introduces:
- the 'revlog._processflags()' method that looks at revision flags and applies
  flag processors registered on them. Due to the need to handle non-commutative
  operations, flag transforms are applied in stable order but the order in which
  the transforms are applied is reversed between read and write operations.
- the 'addflagprocessor()' method allowing to register processors on flags.
  Flag processors are defined as a 3-tuple of (read, write, raw) functions to be
  applied depending on the operation being performed.
- an update on 'revlog.addrevision()' behavior. The current flagprocessor design
  relies on extensions to wrap around 'addrevision()' to set flags on revision
  data, and on the flagprocessor to perform the actual transformation of its
  contents. In the lfs case, this means we need to process flags before we meet
  the 2GB size check, leading to performing some operations before it happens:
  - if flags are set on the revision data, we assume some extensions might be
    modifying the contents using the flag processor next, and we compute the
    node for the original revision data (still allowing extension to override
    the node by wrapping around 'addrevision()').
  - we then invoke the flag processor to apply registered transforms (in lfs's
    case, drastically reducing the size of large blobs).
  - finally, we proceed with the 2GB size check.

Note: In the case a cachedelta is passed to 'addrevision()' and we detect the
flag processor modified the revision data, we chose to trust the flag processor
and drop the cachedelta.

tests/flagprocessorext.py

@@ -0,0 +1,137 b''
+# coding=UTF-8
+
+from __future__ import absolute_import
+
+import base64
+import zlib
+
+from mercurial import (
+    changegroup,
+    extensions,
+    filelog,
+    revlog,
+    util,
+)
+
+# Test only: These flags are defined here only in the context of testing the
+# behavior of the flag processor. The canonical way to add flags is to get in
+# touch with the community and make them known in revlog.
+REVIDX_NOOP = (1 << 3)
+REVIDX_BASE64 = (1 << 2)
+REVIDX_GZIP = (1 << 1)
+REVIDX_FAIL = 1
+
+def validatehash(self, text):
+    return True
+
+def bypass(self, text):
+    return False
+
+def noopdonothing(self, text):
+    return (text, True)
+
+def b64encode(self, text):
+    return (base64.b64encode(text), False)
+
+def b64decode(self, text):
+    return (base64.b64decode(text), True)
+
+def gzipcompress(self, text):
+    return (zlib.compress(text), False)
+
+def gzipdecompress(self, text):
+    return (zlib.decompress(text), True)
+
+def supportedoutgoingversions(orig, repo):
+    versions = orig(repo)
+    versions.discard('01')
+    versions.discard('02')
+    versions.add('03')
+    return versions
+
+def allsupportedversions(orig, ui):
+    versions = orig(ui)
+    versions.add('03')
+    return versions
+
+def noopaddrevision(orig, self, text, transaction, link, p1, p2,
+                    cachedelta=None, node=None,
+                    flags=revlog.REVIDX_DEFAULT_FLAGS):
+    if '[NOOP]' in text:
+        flags |= REVIDX_NOOP
+    return orig(self, text, transaction, link, p1, p2, cachedelta=cachedelta,
+                node=node, flags=flags)
+
+def b64addrevision(orig, self, text, transaction, link, p1, p2,
+                   cachedelta=None, node=None,
+                   flags=revlog.REVIDX_DEFAULT_FLAGS):
+    if '[BASE64]' in text:
+        flags |= REVIDX_BASE64
+    return orig(self, text, transaction, link, p1, p2, cachedelta=cachedelta,
+                node=node, flags=flags)
+
+def gzipaddrevision(orig, self, text, transaction, link, p1, p2,
+                    cachedelta=None, node=None,
+                    flags=revlog.REVIDX_DEFAULT_FLAGS):
+    if '[GZIP]' in text:
+        flags |= REVIDX_GZIP
+    return orig(self, text, transaction, link, p1, p2, cachedelta=cachedelta,
+                node=node, flags=flags)
+
+def failaddrevision(orig, self, text, transaction, link, p1, p2,
+                    cachedelta=None, node=None,
+                    flags=revlog.REVIDX_DEFAULT_FLAGS):
+    # This addrevision wrapper is meant to add a flag we will not have
+    # transforms registered for, ensuring we handle this error case.
+    if '[FAIL]' in text:
+        flags |= REVIDX_FAIL
+    return orig(self, text, transaction, link, p1, p2, cachedelta=cachedelta,
+                node=node, flags=flags)
+
+def extsetup(ui):
+    # Enable changegroup3 for flags to be sent over the wire
+    wrapfunction = extensions.wrapfunction
+    wrapfunction(changegroup,
+                 'supportedoutgoingversions',
+                 supportedoutgoingversions)
+    wrapfunction(changegroup,
+                 'allsupportedversions',
+                 allsupportedversions)
+
+    # Teach revlog about our test flags
+    flags = [REVIDX_NOOP, REVIDX_BASE64, REVIDX_GZIP, REVIDX_FAIL]
+    revlog.REVIDX_KNOWN_FLAGS |= util.bitsfrom(flags)
+    revlog.REVIDX_FLAGS_ORDER.extend(flags)
+
+    # Add wrappers for addrevision, responsible to set flags depending on the
+    # revision data contents.
+    wrapfunction(filelog.filelog, 'addrevision', noopaddrevision)
+    wrapfunction(filelog.filelog, 'addrevision', b64addrevision)
+    wrapfunction(filelog.filelog, 'addrevision', gzipaddrevision)
+    wrapfunction(filelog.filelog, 'addrevision', failaddrevision)
+
+    # Register flag processors for each extension
+    revlog.addflagprocessor(
+        REVIDX_NOOP,
+        (
+            noopdonothing,
+            noopdonothing,
+            validatehash,
+        )
+    )
+    revlog.addflagprocessor(
+        REVIDX_BASE64,
+        (
+            b64decode,
+            b64encode,
+            bypass,
+        ),
+    )
+    revlog.addflagprocessor(
+        REVIDX_GZIP,
+        (
+            gzipdecompress,
+            gzipcompress,
+            bypass
+        )
+    )

tests/test-flagprocessor.t

@@ -0,0 +1,165 b''
+# Create server
+  $ hg init server
+  $ cd server
+  $ cat >> .hg/hgrc << EOF
+  > [extensions]
+  > extension=$TESTDIR/flagprocessorext.py
+  > EOF
+  $ cd ../
+
+# Clone server and enable extensions
+  $ hg clone -q server client
+  $ cd client
+  $ cat >> .hg/hgrc << EOF
+  > [extensions]
+  > extension=$TESTDIR/flagprocessorext.py
+  > EOF
+
+# Commit file that will trigger the noop extension
+  $ echo '[NOOP]' > noop
+  $ hg commit -Aqm "noop"
+
+# Commit file that will trigger the base64 extension
+  $ echo '[BASE64]' > base64
+  $ hg commit -Aqm 'base64'
+
+# Commit file that will trigger the gzip extension
+  $ echo '[GZIP]' > gzip
+  $ hg commit -Aqm 'gzip'
+
+# Commit file that will trigger noop and base64
+  $ echo '[NOOP][BASE64]' > noop-base64
+  $ hg commit -Aqm 'noop+base64'
+
+# Commit file that will trigger noop and gzip
+  $ echo '[NOOP][GZIP]' > noop-gzip
+  $ hg commit -Aqm 'noop+gzip'
+
+# Commit file that will trigger base64 and gzip
+  $ echo '[BASE64][GZIP]' > base64-gzip
+  $ hg commit -Aqm 'base64+gzip'
+
+# Commit file that will trigger base64, gzip and noop
+  $ echo '[BASE64][GZIP][NOOP]' > base64-gzip-noop
+  $ hg commit -Aqm 'base64+gzip+noop'
+
+# TEST: ensure the revision data is consistent
+  $ hg cat noop
+  [NOOP]
+  $ hg debugdata noop 0
+  [NOOP]
+
+  $ hg cat -r . base64
+  [BASE64]
+  $ hg debugdata base64 0
+  W0JBU0U2NF0K (no-eol)
+
+  $ hg cat -r . gzip
+  [GZIP]
+  $ hg debugdata gzip 0
+  x\x9c\x8bv\x8f\xf2\x0c\x88\xe5\x02\x00\x08\xc8\x01\xfd (no-eol) (esc)
+
+  $ hg cat -r . noop-base64
+  [NOOP][BASE64]
+  $ hg debugdata noop-base64 0
+  W05PT1BdW0JBU0U2NF0K (no-eol)
+
+  $ hg cat -r . noop-gzip
+  [NOOP][GZIP]
+  $ hg debugdata noop-gzip 0
+  x\x9c\x8b\xf6\xf3\xf7\x0f\x88\x8dv\x8f\xf2\x0c\x88\xe5\x02\x00\x1dH\x03\xf1 (no-eol) (esc)
+
+  $ hg cat -r . base64-gzip
+  [BASE64][GZIP]
+  $ hg debugdata base64-gzip 0
+  eJyLdnIMdjUziY12j/IMiOUCACLBBDo= (no-eol)
+
+  $ hg cat -r . base64-gzip-noop
+  [BASE64][GZIP][NOOP]
+  $ hg debugdata base64-gzip-noop 0
+  eJyLdnIMdjUziY12j/IMiI328/cPiOUCAESjBi4= (no-eol)
+
+# Push to the server
+  $ hg push
+  pushing to $TESTTMP/server (glob)
+  searching for changes
+  adding changesets
+  adding manifests
+  adding file changes
+  added 7 changesets with 7 changes to 7 files
+
+# Initialize new client (not cloning) and setup extension
+  $ cd ..
+  $ hg init client2
+  $ cd client2
+  $ cat >> .hg/hgrc << EOF
+  > [paths]
+  > default = $TESTTMP/server
+  > [extensions]
+  > extension=$TESTDIR/flagprocessorext.py
+  > EOF
+
+# Pull from server and update to latest revision
+  $ hg pull default
+  pulling from $TESTTMP/server (glob)
+  requesting all changes
+  adding changesets
+  adding manifests
+  adding file changes
+  added 7 changesets with 7 changes to 7 files
+  (run 'hg update' to get a working copy)
+  $ hg update
+  7 files updated, 0 files merged, 0 files removed, 0 files unresolved
+
+# TEST: ensure the revision data is consistent
+  $ hg cat noop
+  [NOOP]
+  $ hg debugdata noop 0
+  [NOOP]
+
+  $ hg cat -r . base64
+  [BASE64]
+  $ hg debugdata base64 0
+  W0JBU0U2NF0K (no-eol)
+
+  $ hg cat -r . gzip
+  [GZIP]
+  $ hg debugdata gzip 0
+  x\x9c\x8bv\x8f\xf2\x0c\x88\xe5\x02\x00\x08\xc8\x01\xfd (no-eol) (esc)
+
+  $ hg cat -r . noop-base64
+  [NOOP][BASE64]
+  $ hg debugdata noop-base64 0
+  W05PT1BdW0JBU0U2NF0K (no-eol)
+
+  $ hg cat -r . noop-gzip
+  [NOOP][GZIP]
+  $ hg debugdata noop-gzip 0
+  x\x9c\x8b\xf6\xf3\xf7\x0f\x88\x8dv\x8f\xf2\x0c\x88\xe5\x02\x00\x1dH\x03\xf1 (no-eol) (esc)
+
+  $ hg cat -r . base64-gzip
+  [BASE64][GZIP]
+  $ hg debugdata base64-gzip 0
+  eJyLdnIMdjUziY12j/IMiOUCACLBBDo= (no-eol)
+
+  $ hg cat -r . base64-gzip-noop
+  [BASE64][GZIP][NOOP]
+  $ hg debugdata base64-gzip-noop 0
+  eJyLdnIMdjUziY12j/IMiI328/cPiOUCAESjBi4= (no-eol)
+
+# TEST: ensure a missing processor is handled
+  $ echo '[FAIL][BASE64][GZIP][NOOP]' > fail-base64-gzip-noop
+  $ hg commit -Aqm 'fail+base64+gzip+noop'
+  abort: missing processor for flag '0x1'!
+  [255]
+
+# TEST: ensure we cannot register several flag processors on the same flag
+  $ cat >> .hg/hgrc << EOF
+  > [extensions]
+  > extension=$TESTDIR/flagprocessorext.py
+  > duplicate=$TESTDIR/flagprocessorext.py
+  > EOF
+  $ echo 'this should fail' > file
+  $ hg commit -Aqm 'add file'
+  abort: cannot register multiple processors on flag '0x8'.
+  [255]

mercurial/bundlerepo.py

@@ -1,554 +1,557 b''
 # bundlerepo.py - repository class for viewing uncompressed bundles
 #
 # Copyright 2006, 2007 Benoit Boissinot <[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.
 
 """Repository class for viewing uncompressed bundles.
 
 This provides a read-only repository interface to bundles as if they
 were part of the actual repository.
 """
 
 from __future__ import absolute_import
 
 import os
 import shutil
 import tempfile
 
 from .i18n import _
 from .node import nullid
 
 from . import (
     bundle2,
     changegroup,
     changelog,
     cmdutil,
     discovery,
     error,
     exchange,
     filelog,
     localrepo,
     manifest,
     mdiff,
     node as nodemod,
     pathutil,
     phases,
     pycompat,
     revlog,
     scmutil,
     util,
 )
 
 class bundlerevlog(revlog.revlog):
     def __init__(self, opener, indexfile, bundle, linkmapper):
         # How it works:
         # To retrieve a revision, we need to know the offset of the revision in
         # the bundle (an unbundle object). We store this offset in the index
         # (start). The base of the delta is stored in the base field.
         #
         # To differentiate a rev in the bundle from a rev in the revlog, we
         # check revision against repotiprev.
         opener = scmutil.readonlyvfs(opener)
         revlog.revlog.__init__(self, opener, indexfile)
         self.bundle = bundle
         n = len(self)
         self.repotiprev = n - 1
         chain = None
         self.bundlerevs = set() # used by 'bundle()' revset expression
         getchunk = lambda: bundle.deltachunk(chain)
         for chunkdata in iter(getchunk, {}):
             node = chunkdata['node']
             p1 = chunkdata['p1']
             p2 = chunkdata['p2']
             cs = chunkdata['cs']
             deltabase = chunkdata['deltabase']
             delta = chunkdata['delta']
 
             size = len(delta)
             start = bundle.tell() - size
 
             link = linkmapper(cs)
             if node in self.nodemap:
                 # this can happen if two branches make the same change
                 chain = node
                 self.bundlerevs.add(self.nodemap[node])
                 continue
 
             for p in (p1, p2):
                 if p not in self.nodemap:
                     raise error.LookupError(p, self.indexfile,
                                             _("unknown parent"))
 
             if deltabase not in self.nodemap:
                 raise LookupError(deltabase, self.indexfile,
                                   _('unknown delta base'))
 
             baserev = self.rev(deltabase)
             # start, size, full unc. size, base (unused), link, p1, p2, node
             e = (revlog.offset_type(start, 0), size, -1, baserev, link,
                  self.rev(p1), self.rev(p2), node)
             self.index.insert(-1, e)
             self.nodemap[node] = n
             self.bundlerevs.add(n)
             chain = node
             n += 1
 
     def _chunk(self, rev):
         # Warning: in case of bundle, the diff is against what we stored as
         # delta base, not against rev - 1
         # XXX: could use some caching
         if rev <= self.repotiprev:
             return revlog.revlog._chunk(self, rev)
         self.bundle.seek(self.start(rev))
         return self.bundle.read(self.length(rev))
 
     def revdiff(self, rev1, rev2):
         """return or calculate a delta between two revisions"""
         if rev1 > self.repotiprev and rev2 > self.repotiprev:
             # hot path for bundle
             revb = self.index[rev2][3]
             if revb == rev1:
                 return self._chunk(rev2)
         elif rev1 <= self.repotiprev and rev2 <= self.repotiprev:
             return revlog.revlog.revdiff(self, rev1, rev2)
 
         return mdiff.textdiff(self.revision(self.node(rev1)),
                               self.revision(self.node(rev2)))
 
     def revision(self, nodeorrev, raw=False):
         """return an uncompressed revision of a given node or revision
         number.
         """
         if isinstance(nodeorrev, int):
             rev = nodeorrev
             node = self.node(rev)
         else:
             node = nodeorrev
             rev = self.rev(node)
 
         if node == nullid:
             return ""
 
         text = None
         chain = []
         iterrev = rev
         # reconstruct the revision if it is from a changegroup
         while iterrev > self.repotiprev:
             if self._cache and self._cache[1] == iterrev:
                 text = self._cache[2]
                 break
             chain.append(iterrev)
             iterrev = self.index[iterrev][3]
         if text is None:
             text = self.baserevision(iterrev)
 
         while chain:
             delta = self._chunk(chain.pop())
             text = mdiff.patches(text, [delta])
 
-        self.checkhash(text, node, rev=rev)
+        text, validatehash = self._processflags(text, self.flags(rev),
+                                                'read', raw=raw)
+        if validatehash:
+            self.checkhash(text, node, rev=rev)
         self._cache = (node, rev, text)
         return text
 
     def baserevision(self, nodeorrev):
         # Revlog subclasses may override 'revision' method to modify format of
         # content retrieved from revlog. To use bundlerevlog with such class one
         # needs to override 'baserevision' and make more specific call here.
         return revlog.revlog.revision(self, nodeorrev)
 
     def addrevision(self, text, transaction, link, p1=None, p2=None, d=None):
         raise NotImplementedError
     def addgroup(self, revs, linkmapper, transaction):
         raise NotImplementedError
     def strip(self, rev, minlink):
         raise NotImplementedError
     def checksize(self):
         raise NotImplementedError
 
 class bundlechangelog(bundlerevlog, changelog.changelog):
     def __init__(self, opener, bundle):
         changelog.changelog.__init__(self, opener)
         linkmapper = lambda x: x
         bundlerevlog.__init__(self, opener, self.indexfile, bundle,
                               linkmapper)
 
     def baserevision(self, nodeorrev):
         # Although changelog doesn't override 'revision' method, some extensions
         # may replace this class with another that does. Same story with
         # manifest and filelog classes.
 
         # This bypasses filtering on changelog.node() and rev() because we need
         # revision text of the bundle base even if it is hidden.
         oldfilter = self.filteredrevs
         try:
             self.filteredrevs = ()
             return changelog.changelog.revision(self, nodeorrev)
         finally:
             self.filteredrevs = oldfilter
 
 class bundlemanifest(bundlerevlog, manifest.manifestrevlog):
     def __init__(self, opener, bundle, linkmapper, dirlogstarts=None, dir=''):
         manifest.manifestrevlog.__init__(self, opener, dir=dir)
         bundlerevlog.__init__(self, opener, self.indexfile, bundle,
                               linkmapper)
         if dirlogstarts is None:
             dirlogstarts = {}
             if self.bundle.version == "03":
                 dirlogstarts = _getfilestarts(self.bundle)
         self._dirlogstarts = dirlogstarts
         self._linkmapper = linkmapper
 
     def baserevision(self, nodeorrev):
         node = nodeorrev
         if isinstance(node, int):
             node = self.node(node)
 
         if node in self.fulltextcache:
             result = self.fulltextcache[node].tostring()
         else:
             result = manifest.manifestrevlog.revision(self, nodeorrev)
         return result
 
     def dirlog(self, d):
         if d in self._dirlogstarts:
             self.bundle.seek(self._dirlogstarts[d])
             return bundlemanifest(
                 self.opener, self.bundle, self._linkmapper,
                 self._dirlogstarts, dir=d)
         return super(bundlemanifest, self).dirlog(d)
 
 class bundlefilelog(bundlerevlog, filelog.filelog):
     def __init__(self, opener, path, bundle, linkmapper):
         filelog.filelog.__init__(self, opener, path)
         bundlerevlog.__init__(self, opener, self.indexfile, bundle,
                               linkmapper)
 
     def baserevision(self, nodeorrev):
         return filelog.filelog.revision(self, nodeorrev)
 
 class bundlepeer(localrepo.localpeer):
     def canpush(self):
         return False
 
 class bundlephasecache(phases.phasecache):
     def __init__(self, *args, **kwargs):
         super(bundlephasecache, self).__init__(*args, **kwargs)
         if util.safehasattr(self, 'opener'):
             self.opener = scmutil.readonlyvfs(self.opener)
 
     def write(self):
         raise NotImplementedError
 
     def _write(self, fp):
         raise NotImplementedError
 
     def _updateroots(self, phase, newroots, tr):
         self.phaseroots[phase] = newroots
         self.invalidate()
         self.dirty = True
 
 def _getfilestarts(bundle):
     bundlefilespos = {}
     for chunkdata in iter(bundle.filelogheader, {}):
         fname = chunkdata['filename']
         bundlefilespos[fname] = bundle.tell()
         for chunk in iter(lambda: bundle.deltachunk(None), {}):
             pass
     return bundlefilespos
 
 class bundlerepository(localrepo.localrepository):
     def __init__(self, ui, path, bundlename):
         def _writetempbundle(read, suffix, header=''):
             """Write a temporary file to disk
 
             This is closure because we need to make sure this tracked by
             self.tempfile for cleanup purposes."""
             fdtemp, temp = self.vfs.mkstemp(prefix="hg-bundle-",
                                             suffix=".hg10un")
             self.tempfile = temp
 
             with os.fdopen(fdtemp, 'wb') as fptemp:
                 fptemp.write(header)
                 while True:
                     chunk = read(2**18)
                     if not chunk:
                         break
                     fptemp.write(chunk)
 
             return self.vfs.open(self.tempfile, mode="rb")
         self._tempparent = None
         try:
             localrepo.localrepository.__init__(self, ui, path)
         except error.RepoError:
             self._tempparent = tempfile.mkdtemp()
             localrepo.instance(ui, self._tempparent, 1)
             localrepo.localrepository.__init__(self, ui, self._tempparent)
         self.ui.setconfig('phases', 'publish', False, 'bundlerepo')
 
         if path:
             self._url = 'bundle:' + util.expandpath(path) + '+' + bundlename
         else:
             self._url = 'bundle:' + bundlename
 
         self.tempfile = None
         f = util.posixfile(bundlename, "rb")
         self.bundlefile = self.bundle = exchange.readbundle(ui, f, bundlename)
 
         if isinstance(self.bundle, bundle2.unbundle20):
             cgstream = None
             for part in self.bundle.iterparts():
                 if part.type == 'changegroup':
                     if cgstream is not None:
                         raise NotImplementedError("can't process "
                                                   "multiple changegroups")
                     cgstream = part
                     version = part.params.get('version', '01')
                     legalcgvers = changegroup.supportedincomingversions(self)
                     if version not in legalcgvers:
                         msg = _('Unsupported changegroup version: %s')
                         raise error.Abort(msg % version)
                     if self.bundle.compressed():
                         cgstream = _writetempbundle(part.read,
                                                     ".cg%sun" % version)
 
             if cgstream is None:
                 raise error.Abort(_('No changegroups found'))
             cgstream.seek(0)
 
             self.bundle = changegroup.getunbundler(version, cgstream, 'UN')
 
         elif self.bundle.compressed():
             f = _writetempbundle(self.bundle.read, '.hg10un', header='HG10UN')
             self.bundlefile = self.bundle = exchange.readbundle(ui, f,
                                                                 bundlename,
                                                                 self.vfs)
 
         # dict with the mapping 'filename' -> position in the bundle
         self.bundlefilespos = {}
 
         self.firstnewrev = self.changelog.repotiprev + 1
         phases.retractboundary(self, None, phases.draft,
                                [ctx.node() for ctx in self[self.firstnewrev:]])
 
     @localrepo.unfilteredpropertycache
     def _phasecache(self):
         return bundlephasecache(self, self._phasedefaults)
 
     @localrepo.unfilteredpropertycache
     def changelog(self):
         # consume the header if it exists
         self.bundle.changelogheader()
         c = bundlechangelog(self.svfs, self.bundle)
         self.manstart = self.bundle.tell()
         return c
 
     def _constructmanifest(self):
         self.bundle.seek(self.manstart)
         # consume the header if it exists
         self.bundle.manifestheader()
         linkmapper = self.unfiltered().changelog.rev
         m = bundlemanifest(self.svfs, self.bundle, linkmapper)
         self.filestart = self.bundle.tell()
         return m
 
     @localrepo.unfilteredpropertycache
     def manstart(self):
         self.changelog
         return self.manstart
 
     @localrepo.unfilteredpropertycache
     def filestart(self):
         self.manifestlog
         return self.filestart
 
     def url(self):
         return self._url
 
     def file(self, f):
         if not self.bundlefilespos:
             self.bundle.seek(self.filestart)
             self.bundlefilespos = _getfilestarts(self.bundle)
 
         if f in self.bundlefilespos:
             self.bundle.seek(self.bundlefilespos[f])
             linkmapper = self.unfiltered().changelog.rev
             return bundlefilelog(self.svfs, f, self.bundle, linkmapper)
         else:
             return filelog.filelog(self.svfs, f)
 
     def close(self):
         """Close assigned bundle file immediately."""
         self.bundlefile.close()
         if self.tempfile is not None:
             self.vfs.unlink(self.tempfile)
         if self._tempparent:
             shutil.rmtree(self._tempparent, True)
 
     def cancopy(self):
         return False
 
     def peer(self):
         return bundlepeer(self)
 
     def getcwd(self):
         return pycompat.getcwd() # always outside the repo
 
     # Check if parents exist in localrepo before setting
     def setparents(self, p1, p2=nullid):
         p1rev = self.changelog.rev(p1)
         p2rev = self.changelog.rev(p2)
         msg = _("setting parent to node %s that only exists in the bundle\n")
         if self.changelog.repotiprev < p1rev:
             self.ui.warn(msg % nodemod.hex(p1))
         if self.changelog.repotiprev < p2rev:
             self.ui.warn(msg % nodemod.hex(p2))
         return super(bundlerepository, self).setparents(p1, p2)
 
 def instance(ui, path, create):
     if create:
         raise error.Abort(_('cannot create new bundle repository'))
     # internal config: bundle.mainreporoot
     parentpath = ui.config("bundle", "mainreporoot", "")
     if not parentpath:
         # try to find the correct path to the working directory repo
         parentpath = cmdutil.findrepo(pycompat.getcwd())
         if parentpath is None:
             parentpath = ''
     if parentpath:
         # Try to make the full path relative so we get a nice, short URL.
         # In particular, we don't want temp dir names in test outputs.
         cwd = pycompat.getcwd()
         if parentpath == cwd:
             parentpath = ''
         else:
             cwd = pathutil.normasprefix(cwd)
             if parentpath.startswith(cwd):
                 parentpath = parentpath[len(cwd):]
     u = util.url(path)
     path = u.localpath()
     if u.scheme == 'bundle':
         s = path.split("+", 1)
         if len(s) == 1:
             repopath, bundlename = parentpath, s[0]
         else:
             repopath, bundlename = s
     else:
         repopath, bundlename = parentpath, path
     return bundlerepository(ui, repopath, bundlename)
 
 class bundletransactionmanager(object):
     def transaction(self):
         return None
 
     def close(self):
         raise NotImplementedError
 
     def release(self):
         raise NotImplementedError
 
 def getremotechanges(ui, repo, other, onlyheads=None, bundlename=None,
                      force=False):
     '''obtains a bundle of changes incoming from other
 
     "onlyheads" restricts the returned changes to those reachable from the
       specified heads.
     "bundlename", if given, stores the bundle to this file path permanently;
       otherwise it's stored to a temp file and gets deleted again when you call
       the returned "cleanupfn".
     "force" indicates whether to proceed on unrelated repos.
 
     Returns a tuple (local, csets, cleanupfn):
 
     "local" is a local repo from which to obtain the actual incoming
       changesets; it is a bundlerepo for the obtained bundle when the
       original "other" is remote.
     "csets" lists the incoming changeset node ids.
     "cleanupfn" must be called without arguments when you're done processing
       the changes; it closes both the original "other" and the one returned
       here.
     '''
     tmp = discovery.findcommonincoming(repo, other, heads=onlyheads,
                                        force=force)
     common, incoming, rheads = tmp
     if not incoming:
         try:
             if bundlename:
                 os.unlink(bundlename)
         except OSError:
             pass
         return repo, [], other.close
 
     commonset = set(common)
     rheads = [x for x in rheads if x not in commonset]
 
     bundle = None
     bundlerepo = None
     localrepo = other.local()
     if bundlename or not localrepo:
         # create a bundle (uncompressed if other repo is not local)
 
         # developer config: devel.legacy.exchange
         legexc = ui.configlist('devel', 'legacy.exchange')
         forcebundle1 = 'bundle2' not in legexc and 'bundle1' in legexc
         canbundle2 = (not forcebundle1
                       and other.capable('getbundle')
                       and other.capable('bundle2'))
         if canbundle2:
             kwargs = {}
             kwargs['common'] = common
             kwargs['heads'] = rheads
             kwargs['bundlecaps'] = exchange.caps20to10(repo)
             kwargs['cg'] = True
             b2 = other.getbundle('incoming', **kwargs)
             fname = bundle = changegroup.writechunks(ui, b2._forwardchunks(),
                                                      bundlename)
         else:
             if other.capable('getbundle'):
                 cg = other.getbundle('incoming', common=common, heads=rheads)
             elif onlyheads is None and not other.capable('changegroupsubset'):
                 # compat with older servers when pulling all remote heads
                 cg = other.changegroup(incoming, "incoming")
                 rheads = None
             else:
                 cg = other.changegroupsubset(incoming, rheads, 'incoming')
             if localrepo:
                 bundletype = "HG10BZ"
             else:
                 bundletype = "HG10UN"
             fname = bundle = bundle2.writebundle(ui, cg, bundlename,
                                                      bundletype)
         # keep written bundle?
         if bundlename:
             bundle = None
         if not localrepo:
             # use the created uncompressed bundlerepo
             localrepo = bundlerepo = bundlerepository(repo.baseui, repo.root,
                                                       fname)
             # this repo contains local and other now, so filter out local again
             common = repo.heads()
     if localrepo:
         # Part of common may be remotely filtered
         # So use an unfiltered version
         # The discovery process probably need cleanup to avoid that
         localrepo = localrepo.unfiltered()
 
     csets = localrepo.changelog.findmissing(common, rheads)
 
     if bundlerepo:
         reponodes = [ctx.node() for ctx in bundlerepo[bundlerepo.firstnewrev:]]
         remotephases = other.listkeys('phases')
 
         pullop = exchange.pulloperation(bundlerepo, other, heads=reponodes)
         pullop.trmanager = bundletransactionmanager()
         exchange._pullapplyphases(pullop, remotephases)
 
     def cleanup():
         if bundlerepo:
             bundlerepo.close()
         if bundle:
             os.unlink(bundle)
         other.close()
 
     return (localrepo, csets, cleanup)

mercurial/revlog.py

@@ -1,1833 +1,1948 b''
 # revlog.py - storage back-end for mercurial
 #
 # Copyright 2005-2007 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.
 
 """Storage back-end for Mercurial.
 
 This provides efficient delta storage with O(1) retrieve and append
 and O(changes) merge between branches.
 """
 
 from __future__ import absolute_import
 
 import collections
 import errno
 import hashlib
 import os
 import struct
 import zlib
 
 # import stuff from node for others to import from revlog
 from .node import (
     bin,
     hex,
     nullid,
     nullrev,
 )
 from .i18n import _
 from . import (
     ancestor,
     error,
     mdiff,
     parsers,
     templatefilters,
     util,
 )
 
 _pack = struct.pack
 _unpack = struct.unpack
 _compress = zlib.compress
 _decompress = zlib.decompress
 
 # revlog header flags
 REVLOGV0 = 0
 REVLOGNG = 1
 REVLOGNGINLINEDATA = (1 << 16)
 REVLOGGENERALDELTA = (1 << 17)
 REVLOG_DEFAULT_FLAGS = REVLOGNGINLINEDATA
 REVLOG_DEFAULT_FORMAT = REVLOGNG
 REVLOG_DEFAULT_VERSION = REVLOG_DEFAULT_FORMAT | REVLOG_DEFAULT_FLAGS
 REVLOGNG_FLAGS = REVLOGNGINLINEDATA | REVLOGGENERALDELTA
 
 # revlog index flags
 REVIDX_ISCENSORED = (1 << 15) # revision has censor metadata, must be verified
 REVIDX_DEFAULT_FLAGS = 0
-REVIDX_KNOWN_FLAGS = REVIDX_ISCENSORED
+# stable order in which flags need to be processed and their processors applied
+REVIDX_FLAGS_ORDER = [
+    REVIDX_ISCENSORED,
+]
+REVIDX_KNOWN_FLAGS = util.bitsfrom(REVIDX_FLAGS_ORDER)
 
 # max size of revlog with inline data
 _maxinline = 131072
 _chunksize = 1048576
 
 RevlogError = error.RevlogError
 LookupError = error.LookupError
 CensoredNodeError = error.CensoredNodeError
+ProgrammingError = error.ProgrammingError
+
+# Store flag processors (cf. 'addflagprocessor()' to register)
+_flagprocessors = {
+    REVIDX_ISCENSORED: None,
+}
+
+def addflagprocessor(flag, processor):
+    """Register a flag processor on a revision data flag.
+
+    Invariant:
+    - Flags need to be defined in REVIDX_KNOWN_FLAGS and REVIDX_FLAGS_ORDER.
+    - Only one flag processor can be registered on a specific flag.
+    - flagprocessors must be 3-tuples of functions (read, write, raw) with the
+      following signatures:
+          - (read)  f(self, text) -> newtext, bool
+          - (write) f(self, text) -> newtext, bool
+          - (raw)   f(self, text) -> bool
+      The boolean returned by these transforms is used to determine whether
+      'newtext' can be used for hash integrity checking.
+
+      Note: The 'raw' transform is used for changegroup generation and in some
+      debug commands. In this case the transform only indicates whether the
+      contents can be used for hash integrity checks.
+    """
+    if not flag & REVIDX_KNOWN_FLAGS:
+        msg = _("cannot register processor on unknown flag '%#x'.") % (flag)
+        raise ProgrammingError(msg)
+    if flag not in REVIDX_FLAGS_ORDER:
+        msg = _("flag '%#x' undefined in REVIDX_FLAGS_ORDER.") % (flag)
+        raise ProgrammingError(msg)
+    if flag in _flagprocessors:
+        msg = _("cannot register multiple processors on flag '%#x'.") % (flag)
+        raise error.Abort(msg)
+    _flagprocessors[flag] = processor
 
 def getoffset(q):
     return int(q >> 16)
 
 def gettype(q):
     return int(q & 0xFFFF)
 
 def offset_type(offset, type):
     if (type & ~REVIDX_KNOWN_FLAGS) != 0:
         raise ValueError('unknown revlog index flags')
     return long(long(offset) << 16 | type)
 
 _nullhash = hashlib.sha1(nullid)
 
 def hash(text, p1, p2):
     """generate a hash from the given text and its parent hashes
 
     This hash combines both the current file contents and its history
     in a manner that makes it easy to distinguish nodes with the same
     content in the revision graph.
     """
     # As of now, if one of the parent node is null, p2 is null
     if p2 == nullid:
         # deep copy of a hash is faster than creating one
         s = _nullhash.copy()
         s.update(p1)
     else:
         # none of the parent nodes are nullid
         l = [p1, p2]
         l.sort()
         s = hashlib.sha1(l[0])
         s.update(l[1])
     s.update(text)
     return s.digest()
 
 def decompress(bin):
     """ decompress the given input """
     if not bin:
         return bin
     t = bin[0]
     if t == '\0':
         return bin
     if t == 'x':
         try:
             return _decompress(bin)
         except zlib.error as e:
             raise RevlogError(_("revlog decompress error: %s") % str(e))
     if t == 'u':
         return util.buffer(bin, 1)
     raise RevlogError(_("unknown compression type %r") % t)
 
 # index v0:
 #  4 bytes: offset
 #  4 bytes: compressed length
 #  4 bytes: base rev
 #  4 bytes: link rev
 # 20 bytes: parent 1 nodeid
 # 20 bytes: parent 2 nodeid
 # 20 bytes: nodeid
 indexformatv0 = ">4l20s20s20s"
 
 class revlogoldio(object):
     def __init__(self):
         self.size = struct.calcsize(indexformatv0)
 
     def parseindex(self, data, inline):
         s = self.size
         index = []
         nodemap = {nullid: nullrev}
         n = off = 0
         l = len(data)
         while off + s <= l:
             cur = data[off:off + s]
             off += s
             e = _unpack(indexformatv0, cur)
             # transform to revlogv1 format
             e2 = (offset_type(e[0], 0), e[1], -1, e[2], e[3],
                   nodemap.get(e[4], nullrev), nodemap.get(e[5], nullrev), e[6])
             index.append(e2)
             nodemap[e[6]] = n
             n += 1
 
         # add the magic null revision at -1
         index.append((0, 0, 0, -1, -1, -1, -1, nullid))
 
         return index, nodemap, None
 
     def packentry(self, entry, node, version, rev):
         if gettype(entry[0]):
             raise RevlogError(_("index entry flags need RevlogNG"))
         e2 = (getoffset(entry[0]), entry[1], entry[3], entry[4],
               node(entry[5]), node(entry[6]), entry[7])
         return _pack(indexformatv0, *e2)
 
 # index ng:
 #  6 bytes: offset
 #  2 bytes: flags
 #  4 bytes: compressed length
 #  4 bytes: uncompressed length
 #  4 bytes: base rev
 #  4 bytes: link rev
 #  4 bytes: parent 1 rev
 #  4 bytes: parent 2 rev
 # 32 bytes: nodeid
 indexformatng = ">Qiiiiii20s12x"
 versionformat = ">I"
 
 # corresponds to uncompressed length of indexformatng (2 gigs, 4-byte
 # signed integer)
 _maxentrysize = 0x7fffffff
 
 class revlogio(object):
     def __init__(self):
         self.size = struct.calcsize(indexformatng)
 
     def parseindex(self, data, inline):
         # call the C implementation to parse the index data
         index, cache = parsers.parse_index2(data, inline)
         return index, getattr(index, 'nodemap', None), cache
 
     def packentry(self, entry, node, version, rev):
         p = _pack(indexformatng, *entry)
         if rev == 0:
             p = _pack(versionformat, version) + p[4:]
         return p
 
 class revlog(object):
     """
     the underlying revision storage object
 
     A revlog consists of two parts, an index and the revision data.
 
     The index is a file with a fixed record size containing
     information on each revision, including its nodeid (hash), the
     nodeids of its parents, the position and offset of its data within
     the data file, and the revision it's based on. Finally, each entry
     contains a linkrev entry that can serve as a pointer to external
     data.
 
     The revision data itself is a linear collection of data chunks.
     Each chunk represents a revision and is usually represented as a
     delta against the previous chunk. To bound lookup time, runs of
     deltas are limited to about 2 times the length of the original
     version data. This makes retrieval of a version proportional to
     its size, or O(1) relative to the number of revisions.
 
     Both pieces of the revlog are written to in an append-only
     fashion, which means we never need to rewrite a file to insert or
     remove data, and can use some simple techniques to avoid the need
     for locking while reading.
 
     If checkambig, indexfile is opened with checkambig=True at
     writing, to avoid file stat ambiguity.
     """
     def __init__(self, opener, indexfile, checkambig=False):
         """
         create a revlog object
 
         opener is a function that abstracts the file opening operation
         and can be used to implement COW semantics or the like.
         """
         self.indexfile = indexfile
         self.datafile = indexfile[:-2] + ".d"
         self.opener = opener
         #  When True, indexfile is opened with checkambig=True at writing, to
         #  avoid file stat ambiguity.
         self._checkambig = checkambig
         # 3-tuple of (node, rev, text) for a raw revision.
         self._cache = None
         # Maps rev to chain base rev.
         self._chainbasecache = util.lrucachedict(100)
         # 2-tuple of (offset, data) of raw data from the revlog at an offset.
         self._chunkcache = (0, '')
         # How much data to read and cache into the raw revlog data cache.
         self._chunkcachesize = 65536
         self._maxchainlen = None
         self._aggressivemergedeltas = False
         self.index = []
         # Mapping of partial identifiers to full nodes.
         self._pcache = {}
         # Mapping of revision integer to full node.
         self._nodecache = {nullid: nullrev}
         self._nodepos = None
 
         v = REVLOG_DEFAULT_VERSION
         opts = getattr(opener, 'options', None)
         if opts is not None:
             if 'revlogv1' in opts:
                 if 'generaldelta' in opts:
                     v |= REVLOGGENERALDELTA
             else:
                 v = 0
             if 'chunkcachesize' in opts:
                 self._chunkcachesize = opts['chunkcachesize']
             if 'maxchainlen' in opts:
                 self._maxchainlen = opts['maxchainlen']
             if 'aggressivemergedeltas' in opts:
                 self._aggressivemergedeltas = opts['aggressivemergedeltas']
             self._lazydeltabase = bool(opts.get('lazydeltabase', False))
 
         if self._chunkcachesize <= 0:
             raise RevlogError(_('revlog chunk cache size %r is not greater '
                                 'than 0') % self._chunkcachesize)
         elif self._chunkcachesize & (self._chunkcachesize - 1):
             raise RevlogError(_('revlog chunk cache size %r is not a power '
                                 'of 2') % self._chunkcachesize)
 
         indexdata = ''
         self._initempty = True
         try:
             f = self.opener(self.indexfile)
             indexdata = f.read()
             f.close()
             if len(indexdata) > 0:
                 v = struct.unpack(versionformat, indexdata[:4])[0]
                 self._initempty = False
         except IOError as inst:
             if inst.errno != errno.ENOENT:
                 raise
 
         self.version = v
         self._inline = v & REVLOGNGINLINEDATA
         self._generaldelta = v & REVLOGGENERALDELTA
         flags = v & ~0xFFFF
         fmt = v & 0xFFFF
         if fmt == REVLOGV0 and flags:
             raise RevlogError(_("index %s unknown flags %#04x for format v0")
                               % (self.indexfile, flags >> 16))
         elif fmt == REVLOGNG and flags & ~REVLOGNG_FLAGS:
             raise RevlogError(_("index %s unknown flags %#04x for revlogng")
                               % (self.indexfile, flags >> 16))
         elif fmt > REVLOGNG:
             raise RevlogError(_("index %s unknown format %d")
                               % (self.indexfile, fmt))
 
         self.storedeltachains = True
 
         self._io = revlogio()
         if self.version == REVLOGV0:
             self._io = revlogoldio()
         try:
             d = self._io.parseindex(indexdata, self._inline)
         except (ValueError, IndexError):
             raise RevlogError(_("index %s is corrupted") % (self.indexfile))
         self.index, nodemap, self._chunkcache = d
         if nodemap is not None:
             self.nodemap = self._nodecache = nodemap
         if not self._chunkcache:
             self._chunkclear()
         # revnum -> (chain-length, sum-delta-length)
         self._chaininfocache = {}
 
     def tip(self):
         return self.node(len(self.index) - 2)
     def __contains__(self, rev):
         return 0 <= rev < len(self)
     def __len__(self):
         return len(self.index) - 1
     def __iter__(self):
         return iter(xrange(len(self)))
     def revs(self, start=0, stop=None):
         """iterate over all rev in this revlog (from start to stop)"""
         step = 1
         if stop is not None:
             if start > stop:
                 step = -1
             stop += step
         else:
             stop = len(self)
         return xrange(start, stop, step)
 
     @util.propertycache
     def nodemap(self):
         self.rev(self.node(0))
         return self._nodecache
 
     def hasnode(self, node):
         try:
             self.rev(node)
             return True
         except KeyError:
             return False
 
     def clearcaches(self):
         self._cache = None
         self._chainbasecache.clear()
         self._chunkcache = (0, '')
         self._pcache = {}
 
         try:
             self._nodecache.clearcaches()
         except AttributeError:
             self._nodecache = {nullid: nullrev}
             self._nodepos = None
 
     def rev(self, node):
         try:
             return self._nodecache[node]
         except TypeError:
             raise
         except RevlogError:
             # parsers.c radix tree lookup failed
             raise LookupError(node, self.indexfile, _('no node'))
         except KeyError:
             # pure python cache lookup failed
             n = self._nodecache
             i = self.index
             p = self._nodepos
             if p is None:
                 p = len(i) - 2
             for r in xrange(p, -1, -1):
                 v = i[r][7]
                 n[v] = r
                 if v == node:
                     self._nodepos = r - 1
                     return r
             raise LookupError(node, self.indexfile, _('no node'))
 
     # Accessors for index entries.
 
     # First tuple entry is 8 bytes. First 6 bytes are offset. Last 2 bytes
     # are flags.
     def start(self, rev):
         return int(self.index[rev][0] >> 16)
 
     def flags(self, rev):
         return self.index[rev][0] & 0xFFFF
 
     def length(self, rev):
         return self.index[rev][1]
 
     def rawsize(self, rev):
         """return the length of the uncompressed text for a given revision"""
         l = self.index[rev][2]
         if l >= 0:
             return l
 
         t = self.revision(self.node(rev))
         return len(t)
     size = rawsize
 
     def chainbase(self, rev):
         base = self._chainbasecache.get(rev)
         if base is not None:
             return base
 
         index = self.index
         base = index[rev][3]
         while base != rev:
             rev = base
             base = index[rev][3]
 
         self._chainbasecache[rev] = base
         return base
 
     def linkrev(self, rev):
         return self.index[rev][4]
 
     def parentrevs(self, rev):
         return self.index[rev][5:7]
 
     def node(self, rev):
         return self.index[rev][7]
 
     # Derived from index values.
 
     def end(self, rev):
         return self.start(rev) + self.length(rev)
 
     def parents(self, node):
         i = self.index
         d = i[self.rev(node)]
         return i[d[5]][7], i[d[6]][7] # map revisions to nodes inline
 
     def chainlen(self, rev):
         return self._chaininfo(rev)[0]
 
     def _chaininfo(self, rev):
         chaininfocache = self._chaininfocache
         if rev in chaininfocache:
             return chaininfocache[rev]
         index = self.index
         generaldelta = self._generaldelta
         iterrev = rev
         e = index[iterrev]
         clen = 0
         compresseddeltalen = 0
         while iterrev != e[3]:
             clen += 1
             compresseddeltalen += e[1]
             if generaldelta:
                 iterrev = e[3]
             else:
                 iterrev -= 1
             if iterrev in chaininfocache:
                 t = chaininfocache[iterrev]
                 clen += t[0]
                 compresseddeltalen += t[1]
                 break
             e = index[iterrev]
         else:
             # Add text length of base since decompressing that also takes
             # work. For cache hits the length is already included.
             compresseddeltalen += e[1]
         r = (clen, compresseddeltalen)
         chaininfocache[rev] = r
         return r
 
     def _deltachain(self, rev, stoprev=None):
         """Obtain the delta chain for a revision.
 
         ``stoprev`` specifies a revision to stop at. If not specified, we
         stop at the base of the chain.
 
         Returns a 2-tuple of (chain, stopped) where ``chain`` is a list of
         revs in ascending order and ``stopped`` is a bool indicating whether
         ``stoprev`` was hit.
         """
         chain = []
 
         # Alias to prevent attribute lookup in tight loop.
         index = self.index
         generaldelta = self._generaldelta
 
         iterrev = rev
         e = index[iterrev]
         while iterrev != e[3] and iterrev != stoprev:
             chain.append(iterrev)
             if generaldelta:
                 iterrev = e[3]
             else:
                 iterrev -= 1
             e = index[iterrev]
 
         if iterrev == stoprev:
             stopped = True
         else:
             chain.append(iterrev)
             stopped = False
 
         chain.reverse()
         return chain, stopped
 
     def ancestors(self, revs, stoprev=0, inclusive=False):
         """Generate the ancestors of 'revs' in reverse topological order.
         Does not generate revs lower than stoprev.
 
         See the documentation for ancestor.lazyancestors for more details."""
 
         return ancestor.lazyancestors(self.parentrevs, revs, stoprev=stoprev,
                                       inclusive=inclusive)
 
     def descendants(self, revs):
         """Generate the descendants of 'revs' in revision order.
 
         Yield a sequence of revision numbers starting with a child of
         some rev in revs, i.e., each revision is *not* considered a
         descendant of itself.  Results are ordered by revision number (a
         topological sort)."""
         first = min(revs)
         if first == nullrev:
             for i in self:
                 yield i
             return
 
         seen = set(revs)
         for i in self.revs(start=first + 1):
             for x in self.parentrevs(i):
                 if x != nullrev and x in seen:
                     seen.add(i)
                     yield i
                     break
 
     def findcommonmissing(self, common=None, heads=None):
         """Return a tuple of the ancestors of common and the ancestors of heads
         that are not ancestors of common. In revset terminology, we return the
         tuple:
 
           ::common, (::heads) - (::common)
 
         The list is sorted by revision number, meaning it is
         topologically sorted.
 
         'heads' and 'common' are both lists of node IDs.  If heads is
         not supplied, uses all of the revlog's heads.  If common is not
         supplied, uses nullid."""
         if common is None:
             common = [nullid]
         if heads is None:
             heads = self.heads()
 
         common = [self.rev(n) for n in common]
         heads = [self.rev(n) for n in heads]
 
         # we want the ancestors, but inclusive
         class lazyset(object):
             def __init__(self, lazyvalues):
                 self.addedvalues = set()
                 self.lazyvalues = lazyvalues
 
             def __contains__(self, value):
                 return value in self.addedvalues or value in self.lazyvalues
 
             def __iter__(self):
                 added = self.addedvalues
                 for r in added:
                     yield r
                 for r in self.lazyvalues:
                     if not r in added:
                         yield r
 
             def add(self, value):
                 self.addedvalues.add(value)
 
             def update(self, values):
                 self.addedvalues.update(values)
 
         has = lazyset(self.ancestors(common))
         has.add(nullrev)
         has.update(common)
 
         # take all ancestors from heads that aren't in has
         missing = set()
         visit = collections.deque(r for r in heads if r not in has)
         while visit:
             r = visit.popleft()
             if r in missing:
                 continue
             else:
                 missing.add(r)
                 for p in self.parentrevs(r):
                     if p not in has:
                         visit.append(p)
         missing = list(missing)
         missing.sort()
         return has, [self.node(miss) for miss in missing]
 
     def incrementalmissingrevs(self, common=None):
         """Return an object that can be used to incrementally compute the
         revision numbers of the ancestors of arbitrary sets that are not
         ancestors of common. This is an ancestor.incrementalmissingancestors
         object.
 
         'common' is a list of revision numbers. If common is not supplied, uses
         nullrev.
         """
         if common is None:
             common = [nullrev]
 
         return ancestor.incrementalmissingancestors(self.parentrevs, common)
 
     def findmissingrevs(self, common=None, heads=None):
         """Return the revision numbers of the ancestors of heads that
         are not ancestors of common.
 
         More specifically, return a list of revision numbers corresponding to
         nodes N such that every N satisfies the following constraints:
 
           1. N is an ancestor of some node in 'heads'
           2. N is not an ancestor of any node in 'common'
 
         The list is sorted by revision number, meaning it is
         topologically sorted.
 
         'heads' and 'common' are both lists of revision numbers.  If heads is
         not supplied, uses all of the revlog's heads.  If common is not
         supplied, uses nullid."""
         if common is None:
             common = [nullrev]
         if heads is None:
             heads = self.headrevs()
 
         inc = self.incrementalmissingrevs(common=common)
         return inc.missingancestors(heads)
 
     def findmissing(self, common=None, heads=None):
         """Return the ancestors of heads that are not ancestors of common.
 
         More specifically, return a list of nodes N such that every N
         satisfies the following constraints:
 
           1. N is an ancestor of some node in 'heads'
           2. N is not an ancestor of any node in 'common'
 
         The list is sorted by revision number, meaning it is
         topologically sorted.
 
         'heads' and 'common' are both lists of node IDs.  If heads is
         not supplied, uses all of the revlog's heads.  If common is not
         supplied, uses nullid."""
         if common is None:
             common = [nullid]
         if heads is None:
             heads = self.heads()
 
         common = [self.rev(n) for n in common]
         heads = [self.rev(n) for n in heads]
 
         inc = self.incrementalmissingrevs(common=common)
         return [self.node(r) for r in inc.missingancestors(heads)]
 
     def nodesbetween(self, roots=None, heads=None):
         """Return a topological path from 'roots' to 'heads'.
 
         Return a tuple (nodes, outroots, outheads) where 'nodes' is a
         topologically sorted list of all nodes N that satisfy both of
         these constraints:
 
           1. N is a descendant of some node in 'roots'
           2. N is an ancestor of some node in 'heads'
 
         Every node is considered to be both a descendant and an ancestor
         of itself, so every reachable node in 'roots' and 'heads' will be
         included in 'nodes'.
 
         'outroots' is the list of reachable nodes in 'roots', i.e., the
         subset of 'roots' that is returned in 'nodes'.  Likewise,
         'outheads' is the subset of 'heads' that is also in 'nodes'.
 
         'roots' and 'heads' are both lists of node IDs.  If 'roots' is
         unspecified, uses nullid as the only root.  If 'heads' is
         unspecified, uses list of all of the revlog's heads."""
         nonodes = ([], [], [])
         if roots is not None:
             roots = list(roots)
             if not roots:
                 return nonodes
             lowestrev = min([self.rev(n) for n in roots])
         else:
             roots = [nullid] # Everybody's a descendant of nullid
             lowestrev = nullrev
         if (lowestrev == nullrev) and (heads is None):
             # We want _all_ the nodes!
             return ([self.node(r) for r in self], [nullid], list(self.heads()))
         if heads is None:
             # All nodes are ancestors, so the latest ancestor is the last
             # node.
             highestrev = len(self) - 1
             # Set ancestors to None to signal that every node is an ancestor.
             ancestors = None
             # Set heads to an empty dictionary for later discovery of heads
             heads = {}
         else:
             heads = list(heads)
             if not heads:
                 return nonodes
             ancestors = set()
             # Turn heads into a dictionary so we can remove 'fake' heads.
             # Also, later we will be using it to filter out the heads we can't
             # find from roots.
             heads = dict.fromkeys(heads, False)
             # Start at the top and keep marking parents until we're done.
             nodestotag = set(heads)
             # Remember where the top was so we can use it as a limit later.
             highestrev = max([self.rev(n) for n in nodestotag])
             while nodestotag:
                 # grab a node to tag
                 n = nodestotag.pop()
                 # Never tag nullid
                 if n == nullid:
                     continue
                 # A node's revision number represents its place in a
                 # topologically sorted list of nodes.
                 r = self.rev(n)
                 if r >= lowestrev:
                     if n not in ancestors:
                         # If we are possibly a descendant of one of the roots
                         # and we haven't already been marked as an ancestor
                         ancestors.add(n) # Mark as ancestor
                         # Add non-nullid parents to list of nodes to tag.
                         nodestotag.update([p for p in self.parents(n) if
                                            p != nullid])
                     elif n in heads: # We've seen it before, is it a fake head?
                         # So it is, real heads should not be the ancestors of
                         # any other heads.
                         heads.pop(n)
             if not ancestors:
                 return nonodes
             # Now that we have our set of ancestors, we want to remove any
             # roots that are not ancestors.
 
             # If one of the roots was nullid, everything is included anyway.
             if lowestrev > nullrev:
                 # But, since we weren't, let's recompute the lowest rev to not
                 # include roots that aren't ancestors.
 
                 # Filter out roots that aren't ancestors of heads
                 roots = [root for root in roots if root in ancestors]
                 # Recompute the lowest revision
                 if roots:
                     lowestrev = min([self.rev(root) for root in roots])
                 else:
                     # No more roots?  Return empty list
                     return nonodes
             else:
                 # We are descending from nullid, and don't need to care about
                 # any other roots.
                 lowestrev = nullrev
                 roots = [nullid]
         # Transform our roots list into a set.
         descendants = set(roots)
         # Also, keep the original roots so we can filter out roots that aren't
         # 'real' roots (i.e. are descended from other roots).
         roots = descendants.copy()
         # Our topologically sorted list of output nodes.
         orderedout = []
         # Don't start at nullid since we don't want nullid in our output list,
         # and if nullid shows up in descendants, empty parents will look like
         # they're descendants.
         for r in self.revs(start=max(lowestrev, 0), stop=highestrev + 1):
             n = self.node(r)
             isdescendant = False
             if lowestrev == nullrev:  # Everybody is a descendant of nullid
                 isdescendant = True
             elif n in descendants:
                 # n is already a descendant
                 isdescendant = True
                 # This check only needs to be done here because all the roots
                 # will start being marked is descendants before the loop.
                 if n in roots:
                     # If n was a root, check if it's a 'real' root.
                     p = tuple(self.parents(n))
                     # If any of its parents are descendants, it's not a root.
                     if (p[0] in descendants) or (p[1] in descendants):
                         roots.remove(n)
             else:
                 p = tuple(self.parents(n))
                 # A node is a descendant if either of its parents are
                 # descendants.  (We seeded the dependents list with the roots
                 # up there, remember?)
                 if (p[0] in descendants) or (p[1] in descendants):
                     descendants.add(n)
                     isdescendant = True
             if isdescendant and ((ancestors is None) or (n in ancestors)):
                 # Only include nodes that are both descendants and ancestors.
                 orderedout.append(n)
                 if (ancestors is not None) and (n in heads):
                     # We're trying to figure out which heads are reachable
                     # from roots.
                     # Mark this head as having been reached
                     heads[n] = True
                 elif ancestors is None:
                     # Otherwise, we're trying to discover the heads.
                     # Assume this is a head because if it isn't, the next step
                     # will eventually remove it.
                     heads[n] = True
                     # But, obviously its parents aren't.
                     for p in self.parents(n):
                         heads.pop(p, None)
         heads = [head for head, flag in heads.iteritems() if flag]
         roots = list(roots)
         assert orderedout
         assert roots
         assert heads
         return (orderedout, roots, heads)
 
     def headrevs(self):
         try:
             return self.index.headrevs()
         except AttributeError:
             return self._headrevs()
 
     def computephases(self, roots):
         return self.index.computephasesmapsets(roots)
 
     def _headrevs(self):
         count = len(self)
         if not count:
             return [nullrev]
         # we won't iter over filtered rev so nobody is a head at start
         ishead = [0] * (count + 1)
         index = self.index
         for r in self:
             ishead[r] = 1  # I may be an head
             e = index[r]
             ishead[e[5]] = ishead[e[6]] = 0  # my parent are not
         return [r for r, val in enumerate(ishead) if val]
 
     def heads(self, start=None, stop=None):
         """return the list of all nodes that have no children
 
         if start is specified, only heads that are descendants of
         start will be returned
         if stop is specified, it will consider all the revs from stop
         as if they had no children
         """
         if start is None and stop is None:
             if not len(self):
                 return [nullid]
             return [self.node(r) for r in self.headrevs()]
 
         if start is None:
             start = nullid
         if stop is None:
             stop = []
         stoprevs = set([self.rev(n) for n in stop])
         startrev = self.rev(start)
         reachable = set((startrev,))
         heads = set((startrev,))
 
         parentrevs = self.parentrevs
         for r in self.revs(start=startrev + 1):
             for p in parentrevs(r):
                 if p in reachable:
                     if r not in stoprevs:
                         reachable.add(r)
                     heads.add(r)
                 if p in heads and p not in stoprevs:
                     heads.remove(p)
 
         return [self.node(r) for r in heads]
 
     def children(self, node):
         """find the children of a given node"""
         c = []
         p = self.rev(node)
         for r in self.revs(start=p + 1):
             prevs = [pr for pr in self.parentrevs(r) if pr != nullrev]
             if prevs:
                 for pr in prevs:
                     if pr == p:
                         c.append(self.node(r))
             elif p == nullrev:
                 c.append(self.node(r))
         return c
 
     def descendant(self, start, end):
         if start == nullrev:
             return True
         for i in self.descendants([start]):
             if i == end:
                 return True
             elif i > end:
                 break
         return False
 
     def commonancestorsheads(self, a, b):
         """calculate all the heads of the common ancestors of nodes a and b"""
         a, b = self.rev(a), self.rev(b)
         try:
             ancs = self.index.commonancestorsheads(a, b)
         except (AttributeError, OverflowError): # C implementation failed
             ancs = ancestor.commonancestorsheads(self.parentrevs, a, b)
         return map(self.node, ancs)
 
     def isancestor(self, a, b):
         """return True if node a is an ancestor of node b
 
         The implementation of this is trivial but the use of
         commonancestorsheads is not."""
         return a in self.commonancestorsheads(a, b)
 
     def ancestor(self, a, b):
         """calculate the "best" common ancestor of nodes a and b"""
 
         a, b = self.rev(a), self.rev(b)
         try:
             ancs = self.index.ancestors(a, b)
         except (AttributeError, OverflowError):
             ancs = ancestor.ancestors(self.parentrevs, a, b)
         if ancs:
             # choose a consistent winner when there's a tie
             return min(map(self.node, ancs))
         return nullid
 
     def _match(self, id):
         if isinstance(id, int):
             # rev
             return self.node(id)
         if len(id) == 20:
             # possibly a binary node
             # odds of a binary node being all hex in ASCII are 1 in 10**25
             try:
                 node = id
                 self.rev(node) # quick search the index
                 return node
             except LookupError:
                 pass # may be partial hex id
         try:
             # str(rev)
             rev = int(id)
             if str(rev) != id:
                 raise ValueError
             if rev < 0:
                 rev = len(self) + rev
             if rev < 0 or rev >= len(self):
                 raise ValueError
             return self.node(rev)
         except (ValueError, OverflowError):
             pass
         if len(id) == 40:
             try:
                 # a full hex nodeid?
                 node = bin(id)
                 self.rev(node)
                 return node
             except (TypeError, LookupError):
                 pass
 
     def _partialmatch(self, id):
         try:
             partial = self.index.partialmatch(id)
             if partial and self.hasnode(partial):
                 return partial
             return None
         except RevlogError:
             # parsers.c radix tree lookup gave multiple matches
             # fast path: for unfiltered changelog, radix tree is accurate
             if not getattr(self, 'filteredrevs', None):
                 raise LookupError(id, self.indexfile,
                                   _('ambiguous identifier'))
             # fall through to slow path that filters hidden revisions
         except (AttributeError, ValueError):
             # we are pure python, or key was too short to search radix tree
             pass
 
         if id in self._pcache:
             return self._pcache[id]
 
         if len(id) < 40:
             try:
                 # hex(node)[:...]
                 l = len(id) // 2  # grab an even number of digits
                 prefix = bin(id[:l * 2])
                 nl = [e[7] for e in self.index if e[7].startswith(prefix)]
                 nl = [n for n in nl if hex(n).startswith(id) and
                       self.hasnode(n)]
                 if len(nl) > 0:
                     if len(nl) == 1:
                         self._pcache[id] = nl[0]
                         return nl[0]
                     raise LookupError(id, self.indexfile,
                                       _('ambiguous identifier'))
                 return None
             except TypeError:
                 pass
 
     def lookup(self, id):
         """locate a node based on:
             - revision number or str(revision number)
             - nodeid or subset of hex nodeid
         """
         n = self._match(id)
         if n is not None:
             return n
         n = self._partialmatch(id)
         if n:
             return n
 
         raise LookupError(id, self.indexfile, _('no match found'))
 
     def cmp(self, node, text):
         """compare text with a given file revision
 
         returns True if text is different than what is stored.
         """
         p1, p2 = self.parents(node)
         return hash(text, p1, p2) != node
 
     def _addchunk(self, offset, data):
         """Add a segment to the revlog cache.
 
         Accepts an absolute offset and the data that is at that location.
         """
         o, d = self._chunkcache
         # try to add to existing cache
         if o + len(d) == offset and len(d) + len(data) < _chunksize:
             self._chunkcache = o, d + data
         else:
             self._chunkcache = offset, data
 
     def _loadchunk(self, offset, length, df=None):
         """Load a segment of raw data from the revlog.
 
         Accepts an absolute offset, length to read, and an optional existing
         file handle to read from.
 
         If an existing file handle is passed, it will be seeked and the
         original seek position will NOT be restored.
 
         Returns a str or buffer of raw byte data.
         """
         if df is not None:
             closehandle = False
         else:
             if self._inline:
                 df = self.opener(self.indexfile)
             else:
                 df = self.opener(self.datafile)
             closehandle = True
 
         # Cache data both forward and backward around the requested
         # data, in a fixed size window. This helps speed up operations
         # involving reading the revlog backwards.
         cachesize = self._chunkcachesize
         realoffset = offset & ~(cachesize - 1)
         reallength = (((offset + length + cachesize) & ~(cachesize - 1))
                       - realoffset)
         df.seek(realoffset)
         d = df.read(reallength)
         if closehandle:
             df.close()
         self._addchunk(realoffset, d)
         if offset != realoffset or reallength != length:
             return util.buffer(d, offset - realoffset, length)
         return d
 
     def _getchunk(self, offset, length, df=None):
         """Obtain a segment of raw data from the revlog.
 
         Accepts an absolute offset, length of bytes to obtain, and an
         optional file handle to the already-opened revlog. If the file
         handle is used, it's original seek position will not be preserved.
 
         Requests for data may be returned from a cache.
 
         Returns a str or a buffer instance of raw byte data.
         """
         o, d = self._chunkcache
         l = len(d)
 
         # is it in the cache?
         cachestart = offset - o
         cacheend = cachestart + length
         if cachestart >= 0 and cacheend <= l:
             if cachestart == 0 and cacheend == l:
                 return d # avoid a copy
             return util.buffer(d, cachestart, cacheend - cachestart)
 
         return self._loadchunk(offset, length, df=df)
 
     def _chunkraw(self, startrev, endrev, df=None):
         """Obtain a segment of raw data corresponding to a range of revisions.
 
         Accepts the start and end revisions and an optional already-open
         file handle to be used for reading. If the file handle is read, its
         seek position will not be preserved.
 
         Requests for data may be satisfied by a cache.
 
         Returns a 2-tuple of (offset, data) for the requested range of
         revisions. Offset is the integer offset from the beginning of the
         revlog and data is a str or buffer of the raw byte data.
 
         Callers will need to call ``self.start(rev)`` and ``self.length(rev)``
         to determine where each revision's data begins and ends.
         """
         # Inlined self.start(startrev) & self.end(endrev) for perf reasons
         # (functions are expensive).
         index = self.index
         istart = index[startrev]
         start = int(istart[0] >> 16)
         if startrev == endrev:
             end = start + istart[1]
         else:
             iend = index[endrev]
             end = int(iend[0] >> 16) + iend[1]
 
         if self._inline:
             start += (startrev + 1) * self._io.size
             end += (endrev + 1) * self._io.size
         length = end - start
 
         return start, self._getchunk(start, length, df=df)
 
     def _chunk(self, rev, df=None):
         """Obtain a single decompressed chunk for a revision.
 
         Accepts an integer revision and an optional already-open file handle
         to be used for reading. If used, the seek position of the file will not
         be preserved.
 
         Returns a str holding uncompressed data for the requested revision.
         """
         return decompress(self._chunkraw(rev, rev, df=df)[1])
 
     def _chunks(self, revs, df=None):
         """Obtain decompressed chunks for the specified revisions.
 
         Accepts an iterable of numeric revisions that are assumed to be in
         ascending order. Also accepts an optional already-open file handle
         to be used for reading. If used, the seek position of the file will
         not be preserved.
 
         This function is similar to calling ``self._chunk()`` multiple times,
         but is faster.
 
         Returns a list with decompressed data for each requested revision.
         """
         if not revs:
             return []
         start = self.start
         length = self.length
         inline = self._inline
         iosize = self._io.size
         buffer = util.buffer
 
         l = []
         ladd = l.append
 
         try:
             offset, data = self._chunkraw(revs[0], revs[-1], df=df)
         except OverflowError:
             # issue4215 - we can't cache a run of chunks greater than
             # 2G on Windows
             return [self._chunk(rev, df=df) for rev in revs]
 
         for rev in revs:
             chunkstart = start(rev)
             if inline:
                 chunkstart += (rev + 1) * iosize
             chunklength = length(rev)
             ladd(decompress(buffer(data, chunkstart - offset, chunklength)))
 
         return l
 
     def _chunkclear(self):
         """Clear the raw chunk cache."""
         self._chunkcache = (0, '')
 
     def deltaparent(self, rev):
         """return deltaparent of the given revision"""
         base = self.index[rev][3]
         if base == rev:
             return nullrev
         elif self._generaldelta:
             return base
         else:
             return rev - 1
 
     def revdiff(self, rev1, rev2):
         """return or calculate a delta between two revisions"""
         if rev1 != nullrev and self.deltaparent(rev2) == rev1:
             return str(self._chunk(rev2))
 
         return mdiff.textdiff(self.revision(rev1),
                               self.revision(rev2))
 
     def revision(self, nodeorrev, _df=None, raw=False):
         """return an uncompressed revision of a given node or revision
         number.
 
         _df - an existing file handle to read from. (internal-only)
         raw - an optional argument specifying if the revision data is to be
         treated as raw data when applying flag transforms. 'raw' should be set
         to True when generating changegroups or in debug commands.
         """
         if isinstance(nodeorrev, int):
             rev = nodeorrev
             node = self.node(rev)
         else:
             node = nodeorrev
             rev = None
 
         cachedrev = None
         if node == nullid:
             return ""
         if self._cache:
             if self._cache[0] == node:
                 return self._cache[2]
             cachedrev = self._cache[1]
 
         # look up what we need to read
         text = None
         if rev is None:
             rev = self.rev(node)
 
-        # check rev flags
-        if self.flags(rev) & ~REVIDX_KNOWN_FLAGS:
-            raise RevlogError(_('incompatible revision flag %x') %
-                              (self.flags(rev) & ~REVIDX_KNOWN_FLAGS))
-
         chain, stopped = self._deltachain(rev, stoprev=cachedrev)
         if stopped:
             text = self._cache[2]
 
         # drop cache to save memory
         self._cache = None
 
         bins = self._chunks(chain, df=_df)
         if text is None:
             text = str(bins[0])
             bins = bins[1:]
 
         text = mdiff.patches(text, bins)
-        self.checkhash(text, node, rev=rev)
+
+        text, validatehash = self._processflags(text, self.flags(rev), 'read',
+                                                raw=raw)
+        if validatehash:
+            self.checkhash(text, node, rev=rev)
+
         self._cache = (node, rev, text)
         return text
 
     def hash(self, text, p1, p2):
         """Compute a node hash.
 
         Available as a function so that subclasses can replace the hash
         as needed.
         """
         return hash(text, p1, p2)
 
+    def _processflags(self, text, flags, operation, raw=False):
+        """Inspect revision data flags and applies transforms defined by
+        registered flag processors.
+
+        ``text`` - the revision data to process
+        ``flags`` - the revision flags
+        ``operation`` - the operation being performed (read or write)
+        ``raw`` - an optional argument describing if the raw transform should be
+        applied.
+
+        This method processes the flags in the order (or reverse order if
+        ``operation`` is 'write') defined by REVIDX_FLAGS_ORDER, applying the
+        flag processors registered for present flags. The order of flags defined
+        in REVIDX_FLAGS_ORDER needs to be stable to allow non-commutativity.
+
+        Returns a 2-tuple of ``(text, validatehash)`` where ``text`` is the
+        processed text and ``validatehash`` is a bool indicating whether the
+        returned text should be checked for hash integrity.
+
+        Note: If the ``raw`` argument is set, it has precedence over the
+        operation and will only update the value of ``validatehash``.
+        """
+        if not operation in ('read', 'write'):
+            raise ProgrammingError(_("invalid '%s' operation ") % (operation))
+        # Check all flags are known.
+        if flags & ~REVIDX_KNOWN_FLAGS:
+            raise RevlogError(_("incompatible revision flag '%#x'") %
+                              (flags & ~REVIDX_KNOWN_FLAGS))
+        validatehash = True
+        # Depending on the operation (read or write), the order might be
+        # reversed due to non-commutative transforms.
+        orderedflags = REVIDX_FLAGS_ORDER
+        if operation == 'write':
+            orderedflags = reversed(orderedflags)
+
+        for flag in orderedflags:
+            # If a flagprocessor has been registered for a known flag, apply the
+            # related operation transform and update result tuple.
+            if flag & flags:
+                vhash = True
+
+                if flag not in _flagprocessors:
+                    message = _("missing processor for flag '%#x'") % (flag)
+                    raise RevlogError(message)
+
+                processor = _flagprocessors[flag]
+                if processor is not None:
+                    readtransform, writetransform, rawtransform = processor
+
+                    if raw:
+                        vhash = rawtransform(self, text)
+                    elif operation == 'read':
+                        text, vhash = readtransform(self, text)
+                    else: # write operation
+                        text, vhash = writetransform(self, text)
+                validatehash = validatehash and vhash
+
+        return text, validatehash
+
     def checkhash(self, text, node, p1=None, p2=None, rev=None):
         """Check node hash integrity.
 
         Available as a function so that subclasses can extend hash mismatch
         behaviors as needed.
         """
         if p1 is None and p2 is None:
             p1, p2 = self.parents(node)
         if node != self.hash(text, p1, p2):
             revornode = rev
             if revornode is None:
                 revornode = templatefilters.short(hex(node))
             raise RevlogError(_("integrity check failed on %s:%s")
                 % (self.indexfile, revornode))
 
     def checkinlinesize(self, tr, fp=None):
         """Check if the revlog is too big for inline and convert if so.
 
         This should be called after revisions are added to the revlog. If the
         revlog has grown too large to be an inline revlog, it will convert it
         to use multiple index and data files.
         """
         if not self._inline or (self.start(-2) + self.length(-2)) < _maxinline:
             return
 
         trinfo = tr.find(self.indexfile)
         if trinfo is None:
             raise RevlogError(_("%s not found in the transaction")
                               % self.indexfile)
 
         trindex = trinfo[2]
         if trindex is not None:
             dataoff = self.start(trindex)
         else:
             # revlog was stripped at start of transaction, use all leftover data
             trindex = len(self) - 1
             dataoff = self.end(-2)
 
         tr.add(self.datafile, dataoff)
 
         if fp:
             fp.flush()
             fp.close()
 
         df = self.opener(self.datafile, 'w')
         try:
             for r in self:
                 df.write(self._chunkraw(r, r)[1])
         finally:
             df.close()
 
         fp = self.opener(self.indexfile, 'w', atomictemp=True,
                          checkambig=self._checkambig)
         self.version &= ~(REVLOGNGINLINEDATA)
         self._inline = False
         for i in self:
             e = self._io.packentry(self.index[i], self.node, self.version, i)
             fp.write(e)
 
         # if we don't call close, the temp file will never replace the
         # real index
         fp.close()
 
         tr.replace(self.indexfile, trindex * self._io.size)
         self._chunkclear()
 
     def addrevision(self, text, transaction, link, p1, p2, cachedelta=None,
                     node=None, flags=REVIDX_DEFAULT_FLAGS):
         """add a revision to the log
 
         text - the revision data to add
         transaction - the transaction object used for rollback
         link - the linkrev data to add
         p1, p2 - the parent nodeids of the revision
         cachedelta - an optional precomputed delta
         node - nodeid of revision; typically node is not specified, and it is
             computed by default as hash(text, p1, p2), however subclasses might
             use different hashing method (and override checkhash() in such case)
         flags - the known flags to set on the revision
         """
         if link == nullrev:
             raise RevlogError(_("attempted to add linkrev -1 to %s")
                               % self.indexfile)
 
+        if flags:
+            node = node or self.hash(text, p1, p2)
+
+        newtext, validatehash = self._processflags(text, flags, 'write')
+
+        # If the flag processor modifies the revision data, ignore any provided
+        # cachedelta.
+        if newtext != text:
+            cachedelta = None
+        text = newtext
+
         if len(text) > _maxentrysize:
             raise RevlogError(
                 _("%s: size of %d bytes exceeds maximum revlog storage of 2GiB")
                 % (self.indexfile, len(text)))
 
         node = node or self.hash(text, p1, p2)
         if node in self.nodemap:
             return node
 
+        if validatehash:
+            self.checkhash(text, node, p1=p1, p2=p2)
+
         dfh = None
         if not self._inline:
             dfh = self.opener(self.datafile, "a+")
         ifh = self.opener(self.indexfile, "a+", checkambig=self._checkambig)
         try:
             return self._addrevision(node, text, transaction, link, p1, p2,
                                      flags, cachedelta, ifh, dfh)
         finally:
             if dfh:
                 dfh.close()
             ifh.close()
 
     def compress(self, text):
         """ generate a possibly-compressed representation of text """
         if not text:
             return ("", text)
         l = len(text)
         bin = None
         if l < 44:
             pass
         elif l > 1000000:
             # zlib makes an internal copy, thus doubling memory usage for
             # large files, so lets do this in pieces
             z = zlib.compressobj()
             p = []
             pos = 0
             while pos < l:
                 pos2 = pos + 2**20
                 p.append(z.compress(text[pos:pos2]))
                 pos = pos2
             p.append(z.flush())
             if sum(map(len, p)) < l:
                 bin = "".join(p)
         else:
             bin = _compress(text)
         if bin is None or len(bin) > l:
             if text[0] == '\0':
                 return ("", text)
             return ('u', text)
         return ("", bin)
 
     def _isgooddelta(self, d, textlen):
         """Returns True if the given delta is good. Good means that it is within
         the disk span, disk size, and chain length bounds that we know to be
         performant."""
         if d is None:
             return False
 
         # - 'dist' is the distance from the base revision -- bounding it limits
         #   the amount of I/O we need to do.
         # - 'compresseddeltalen' is the sum of the total size of deltas we need
         #   to apply -- bounding it limits the amount of CPU we consume.
         dist, l, data, base, chainbase, chainlen, compresseddeltalen = d
         if (dist > textlen * 4 or l > textlen or
             compresseddeltalen > textlen * 2 or
             (self._maxchainlen and chainlen > self._maxchainlen)):
             return False
 
         return True
 
     def _addrevision(self, node, text, transaction, link, p1, p2, flags,
                      cachedelta, ifh, dfh, alwayscache=False, raw=False):
         """internal function to add revisions to the log
 
         see addrevision for argument descriptions.
         invariants:
         - text is optional (can be None); if not set, cachedelta must be set.
           if both are set, they must correspond to each other.
         - raw is optional; if set to True, it indicates the revision data is to
           be treated by _processflags() as raw. It is usually set by changegroup
           generation and debug commands.
         """
         btext = [text]
         def buildtext():
             if btext[0] is not None:
                 return btext[0]
             baserev = cachedelta[0]
             delta = cachedelta[1]
             # special case deltas which replace entire base; no need to decode
             # base revision. this neatly avoids censored bases, which throw when
             # they're decoded.
             hlen = struct.calcsize(">lll")
             if delta[:hlen] == mdiff.replacediffheader(self.rawsize(baserev),
                                                        len(delta) - hlen):
                 btext[0] = delta[hlen:]
             else:
                 if self._inline:
                     fh = ifh
                 else:
                     fh = dfh
                 basetext = self.revision(self.node(baserev), _df=fh, raw=raw)
                 btext[0] = mdiff.patch(basetext, delta)
 
             try:
-                self.checkhash(btext[0], node, p1=p1, p2=p2)
+                res = self._processflags(btext[0], flags, 'read', raw=raw)
+                btext[0], validatehash = res
+                if validatehash:
+                    self.checkhash(btext[0], node, p1=p1, p2=p2)
                 if flags & REVIDX_ISCENSORED:
                     raise RevlogError(_('node %s is not censored') % node)
             except CensoredNodeError:
                 # must pass the censored index flag to add censored revisions
                 if not flags & REVIDX_ISCENSORED:
                     raise
             return btext[0]
 
         def builddelta(rev):
             # can we use the cached delta?
             if cachedelta and cachedelta[0] == rev:
                 delta = cachedelta[1]
             else:
                 t = buildtext()
                 if self.iscensored(rev):
                     # deltas based on a censored revision must replace the
                     # full content in one patch, so delta works everywhere
                     header = mdiff.replacediffheader(self.rawsize(rev), len(t))
                     delta = header + t
                 else:
                     if self._inline:
                         fh = ifh
                     else:
                         fh = dfh
                     ptext = self.revision(self.node(rev), _df=fh)
                     delta = mdiff.textdiff(ptext, t)
             header, data = self.compress(delta)
             deltalen = len(header) + len(data)
             chainbase = self.chainbase(rev)
             dist = deltalen + offset - self.start(chainbase)
             if self._generaldelta:
                 base = rev
             else:
                 base = chainbase
             chainlen, compresseddeltalen = self._chaininfo(rev)
             chainlen += 1
             compresseddeltalen += deltalen
             return (dist, deltalen, (header, data), base,
                     chainbase, chainlen, compresseddeltalen)
 
         curr = len(self)
         prev = curr - 1
         offset = self.end(prev)
         delta = None
         p1r, p2r = self.rev(p1), self.rev(p2)
 
         # full versions are inserted when the needed deltas
         # become comparable to the uncompressed text
         if text is None:
             textlen = mdiff.patchedsize(self.rawsize(cachedelta[0]),
                                         cachedelta[1])
         else:
             textlen = len(text)
 
         # should we try to build a delta?
         if prev != nullrev and self.storedeltachains:
             tested = set()
             # This condition is true most of the time when processing
             # changegroup data into a generaldelta repo. The only time it
             # isn't true is if this is the first revision in a delta chain
             # or if ``format.generaldelta=true`` disabled ``lazydeltabase``.
             if cachedelta and self._generaldelta and self._lazydeltabase:
                 # Assume what we received from the server is a good choice
                 # build delta will reuse the cache
                 candidatedelta = builddelta(cachedelta[0])
                 tested.add(cachedelta[0])
                 if self._isgooddelta(candidatedelta, textlen):
                     delta = candidatedelta
             if delta is None and self._generaldelta:
                 # exclude already lazy tested base if any
                 parents = [p for p in (p1r, p2r)
                            if p != nullrev and p not in tested]
                 if parents and not self._aggressivemergedeltas:
                     # Pick whichever parent is closer to us (to minimize the
                     # chance of having to build a fulltext).
                     parents = [max(parents)]
                 tested.update(parents)
                 pdeltas = []
                 for p in parents:
                     pd = builddelta(p)
                     if self._isgooddelta(pd, textlen):
                         pdeltas.append(pd)
                 if pdeltas:
                     delta = min(pdeltas, key=lambda x: x[1])
             if delta is None and prev not in tested:
                 # other approach failed try against prev to hopefully save us a
                 # fulltext.
                 candidatedelta = builddelta(prev)
                 if self._isgooddelta(candidatedelta, textlen):
                     delta = candidatedelta
         if delta is not None:
             dist, l, data, base, chainbase, chainlen, compresseddeltalen = delta
         else:
             text = buildtext()
             data = self.compress(text)
             l = len(data[1]) + len(data[0])
             base = chainbase = curr
 
         e = (offset_type(offset, flags), l, textlen,
              base, link, p1r, p2r, node)
         self.index.insert(-1, e)
         self.nodemap[node] = curr
 
         entry = self._io.packentry(e, self.node, self.version, curr)
         self._writeentry(transaction, ifh, dfh, entry, data, link, offset)
 
         if alwayscache and text is None:
             text = buildtext()
 
         if type(text) == str: # only accept immutable objects
             self._cache = (node, curr, text)
         self._chainbasecache[curr] = chainbase
         return node
 
     def _writeentry(self, transaction, ifh, dfh, entry, data, link, offset):
         # Files opened in a+ mode have inconsistent behavior on various
         # platforms. Windows requires that a file positioning call be made
         # when the file handle transitions between reads and writes. See
         # 3686fa2b8eee and the mixedfilemodewrapper in windows.py. On other
         # platforms, Python or the platform itself can be buggy. Some versions
         # of Solaris have been observed to not append at the end of the file
         # if the file was seeked to before the end. See issue4943 for more.
         #
         # We work around this issue by inserting a seek() before writing.
         # Note: This is likely not necessary on Python 3.
         ifh.seek(0, os.SEEK_END)
         if dfh:
             dfh.seek(0, os.SEEK_END)
 
         curr = len(self) - 1
         if not self._inline:
             transaction.add(self.datafile, offset)
             transaction.add(self.indexfile, curr * len(entry))
             if data[0]:
                 dfh.write(data[0])
             dfh.write(data[1])
             ifh.write(entry)
         else:
             offset += curr * self._io.size
             transaction.add(self.indexfile, offset, curr)
             ifh.write(entry)
             ifh.write(data[0])
             ifh.write(data[1])
             self.checkinlinesize(transaction, ifh)
 
     def addgroup(self, cg, linkmapper, transaction, addrevisioncb=None):
         """
         add a delta group
 
         given a set of deltas, add them to the revision log. the
         first delta is against its parent, which should be in our
         log, the rest are against the previous delta.
 
         If ``addrevisioncb`` is defined, it will be called with arguments of
         this revlog and the node that was added.
         """
 
         # track the base of the current delta log
         content = []
         node = None
 
         r = len(self)
         end = 0
         if r:
             end = self.end(r - 1)
         ifh = self.opener(self.indexfile, "a+", checkambig=self._checkambig)
         isize = r * self._io.size
         if self._inline:
             transaction.add(self.indexfile, end + isize, r)
             dfh = None
         else:
             transaction.add(self.indexfile, isize, r)
             transaction.add(self.datafile, end)
             dfh = self.opener(self.datafile, "a+")
         def flush():
             if dfh:
                 dfh.flush()
             ifh.flush()
         try:
             # loop through our set of deltas
             chain = None
             for chunkdata in iter(lambda: cg.deltachunk(chain), {}):
                 node = chunkdata['node']
                 p1 = chunkdata['p1']
                 p2 = chunkdata['p2']
                 cs = chunkdata['cs']
                 deltabase = chunkdata['deltabase']
                 delta = chunkdata['delta']
                 flags = chunkdata['flags'] or REVIDX_DEFAULT_FLAGS
 
                 content.append(node)
 
                 link = linkmapper(cs)
                 if node in self.nodemap:
                     # this can happen if two branches make the same change
                     chain = node
                     continue
 
                 for p in (p1, p2):
                     if p not in self.nodemap:
                         raise LookupError(p, self.indexfile,
                                           _('unknown parent'))
 
                 if deltabase not in self.nodemap:
                     raise LookupError(deltabase, self.indexfile,
                                       _('unknown delta base'))
 
                 baserev = self.rev(deltabase)
 
                 if baserev != nullrev and self.iscensored(baserev):
                     # if base is censored, delta must be full replacement in a
                     # single patch operation
                     hlen = struct.calcsize(">lll")
                     oldlen = self.rawsize(baserev)
                     newlen = len(delta) - hlen
                     if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen):
                         raise error.CensoredBaseError(self.indexfile,
                                                       self.node(baserev))
 
                 if not flags and self._peek_iscensored(baserev, delta, flush):
                     flags |= REVIDX_ISCENSORED
 
                 # We assume consumers of addrevisioncb will want to retrieve
                 # the added revision, which will require a call to
                 # revision(). revision() will fast path if there is a cache
                 # hit. So, we tell _addrevision() to always cache in this case.
                 # We're only using addgroup() in the context of changegroup
                 # generation so the revision data can always be handled as raw
                 # by the flagprocessor.
                 chain = self._addrevision(node, None, transaction, link,
                                           p1, p2, flags, (baserev, delta),
                                           ifh, dfh,
                                           alwayscache=bool(addrevisioncb),
                                           raw=True)
 
                 if addrevisioncb:
                     addrevisioncb(self, chain)
 
                 if not dfh and not self._inline:
                     # addrevision switched from inline to conventional
                     # reopen the index
                     ifh.close()
                     dfh = self.opener(self.datafile, "a+")
                     ifh = self.opener(self.indexfile, "a+",
                                       checkambig=self._checkambig)
         finally:
             if dfh:
                 dfh.close()
             ifh.close()
 
         return content
 
     def iscensored(self, rev):
         """Check if a file revision is censored."""
         return False
 
     def _peek_iscensored(self, baserev, delta, flush):
         """Quickly check if a delta produces a censored revision."""
         return False
 
     def getstrippoint(self, minlink):
         """find the minimum rev that must be stripped to strip the linkrev
 
         Returns a tuple containing the minimum rev and a set of all revs that
         have linkrevs that will be broken by this strip.
         """
         brokenrevs = set()
         strippoint = len(self)
 
         heads = {}
         futurelargelinkrevs = set()
         for head in self.headrevs():
             headlinkrev = self.linkrev(head)
             heads[head] = headlinkrev
             if headlinkrev >= minlink:
                 futurelargelinkrevs.add(headlinkrev)
 
         # This algorithm involves walking down the rev graph, starting at the
         # heads. Since the revs are topologically sorted according to linkrev,
         # once all head linkrevs are below the minlink, we know there are
         # no more revs that could have a linkrev greater than minlink.
         # So we can stop walking.
         while futurelargelinkrevs:
             strippoint -= 1
             linkrev = heads.pop(strippoint)
 
             if linkrev < minlink:
                 brokenrevs.add(strippoint)
             else:
                 futurelargelinkrevs.remove(linkrev)
 
             for p in self.parentrevs(strippoint):
                 if p != nullrev:
                     plinkrev = self.linkrev(p)
                     heads[p] = plinkrev
                     if plinkrev >= minlink:
                         futurelargelinkrevs.add(plinkrev)
 
         return strippoint, brokenrevs
 
     def strip(self, minlink, transaction):
         """truncate the revlog on the first revision with a linkrev >= minlink
 
         This function is called when we're stripping revision minlink and
         its descendants from the repository.
 
         We have to remove all revisions with linkrev >= minlink, because
         the equivalent changelog revisions will be renumbered after the
         strip.
 
         So we truncate the revlog on the first of these revisions, and
         trust that the caller has saved the revisions that shouldn't be
         removed and that it'll re-add them after this truncation.
         """
         if len(self) == 0:
             return
 
         rev, _ = self.getstrippoint(minlink)
         if rev == len(self):
             return
 
         # first truncate the files on disk
         end = self.start(rev)
         if not self._inline:
             transaction.add(self.datafile, end)
             end = rev * self._io.size
         else:
             end += rev * self._io.size
 
         transaction.add(self.indexfile, end)
 
         # then reset internal state in memory to forget those revisions
         self._cache = None
         self._chaininfocache = {}
         self._chunkclear()
         for x in xrange(rev, len(self)):
             del self.nodemap[self.node(x)]
 
         del self.index[rev:-1]
 
     def checksize(self):
         expected = 0
         if len(self):
             expected = max(0, self.end(len(self) - 1))
 
         try:
             f = self.opener(self.datafile)
             f.seek(0, 2)
             actual = f.tell()
             f.close()
             dd = actual - expected
         except IOError as inst:
             if inst.errno != errno.ENOENT:
                 raise
             dd = 0
 
         try:
             f = self.opener(self.indexfile)
             f.seek(0, 2)
             actual = f.tell()
             f.close()
             s = self._io.size
             i = max(0, actual // s)
             di = actual - (i * s)
             if self._inline:
                 databytes = 0
                 for r in self:
                     databytes += max(0, self.length(r))
                 dd = 0
                 di = actual - len(self) * s - databytes
         except IOError as inst:
             if inst.errno != errno.ENOENT:
                 raise
             di = 0
 
         return (dd, di)
 
     def files(self):
         res = [self.indexfile]
         if not self._inline:
             res.append(self.datafile)
         return res

mercurial/util.py

@@ -1,3250 +1,3256 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 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 zlib
 
 from . import (
     encoding,
     error,
     i18n,
     osutil,
     parsers,
     pycompat,
 )
 
 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
 urlparse = pycompat.urlparse
 urlreq = pycompat.urlreq
 xmlrpclib = pycompat.xmlrpclib
 
 if pycompat.osname == 'nt':
     from . import windows as platform
     stdout = platform.winstdout(pycompat.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
 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
 unlinkpath = platform.unlinkpath
 username = platform.username
 
 # 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
+
 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):
             return sliceable[offset:]
     else:
         def buffer(sliceable, offset=0):
             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(dict):
     '''a simple sorted dictionary'''
     def __init__(self, data=None):
         self._list = []
         if data:
             self.update(data)
     def copy(self):
         return sortdict(self)
     def __setitem__(self, key, val):
         if key in self:
             self._list.remove(key)
         self._list.append(key)
         dict.__setitem__(self, key, val)
     def __iter__(self):
         return self._list.__iter__()
     def update(self, src):
         if isinstance(src, dict):
             src = src.iteritems()
         for k, v in src:
             self[k] = v
     def clear(self):
         dict.clear(self)
         self._list = []
     def items(self):
         return [(k, self[k]) for k in self._list]
     def __delitem__(self, key):
         dict.__delitem__(self, key)
         self._list.remove(key)
     def pop(self, key, *args, **kwargs):
         dict.pop(self, key, *args, **kwargs)
         try:
             self._list.remove(key)
         except ValueError:
             pass
     def keys(self):
         return self._list
     def iterkeys(self):
         return self._list.__iter__()
     def iteritems(self):
         for k in self._list:
             yield k, self[k]
     def insert(self, index, key, val):
         self._list.insert(index, key)
         dict.__setitem__(self, key, val)
     def __repr__(self):
         if not self:
             return '%s()' % self.__class__.__name__
         return '%s(%r)' % (self.__class__.__name__, self.items())
 
 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, '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(__file__)
 
 if not isinstance(datapath, bytes):
     datapath = pycompat.fsencode(datapath)
 
 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['__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(getattr(mainmod, '__file__', '')) == 'hg':
             _sethgexecutable(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, onerr=None, errprefix=None, out=None):
     '''enhanced shell command execution.
     run with environment maybe modified, maybe in different dir.
 
     if command fails and onerr is None, return status, else raise onerr
     object as exception.
 
     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
     origcmd = cmd
     cmd = quotecommand(cmd)
     if pycompat.sysplatform == 'plan9' and (sys.version_info[0] == 2
                                     and sys.version_info[1] < 7):
         # subprocess kludge to work around issues in half-baked Python
         # ports, notably bichued/python:
         if not cwd is None:
             os.chdir(cwd)
         rc = os.system(cmd)
     else:
         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
     if rc and onerr:
         errmsg = '%s %s' % (os.path.basename(origcmd.split(None, 1)[0]),
                             explainexit(rc)[0])
         if errprefix:
             errmsg = '%s: %s' % (errprefix, errmsg)
         raise onerr(errmsg)
     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
 
 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(dest)
         unlink(dest)
     # hardlinks are problematic on CIFS, quietly ignore this flag
     # until we find a way to work around it cleanly (issue4546)
     if False and 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(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
 
     if hardlink is None:
         hardlink = (os.stat(src).st_dev ==
                     os.stat(os.path.dirname(dst)).st_dev)
     if hardlink:
         topic = _('linking')
     else:
         topic = _('copying')
 
     if os.path.isdir(src):
         os.mkdir(dst)
         for name, kind in osutil.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:
             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 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
 else:
     checkosfilename = platform.checkosfilename
 
 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(r'([^%s]+)|([%s]+)' % (seps, seps))
     dir = os.path.normpath(root)
     result = []
     for part, sep in pattern.findall(name):
         if sep:
             result.append(sep)
             continue
 
         if dir not in _fspathcache:
             _fspathcache[dir] = _makefspathcacheentry(dir)
         contents = _fspathcache[dir]
 
         found = contents.get(part)
         if not found:
             # retry "once per directory" per "dirstate.walk" which
             # may take place for each patches of "hg qpush", for example
             _fspathcache[dir] = contents = _makefspathcacheentry(dir)
             found = contents.get(part)
 
         result.append(found or part)
         dir = os.path.join(dir, part)
 
     return ''.join(result)
 
 def checknlink(testfile):
     '''check whether hardlink count reporting works properly'''
 
     # testfile may be open, so we need a separate file for checking to
     # work around issue2543 (or testfile may get lost on Samba shares)
     f1 = 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, path):
         try:
             self.stat = os.stat(path)
         except OSError as err:
             if err.errno != errno.ENOENT:
                 raise
             self.stat = None
 
     __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:
             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'.
         """
         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
             raise
 
     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(filename)
             if oldstat and oldstat.stat:
                 rename(self._tempname, filename)
                 newstat = filestat(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 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.
         targetsize is how big a buffer to try to maintain."""
         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 = t.strftime(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=[]):
     """parse a localized time string and return a (unixtime, offset) tuple.
     if the string cannot be parsed, ValueError is raised."""
     # 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
         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(date, 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] in "HMS":
                     b = "00"
                 else:
                     b = "0"
 
             # this piece is for matching the generic end to today's date
             n = datestr(now, "%" + part[0])
 
             defaults[part] = (b, n)
 
         for format in formats:
             try:
                 when, offset = strdate(date, format, defaults)
             except (ValueError, OverflowError):
                 pass
             else:
                 break
         else:
             raise Abort(_('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 Abort(_('date exceeds 32 bits: %d') % when)
     if offset < -50400 or offset > 43200:
         raise Abort(_('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):
     """
     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])
 
     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])
     """
     if pattern.startswith('re:'):
         pattern = pattern[3:]
         try:
             regex = remod.compile(pattern)
         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:]
     return 'literal', pattern, pattern.__eq__
 
 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 count >= divisor * multiplier:
                 return format % (count / float(divisor))
         return unittable[-1][2] % count
 
     return go
 
 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')),
     )
 
 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() == '' 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() == ''):
                     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 + ''.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(encoding.encoding, encoding.encodingmode)
     initindent = initindent.decode(encoding.encoding, encoding.encodingmode)
     hangindent = hangindent.decode(encoding.encoding, encoding.encodingmode)
     wrapper = MBTextWrapper(width=width,
                             initial_indent=initindent,
                             subsequent_indent=hangindent)
     return wrapper.fill(line).encode(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, pycompat.urlunquote(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):
         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
         """
         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
 
     def authinfo(self):
         user, passwd = self.user, self.passwd
         try:
             self.user, self.passwd = None, None
             s = str(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 str(u)
 
 def removeauth(u):
     '''remove all authentication information from a url string'''
     u = url(u)
     u.user = u.passwd = None
     return str(u)
 
 def isatty(fp):
     try:
         return fp.isatty()
     except AttributeError:
         return False
 
 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 = time.time()
         indent = 2
         _timenesting[0] += indent
         try:
             return func(*args, **kwargs)
         finally:
             elapsed = time.time() - 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'):
     '''Yields lines for a nicely formatted stacktrace.
     Skips the 'skip' last 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]]
     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):
     '''Writes a message to f (stderr) with a nicely formatted stacktrace.
     Skips the 'skip' last 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)
     for line in getstackframes(skip + 1):
         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 self._dirs.iterkeys()
 
     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
 
 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 = {}
 
     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
 
         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
 
 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``.
         """
         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()
 
 class _zlibengine(compressionengine):
     def name(self):
         return 'zlib'
 
     def bundletype(self):
         return 'gzip', 'GZ'
 
     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())
 
 compengines.register(_zlibengine())
 
 class _bz2engine(compressionengine):
     def name(self):
         return 'bz2'
 
     def bundletype(self):
         return 'bzip2', 'BZ'
 
     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):
         return 'none', 'UN'
 
     def compressstream(self, it, opts=None):
         return it
 
     def decompressorreader(self, fh):
         return fh
 
 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):
         return 'zstd', 'ZS'
 
     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))
 
 compengines.register(_zstdengine())
 
 # convenient shortcut
 dst = debugstacktrace