exchange: refactor APIs to obtain bundle data (API)

Currently, exchange.getbundle() returns either a cg1unpacker or a
util.chunkbuffer (in the case of bundle2). This is kinda OK, as
both expose a .read() to consumers. However, localpeer.getbundle()
has code inferring what the response type is based on arguments and
converts the util.chunkbuffer returned in the bundle2 case to a
bundle2.unbundle20 instance. This is a sign that the API for
exchange.getbundle() is not ideal because it doesn't consistently
return an "unbundler" instance.

In addition, unbundlers mask the fact that there is an underlying
generator of changegroup data. In both cg1 and bundle2, this generator
is being fed into a util.chunkbuffer so it can be re-exposed as a
file object.

util.chunkbuffer is a nice abstraction. However, it should only be
used "at the edges." This is because keeping data as a generator is
more efficient than converting it to a chunkbuffer, especially if we
convert that chunkbuffer back to a generator (as is the case in some
code paths currently).

This patch refactors exchange.getbundle() into
exchange.getbundlechunks(). The new API returns an iterator of chunks
instead of a file-like object.

Callers of exchange.getbundle() have been updated to use the new API.

There is a minor change of behavior in test-getbundle.t. This is
because `hg debuggetbundle` isn't defining bundlecaps. As a result,
a cg1 data stream and unpacker is being produced. This is getting fed
into a new bundle20 instance via bundle2.writebundle(), which uses
a backchannel mechanism between changegroup generation to add the
"nbchanges" part parameter. I never liked this backchannel mechanism
and I plan to remove it someday. `hg bundle` still produces the
"nbchanges" part parameter, so there should be no user-visible
change of behavior. I consider this "regression" a bug in
`hg debuggetbundle`. And that bug is captured by an existing
"TODO" in the code to use bundle2 capabilities.

mercurial/exchange.py

@@ -1,1962 +1,1956
 # exchange.py - utility to exchange data between repos.
 #
 # 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.
 
 from __future__ import absolute_import
 
 import errno
 import hashlib
 
 from .i18n import _
 from .node import (
     hex,
     nullid,
 )
 from . import (
     base85,
     bookmarks as bookmod,
     bundle2,
     changegroup,
     discovery,
     error,
     lock as lockmod,
     obsolete,
     phases,
     pushkey,
     scmutil,
     sslutil,
     streamclone,
     tags,
     url as urlmod,
     util,
 )
 
 urlerr = util.urlerr
 urlreq = util.urlreq
 
 # Maps bundle compression human names to internal representation.
 _bundlespeccompressions = {'none': None,
                            'bzip2': 'BZ',
                            'gzip': 'GZ',
                           }
 
 # Maps bundle version human names to changegroup versions.
 _bundlespeccgversions = {'v1': '01',
                          'v2': '02',
                          'packed1': 's1',
                          'bundle2': '02', #legacy
                         }
 
 def parsebundlespec(repo, spec, strict=True, externalnames=False):
     """Parse a bundle string specification into parts.
 
     Bundle specifications denote a well-defined bundle/exchange format.
     The content of a given specification should not change over time in
     order to ensure that bundles produced by a newer version of Mercurial are
     readable from an older version.
 
     The string currently has the form:
 
        <compression>-<type>[;<parameter0>[;<parameter1>]]
 
     Where <compression> is one of the supported compression formats
     and <type> is (currently) a version string. A ";" can follow the type and
     all text afterwards is interpretted as URI encoded, ";" delimited key=value
     pairs.
 
     If ``strict`` is True (the default) <compression> is required. Otherwise,
     it is optional.
 
     If ``externalnames`` is False (the default), the human-centric names will
     be converted to their internal representation.
 
     Returns a 3-tuple of (compression, version, parameters). Compression will
     be ``None`` if not in strict mode and a compression isn't defined.
 
     An ``InvalidBundleSpecification`` is raised when the specification is
     not syntactically well formed.
 
     An ``UnsupportedBundleSpecification`` is raised when the compression or
     bundle type/version is not recognized.
 
     Note: this function will likely eventually return a more complex data
     structure, including bundle2 part information.
     """
     def parseparams(s):
         if ';' not in s:
             return s, {}
 
         params = {}
         version, paramstr = s.split(';', 1)
 
         for p in paramstr.split(';'):
             if '=' not in p:
                 raise error.InvalidBundleSpecification(
                     _('invalid bundle specification: '
                       'missing "=" in parameter: %s') % p)
 
             key, value = p.split('=', 1)
             key = urlreq.unquote(key)
             value = urlreq.unquote(value)
             params[key] = value
 
         return version, params
 
 
     if strict and '-' not in spec:
         raise error.InvalidBundleSpecification(
                 _('invalid bundle specification; '
                   'must be prefixed with compression: %s') % spec)
 
     if '-' in spec:
         compression, version = spec.split('-', 1)
 
         if compression not in _bundlespeccompressions:
             raise error.UnsupportedBundleSpecification(
                     _('%s compression is not supported') % compression)
 
         version, params = parseparams(version)
 
         if version not in _bundlespeccgversions:
             raise error.UnsupportedBundleSpecification(
                     _('%s is not a recognized bundle version') % version)
     else:
         # Value could be just the compression or just the version, in which
         # case some defaults are assumed (but only when not in strict mode).
         assert not strict
 
         spec, params = parseparams(spec)
 
         if spec in _bundlespeccompressions:
             compression = spec
             version = 'v1'
             if 'generaldelta' in repo.requirements:
                 version = 'v2'
         elif spec in _bundlespeccgversions:
             if spec == 'packed1':
                 compression = 'none'
             else:
                 compression = 'bzip2'
             version = spec
         else:
             raise error.UnsupportedBundleSpecification(
                     _('%s is not a recognized bundle specification') % spec)
 
     # The specification for packed1 can optionally declare the data formats
     # required to apply it. If we see this metadata, compare against what the
     # repo supports and error if the bundle isn't compatible.
     if version == 'packed1' and 'requirements' in params:
         requirements = set(params['requirements'].split(','))
         missingreqs = requirements - repo.supportedformats
         if missingreqs:
             raise error.UnsupportedBundleSpecification(
                     _('missing support for repository features: %s') %
                       ', '.join(sorted(missingreqs)))
 
     if not externalnames:
         compression = _bundlespeccompressions[compression]
         version = _bundlespeccgversions[version]
     return compression, version, params
 
 def readbundle(ui, fh, fname, vfs=None):
     header = changegroup.readexactly(fh, 4)
 
     alg = None
     if not fname:
         fname = "stream"
         if not header.startswith('HG') and header.startswith('\0'):
             fh = changegroup.headerlessfixup(fh, header)
             header = "HG10"
             alg = 'UN'
     elif vfs:
         fname = vfs.join(fname)
 
     magic, version = header[0:2], header[2:4]
 
     if magic != 'HG':
         raise error.Abort(_('%s: not a Mercurial bundle') % fname)
     if version == '10':
         if alg is None:
             alg = changegroup.readexactly(fh, 2)
         return changegroup.cg1unpacker(fh, alg)
     elif version.startswith('2'):
         return bundle2.getunbundler(ui, fh, magicstring=magic + version)
     elif version == 'S1':
         return streamclone.streamcloneapplier(fh)
     else:
         raise error.Abort(_('%s: unknown bundle version %s') % (fname, version))
 
 def getbundlespec(ui, fh):
     """Infer the bundlespec from a bundle file handle.
 
     The input file handle is seeked and the original seek position is not
     restored.
     """
     def speccompression(alg):
         for k, v in _bundlespeccompressions.items():
             if v == alg:
                 return k
         return None
 
     b = readbundle(ui, fh, None)
     if isinstance(b, changegroup.cg1unpacker):
         alg = b._type
         if alg == '_truncatedBZ':
             alg = 'BZ'
         comp = speccompression(alg)
         if not comp:
             raise error.Abort(_('unknown compression algorithm: %s') % alg)
         return '%s-v1' % comp
     elif isinstance(b, bundle2.unbundle20):
         if 'Compression' in b.params:
             comp = speccompression(b.params['Compression'])
             if not comp:
                 raise error.Abort(_('unknown compression algorithm: %s') % comp)
         else:
             comp = 'none'
 
         version = None
         for part in b.iterparts():
             if part.type == 'changegroup':
                 version = part.params['version']
                 if version in ('01', '02'):
                     version = 'v2'
                 else:
                     raise error.Abort(_('changegroup version %s does not have '
                                         'a known bundlespec') % version,
                                       hint=_('try upgrading your Mercurial '
                                               'client'))
 
         if not version:
             raise error.Abort(_('could not identify changegroup version in '
                                 'bundle'))
 
         return '%s-%s' % (comp, version)
     elif isinstance(b, streamclone.streamcloneapplier):
         requirements = streamclone.readbundle1header(fh)[2]
         params = 'requirements=%s' % ','.join(sorted(requirements))
         return 'none-packed1;%s' % urlreq.quote(params)
     else:
         raise error.Abort(_('unknown bundle type: %s') % b)
 
 def buildobsmarkerspart(bundler, markers):
     """add an obsmarker part to the bundler with <markers>
 
     No part is created if markers is empty.
     Raises ValueError if the bundler doesn't support any known obsmarker format.
     """
     if markers:
         remoteversions = bundle2.obsmarkersversion(bundler.capabilities)
         version = obsolete.commonversion(remoteversions)
         if version is None:
             raise ValueError('bundler does not support common obsmarker format')
         stream = obsolete.encodemarkers(markers, True, version=version)
         return bundler.newpart('obsmarkers', data=stream)
     return None
 
 def _computeoutgoing(repo, heads, common):
     """Computes which revs are outgoing given a set of common
     and a set of heads.
 
     This is a separate function so extensions can have access to
     the logic.
 
     Returns a discovery.outgoing object.
     """
     cl = repo.changelog
     if common:
         hasnode = cl.hasnode
         common = [n for n in common if hasnode(n)]
     else:
         common = [nullid]
     if not heads:
         heads = cl.heads()
     return discovery.outgoing(repo, common, heads)
 
 def _forcebundle1(op):
     """return true if a pull/push must use bundle1
 
     This function is used to allow testing of the older bundle version"""
     ui = op.repo.ui
     forcebundle1 = False
     # The goal is this config is to allow developper to choose the bundle
     # version used during exchanged. This is especially handy during test.
     # Value is a list of bundle version to be picked from, highest version
     # should be used.
     #
     # developer config: devel.legacy.exchange
     exchange = ui.configlist('devel', 'legacy.exchange')
     forcebundle1 = 'bundle2' not in exchange and 'bundle1' in exchange
     return forcebundle1 or not op.remote.capable('bundle2')
 
 class pushoperation(object):
     """A object that represent a single push operation
 
     Its purpose is to carry push related state and very common operations.
 
     A new pushoperation should be created at the beginning of each push and
     discarded afterward.
     """
 
     def __init__(self, repo, remote, force=False, revs=None, newbranch=False,
                  bookmarks=()):
         # repo we push from
         self.repo = repo
         self.ui = repo.ui
         # repo we push to
         self.remote = remote
         # force option provided
         self.force = force
         # revs to be pushed (None is "all")
         self.revs = revs
         # bookmark explicitly pushed
         self.bookmarks = bookmarks
         # allow push of new branch
         self.newbranch = newbranch
         # did a local lock get acquired?
         self.locallocked = None
         # step already performed
         # (used to check what steps have been already performed through bundle2)
         self.stepsdone = set()
         # Integer version of the changegroup push result
         # - None means nothing to push
         # - 0 means HTTP error
         # - 1 means we pushed and remote head count is unchanged *or*
         #   we have outgoing changesets but refused to push
         # - other values as described by addchangegroup()
         self.cgresult = None
         # Boolean value for the bookmark push
         self.bkresult = None
         # discover.outgoing object (contains common and outgoing data)
         self.outgoing = None
         # all remote heads before the push
         self.remoteheads = None
         # testable as a boolean indicating if any nodes are missing locally.
         self.incoming = None
         # phases changes that must be pushed along side the changesets
         self.outdatedphases = None
         # phases changes that must be pushed if changeset push fails
         self.fallbackoutdatedphases = None
         # outgoing obsmarkers
         self.outobsmarkers = set()
         # outgoing bookmarks
         self.outbookmarks = []
         # transaction manager
         self.trmanager = None
         # map { pushkey partid -> callback handling failure}
         # used to handle exception from mandatory pushkey part failure
         self.pkfailcb = {}
 
     @util.propertycache
     def futureheads(self):
         """future remote heads if the changeset push succeeds"""
         return self.outgoing.missingheads
 
     @util.propertycache
     def fallbackheads(self):
         """future remote heads if the changeset push fails"""
         if self.revs is None:
             # not target to push, all common are relevant
             return self.outgoing.commonheads
         unfi = self.repo.unfiltered()
         # I want cheads = heads(::missingheads and ::commonheads)
         # (missingheads is revs with secret changeset filtered out)
         #
         # This can be expressed as:
         #     cheads = ( (missingheads and ::commonheads)
         #              + (commonheads and ::missingheads))"
         #              )
         #
         # while trying to push we already computed the following:
         #     common = (::commonheads)
         #     missing = ((commonheads::missingheads) - commonheads)
         #
         # We can pick:
         # * missingheads part of common (::commonheads)
         common = self.outgoing.common
         nm = self.repo.changelog.nodemap
         cheads = [node for node in self.revs if nm[node] in common]
         # and
         # * commonheads parents on missing
         revset = unfi.set('%ln and parents(roots(%ln))',
                          self.outgoing.commonheads,
                          self.outgoing.missing)
         cheads.extend(c.node() for c in revset)
         return cheads
 
     @property
     def commonheads(self):
         """set of all common heads after changeset bundle push"""
         if self.cgresult:
             return self.futureheads
         else:
             return self.fallbackheads
 
 # mapping of message used when pushing bookmark
 bookmsgmap = {'update': (_("updating bookmark %s\n"),
                          _('updating bookmark %s failed!\n')),
               'export': (_("exporting bookmark %s\n"),
                          _('exporting bookmark %s failed!\n')),
               'delete': (_("deleting remote bookmark %s\n"),
                          _('deleting remote bookmark %s failed!\n')),
               }
 
 
 def push(repo, remote, force=False, revs=None, newbranch=False, bookmarks=(),
          opargs=None):
     '''Push outgoing changesets (limited by revs) from a local
     repository to remote. Return an integer:
       - None means nothing to push
       - 0 means HTTP error
       - 1 means we pushed and remote head count is unchanged *or*
         we have outgoing changesets but refused to push
       - other values as described by addchangegroup()
     '''
     if opargs is None:
         opargs = {}
     pushop = pushoperation(repo, remote, force, revs, newbranch, bookmarks,
                            **opargs)
     if pushop.remote.local():
         missing = (set(pushop.repo.requirements)
                    - pushop.remote.local().supported)
         if missing:
             msg = _("required features are not"
                     " supported in the destination:"
                     " %s") % (', '.join(sorted(missing)))
             raise error.Abort(msg)
 
     # there are two ways to push to remote repo:
     #
     # addchangegroup assumes local user can lock remote
     # repo (local filesystem, old ssh servers).
     #
     # unbundle assumes local user cannot lock remote repo (new ssh
     # servers, http servers).
 
     if not pushop.remote.canpush():
         raise error.Abort(_("destination does not support push"))
     # get local lock as we might write phase data
     localwlock = locallock = None
     try:
         # bundle2 push may receive a reply bundle touching bookmarks or other
         # things requiring the wlock. Take it now to ensure proper ordering.
         maypushback = pushop.ui.configbool('experimental', 'bundle2.pushback')
         if (not _forcebundle1(pushop)) and maypushback:
             localwlock = pushop.repo.wlock()
         locallock = pushop.repo.lock()
         pushop.locallocked = True
     except IOError as err:
         pushop.locallocked = False
         if err.errno != errno.EACCES:
             raise
         # source repo cannot be locked.
         # We do not abort the push, but just disable the local phase
         # synchronisation.
         msg = 'cannot lock source repository: %s\n' % err
         pushop.ui.debug(msg)
     try:
         if pushop.locallocked:
             pushop.trmanager = transactionmanager(pushop.repo,
                                                   'push-response',
                                                   pushop.remote.url())
         pushop.repo.checkpush(pushop)
         lock = None
         unbundle = pushop.remote.capable('unbundle')
         if not unbundle:
             lock = pushop.remote.lock()
         try:
             _pushdiscovery(pushop)
             if not _forcebundle1(pushop):
                 _pushbundle2(pushop)
             _pushchangeset(pushop)
             _pushsyncphase(pushop)
             _pushobsolete(pushop)
             _pushbookmark(pushop)
         finally:
             if lock is not None:
                 lock.release()
         if pushop.trmanager:
             pushop.trmanager.close()
     finally:
         if pushop.trmanager:
             pushop.trmanager.release()
         if locallock is not None:
             locallock.release()
         if localwlock is not None:
             localwlock.release()
 
     return pushop
 
 # list of steps to perform discovery before push
 pushdiscoveryorder = []
 
 # Mapping between step name and function
 #
 # This exists to help extensions wrap steps if necessary
 pushdiscoverymapping = {}
 
 def pushdiscovery(stepname):
     """decorator for function performing discovery before push
 
     The function is added to the step -> function mapping and appended to the
     list of steps.  Beware that decorated function will be added in order (this
     may matter).
 
     You can only use this decorator for a new step, if you want to wrap a step
     from an extension, change the pushdiscovery dictionary directly."""
     def dec(func):
         assert stepname not in pushdiscoverymapping
         pushdiscoverymapping[stepname] = func
         pushdiscoveryorder.append(stepname)
         return func
     return dec
 
 def _pushdiscovery(pushop):
     """Run all discovery steps"""
     for stepname in pushdiscoveryorder:
         step = pushdiscoverymapping[stepname]
         step(pushop)
 
 @pushdiscovery('changeset')
 def _pushdiscoverychangeset(pushop):
     """discover the changeset that need to be pushed"""
     fci = discovery.findcommonincoming
     commoninc = fci(pushop.repo, pushop.remote, force=pushop.force)
     common, inc, remoteheads = commoninc
     fco = discovery.findcommonoutgoing
     outgoing = fco(pushop.repo, pushop.remote, onlyheads=pushop.revs,
                    commoninc=commoninc, force=pushop.force)
     pushop.outgoing = outgoing
     pushop.remoteheads = remoteheads
     pushop.incoming = inc
 
 @pushdiscovery('phase')
 def _pushdiscoveryphase(pushop):
     """discover the phase that needs to be pushed
 
     (computed for both success and failure case for changesets push)"""
     outgoing = pushop.outgoing
     unfi = pushop.repo.unfiltered()
     remotephases = pushop.remote.listkeys('phases')
     publishing = remotephases.get('publishing', False)
     if (pushop.ui.configbool('ui', '_usedassubrepo', False)
         and remotephases    # server supports phases
         and not pushop.outgoing.missing # no changesets to be pushed
         and publishing):
         # When:
         # - this is a subrepo push
         # - and remote support phase
         # - and no changeset are to be pushed
         # - and remote is publishing
         # We may be in issue 3871 case!
         # We drop the possible phase synchronisation done by
         # courtesy to publish changesets possibly locally draft
         # on the remote.
         remotephases = {'publishing': 'True'}
     ana = phases.analyzeremotephases(pushop.repo,
                                      pushop.fallbackheads,
                                      remotephases)
     pheads, droots = ana
     extracond = ''
     if not publishing:
         extracond = ' and public()'
     revset = 'heads((%%ln::%%ln) %s)' % extracond
     # Get the list of all revs draft on remote by public here.
     # XXX Beware that revset break if droots is not strictly
     # XXX root we may want to ensure it is but it is costly
     fallback = list(unfi.set(revset, droots, pushop.fallbackheads))
     if not outgoing.missing:
         future = fallback
     else:
         # adds changeset we are going to push as draft
         #
         # should not be necessary for publishing server, but because of an
         # issue fixed in xxxxx we have to do it anyway.
         fdroots = list(unfi.set('roots(%ln  + %ln::)',
                        outgoing.missing, droots))
         fdroots = [f.node() for f in fdroots]
         future = list(unfi.set(revset, fdroots, pushop.futureheads))
     pushop.outdatedphases = future
     pushop.fallbackoutdatedphases = fallback
 
 @pushdiscovery('obsmarker')
 def _pushdiscoveryobsmarkers(pushop):
     if (obsolete.isenabled(pushop.repo, obsolete.exchangeopt)
         and pushop.repo.obsstore
         and 'obsolete' in pushop.remote.listkeys('namespaces')):
         repo = pushop.repo
         # very naive computation, that can be quite expensive on big repo.
         # However: evolution is currently slow on them anyway.
         nodes = (c.node() for c in repo.set('::%ln', pushop.futureheads))
         pushop.outobsmarkers = pushop.repo.obsstore.relevantmarkers(nodes)
 
 @pushdiscovery('bookmarks')
 def _pushdiscoverybookmarks(pushop):
     ui = pushop.ui
     repo = pushop.repo.unfiltered()
     remote = pushop.remote
     ui.debug("checking for updated bookmarks\n")
     ancestors = ()
     if pushop.revs:
         revnums = map(repo.changelog.rev, pushop.revs)
         ancestors = repo.changelog.ancestors(revnums, inclusive=True)
     remotebookmark = remote.listkeys('bookmarks')
 
     explicit = set([repo._bookmarks.expandname(bookmark)
                     for bookmark in pushop.bookmarks])
 
     comp = bookmod.compare(repo, repo._bookmarks, remotebookmark, srchex=hex)
     addsrc, adddst, advsrc, advdst, diverge, differ, invalid, same = comp
     for b, scid, dcid in advsrc:
         if b in explicit:
             explicit.remove(b)
         if not ancestors or repo[scid].rev() in ancestors:
             pushop.outbookmarks.append((b, dcid, scid))
     # search added bookmark
     for b, scid, dcid in addsrc:
         if b in explicit:
             explicit.remove(b)
             pushop.outbookmarks.append((b, '', scid))
     # search for overwritten bookmark
     for b, scid, dcid in advdst + diverge + differ:
         if b in explicit:
             explicit.remove(b)
             pushop.outbookmarks.append((b, dcid, scid))
     # search for bookmark to delete
     for b, scid, dcid in adddst:
         if b in explicit:
             explicit.remove(b)
             # treat as "deleted locally"
             pushop.outbookmarks.append((b, dcid, ''))
     # identical bookmarks shouldn't get reported
     for b, scid, dcid in same:
         if b in explicit:
             explicit.remove(b)
 
     if explicit:
         explicit = sorted(explicit)
         # we should probably list all of them
         ui.warn(_('bookmark %s does not exist on the local '
                   'or remote repository!\n') % explicit[0])
         pushop.bkresult = 2
 
     pushop.outbookmarks.sort()
 
 def _pushcheckoutgoing(pushop):
     outgoing = pushop.outgoing
     unfi = pushop.repo.unfiltered()
     if not outgoing.missing:
         # nothing to push
         scmutil.nochangesfound(unfi.ui, unfi, outgoing.excluded)
         return False
     # something to push
     if not pushop.force:
         # if repo.obsstore == False --> no obsolete
         # then, save the iteration
         if unfi.obsstore:
             # this message are here for 80 char limit reason
             mso = _("push includes obsolete changeset: %s!")
             mst = {"unstable": _("push includes unstable changeset: %s!"),
                    "bumped": _("push includes bumped changeset: %s!"),
                    "divergent": _("push includes divergent changeset: %s!")}
             # If we are to push if there is at least one
             # obsolete or unstable changeset in missing, at
             # least one of the missinghead will be obsolete or
             # unstable. So checking heads only is ok
             for node in outgoing.missingheads:
                 ctx = unfi[node]
                 if ctx.obsolete():
                     raise error.Abort(mso % ctx)
                 elif ctx.troubled():
                     raise error.Abort(mst[ctx.troubles()[0]] % ctx)
 
         discovery.checkheads(pushop)
     return True
 
 # List of names of steps to perform for an outgoing bundle2, order matters.
 b2partsgenorder = []
 
 # Mapping between step name and function
 #
 # This exists to help extensions wrap steps if necessary
 b2partsgenmapping = {}
 
 def b2partsgenerator(stepname, idx=None):
     """decorator for function generating bundle2 part
 
     The function is added to the step -> function mapping and appended to the
     list of steps.  Beware that decorated functions will be added in order
     (this may matter).
 
     You can only use this decorator for new steps, if you want to wrap a step
     from an extension, attack the b2partsgenmapping dictionary directly."""
     def dec(func):
         assert stepname not in b2partsgenmapping
         b2partsgenmapping[stepname] = func
         if idx is None:
             b2partsgenorder.append(stepname)
         else:
             b2partsgenorder.insert(idx, stepname)
         return func
     return dec
 
 def _pushb2ctxcheckheads(pushop, bundler):
     """Generate race condition checking parts
 
     Exists as an independent function to aid extensions
     """
     if not pushop.force:
         bundler.newpart('check:heads', data=iter(pushop.remoteheads))
 
 @b2partsgenerator('changeset')
 def _pushb2ctx(pushop, bundler):
     """handle changegroup push through bundle2
 
     addchangegroup result is stored in the ``pushop.cgresult`` attribute.
     """
     if 'changesets' in pushop.stepsdone:
         return
     pushop.stepsdone.add('changesets')
     # Send known heads to the server for race detection.
     if not _pushcheckoutgoing(pushop):
         return
     pushop.repo.prepushoutgoinghooks(pushop)
 
     _pushb2ctxcheckheads(pushop, bundler)
 
     b2caps = bundle2.bundle2caps(pushop.remote)
     version = '01'
     cgversions = b2caps.get('changegroup')
     if cgversions:  # 3.1 and 3.2 ship with an empty value
         cgversions = [v for v in cgversions
                       if v in changegroup.supportedoutgoingversions(
                           pushop.repo)]
         if not cgversions:
             raise ValueError(_('no common changegroup version'))
         version = max(cgversions)
     cg = changegroup.getlocalchangegroupraw(pushop.repo, 'push',
                                             pushop.outgoing,
                                             version=version)
     cgpart = bundler.newpart('changegroup', data=cg)
     if cgversions:
         cgpart.addparam('version', version)
     if 'treemanifest' in pushop.repo.requirements:
         cgpart.addparam('treemanifest', '1')
     def handlereply(op):
         """extract addchangegroup returns from server reply"""
         cgreplies = op.records.getreplies(cgpart.id)
         assert len(cgreplies['changegroup']) == 1
         pushop.cgresult = cgreplies['changegroup'][0]['return']
     return handlereply
 
 @b2partsgenerator('phase')
 def _pushb2phases(pushop, bundler):
     """handle phase push through bundle2"""
     if 'phases' in pushop.stepsdone:
         return
     b2caps = bundle2.bundle2caps(pushop.remote)
     if not 'pushkey' in b2caps:
         return
     pushop.stepsdone.add('phases')
     part2node = []
 
     def handlefailure(pushop, exc):
         targetid = int(exc.partid)
         for partid, node in part2node:
             if partid == targetid:
                 raise error.Abort(_('updating %s to public failed') % node)
 
     enc = pushkey.encode
     for newremotehead in pushop.outdatedphases:
         part = bundler.newpart('pushkey')
         part.addparam('namespace', enc('phases'))
         part.addparam('key', enc(newremotehead.hex()))
         part.addparam('old', enc(str(phases.draft)))
         part.addparam('new', enc(str(phases.public)))
         part2node.append((part.id, newremotehead))
         pushop.pkfailcb[part.id] = handlefailure
 
     def handlereply(op):
         for partid, node in part2node:
             partrep = op.records.getreplies(partid)
             results = partrep['pushkey']
             assert len(results) <= 1
             msg = None
             if not results:
                 msg = _('server ignored update of %s to public!\n') % node
             elif not int(results[0]['return']):
                 msg = _('updating %s to public failed!\n') % node
             if msg is not None:
                 pushop.ui.warn(msg)
     return handlereply
 
 @b2partsgenerator('obsmarkers')
 def _pushb2obsmarkers(pushop, bundler):
     if 'obsmarkers' in pushop.stepsdone:
         return
     remoteversions = bundle2.obsmarkersversion(bundler.capabilities)
     if obsolete.commonversion(remoteversions) is None:
         return
     pushop.stepsdone.add('obsmarkers')
     if pushop.outobsmarkers:
         markers = sorted(pushop.outobsmarkers)
         buildobsmarkerspart(bundler, markers)
 
 @b2partsgenerator('bookmarks')
 def _pushb2bookmarks(pushop, bundler):
     """handle bookmark push through bundle2"""
     if 'bookmarks' in pushop.stepsdone:
         return
     b2caps = bundle2.bundle2caps(pushop.remote)
     if 'pushkey' not in b2caps:
         return
     pushop.stepsdone.add('bookmarks')
     part2book = []
     enc = pushkey.encode
 
     def handlefailure(pushop, exc):
         targetid = int(exc.partid)
         for partid, book, action in part2book:
             if partid == targetid:
                 raise error.Abort(bookmsgmap[action][1].rstrip() % book)
         # we should not be called for part we did not generated
         assert False
 
     for book, old, new in pushop.outbookmarks:
         part = bundler.newpart('pushkey')
         part.addparam('namespace', enc('bookmarks'))
         part.addparam('key', enc(book))
         part.addparam('old', enc(old))
         part.addparam('new', enc(new))
         action = 'update'
         if not old:
             action = 'export'
         elif not new:
             action = 'delete'
         part2book.append((part.id, book, action))
         pushop.pkfailcb[part.id] = handlefailure
 
     def handlereply(op):
         ui = pushop.ui
         for partid, book, action in part2book:
             partrep = op.records.getreplies(partid)
             results = partrep['pushkey']
             assert len(results) <= 1
             if not results:
                 pushop.ui.warn(_('server ignored bookmark %s update\n') % book)
             else:
                 ret = int(results[0]['return'])
                 if ret:
                     ui.status(bookmsgmap[action][0] % book)
                 else:
                     ui.warn(bookmsgmap[action][1] % book)
                     if pushop.bkresult is not None:
                         pushop.bkresult = 1
     return handlereply
 
 
 def _pushbundle2(pushop):
     """push data to the remote using bundle2
 
     The only currently supported type of data is changegroup but this will
     evolve in the future."""
     bundler = bundle2.bundle20(pushop.ui, bundle2.bundle2caps(pushop.remote))
     pushback = (pushop.trmanager
                 and pushop.ui.configbool('experimental', 'bundle2.pushback'))
 
     # create reply capability
     capsblob = bundle2.encodecaps(bundle2.getrepocaps(pushop.repo,
                                                       allowpushback=pushback))
     bundler.newpart('replycaps', data=capsblob)
     replyhandlers = []
     for partgenname in b2partsgenorder:
         partgen = b2partsgenmapping[partgenname]
         ret = partgen(pushop, bundler)
         if callable(ret):
             replyhandlers.append(ret)
     # do not push if nothing to push
     if bundler.nbparts <= 1:
         return
     stream = util.chunkbuffer(bundler.getchunks())
     try:
         try:
             reply = pushop.remote.unbundle(
                 stream, ['force'], pushop.remote.url())
         except error.BundleValueError as exc:
             raise error.Abort(_('missing support for %s') % exc)
         try:
             trgetter = None
             if pushback:
                 trgetter = pushop.trmanager.transaction
             op = bundle2.processbundle(pushop.repo, reply, trgetter)
         except error.BundleValueError as exc:
             raise error.Abort(_('missing support for %s') % exc)
         except bundle2.AbortFromPart as exc:
             pushop.ui.status(_('remote: %s\n') % exc)
             raise error.Abort(_('push failed on remote'), hint=exc.hint)
     except error.PushkeyFailed as exc:
         partid = int(exc.partid)
         if partid not in pushop.pkfailcb:
             raise
         pushop.pkfailcb[partid](pushop, exc)
     for rephand in replyhandlers:
         rephand(op)
 
 def _pushchangeset(pushop):
     """Make the actual push of changeset bundle to remote repo"""
     if 'changesets' in pushop.stepsdone:
         return
     pushop.stepsdone.add('changesets')
     if not _pushcheckoutgoing(pushop):
         return
     pushop.repo.prepushoutgoinghooks(pushop)
     outgoing = pushop.outgoing
     unbundle = pushop.remote.capable('unbundle')
     # TODO: get bundlecaps from remote
     bundlecaps = None
     # create a changegroup from local
     if pushop.revs is None and not (outgoing.excluded
                             or pushop.repo.changelog.filteredrevs):
         # push everything,
         # use the fast path, no race possible on push
         bundler = changegroup.cg1packer(pushop.repo, bundlecaps)
         cg = changegroup.getsubset(pushop.repo,
                                    outgoing,
                                    bundler,
                                    'push',
                                    fastpath=True)
     else:
         cg = changegroup.getlocalchangegroup(pushop.repo, 'push', outgoing,
                                         bundlecaps)
 
     # apply changegroup to remote
     if unbundle:
         # local repo finds heads on server, finds out what
         # revs it must push. once revs transferred, if server
         # finds it has different heads (someone else won
         # commit/push race), server aborts.
         if pushop.force:
             remoteheads = ['force']
         else:
             remoteheads = pushop.remoteheads
         # ssh: return remote's addchangegroup()
         # http: return remote's addchangegroup() or 0 for error
         pushop.cgresult = pushop.remote.unbundle(cg, remoteheads,
                                             pushop.repo.url())
     else:
         # we return an integer indicating remote head count
         # change
         pushop.cgresult = pushop.remote.addchangegroup(cg, 'push',
                                                        pushop.repo.url())
 
 def _pushsyncphase(pushop):
     """synchronise phase information locally and remotely"""
     cheads = pushop.commonheads
     # even when we don't push, exchanging phase data is useful
     remotephases = pushop.remote.listkeys('phases')
     if (pushop.ui.configbool('ui', '_usedassubrepo', False)
         and remotephases    # server supports phases
         and pushop.cgresult is None # nothing was pushed
         and remotephases.get('publishing', False)):
         # When:
         # - this is a subrepo push
         # - and remote support phase
         # - and no changeset was pushed
         # - and remote is publishing
         # We may be in issue 3871 case!
         # We drop the possible phase synchronisation done by
         # courtesy to publish changesets possibly locally draft
         # on the remote.
         remotephases = {'publishing': 'True'}
     if not remotephases: # old server or public only reply from non-publishing
         _localphasemove(pushop, cheads)
         # don't push any phase data as there is nothing to push
     else:
         ana = phases.analyzeremotephases(pushop.repo, cheads,
                                          remotephases)
         pheads, droots = ana
         ### Apply remote phase on local
         if remotephases.get('publishing', False):
             _localphasemove(pushop, cheads)
         else: # publish = False
             _localphasemove(pushop, pheads)
             _localphasemove(pushop, cheads, phases.draft)
         ### Apply local phase on remote
 
         if pushop.cgresult:
             if 'phases' in pushop.stepsdone:
                 # phases already pushed though bundle2
                 return
             outdated = pushop.outdatedphases
         else:
             outdated = pushop.fallbackoutdatedphases
 
         pushop.stepsdone.add('phases')
 
         # filter heads already turned public by the push
         outdated = [c for c in outdated if c.node() not in pheads]
         # fallback to independent pushkey command
         for newremotehead in outdated:
             r = pushop.remote.pushkey('phases',
                                       newremotehead.hex(),
                                       str(phases.draft),
                                       str(phases.public))
             if not r:
                 pushop.ui.warn(_('updating %s to public failed!\n')
                                % newremotehead)
 
 def _localphasemove(pushop, nodes, phase=phases.public):
     """move <nodes> to <phase> in the local source repo"""
     if pushop.trmanager:
         phases.advanceboundary(pushop.repo,
                                pushop.trmanager.transaction(),
                                phase,
                                nodes)
     else:
         # repo is not locked, do not change any phases!
         # Informs the user that phases should have been moved when
         # applicable.
         actualmoves = [n for n in nodes if phase < pushop.repo[n].phase()]
         phasestr = phases.phasenames[phase]
         if actualmoves:
             pushop.ui.status(_('cannot lock source repo, skipping '
                                'local %s phase update\n') % phasestr)
 
 def _pushobsolete(pushop):
     """utility function to push obsolete markers to a remote"""
     if 'obsmarkers' in pushop.stepsdone:
         return
     repo = pushop.repo
     remote = pushop.remote
     pushop.stepsdone.add('obsmarkers')
     if pushop.outobsmarkers:
         pushop.ui.debug('try to push obsolete markers to remote\n')
         rslts = []
         remotedata = obsolete._pushkeyescape(sorted(pushop.outobsmarkers))
         for key in sorted(remotedata, reverse=True):
             # reverse sort to ensure we end with dump0
             data = remotedata[key]
             rslts.append(remote.pushkey('obsolete', key, '', data))
         if [r for r in rslts if not r]:
             msg = _('failed to push some obsolete markers!\n')
             repo.ui.warn(msg)
 
 def _pushbookmark(pushop):
     """Update bookmark position on remote"""
     if pushop.cgresult == 0 or 'bookmarks' in pushop.stepsdone:
         return
     pushop.stepsdone.add('bookmarks')
     ui = pushop.ui
     remote = pushop.remote
 
     for b, old, new in pushop.outbookmarks:
         action = 'update'
         if not old:
             action = 'export'
         elif not new:
             action = 'delete'
         if remote.pushkey('bookmarks', b, old, new):
             ui.status(bookmsgmap[action][0] % b)
         else:
             ui.warn(bookmsgmap[action][1] % b)
             # discovery can have set the value form invalid entry
             if pushop.bkresult is not None:
                 pushop.bkresult = 1
 
 class pulloperation(object):
     """A object that represent a single pull operation
 
     It purpose is to carry pull related state and very common operation.
 
     A new should be created at the beginning of each pull and discarded
     afterward.
     """
 
     def __init__(self, repo, remote, heads=None, force=False, bookmarks=(),
                  remotebookmarks=None, streamclonerequested=None):
         # repo we pull into
         self.repo = repo
         # repo we pull from
         self.remote = remote
         # revision we try to pull (None is "all")
         self.heads = heads
         # bookmark pulled explicitly
         self.explicitbookmarks = [repo._bookmarks.expandname(bookmark)
                                   for bookmark in bookmarks]
         # do we force pull?
         self.force = force
         # whether a streaming clone was requested
         self.streamclonerequested = streamclonerequested
         # transaction manager
         self.trmanager = None
         # set of common changeset between local and remote before pull
         self.common = None
         # set of pulled head
         self.rheads = None
         # list of missing changeset to fetch remotely
         self.fetch = None
         # remote bookmarks data
         self.remotebookmarks = remotebookmarks
         # result of changegroup pulling (used as return code by pull)
         self.cgresult = None
         # list of step already done
         self.stepsdone = set()
         # Whether we attempted a clone from pre-generated bundles.
         self.clonebundleattempted = False
 
     @util.propertycache
     def pulledsubset(self):
         """heads of the set of changeset target by the pull"""
         # compute target subset
         if self.heads is None:
             # We pulled every thing possible
             # sync on everything common
             c = set(self.common)
             ret = list(self.common)
             for n in self.rheads:
                 if n not in c:
                     ret.append(n)
             return ret
         else:
             # We pulled a specific subset
             # sync on this subset
             return self.heads
 
     @util.propertycache
     def canusebundle2(self):
         return not _forcebundle1(self)
 
     @util.propertycache
     def remotebundle2caps(self):
         return bundle2.bundle2caps(self.remote)
 
     def gettransaction(self):
         # deprecated; talk to trmanager directly
         return self.trmanager.transaction()
 
 class transactionmanager(object):
     """An object to manage the life cycle of a transaction
 
     It creates the transaction on demand and calls the appropriate hooks when
     closing the transaction."""
     def __init__(self, repo, source, url):
         self.repo = repo
         self.source = source
         self.url = url
         self._tr = None
 
     def transaction(self):
         """Return an open transaction object, constructing if necessary"""
         if not self._tr:
             trname = '%s\n%s' % (self.source, util.hidepassword(self.url))
             self._tr = self.repo.transaction(trname)
             self._tr.hookargs['source'] = self.source
             self._tr.hookargs['url'] = self.url
         return self._tr
 
     def close(self):
         """close transaction if created"""
         if self._tr is not None:
             self._tr.close()
 
     def release(self):
         """release transaction if created"""
         if self._tr is not None:
             self._tr.release()
 
 def pull(repo, remote, heads=None, force=False, bookmarks=(), opargs=None,
          streamclonerequested=None):
     """Fetch repository data from a remote.
 
     This is the main function used to retrieve data from a remote repository.
 
     ``repo`` is the local repository to clone into.
     ``remote`` is a peer instance.
     ``heads`` is an iterable of revisions we want to pull. ``None`` (the
     default) means to pull everything from the remote.
     ``bookmarks`` is an iterable of bookmarks requesting to be pulled. By
     default, all remote bookmarks are pulled.
     ``opargs`` are additional keyword arguments to pass to ``pulloperation``
     initialization.
     ``streamclonerequested`` is a boolean indicating whether a "streaming
     clone" is requested. A "streaming clone" is essentially a raw file copy
     of revlogs from the server. This only works when the local repository is
     empty. The default value of ``None`` means to respect the server
     configuration for preferring stream clones.
 
     Returns the ``pulloperation`` created for this pull.
     """
     if opargs is None:
         opargs = {}
     pullop = pulloperation(repo, remote, heads, force, bookmarks=bookmarks,
                            streamclonerequested=streamclonerequested, **opargs)
     if pullop.remote.local():
         missing = set(pullop.remote.requirements) - pullop.repo.supported
         if missing:
             msg = _("required features are not"
                     " supported in the destination:"
                     " %s") % (', '.join(sorted(missing)))
             raise error.Abort(msg)
 
     wlock = lock = None
     try:
         wlock = pullop.repo.wlock()
         lock = pullop.repo.lock()
         pullop.trmanager = transactionmanager(repo, 'pull', remote.url())
         streamclone.maybeperformlegacystreamclone(pullop)
         # This should ideally be in _pullbundle2(). However, it needs to run
         # before discovery to avoid extra work.
         _maybeapplyclonebundle(pullop)
         _pulldiscovery(pullop)
         if pullop.canusebundle2:
             _pullbundle2(pullop)
         _pullchangeset(pullop)
         _pullphase(pullop)
         _pullbookmarks(pullop)
         _pullobsolete(pullop)
         pullop.trmanager.close()
     finally:
         lockmod.release(pullop.trmanager, lock, wlock)
 
     return pullop
 
 # list of steps to perform discovery before pull
 pulldiscoveryorder = []
 
 # Mapping between step name and function
 #
 # This exists to help extensions wrap steps if necessary
 pulldiscoverymapping = {}
 
 def pulldiscovery(stepname):
     """decorator for function performing discovery before pull
 
     The function is added to the step -> function mapping and appended to the
     list of steps.  Beware that decorated function will be added in order (this
     may matter).
 
     You can only use this decorator for a new step, if you want to wrap a step
     from an extension, change the pulldiscovery dictionary directly."""
     def dec(func):
         assert stepname not in pulldiscoverymapping
         pulldiscoverymapping[stepname] = func
         pulldiscoveryorder.append(stepname)
         return func
     return dec
 
 def _pulldiscovery(pullop):
     """Run all discovery steps"""
     for stepname in pulldiscoveryorder:
         step = pulldiscoverymapping[stepname]
         step(pullop)
 
 @pulldiscovery('b1:bookmarks')
 def _pullbookmarkbundle1(pullop):
     """fetch bookmark data in bundle1 case
 
     If not using bundle2, we have to fetch bookmarks before changeset
     discovery to reduce the chance and impact of race conditions."""
     if pullop.remotebookmarks is not None:
         return
     if pullop.canusebundle2 and 'listkeys' in pullop.remotebundle2caps:
         # all known bundle2 servers now support listkeys, but lets be nice with
         # new implementation.
         return
     pullop.remotebookmarks = pullop.remote.listkeys('bookmarks')
 
 
 @pulldiscovery('changegroup')
 def _pulldiscoverychangegroup(pullop):
     """discovery phase for the pull
 
     Current handle changeset discovery only, will change handle all discovery
     at some point."""
     tmp = discovery.findcommonincoming(pullop.repo,
                                        pullop.remote,
                                        heads=pullop.heads,
                                        force=pullop.force)
     common, fetch, rheads = tmp
     nm = pullop.repo.unfiltered().changelog.nodemap
     if fetch and rheads:
         # If a remote heads in filtered locally, lets drop it from the unknown
         # remote heads and put in back in common.
         #
         # This is a hackish solution to catch most of "common but locally
         # hidden situation".  We do not performs discovery on unfiltered
         # repository because it end up doing a pathological amount of round
         # trip for w huge amount of changeset we do not care about.
         #
         # If a set of such "common but filtered" changeset exist on the server
         # but are not including a remote heads, we'll not be able to detect it,
         scommon = set(common)
         filteredrheads = []
         for n in rheads:
             if n in nm:
                 if n not in scommon:
                     common.append(n)
             else:
                 filteredrheads.append(n)
         if not filteredrheads:
             fetch = []
         rheads = filteredrheads
     pullop.common = common
     pullop.fetch = fetch
     pullop.rheads = rheads
 
 def _pullbundle2(pullop):
     """pull data using bundle2
 
     For now, the only supported data are changegroup."""
     kwargs = {'bundlecaps': caps20to10(pullop.repo)}
 
     streaming, streamreqs = streamclone.canperformstreamclone(pullop)
 
     # pulling changegroup
     pullop.stepsdone.add('changegroup')
 
     kwargs['common'] = pullop.common
     kwargs['heads'] = pullop.heads or pullop.rheads
     kwargs['cg'] = pullop.fetch
     if 'listkeys' in pullop.remotebundle2caps:
         kwargs['listkeys'] = ['phases']
         if pullop.remotebookmarks is None:
             # make sure to always includes bookmark data when migrating
             # `hg incoming --bundle` to using this function.
             kwargs['listkeys'].append('bookmarks')
 
     # If this is a full pull / clone and the server supports the clone bundles
     # feature, tell the server whether we attempted a clone bundle. The
     # presence of this flag indicates the client supports clone bundles. This
     # will enable the server to treat clients that support clone bundles
     # differently from those that don't.
     if (pullop.remote.capable('clonebundles')
         and pullop.heads is None and list(pullop.common) == [nullid]):
         kwargs['cbattempted'] = pullop.clonebundleattempted
 
     if streaming:
         pullop.repo.ui.status(_('streaming all changes\n'))
     elif not pullop.fetch:
         pullop.repo.ui.status(_("no changes found\n"))
         pullop.cgresult = 0
     else:
         if pullop.heads is None and list(pullop.common) == [nullid]:
             pullop.repo.ui.status(_("requesting all changes\n"))
     if obsolete.isenabled(pullop.repo, obsolete.exchangeopt):
         remoteversions = bundle2.obsmarkersversion(pullop.remotebundle2caps)
         if obsolete.commonversion(remoteversions) is not None:
             kwargs['obsmarkers'] = True
             pullop.stepsdone.add('obsmarkers')
     _pullbundle2extraprepare(pullop, kwargs)
     bundle = pullop.remote.getbundle('pull', **kwargs)
     try:
         op = bundle2.processbundle(pullop.repo, bundle, pullop.gettransaction)
     except error.BundleValueError as exc:
         raise error.Abort(_('missing support for %s') % exc)
 
     if pullop.fetch:
         results = [cg['return'] for cg in op.records['changegroup']]
         pullop.cgresult = changegroup.combineresults(results)
 
     # processing phases change
     for namespace, value in op.records['listkeys']:
         if namespace == 'phases':
             _pullapplyphases(pullop, value)
 
     # processing bookmark update
     for namespace, value in op.records['listkeys']:
         if namespace == 'bookmarks':
             pullop.remotebookmarks = value
 
     # bookmark data were either already there or pulled in the bundle
     if pullop.remotebookmarks is not None:
         _pullbookmarks(pullop)
 
 def _pullbundle2extraprepare(pullop, kwargs):
     """hook function so that extensions can extend the getbundle call"""
     pass
 
 def _pullchangeset(pullop):
     """pull changeset from unbundle into the local repo"""
     # We delay the open of the transaction as late as possible so we
     # don't open transaction for nothing or you break future useful
     # rollback call
     if 'changegroup' in pullop.stepsdone:
         return
     pullop.stepsdone.add('changegroup')
     if not pullop.fetch:
         pullop.repo.ui.status(_("no changes found\n"))
         pullop.cgresult = 0
         return
     pullop.gettransaction()
     if pullop.heads is None and list(pullop.common) == [nullid]:
         pullop.repo.ui.status(_("requesting all changes\n"))
     elif pullop.heads is None and pullop.remote.capable('changegroupsubset'):
         # issue1320, avoid a race if remote changed after discovery
         pullop.heads = pullop.rheads
 
     if pullop.remote.capable('getbundle'):
         # TODO: get bundlecaps from remote
         cg = pullop.remote.getbundle('pull', common=pullop.common,
                                      heads=pullop.heads or pullop.rheads)
     elif pullop.heads is None:
         cg = pullop.remote.changegroup(pullop.fetch, 'pull')
     elif not pullop.remote.capable('changegroupsubset'):
         raise error.Abort(_("partial pull cannot be done because "
                            "other repository doesn't support "
                            "changegroupsubset."))
     else:
         cg = pullop.remote.changegroupsubset(pullop.fetch, pullop.heads, 'pull')
     pullop.cgresult = cg.apply(pullop.repo, 'pull', pullop.remote.url())
 
 def _pullphase(pullop):
     # Get remote phases data from remote
     if 'phases' in pullop.stepsdone:
         return
     remotephases = pullop.remote.listkeys('phases')
     _pullapplyphases(pullop, remotephases)
 
 def _pullapplyphases(pullop, remotephases):
     """apply phase movement from observed remote state"""
     if 'phases' in pullop.stepsdone:
         return
     pullop.stepsdone.add('phases')
     publishing = bool(remotephases.get('publishing', False))
     if remotephases and not publishing:
         # remote is new and unpublishing
         pheads, _dr = phases.analyzeremotephases(pullop.repo,
                                                  pullop.pulledsubset,
                                                  remotephases)
         dheads = pullop.pulledsubset
     else:
         # Remote is old or publishing all common changesets
         # should be seen as public
         pheads = pullop.pulledsubset
         dheads = []
     unfi = pullop.repo.unfiltered()
     phase = unfi._phasecache.phase
     rev = unfi.changelog.nodemap.get
     public = phases.public
     draft = phases.draft
 
     # exclude changesets already public locally and update the others
     pheads = [pn for pn in pheads if phase(unfi, rev(pn)) > public]
     if pheads:
         tr = pullop.gettransaction()
         phases.advanceboundary(pullop.repo, tr, public, pheads)
 
     # exclude changesets already draft locally and update the others
     dheads = [pn for pn in dheads if phase(unfi, rev(pn)) > draft]
     if dheads:
         tr = pullop.gettransaction()
         phases.advanceboundary(pullop.repo, tr, draft, dheads)
 
 def _pullbookmarks(pullop):
     """process the remote bookmark information to update the local one"""
     if 'bookmarks' in pullop.stepsdone:
         return
     pullop.stepsdone.add('bookmarks')
     repo = pullop.repo
     remotebookmarks = pullop.remotebookmarks
     bookmod.updatefromremote(repo.ui, repo, remotebookmarks,
                              pullop.remote.url(),
                              pullop.gettransaction,
                              explicit=pullop.explicitbookmarks)
 
 def _pullobsolete(pullop):
     """utility function to pull obsolete markers from a remote
 
     The `gettransaction` is function that return the pull transaction, creating
     one if necessary. We return the transaction to inform the calling code that
     a new transaction have been created (when applicable).
 
     Exists mostly to allow overriding for experimentation purpose"""
     if 'obsmarkers' in pullop.stepsdone:
         return
     pullop.stepsdone.add('obsmarkers')
     tr = None
     if obsolete.isenabled(pullop.repo, obsolete.exchangeopt):
         pullop.repo.ui.debug('fetching remote obsolete markers\n')
         remoteobs = pullop.remote.listkeys('obsolete')
         if 'dump0' in remoteobs:
             tr = pullop.gettransaction()
             markers = []
             for key in sorted(remoteobs, reverse=True):
                 if key.startswith('dump'):
                     data = base85.b85decode(remoteobs[key])
                     version, newmarks = obsolete._readmarkers(data)
                     markers += newmarks
             if markers:
                 pullop.repo.obsstore.add(tr, markers)
             pullop.repo.invalidatevolatilesets()
     return tr
 
 def caps20to10(repo):
     """return a set with appropriate options to use bundle20 during getbundle"""
     caps = set(['HG20'])
     capsblob = bundle2.encodecaps(bundle2.getrepocaps(repo))
     caps.add('bundle2=' + urlreq.quote(capsblob))
     return caps
 
 # List of names of steps to perform for a bundle2 for getbundle, order matters.
 getbundle2partsorder = []
 
 # Mapping between step name and function
 #
 # This exists to help extensions wrap steps if necessary
 getbundle2partsmapping = {}
 
 def getbundle2partsgenerator(stepname, idx=None):
     """decorator for function generating bundle2 part for getbundle
 
     The function is added to the step -> function mapping and appended to the
     list of steps.  Beware that decorated functions will be added in order
     (this may matter).
 
     You can only use this decorator for new steps, if you want to wrap a step
     from an extension, attack the getbundle2partsmapping dictionary directly."""
     def dec(func):
         assert stepname not in getbundle2partsmapping
         getbundle2partsmapping[stepname] = func
         if idx is None:
             getbundle2partsorder.append(stepname)
         else:
             getbundle2partsorder.insert(idx, stepname)
         return func
     return dec
 
 def bundle2requested(bundlecaps):
     if bundlecaps is not None:
         return any(cap.startswith('HG2') for cap in bundlecaps)
     return False
 
-def getbundle(repo, source, heads=None, common=None, bundlecaps=None,
-              **kwargs):
-    """return a full bundle (with potentially multiple kind of parts)
+def getbundlechunks(repo, source, heads=None, common=None, bundlecaps=None,
+                    **kwargs):
+    """Return chunks constituting a bundle's raw data.
 
     Could be a bundle HG10 or a bundle HG20 depending on bundlecaps
-    passed. For now, the bundle can contain only changegroup, but this will
-    changes when more part type will be available for bundle2.
+    passed.
 
-    This is different from changegroup.getchangegroup that only returns an HG10
-    changegroup bundle. They may eventually get reunited in the future when we
-    have a clearer idea of the API we what to query different data.
-
-    The implementation is at a very early stage and will get massive rework
-    when the API of bundle is refined.
+    Returns an iterator over raw chunks (of varying sizes).
     """
     usebundle2 = bundle2requested(bundlecaps)
     # bundle10 case
     if not usebundle2:
         if bundlecaps and not kwargs.get('cg', True):
             raise ValueError(_('request for bundle10 must include changegroup'))
 
         if kwargs:
             raise ValueError(_('unsupported getbundle arguments: %s')
                              % ', '.join(sorted(kwargs.keys())))
         outgoing = _computeoutgoing(repo, heads, common)
-        return changegroup.getchangegroup(repo, source, outgoing,
-                                          bundlecaps=bundlecaps)
+        bundler = changegroup.getbundler('01', repo, bundlecaps)
+        return changegroup.getsubsetraw(repo, outgoing, bundler, source)
 
     # bundle20 case
     b2caps = {}
     for bcaps in bundlecaps:
         if bcaps.startswith('bundle2='):
             blob = urlreq.unquote(bcaps[len('bundle2='):])
             b2caps.update(bundle2.decodecaps(blob))
     bundler = bundle2.bundle20(repo.ui, b2caps)
 
     kwargs['heads'] = heads
     kwargs['common'] = common
 
     for name in getbundle2partsorder:
         func = getbundle2partsmapping[name]
         func(bundler, repo, source, bundlecaps=bundlecaps, b2caps=b2caps,
              **kwargs)
 
-    return util.chunkbuffer(bundler.getchunks())
+    return bundler.getchunks()
 
 @getbundle2partsgenerator('changegroup')
 def _getbundlechangegrouppart(bundler, repo, source, bundlecaps=None,
                               b2caps=None, heads=None, common=None, **kwargs):
     """add a changegroup part to the requested bundle"""
     cg = None
     if kwargs.get('cg', True):
         # build changegroup bundle here.
         version = '01'
         cgversions = b2caps.get('changegroup')
         if cgversions:  # 3.1 and 3.2 ship with an empty value
             cgversions = [v for v in cgversions
                           if v in changegroup.supportedoutgoingversions(repo)]
             if not cgversions:
                 raise ValueError(_('no common changegroup version'))
             version = max(cgversions)
         outgoing = _computeoutgoing(repo, heads, common)
         cg = changegroup.getlocalchangegroupraw(repo, source, outgoing,
                                                 bundlecaps=bundlecaps,
                                                 version=version)
 
     if cg:
         part = bundler.newpart('changegroup', data=cg)
         if cgversions:
             part.addparam('version', version)
         part.addparam('nbchanges', str(len(outgoing.missing)), mandatory=False)
         if 'treemanifest' in repo.requirements:
             part.addparam('treemanifest', '1')
 
 @getbundle2partsgenerator('listkeys')
 def _getbundlelistkeysparts(bundler, repo, source, bundlecaps=None,
                             b2caps=None, **kwargs):
     """add parts containing listkeys namespaces to the requested bundle"""
     listkeys = kwargs.get('listkeys', ())
     for namespace in listkeys:
         part = bundler.newpart('listkeys')
         part.addparam('namespace', namespace)
         keys = repo.listkeys(namespace).items()
         part.data = pushkey.encodekeys(keys)
 
 @getbundle2partsgenerator('obsmarkers')
 def _getbundleobsmarkerpart(bundler, repo, source, bundlecaps=None,
                             b2caps=None, heads=None, **kwargs):
     """add an obsolescence markers part to the requested bundle"""
     if kwargs.get('obsmarkers', False):
         if heads is None:
             heads = repo.heads()
         subset = [c.node() for c in repo.set('::%ln', heads)]
         markers = repo.obsstore.relevantmarkers(subset)
         markers = sorted(markers)
         buildobsmarkerspart(bundler, markers)
 
 @getbundle2partsgenerator('hgtagsfnodes')
 def _getbundletagsfnodes(bundler, repo, source, bundlecaps=None,
                          b2caps=None, heads=None, common=None,
                          **kwargs):
     """Transfer the .hgtags filenodes mapping.
 
     Only values for heads in this bundle will be transferred.
 
     The part data consists of pairs of 20 byte changeset node and .hgtags
     filenodes raw values.
     """
     # Don't send unless:
     # - changeset are being exchanged,
     # - the client supports it.
     if not (kwargs.get('cg', True) and 'hgtagsfnodes' in b2caps):
         return
 
     outgoing = _computeoutgoing(repo, heads, common)
 
     if not outgoing.missingheads:
         return
 
     cache = tags.hgtagsfnodescache(repo.unfiltered())
     chunks = []
 
     # .hgtags fnodes are only relevant for head changesets. While we could
     # transfer values for all known nodes, there will likely be little to
     # no benefit.
     #
     # We don't bother using a generator to produce output data because
     # a) we only have 40 bytes per head and even esoteric numbers of heads
     # consume little memory (1M heads is 40MB) b) we don't want to send the
     # part if we don't have entries and knowing if we have entries requires
     # cache lookups.
     for node in outgoing.missingheads:
         # Don't compute missing, as this may slow down serving.
         fnode = cache.getfnode(node, computemissing=False)
         if fnode is not None:
             chunks.extend([node, fnode])
 
     if chunks:
         bundler.newpart('hgtagsfnodes', data=''.join(chunks))
 
 def check_heads(repo, their_heads, context):
     """check if the heads of a repo have been modified
 
     Used by peer for unbundling.
     """
     heads = repo.heads()
     heads_hash = hashlib.sha1(''.join(sorted(heads))).digest()
     if not (their_heads == ['force'] or their_heads == heads or
             their_heads == ['hashed', heads_hash]):
         # someone else committed/pushed/unbundled while we
         # were transferring data
         raise error.PushRaced('repository changed while %s - '
                               'please try again' % context)
 
 def unbundle(repo, cg, heads, source, url):
     """Apply a bundle to a repo.
 
     this function makes sure the repo is locked during the application and have
     mechanism to check that no push race occurred between the creation of the
     bundle and its application.
 
     If the push was raced as PushRaced exception is raised."""
     r = 0
     # need a transaction when processing a bundle2 stream
     # [wlock, lock, tr] - needs to be an array so nested functions can modify it
     lockandtr = [None, None, None]
     recordout = None
     # quick fix for output mismatch with bundle2 in 3.4
     captureoutput = repo.ui.configbool('experimental', 'bundle2-output-capture',
                                        False)
     if url.startswith('remote:http:') or url.startswith('remote:https:'):
         captureoutput = True
     try:
         check_heads(repo, heads, 'uploading changes')
         # push can proceed
         if util.safehasattr(cg, 'params'):
             r = None
             try:
                 def gettransaction():
                     if not lockandtr[2]:
                         lockandtr[0] = repo.wlock()
                         lockandtr[1] = repo.lock()
                         lockandtr[2] = repo.transaction(source)
                         lockandtr[2].hookargs['source'] = source
                         lockandtr[2].hookargs['url'] = url
                         lockandtr[2].hookargs['bundle2'] = '1'
                     return lockandtr[2]
 
                 # Do greedy locking by default until we're satisfied with lazy
                 # locking.
                 if not repo.ui.configbool('experimental', 'bundle2lazylocking'):
                     gettransaction()
 
                 op = bundle2.bundleoperation(repo, gettransaction,
                                              captureoutput=captureoutput)
                 try:
                     op = bundle2.processbundle(repo, cg, op=op)
                 finally:
                     r = op.reply
                     if captureoutput and r is not None:
                         repo.ui.pushbuffer(error=True, subproc=True)
                         def recordout(output):
                             r.newpart('output', data=output, mandatory=False)
                 if lockandtr[2] is not None:
                     lockandtr[2].close()
             except BaseException as exc:
                 exc.duringunbundle2 = True
                 if captureoutput and r is not None:
                     parts = exc._bundle2salvagedoutput = r.salvageoutput()
                     def recordout(output):
                         part = bundle2.bundlepart('output', data=output,
                                                   mandatory=False)
                         parts.append(part)
                 raise
         else:
             lockandtr[1] = repo.lock()
             r = cg.apply(repo, source, url)
     finally:
         lockmod.release(lockandtr[2], lockandtr[1], lockandtr[0])
         if recordout is not None:
             recordout(repo.ui.popbuffer())
     return r
 
 def _maybeapplyclonebundle(pullop):
     """Apply a clone bundle from a remote, if possible."""
 
     repo = pullop.repo
     remote = pullop.remote
 
     if not repo.ui.configbool('ui', 'clonebundles', True):
         return
 
     # Only run if local repo is empty.
     if len(repo):
         return
 
     if pullop.heads:
         return
 
     if not remote.capable('clonebundles'):
         return
 
     res = remote._call('clonebundles')
 
     # If we call the wire protocol command, that's good enough to record the
     # attempt.
     pullop.clonebundleattempted = True
 
     entries = parseclonebundlesmanifest(repo, res)
     if not entries:
         repo.ui.note(_('no clone bundles available on remote; '
                        'falling back to regular clone\n'))
         return
 
     entries = filterclonebundleentries(repo, entries)
     if not entries:
         # There is a thundering herd concern here. However, if a server
         # operator doesn't advertise bundles appropriate for its clients,
         # they deserve what's coming. Furthermore, from a client's
         # perspective, no automatic fallback would mean not being able to
         # clone!
         repo.ui.warn(_('no compatible clone bundles available on server; '
                        'falling back to regular clone\n'))
         repo.ui.warn(_('(you may want to report this to the server '
                        'operator)\n'))
         return
 
     entries = sortclonebundleentries(repo.ui, entries)
 
     url = entries[0]['URL']
     repo.ui.status(_('applying clone bundle from %s\n') % url)
     if trypullbundlefromurl(repo.ui, repo, url):
         repo.ui.status(_('finished applying clone bundle\n'))
     # Bundle failed.
     #
     # We abort by default to avoid the thundering herd of
     # clients flooding a server that was expecting expensive
     # clone load to be offloaded.
     elif repo.ui.configbool('ui', 'clonebundlefallback', False):
         repo.ui.warn(_('falling back to normal clone\n'))
     else:
         raise error.Abort(_('error applying bundle'),
                           hint=_('if this error persists, consider contacting '
                                  'the server operator or disable clone '
                                  'bundles via '
                                  '"--config ui.clonebundles=false"'))
 
 def parseclonebundlesmanifest(repo, s):
     """Parses the raw text of a clone bundles manifest.
 
     Returns a list of dicts. The dicts have a ``URL`` key corresponding
     to the URL and other keys are the attributes for the entry.
     """
     m = []
     for line in s.splitlines():
         fields = line.split()
         if not fields:
             continue
         attrs = {'URL': fields[0]}
         for rawattr in fields[1:]:
             key, value = rawattr.split('=', 1)
             key = urlreq.unquote(key)
             value = urlreq.unquote(value)
             attrs[key] = value
 
             # Parse BUNDLESPEC into components. This makes client-side
             # preferences easier to specify since you can prefer a single
             # component of the BUNDLESPEC.
             if key == 'BUNDLESPEC':
                 try:
                     comp, version, params = parsebundlespec(repo, value,
                                                             externalnames=True)
                     attrs['COMPRESSION'] = comp
                     attrs['VERSION'] = version
                 except error.InvalidBundleSpecification:
                     pass
                 except error.UnsupportedBundleSpecification:
                     pass
 
         m.append(attrs)
 
     return m
 
 def filterclonebundleentries(repo, entries):
     """Remove incompatible clone bundle manifest entries.
 
     Accepts a list of entries parsed with ``parseclonebundlesmanifest``
     and returns a new list consisting of only the entries that this client
     should be able to apply.
 
     There is no guarantee we'll be able to apply all returned entries because
     the metadata we use to filter on may be missing or wrong.
     """
     newentries = []
     for entry in entries:
         spec = entry.get('BUNDLESPEC')
         if spec:
             try:
                 parsebundlespec(repo, spec, strict=True)
             except error.InvalidBundleSpecification as e:
                 repo.ui.debug(str(e) + '\n')
                 continue
             except error.UnsupportedBundleSpecification as e:
                 repo.ui.debug('filtering %s because unsupported bundle '
                               'spec: %s\n' % (entry['URL'], str(e)))
                 continue
 
         if 'REQUIRESNI' in entry and not sslutil.hassni:
             repo.ui.debug('filtering %s because SNI not supported\n' %
                           entry['URL'])
             continue
 
         newentries.append(entry)
 
     return newentries
 
 def sortclonebundleentries(ui, entries):
     prefers = ui.configlist('ui', 'clonebundleprefers', default=[])
     if not prefers:
         return list(entries)
 
     prefers = [p.split('=', 1) for p in prefers]
 
     # Our sort function.
     def compareentry(a, b):
         for prefkey, prefvalue in prefers:
             avalue = a.get(prefkey)
             bvalue = b.get(prefkey)
 
             # Special case for b missing attribute and a matches exactly.
             if avalue is not None and bvalue is None and avalue == prefvalue:
                 return -1
 
             # Special case for a missing attribute and b matches exactly.
             if bvalue is not None and avalue is None and bvalue == prefvalue:
                 return 1
 
             # We can't compare unless attribute present on both.
             if avalue is None or bvalue is None:
                 continue
 
             # Same values should fall back to next attribute.
             if avalue == bvalue:
                 continue
 
             # Exact matches come first.
             if avalue == prefvalue:
                 return -1
             if bvalue == prefvalue:
                 return 1
 
             # Fall back to next attribute.
             continue
 
         # If we got here we couldn't sort by attributes and prefers. Fall
         # back to index order.
         return 0
 
     return sorted(entries, cmp=compareentry)
 
 def trypullbundlefromurl(ui, repo, url):
     """Attempt to apply a bundle from a URL."""
     lock = repo.lock()
     try:
         tr = repo.transaction('bundleurl')
         try:
             try:
                 fh = urlmod.open(ui, url)
                 cg = readbundle(ui, fh, 'stream')
 
                 if isinstance(cg, bundle2.unbundle20):
                     bundle2.processbundle(repo, cg, lambda: tr)
                 elif isinstance(cg, streamclone.streamcloneapplier):
                     cg.apply(repo)
                 else:
                     cg.apply(repo, 'clonebundles', url)
                 tr.close()
                 return True
             except urlerr.httperror as e:
                 ui.warn(_('HTTP error fetching bundle: %s\n') % str(e))
             except urlerr.urlerror as e:
                 ui.warn(_('error fetching bundle: %s\n') % e.reason[1])
 
             return False
         finally:
             tr.release()
     finally:
         lock.release()

mercurial/localrepo.py

@@ -1,1996 +1,2000
 # localrepo.py - read/write repository class 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.
 
 from __future__ import absolute_import
 
 import errno
 import hashlib
 import inspect
 import os
 import random
 import time
 import weakref
 
 from .i18n import _
 from .node import (
     hex,
     nullid,
     short,
     wdirrev,
 )
 from . import (
     bookmarks,
     branchmap,
     bundle2,
     changegroup,
     changelog,
     cmdutil,
     context,
     dirstate,
     encoding,
     error,
     exchange,
     extensions,
     filelog,
     hook,
     lock as lockmod,
     manifest,
     match as matchmod,
     merge as mergemod,
     namespaces,
     obsolete,
     pathutil,
     peer,
     phases,
     pushkey,
     repoview,
     revset,
     scmutil,
     store,
     subrepo,
     tags as tagsmod,
     transaction,
     util,
 )
 
 release = lockmod.release
 urlerr = util.urlerr
 urlreq = util.urlreq
 
 class repofilecache(scmutil.filecache):
     """All filecache usage on repo are done for logic that should be unfiltered
     """
 
     def __get__(self, repo, type=None):
         if repo is None:
             return self
         return super(repofilecache, self).__get__(repo.unfiltered(), type)
     def __set__(self, repo, value):
         return super(repofilecache, self).__set__(repo.unfiltered(), value)
     def __delete__(self, repo):
         return super(repofilecache, self).__delete__(repo.unfiltered())
 
 class storecache(repofilecache):
     """filecache for files in the store"""
     def join(self, obj, fname):
         return obj.sjoin(fname)
 
 class unfilteredpropertycache(util.propertycache):
     """propertycache that apply to unfiltered repo only"""
 
     def __get__(self, repo, type=None):
         unfi = repo.unfiltered()
         if unfi is repo:
             return super(unfilteredpropertycache, self).__get__(unfi)
         return getattr(unfi, self.name)
 
 class filteredpropertycache(util.propertycache):
     """propertycache that must take filtering in account"""
 
     def cachevalue(self, obj, value):
         object.__setattr__(obj, self.name, value)
 
 
 def hasunfilteredcache(repo, name):
     """check if a repo has an unfilteredpropertycache value for <name>"""
     return name in vars(repo.unfiltered())
 
 def unfilteredmethod(orig):
     """decorate method that always need to be run on unfiltered version"""
     def wrapper(repo, *args, **kwargs):
         return orig(repo.unfiltered(), *args, **kwargs)
     return wrapper
 
 moderncaps = set(('lookup', 'branchmap', 'pushkey', 'known', 'getbundle',
                   'unbundle'))
 legacycaps = moderncaps.union(set(['changegroupsubset']))
 
 class localpeer(peer.peerrepository):
     '''peer for a local repo; reflects only the most recent API'''
 
     def __init__(self, repo, caps=moderncaps):
         peer.peerrepository.__init__(self)
         self._repo = repo.filtered('served')
         self.ui = repo.ui
         self._caps = repo._restrictcapabilities(caps)
         self.requirements = repo.requirements
         self.supportedformats = repo.supportedformats
 
     def close(self):
         self._repo.close()
 
     def _capabilities(self):
         return self._caps
 
     def local(self):
         return self._repo
 
     def canpush(self):
         return True
 
     def url(self):
         return self._repo.url()
 
     def lookup(self, key):
         return self._repo.lookup(key)
 
     def branchmap(self):
         return self._repo.branchmap()
 
     def heads(self):
         return self._repo.heads()
 
     def known(self, nodes):
         return self._repo.known(nodes)
 
     def getbundle(self, source, heads=None, common=None, bundlecaps=None,
                   **kwargs):
-        cg = exchange.getbundle(self._repo, source, heads=heads,
-                                common=common, bundlecaps=bundlecaps, **kwargs)
+        chunks = exchange.getbundlechunks(self._repo, source, heads=heads,
+                                          common=common, bundlecaps=bundlecaps,
+                                          **kwargs)
+        cb = util.chunkbuffer(chunks)
+
         if bundlecaps is not None and 'HG20' in bundlecaps:
             # When requesting a bundle2, getbundle returns a stream to make the
             # wire level function happier. We need to build a proper object
             # from it in local peer.
-            cg = bundle2.getunbundler(self.ui, cg)
-        return cg
+            return bundle2.getunbundler(self.ui, cb)
+        else:
+            return changegroup.getunbundler('01', cb, None)
 
     # TODO We might want to move the next two calls into legacypeer and add
     # unbundle instead.
 
     def unbundle(self, cg, heads, url):
         """apply a bundle on a repo
 
         This function handles the repo locking itself."""
         try:
             try:
                 cg = exchange.readbundle(self.ui, cg, None)
                 ret = exchange.unbundle(self._repo, cg, heads, 'push', url)
                 if util.safehasattr(ret, 'getchunks'):
                     # This is a bundle20 object, turn it into an unbundler.
                     # This little dance should be dropped eventually when the
                     # API is finally improved.
                     stream = util.chunkbuffer(ret.getchunks())
                     ret = bundle2.getunbundler(self.ui, stream)
                 return ret
             except Exception as exc:
                 # If the exception contains output salvaged from a bundle2
                 # reply, we need to make sure it is printed before continuing
                 # to fail. So we build a bundle2 with such output and consume
                 # it directly.
                 #
                 # This is not very elegant but allows a "simple" solution for
                 # issue4594
                 output = getattr(exc, '_bundle2salvagedoutput', ())
                 if output:
                     bundler = bundle2.bundle20(self._repo.ui)
                     for out in output:
                         bundler.addpart(out)
                     stream = util.chunkbuffer(bundler.getchunks())
                     b = bundle2.getunbundler(self.ui, stream)
                     bundle2.processbundle(self._repo, b)
                 raise
         except error.PushRaced as exc:
             raise error.ResponseError(_('push failed:'), str(exc))
 
     def lock(self):
         return self._repo.lock()
 
     def addchangegroup(self, cg, source, url):
         return cg.apply(self._repo, source, url)
 
     def pushkey(self, namespace, key, old, new):
         return self._repo.pushkey(namespace, key, old, new)
 
     def listkeys(self, namespace):
         return self._repo.listkeys(namespace)
 
     def debugwireargs(self, one, two, three=None, four=None, five=None):
         '''used to test argument passing over the wire'''
         return "%s %s %s %s %s" % (one, two, three, four, five)
 
 class locallegacypeer(localpeer):
     '''peer extension which implements legacy methods too; used for tests with
     restricted capabilities'''
 
     def __init__(self, repo):
         localpeer.__init__(self, repo, caps=legacycaps)
 
     def branches(self, nodes):
         return self._repo.branches(nodes)
 
     def between(self, pairs):
         return self._repo.between(pairs)
 
     def changegroup(self, basenodes, source):
         return changegroup.changegroup(self._repo, basenodes, source)
 
     def changegroupsubset(self, bases, heads, source):
         return changegroup.changegroupsubset(self._repo, bases, heads, source)
 
 class localrepository(object):
 
     supportedformats = set(('revlogv1', 'generaldelta', 'treemanifest',
                             'manifestv2'))
     _basesupported = supportedformats | set(('store', 'fncache', 'shared',
                                              'dotencode'))
     openerreqs = set(('revlogv1', 'generaldelta', 'treemanifest', 'manifestv2'))
     filtername = None
 
     # a list of (ui, featureset) functions.
     # only functions defined in module of enabled extensions are invoked
     featuresetupfuncs = set()
 
     def __init__(self, baseui, path=None, create=False):
         self.requirements = set()
         self.wvfs = scmutil.vfs(path, expandpath=True, realpath=True)
         self.wopener = self.wvfs
         self.root = self.wvfs.base
         self.path = self.wvfs.join(".hg")
         self.origroot = path
         self.auditor = pathutil.pathauditor(self.root, self._checknested)
         self.nofsauditor = pathutil.pathauditor(self.root, self._checknested,
                                                 realfs=False)
         self.vfs = scmutil.vfs(self.path)
         self.opener = self.vfs
         self.baseui = baseui
         self.ui = baseui.copy()
         self.ui.copy = baseui.copy # prevent copying repo configuration
         # A list of callback to shape the phase if no data were found.
         # Callback are in the form: func(repo, roots) --> processed root.
         # This list it to be filled by extension during repo setup
         self._phasedefaults = []
         try:
             self.ui.readconfig(self.join("hgrc"), self.root)
             extensions.loadall(self.ui)
         except IOError:
             pass
 
         if self.featuresetupfuncs:
             self.supported = set(self._basesupported) # use private copy
             extmods = set(m.__name__ for n, m
                           in extensions.extensions(self.ui))
             for setupfunc in self.featuresetupfuncs:
                 if setupfunc.__module__ in extmods:
                     setupfunc(self.ui, self.supported)
         else:
             self.supported = self._basesupported
 
         if not self.vfs.isdir():
             if create:
                 self.requirements = newreporequirements(self)
 
                 if not self.wvfs.exists():
                     self.wvfs.makedirs()
                 self.vfs.makedir(notindexed=True)
 
                 if 'store' in self.requirements:
                     self.vfs.mkdir("store")
 
                     # create an invalid changelog
                     self.vfs.append(
                         "00changelog.i",
                         '\0\0\0\2' # represents revlogv2
                         ' dummy changelog to prevent using the old repo layout'
                     )
             else:
                 raise error.RepoError(_("repository %s not found") % path)
         elif create:
             raise error.RepoError(_("repository %s already exists") % path)
         else:
             try:
                 self.requirements = scmutil.readrequires(
                         self.vfs, self.supported)
             except IOError as inst:
                 if inst.errno != errno.ENOENT:
                     raise
 
         self.sharedpath = self.path
         try:
             vfs = scmutil.vfs(self.vfs.read("sharedpath").rstrip('\n'),
                               realpath=True)
             s = vfs.base
             if not vfs.exists():
                 raise error.RepoError(
                     _('.hg/sharedpath points to nonexistent directory %s') % s)
             self.sharedpath = s
         except IOError as inst:
             if inst.errno != errno.ENOENT:
                 raise
 
         self.store = store.store(
                 self.requirements, self.sharedpath, scmutil.vfs)
         self.spath = self.store.path
         self.svfs = self.store.vfs
         self.sjoin = self.store.join
         self.vfs.createmode = self.store.createmode
         self._applyopenerreqs()
         if create:
             self._writerequirements()
 
         self._dirstatevalidatewarned = False
 
         self._branchcaches = {}
         self._revbranchcache = None
         self.filterpats = {}
         self._datafilters = {}
         self._transref = self._lockref = self._wlockref = None
 
         # A cache for various files under .hg/ that tracks file changes,
         # (used by the filecache decorator)
         #
         # Maps a property name to its util.filecacheentry
         self._filecache = {}
 
         # hold sets of revision to be filtered
         # should be cleared when something might have changed the filter value:
         # - new changesets,
         # - phase change,
         # - new obsolescence marker,
         # - working directory parent change,
         # - bookmark changes
         self.filteredrevcache = {}
 
         # generic mapping between names and nodes
         self.names = namespaces.namespaces()
 
     def close(self):
         self._writecaches()
 
     def _writecaches(self):
         if self._revbranchcache:
             self._revbranchcache.write()
 
     def _restrictcapabilities(self, caps):
         if self.ui.configbool('experimental', 'bundle2-advertise', True):
             caps = set(caps)
             capsblob = bundle2.encodecaps(bundle2.getrepocaps(self))
             caps.add('bundle2=' + urlreq.quote(capsblob))
         return caps
 
     def _applyopenerreqs(self):
         self.svfs.options = dict((r, 1) for r in self.requirements
                                            if r in self.openerreqs)
         # experimental config: format.chunkcachesize
         chunkcachesize = self.ui.configint('format', 'chunkcachesize')
         if chunkcachesize is not None:
             self.svfs.options['chunkcachesize'] = chunkcachesize
         # experimental config: format.maxchainlen
         maxchainlen = self.ui.configint('format', 'maxchainlen')
         if maxchainlen is not None:
             self.svfs.options['maxchainlen'] = maxchainlen
         # experimental config: format.manifestcachesize
         manifestcachesize = self.ui.configint('format', 'manifestcachesize')
         if manifestcachesize is not None:
             self.svfs.options['manifestcachesize'] = manifestcachesize
         # experimental config: format.aggressivemergedeltas
         aggressivemergedeltas = self.ui.configbool('format',
             'aggressivemergedeltas', False)
         self.svfs.options['aggressivemergedeltas'] = aggressivemergedeltas
         self.svfs.options['lazydeltabase'] = not scmutil.gddeltaconfig(self.ui)
 
     def _writerequirements(self):
         scmutil.writerequires(self.vfs, self.requirements)
 
     def _checknested(self, path):
         """Determine if path is a legal nested repository."""
         if not path.startswith(self.root):
             return False
         subpath = path[len(self.root) + 1:]
         normsubpath = util.pconvert(subpath)
 
         # XXX: Checking against the current working copy is wrong in
         # the sense that it can reject things like
         #
         #   $ hg cat -r 10 sub/x.txt
         #
         # if sub/ is no longer a subrepository in the working copy
         # parent revision.
         #
         # However, it can of course also allow things that would have
         # been rejected before, such as the above cat command if sub/
         # is a subrepository now, but was a normal directory before.
         # The old path auditor would have rejected by mistake since it
         # panics when it sees sub/.hg/.
         #
         # All in all, checking against the working copy seems sensible
         # since we want to prevent access to nested repositories on
         # the filesystem *now*.
         ctx = self[None]
         parts = util.splitpath(subpath)
         while parts:
             prefix = '/'.join(parts)
             if prefix in ctx.substate:
                 if prefix == normsubpath:
                     return True
                 else:
                     sub = ctx.sub(prefix)
                     return sub.checknested(subpath[len(prefix) + 1:])
             else:
                 parts.pop()
         return False
 
     def peer(self):
         return localpeer(self) # not cached to avoid reference cycle
 
     def unfiltered(self):
         """Return unfiltered version of the repository
 
         Intended to be overwritten by filtered repo."""
         return self
 
     def filtered(self, name):
         """Return a filtered version of a repository"""
         # build a new class with the mixin and the current class
         # (possibly subclass of the repo)
         class proxycls(repoview.repoview, self.unfiltered().__class__):
             pass
         return proxycls(self, name)
 
     @repofilecache('bookmarks', 'bookmarks.current')
     def _bookmarks(self):
         return bookmarks.bmstore(self)
 
     @property
     def _activebookmark(self):
         return self._bookmarks.active
 
     def bookmarkheads(self, bookmark):
         name = bookmark.split('@', 1)[0]
         heads = []
         for mark, n in self._bookmarks.iteritems():
             if mark.split('@', 1)[0] == name:
                 heads.append(n)
         return heads
 
     # _phaserevs and _phasesets depend on changelog. what we need is to
     # call _phasecache.invalidate() if '00changelog.i' was changed, but it
     # can't be easily expressed in filecache mechanism.
     @storecache('phaseroots', '00changelog.i')
     def _phasecache(self):
         return phases.phasecache(self, self._phasedefaults)
 
     @storecache('obsstore')
     def obsstore(self):
         # read default format for new obsstore.
         # developer config: format.obsstore-version
         defaultformat = self.ui.configint('format', 'obsstore-version', None)
         # rely on obsstore class default when possible.
         kwargs = {}
         if defaultformat is not None:
             kwargs['defaultformat'] = defaultformat
         readonly = not obsolete.isenabled(self, obsolete.createmarkersopt)
         store = obsolete.obsstore(self.svfs, readonly=readonly,
                                   **kwargs)
         if store and readonly:
             self.ui.warn(
                 _('obsolete feature not enabled but %i markers found!\n')
                 % len(list(store)))
         return store
 
     @storecache('00changelog.i')
     def changelog(self):
         c = changelog.changelog(self.svfs)
         if 'HG_PENDING' in os.environ:
             p = os.environ['HG_PENDING']
             if p.startswith(self.root):
                 c.readpending('00changelog.i.a')
         return c
 
     @storecache('00manifest.i')
     def manifest(self):
         return manifest.manifest(self.svfs)
 
     @property
     def manifestlog(self):
         return manifest.manifestlog(self.svfs, self)
 
     @repofilecache('dirstate')
     def dirstate(self):
         return dirstate.dirstate(self.vfs, self.ui, self.root,
                                  self._dirstatevalidate)
 
     def _dirstatevalidate(self, node):
         try:
             self.changelog.rev(node)
             return node
         except error.LookupError:
             if not self._dirstatevalidatewarned:
                 self._dirstatevalidatewarned = True
                 self.ui.warn(_("warning: ignoring unknown"
                                " working parent %s!\n") % short(node))
             return nullid
 
     def __getitem__(self, changeid):
         if changeid is None or changeid == wdirrev:
             return context.workingctx(self)
         if isinstance(changeid, slice):
             return [context.changectx(self, i)
                     for i in xrange(*changeid.indices(len(self)))
                     if i not in self.changelog.filteredrevs]
         return context.changectx(self, changeid)
 
     def __contains__(self, changeid):
         try:
             self[changeid]
             return True
         except error.RepoLookupError:
             return False
 
     def __nonzero__(self):
         return True
 
     def __len__(self):
         return len(self.changelog)
 
     def __iter__(self):
         return iter(self.changelog)
 
     def revs(self, expr, *args):
         '''Find revisions matching a revset.
 
         The revset is specified as a string ``expr`` that may contain
         %-formatting to escape certain types. See ``revset.formatspec``.
 
         Revset aliases from the configuration are not expanded. To expand
         user aliases, consider calling ``scmutil.revrange()``.
 
         Returns a revset.abstractsmartset, which is a list-like interface
         that contains integer revisions.
         '''
         expr = revset.formatspec(expr, *args)
         m = revset.match(None, expr)
         return m(self)
 
     def set(self, expr, *args):
         '''Find revisions matching a revset and emit changectx instances.
 
         This is a convenience wrapper around ``revs()`` that iterates the
         result and is a generator of changectx instances.
 
         Revset aliases from the configuration are not expanded. To expand
         user aliases, consider calling ``scmutil.revrange()``.
         '''
         for r in self.revs(expr, *args):
             yield self[r]
 
     def url(self):
         return 'file:' + self.root
 
     def hook(self, name, throw=False, **args):
         """Call a hook, passing this repo instance.
 
         This a convenience method to aid invoking hooks. Extensions likely
         won't call this unless they have registered a custom hook or are
         replacing code that is expected to call a hook.
         """
         return hook.hook(self.ui, self, name, throw, **args)
 
     @unfilteredmethod
     def _tag(self, names, node, message, local, user, date, extra=None,
              editor=False):
         if isinstance(names, str):
             names = (names,)
 
         branches = self.branchmap()
         for name in names:
             self.hook('pretag', throw=True, node=hex(node), tag=name,
                       local=local)
             if name in branches:
                 self.ui.warn(_("warning: tag %s conflicts with existing"
                 " branch name\n") % name)
 
         def writetags(fp, names, munge, prevtags):
             fp.seek(0, 2)
             if prevtags and prevtags[-1] != '\n':
                 fp.write('\n')
             for name in names:
                 if munge:
                     m = munge(name)
                 else:
                     m = name
 
                 if (self._tagscache.tagtypes and
                     name in self._tagscache.tagtypes):
                     old = self.tags().get(name, nullid)
                     fp.write('%s %s\n' % (hex(old), m))
                 fp.write('%s %s\n' % (hex(node), m))
             fp.close()
 
         prevtags = ''
         if local:
             try:
                 fp = self.vfs('localtags', 'r+')
             except IOError:
                 fp = self.vfs('localtags', 'a')
             else:
                 prevtags = fp.read()
 
             # local tags are stored in the current charset
             writetags(fp, names, None, prevtags)
             for name in names:
                 self.hook('tag', node=hex(node), tag=name, local=local)
             return
 
         try:
             fp = self.wfile('.hgtags', 'rb+')
         except IOError as e:
             if e.errno != errno.ENOENT:
                 raise
             fp = self.wfile('.hgtags', 'ab')
         else:
             prevtags = fp.read()
 
         # committed tags are stored in UTF-8
         writetags(fp, names, encoding.fromlocal, prevtags)
 
         fp.close()
 
         self.invalidatecaches()
 
         if '.hgtags' not in self.dirstate:
             self[None].add(['.hgtags'])
 
         m = matchmod.exact(self.root, '', ['.hgtags'])
         tagnode = self.commit(message, user, date, extra=extra, match=m,
                               editor=editor)
 
         for name in names:
             self.hook('tag', node=hex(node), tag=name, local=local)
 
         return tagnode
 
     def tag(self, names, node, message, local, user, date, editor=False):
         '''tag a revision with one or more symbolic names.
 
         names is a list of strings or, when adding a single tag, names may be a
         string.
 
         if local is True, the tags are stored in a per-repository file.
         otherwise, they are stored in the .hgtags file, and a new
         changeset is committed with the change.
 
         keyword arguments:
 
         local: whether to store tags in non-version-controlled file
         (default False)
 
         message: commit message to use if committing
 
         user: name of user to use if committing
 
         date: date tuple to use if committing'''
 
         if not local:
             m = matchmod.exact(self.root, '', ['.hgtags'])
             if any(self.status(match=m, unknown=True, ignored=True)):
                 raise error.Abort(_('working copy of .hgtags is changed'),
                                  hint=_('please commit .hgtags manually'))
 
         self.tags() # instantiate the cache
         self._tag(names, node, message, local, user, date, editor=editor)
 
     @filteredpropertycache
     def _tagscache(self):
         '''Returns a tagscache object that contains various tags related
         caches.'''
 
         # This simplifies its cache management by having one decorated
         # function (this one) and the rest simply fetch things from it.
         class tagscache(object):
             def __init__(self):
                 # These two define the set of tags for this repository. tags
                 # maps tag name to node; tagtypes maps tag name to 'global' or
                 # 'local'. (Global tags are defined by .hgtags across all
                 # heads, and local tags are defined in .hg/localtags.)
                 # They constitute the in-memory cache of tags.
                 self.tags = self.tagtypes = None
 
                 self.nodetagscache = self.tagslist = None
 
         cache = tagscache()
         cache.tags, cache.tagtypes = self._findtags()
 
         return cache
 
     def tags(self):
         '''return a mapping of tag to node'''
         t = {}
         if self.changelog.filteredrevs:
             tags, tt = self._findtags()
         else:
             tags = self._tagscache.tags
         for k, v in tags.iteritems():
             try:
                 # ignore tags to unknown nodes
                 self.changelog.rev(v)
                 t[k] = v
             except (error.LookupError, ValueError):
                 pass
         return t
 
     def _findtags(self):
         '''Do the hard work of finding tags.  Return a pair of dicts
         (tags, tagtypes) where tags maps tag name to node, and tagtypes
         maps tag name to a string like \'global\' or \'local\'.
         Subclasses or extensions are free to add their own tags, but
         should be aware that the returned dicts will be retained for the
         duration of the localrepo object.'''
 
         # XXX what tagtype should subclasses/extensions use?  Currently
         # mq and bookmarks add tags, but do not set the tagtype at all.
         # Should each extension invent its own tag type?  Should there
         # be one tagtype for all such "virtual" tags?  Or is the status
         # quo fine?
 
         alltags = {}                    # map tag name to (node, hist)
         tagtypes = {}
 
         tagsmod.findglobaltags(self.ui, self, alltags, tagtypes)
         tagsmod.readlocaltags(self.ui, self, alltags, tagtypes)
 
         # Build the return dicts.  Have to re-encode tag names because
         # the tags module always uses UTF-8 (in order not to lose info
         # writing to the cache), but the rest of Mercurial wants them in
         # local encoding.
         tags = {}
         for (name, (node, hist)) in alltags.iteritems():
             if node != nullid:
                 tags[encoding.tolocal(name)] = node
         tags['tip'] = self.changelog.tip()
         tagtypes = dict([(encoding.tolocal(name), value)
                          for (name, value) in tagtypes.iteritems()])
         return (tags, tagtypes)
 
     def tagtype(self, tagname):
         '''
         return the type of the given tag. result can be:
 
         'local'  : a local tag
         'global' : a global tag
         None     : tag does not exist
         '''
 
         return self._tagscache.tagtypes.get(tagname)
 
     def tagslist(self):
         '''return a list of tags ordered by revision'''
         if not self._tagscache.tagslist:
             l = []
             for t, n in self.tags().iteritems():
                 l.append((self.changelog.rev(n), t, n))
             self._tagscache.tagslist = [(t, n) for r, t, n in sorted(l)]
 
         return self._tagscache.tagslist
 
     def nodetags(self, node):
         '''return the tags associated with a node'''
         if not self._tagscache.nodetagscache:
             nodetagscache = {}
             for t, n in self._tagscache.tags.iteritems():
                 nodetagscache.setdefault(n, []).append(t)
             for tags in nodetagscache.itervalues():
                 tags.sort()
             self._tagscache.nodetagscache = nodetagscache
         return self._tagscache.nodetagscache.get(node, [])
 
     def nodebookmarks(self, node):
         """return the list of bookmarks pointing to the specified node"""
         marks = []
         for bookmark, n in self._bookmarks.iteritems():
             if n == node:
                 marks.append(bookmark)
         return sorted(marks)
 
     def branchmap(self):
         '''returns a dictionary {branch: [branchheads]} with branchheads
         ordered by increasing revision number'''
         branchmap.updatecache(self)
         return self._branchcaches[self.filtername]
 
     @unfilteredmethod
     def revbranchcache(self):
         if not self._revbranchcache:
             self._revbranchcache = branchmap.revbranchcache(self.unfiltered())
         return self._revbranchcache
 
     def branchtip(self, branch, ignoremissing=False):
         '''return the tip node for a given branch
 
         If ignoremissing is True, then this method will not raise an error.
         This is helpful for callers that only expect None for a missing branch
         (e.g. namespace).
 
         '''
         try:
             return self.branchmap().branchtip(branch)
         except KeyError:
             if not ignoremissing:
                 raise error.RepoLookupError(_("unknown branch '%s'") % branch)
             else:
                 pass
 
     def lookup(self, key):
         return self[key].node()
 
     def lookupbranch(self, key, remote=None):
         repo = remote or self
         if key in repo.branchmap():
             return key
 
         repo = (remote and remote.local()) and remote or self
         return repo[key].branch()
 
     def known(self, nodes):
         cl = self.changelog
         nm = cl.nodemap
         filtered = cl.filteredrevs
         result = []
         for n in nodes:
             r = nm.get(n)
             resp = not (r is None or r in filtered)
             result.append(resp)
         return result
 
     def local(self):
         return self
 
     def publishing(self):
         # it's safe (and desirable) to trust the publish flag unconditionally
         # so that we don't finalize changes shared between users via ssh or nfs
         return self.ui.configbool('phases', 'publish', True, untrusted=True)
 
     def cancopy(self):
         # so statichttprepo's override of local() works
         if not self.local():
             return False
         if not self.publishing():
             return True
         # if publishing we can't copy if there is filtered content
         return not self.filtered('visible').changelog.filteredrevs
 
     def shared(self):
         '''the type of shared repository (None if not shared)'''
         if self.sharedpath != self.path:
             return 'store'
         return None
 
     def join(self, f, *insidef):
         return self.vfs.join(os.path.join(f, *insidef))
 
     def wjoin(self, f, *insidef):
         return self.vfs.reljoin(self.root, f, *insidef)
 
     def file(self, f):
         if f[0] == '/':
             f = f[1:]
         return filelog.filelog(self.svfs, f)
 
     def changectx(self, changeid):
         return self[changeid]
 
     def setparents(self, p1, p2=nullid):
         self.dirstate.beginparentchange()
         copies = self.dirstate.setparents(p1, p2)
         pctx = self[p1]
         if copies:
             # Adjust copy records, the dirstate cannot do it, it
             # requires access to parents manifests. Preserve them
             # only for entries added to first parent.
             for f in copies:
                 if f not in pctx and copies[f] in pctx:
                     self.dirstate.copy(copies[f], f)
         if p2 == nullid:
             for f, s in sorted(self.dirstate.copies().items()):
                 if f not in pctx and s not in pctx:
                     self.dirstate.copy(None, f)
         self.dirstate.endparentchange()
 
     def filectx(self, path, changeid=None, fileid=None):
         """changeid can be a changeset revision, node, or tag.
            fileid can be a file revision or node."""
         return context.filectx(self, path, changeid, fileid)
 
     def getcwd(self):
         return self.dirstate.getcwd()
 
     def pathto(self, f, cwd=None):
         return self.dirstate.pathto(f, cwd)
 
     def wfile(self, f, mode='r'):
         return self.wvfs(f, mode)
 
     def _link(self, f):
         return self.wvfs.islink(f)
 
     def _loadfilter(self, filter):
         if filter not in self.filterpats:
             l = []
             for pat, cmd in self.ui.configitems(filter):
                 if cmd == '!':
                     continue
                 mf = matchmod.match(self.root, '', [pat])
                 fn = None
                 params = cmd
                 for name, filterfn in self._datafilters.iteritems():
                     if cmd.startswith(name):
                         fn = filterfn
                         params = cmd[len(name):].lstrip()
                         break
                 if not fn:
                     fn = lambda s, c, **kwargs: util.filter(s, c)
                 # Wrap old filters not supporting keyword arguments
                 if not inspect.getargspec(fn)[2]:
                     oldfn = fn
                     fn = lambda s, c, **kwargs: oldfn(s, c)
                 l.append((mf, fn, params))
             self.filterpats[filter] = l
         return self.filterpats[filter]
 
     def _filter(self, filterpats, filename, data):
         for mf, fn, cmd in filterpats:
             if mf(filename):
                 self.ui.debug("filtering %s through %s\n" % (filename, cmd))
                 data = fn(data, cmd, ui=self.ui, repo=self, filename=filename)
                 break
 
         return data
 
     @unfilteredpropertycache
     def _encodefilterpats(self):
         return self._loadfilter('encode')
 
     @unfilteredpropertycache
     def _decodefilterpats(self):
         return self._loadfilter('decode')
 
     def adddatafilter(self, name, filter):
         self._datafilters[name] = filter
 
     def wread(self, filename):
         if self._link(filename):
             data = self.wvfs.readlink(filename)
         else:
             data = self.wvfs.read(filename)
         return self._filter(self._encodefilterpats, filename, data)
 
     def wwrite(self, filename, data, flags, backgroundclose=False):
         """write ``data`` into ``filename`` in the working directory
 
         This returns length of written (maybe decoded) data.
         """
         data = self._filter(self._decodefilterpats, filename, data)
         if 'l' in flags:
             self.wvfs.symlink(data, filename)
         else:
             self.wvfs.write(filename, data, backgroundclose=backgroundclose)
             if 'x' in flags:
                 self.wvfs.setflags(filename, False, True)
         return len(data)
 
     def wwritedata(self, filename, data):
         return self._filter(self._decodefilterpats, filename, data)
 
     def currenttransaction(self):
         """return the current transaction or None if non exists"""
         if self._transref:
             tr = self._transref()
         else:
             tr = None
 
         if tr and tr.running():
             return tr
         return None
 
     def transaction(self, desc, report=None):
         if (self.ui.configbool('devel', 'all-warnings')
                 or self.ui.configbool('devel', 'check-locks')):
             if self._currentlock(self._lockref) is None:
                 raise RuntimeError('programming error: transaction requires '
                                    'locking')
         tr = self.currenttransaction()
         if tr is not None:
             return tr.nest()
 
         # abort here if the journal already exists
         if self.svfs.exists("journal"):
             raise error.RepoError(
                 _("abandoned transaction found"),
                 hint=_("run 'hg recover' to clean up transaction"))
 
         idbase = "%.40f#%f" % (random.random(), time.time())
         txnid = 'TXN:' + hashlib.sha1(idbase).hexdigest()
         self.hook('pretxnopen', throw=True, txnname=desc, txnid=txnid)
 
         self._writejournal(desc)
         renames = [(vfs, x, undoname(x)) for vfs, x in self._journalfiles()]
         if report:
             rp = report
         else:
             rp = self.ui.warn
         vfsmap = {'plain': self.vfs} # root of .hg/
         # we must avoid cyclic reference between repo and transaction.
         reporef = weakref.ref(self)
         def validate(tr):
             """will run pre-closing hooks"""
             reporef().hook('pretxnclose', throw=True,
                            txnname=desc, **tr.hookargs)
         def releasefn(tr, success):
             repo = reporef()
             if success:
                 # this should be explicitly invoked here, because
                 # in-memory changes aren't written out at closing
                 # transaction, if tr.addfilegenerator (via
                 # dirstate.write or so) isn't invoked while
                 # transaction running
                 repo.dirstate.write(None)
             else:
                 # discard all changes (including ones already written
                 # out) in this transaction
                 repo.dirstate.restorebackup(None, prefix='journal.')
 
                 repo.invalidate(clearfilecache=True)
 
         tr = transaction.transaction(rp, self.svfs, vfsmap,
                                      "journal",
                                      "undo",
                                      aftertrans(renames),
                                      self.store.createmode,
                                      validator=validate,
                                      releasefn=releasefn)
 
         tr.hookargs['txnid'] = txnid
         # note: writing the fncache only during finalize mean that the file is
         # outdated when running hooks. As fncache is used for streaming clone,
         # this is not expected to break anything that happen during the hooks.
         tr.addfinalize('flush-fncache', self.store.write)
         def txnclosehook(tr2):
             """To be run if transaction is successful, will schedule a hook run
             """
             # Don't reference tr2 in hook() so we don't hold a reference.
             # This reduces memory consumption when there are multiple
             # transactions per lock. This can likely go away if issue5045
             # fixes the function accumulation.
             hookargs = tr2.hookargs
 
             def hook():
                 reporef().hook('txnclose', throw=False, txnname=desc,
                                **hookargs)
             reporef()._afterlock(hook)
         tr.addfinalize('txnclose-hook', txnclosehook)
         def txnaborthook(tr2):
             """To be run if transaction is aborted
             """
             reporef().hook('txnabort', throw=False, txnname=desc,
                            **tr2.hookargs)
         tr.addabort('txnabort-hook', txnaborthook)
         # avoid eager cache invalidation. in-memory data should be identical
         # to stored data if transaction has no error.
         tr.addpostclose('refresh-filecachestats', self._refreshfilecachestats)
         self._transref = weakref.ref(tr)
         return tr
 
     def _journalfiles(self):
         return ((self.svfs, 'journal'),
                 (self.vfs, 'journal.dirstate'),
                 (self.vfs, 'journal.branch'),
                 (self.vfs, 'journal.desc'),
                 (self.vfs, 'journal.bookmarks'),
                 (self.svfs, 'journal.phaseroots'))
 
     def undofiles(self):
         return [(vfs, undoname(x)) for vfs, x in self._journalfiles()]
 
     def _writejournal(self, desc):
         self.dirstate.savebackup(None, prefix='journal.')
         self.vfs.write("journal.branch",
                           encoding.fromlocal(self.dirstate.branch()))
         self.vfs.write("journal.desc",
                           "%d\n%s\n" % (len(self), desc))
         self.vfs.write("journal.bookmarks",
                           self.vfs.tryread("bookmarks"))
         self.svfs.write("journal.phaseroots",
                            self.svfs.tryread("phaseroots"))
 
     def recover(self):
         with self.lock():
             if self.svfs.exists("journal"):
                 self.ui.status(_("rolling back interrupted transaction\n"))
                 vfsmap = {'': self.svfs,
                           'plain': self.vfs,}
                 transaction.rollback(self.svfs, vfsmap, "journal",
                                      self.ui.warn)
                 self.invalidate()
                 return True
             else:
                 self.ui.warn(_("no interrupted transaction available\n"))
                 return False
 
     def rollback(self, dryrun=False, force=False):
         wlock = lock = dsguard = None
         try:
             wlock = self.wlock()
             lock = self.lock()
             if self.svfs.exists("undo"):
                 dsguard = cmdutil.dirstateguard(self, 'rollback')
 
                 return self._rollback(dryrun, force, dsguard)
             else:
                 self.ui.warn(_("no rollback information available\n"))
                 return 1
         finally:
             release(dsguard, lock, wlock)
 
     @unfilteredmethod # Until we get smarter cache management
     def _rollback(self, dryrun, force, dsguard):
         ui = self.ui
         try:
             args = self.vfs.read('undo.desc').splitlines()
             (oldlen, desc, detail) = (int(args[0]), args[1], None)
             if len(args) >= 3:
                 detail = args[2]
             oldtip = oldlen - 1
 
             if detail and ui.verbose:
                 msg = (_('repository tip rolled back to revision %s'
                          ' (undo %s: %s)\n')
                        % (oldtip, desc, detail))
             else:
                 msg = (_('repository tip rolled back to revision %s'
                          ' (undo %s)\n')
                        % (oldtip, desc))
         except IOError:
             msg = _('rolling back unknown transaction\n')
             desc = None
 
         if not force and self['.'] != self['tip'] and desc == 'commit':
             raise error.Abort(
                 _('rollback of last commit while not checked out '
                   'may lose data'), hint=_('use -f to force'))
 
         ui.status(msg)
         if dryrun:
             return 0
 
         parents = self.dirstate.parents()
         self.destroying()
         vfsmap = {'plain': self.vfs, '': self.svfs}
         transaction.rollback(self.svfs, vfsmap, 'undo', ui.warn)
         if self.vfs.exists('undo.bookmarks'):
             self.vfs.rename('undo.bookmarks', 'bookmarks', checkambig=True)
         if self.svfs.exists('undo.phaseroots'):
             self.svfs.rename('undo.phaseroots', 'phaseroots', checkambig=True)
         self.invalidate()
 
         parentgone = (parents[0] not in self.changelog.nodemap or
                       parents[1] not in self.changelog.nodemap)
         if parentgone:
             # prevent dirstateguard from overwriting already restored one
             dsguard.close()
 
             self.dirstate.restorebackup(None, prefix='undo.')
             try:
                 branch = self.vfs.read('undo.branch')
                 self.dirstate.setbranch(encoding.tolocal(branch))
             except IOError:
                 ui.warn(_('named branch could not be reset: '
                           'current branch is still \'%s\'\n')
                         % self.dirstate.branch())
 
             parents = tuple([p.rev() for p in self[None].parents()])
             if len(parents) > 1:
                 ui.status(_('working directory now based on '
                             'revisions %d and %d\n') % parents)
             else:
                 ui.status(_('working directory now based on '
                             'revision %d\n') % parents)
             mergemod.mergestate.clean(self, self['.'].node())
 
         # TODO: if we know which new heads may result from this rollback, pass
         # them to destroy(), which will prevent the branchhead cache from being
         # invalidated.
         self.destroyed()
         return 0
 
     def invalidatecaches(self):
 
         if '_tagscache' in vars(self):
             # can't use delattr on proxy
             del self.__dict__['_tagscache']
 
         self.unfiltered()._branchcaches.clear()
         self.invalidatevolatilesets()
 
     def invalidatevolatilesets(self):
         self.filteredrevcache.clear()
         obsolete.clearobscaches(self)
 
     def invalidatedirstate(self):
         '''Invalidates the dirstate, causing the next call to dirstate
         to check if it was modified since the last time it was read,
         rereading it if it has.
 
         This is different to dirstate.invalidate() that it doesn't always
         rereads the dirstate. Use dirstate.invalidate() if you want to
         explicitly read the dirstate again (i.e. restoring it to a previous
         known good state).'''
         if hasunfilteredcache(self, 'dirstate'):
             for k in self.dirstate._filecache:
                 try:
                     delattr(self.dirstate, k)
                 except AttributeError:
                     pass
             delattr(self.unfiltered(), 'dirstate')
 
     def invalidate(self, clearfilecache=False):
         '''Invalidates both store and non-store parts other than dirstate
 
         If a transaction is running, invalidation of store is omitted,
         because discarding in-memory changes might cause inconsistency
         (e.g. incomplete fncache causes unintentional failure, but
         redundant one doesn't).
         '''
         unfiltered = self.unfiltered() # all file caches are stored unfiltered
         for k in self._filecache.keys():
             # dirstate is invalidated separately in invalidatedirstate()
             if k == 'dirstate':
                 continue
 
             if clearfilecache:
                 del self._filecache[k]
             try:
                 delattr(unfiltered, k)
             except AttributeError:
                 pass
         self.invalidatecaches()
         if not self.currenttransaction():
             # TODO: Changing contents of store outside transaction
             # causes inconsistency. We should make in-memory store
             # changes detectable, and abort if changed.
             self.store.invalidatecaches()
 
     def invalidateall(self):
         '''Fully invalidates both store and non-store parts, causing the
         subsequent operation to reread any outside changes.'''
         # extension should hook this to invalidate its caches
         self.invalidate()
         self.invalidatedirstate()
 
     @unfilteredmethod
     def _refreshfilecachestats(self, tr):
         """Reload stats of cached files so that they are flagged as valid"""
         for k, ce in self._filecache.items():
             if k == 'dirstate' or k not in self.__dict__:
                 continue
             ce.refresh()
 
     def _lock(self, vfs, lockname, wait, releasefn, acquirefn, desc,
               inheritchecker=None, parentenvvar=None):
         parentlock = None
         # the contents of parentenvvar are used by the underlying lock to
         # determine whether it can be inherited
         if parentenvvar is not None:
             parentlock = os.environ.get(parentenvvar)
         try:
             l = lockmod.lock(vfs, lockname, 0, releasefn=releasefn,
                              acquirefn=acquirefn, desc=desc,
                              inheritchecker=inheritchecker,
                              parentlock=parentlock)
         except error.LockHeld as inst:
             if not wait:
                 raise
             # show more details for new-style locks
             if ':' in inst.locker:
                 host, pid = inst.locker.split(":", 1)
                 self.ui.warn(
                     _("waiting for lock on %s held by process %r "
                       "on host %r\n") % (desc, pid, host))
             else:
                 self.ui.warn(_("waiting for lock on %s held by %r\n") %
                              (desc, inst.locker))
             # default to 600 seconds timeout
             l = lockmod.lock(vfs, lockname,
                              int(self.ui.config("ui", "timeout", "600")),
                              releasefn=releasefn, acquirefn=acquirefn,
                              desc=desc)
             self.ui.warn(_("got lock after %s seconds\n") % l.delay)
         return l
 
     def _afterlock(self, callback):
         """add a callback to be run when the repository is fully unlocked
 
         The callback will be executed when the outermost lock is released
         (with wlock being higher level than 'lock')."""
         for ref in (self._wlockref, self._lockref):
             l = ref and ref()
             if l and l.held:
                 l.postrelease.append(callback)
                 break
         else: # no lock have been found.
             callback()
 
     def lock(self, wait=True):
         '''Lock the repository store (.hg/store) and return a weak reference
         to the lock. Use this before modifying the store (e.g. committing or
         stripping). If you are opening a transaction, get a lock as well.)
 
         If both 'lock' and 'wlock' must be acquired, ensure you always acquires
         'wlock' first to avoid a dead-lock hazard.'''
         l = self._currentlock(self._lockref)
         if l is not None:
             l.lock()
             return l
 
         l = self._lock(self.svfs, "lock", wait, None,
                        self.invalidate, _('repository %s') % self.origroot)
         self._lockref = weakref.ref(l)
         return l
 
     def _wlockchecktransaction(self):
         if self.currenttransaction() is not None:
             raise error.LockInheritanceContractViolation(
                 'wlock cannot be inherited in the middle of a transaction')
 
     def wlock(self, wait=True):
         '''Lock the non-store parts of the repository (everything under
         .hg except .hg/store) and return a weak reference to the lock.
 
         Use this before modifying files in .hg.
 
         If both 'lock' and 'wlock' must be acquired, ensure you always acquires
         'wlock' first to avoid a dead-lock hazard.'''
         l = self._wlockref and self._wlockref()
         if l is not None and l.held:
             l.lock()
             return l
 
         # We do not need to check for non-waiting lock acquisition.  Such
         # acquisition would not cause dead-lock as they would just fail.
         if wait and (self.ui.configbool('devel', 'all-warnings')
                      or self.ui.configbool('devel', 'check-locks')):
             if self._currentlock(self._lockref) is not None:
                 self.ui.develwarn('"wlock" acquired after "lock"')
 
         def unlock():
             if self.dirstate.pendingparentchange():
                 self.dirstate.invalidate()
             else:
                 self.dirstate.write(None)
 
             self._filecache['dirstate'].refresh()
 
         l = self._lock(self.vfs, "wlock", wait, unlock,
                        self.invalidatedirstate, _('working directory of %s') %
                        self.origroot,
                        inheritchecker=self._wlockchecktransaction,
                        parentenvvar='HG_WLOCK_LOCKER')
         self._wlockref = weakref.ref(l)
         return l
 
     def _currentlock(self, lockref):
         """Returns the lock if it's held, or None if it's not."""
         if lockref is None:
             return None
         l = lockref()
         if l is None or not l.held:
             return None
         return l
 
     def currentwlock(self):
         """Returns the wlock if it's held, or None if it's not."""
         return self._currentlock(self._wlockref)
 
     def _filecommit(self, fctx, manifest1, manifest2, linkrev, tr, changelist):
         """
         commit an individual file as part of a larger transaction
         """
 
         fname = fctx.path()
         fparent1 = manifest1.get(fname, nullid)
         fparent2 = manifest2.get(fname, nullid)
         if isinstance(fctx, context.filectx):
             node = fctx.filenode()
             if node in [fparent1, fparent2]:
                 self.ui.debug('reusing %s filelog entry\n' % fname)
                 if manifest1.flags(fname) != fctx.flags():
                     changelist.append(fname)
                 return node
 
         flog = self.file(fname)
         meta = {}
         copy = fctx.renamed()
         if copy and copy[0] != fname:
             # Mark the new revision of this file as a copy of another
             # file.  This copy data will effectively act as a parent
             # of this new revision.  If this is a merge, the first
             # parent will be the nullid (meaning "look up the copy data")
             # and the second one will be the other parent.  For example:
             #
             # 0 --- 1 --- 3   rev1 changes file foo
             #   \       /     rev2 renames foo to bar and changes it
             #    \- 2 -/      rev3 should have bar with all changes and
             #                      should record that bar descends from
             #                      bar in rev2 and foo in rev1
             #
             # this allows this merge to succeed:
             #
             # 0 --- 1 --- 3   rev4 reverts the content change from rev2
             #   \       /     merging rev3 and rev4 should use bar@rev2
             #    \- 2 --- 4        as the merge base
             #
 
             cfname = copy[0]
             crev = manifest1.get(cfname)
             newfparent = fparent2
 
             if manifest2: # branch merge
                 if fparent2 == nullid or crev is None: # copied on remote side
                     if cfname in manifest2:
                         crev = manifest2[cfname]
                         newfparent = fparent1
 
             # Here, we used to search backwards through history to try to find
             # where the file copy came from if the source of a copy was not in
             # the parent directory. However, this doesn't actually make sense to
             # do (what does a copy from something not in your working copy even
             # mean?) and it causes bugs (eg, issue4476). Instead, we will warn
             # the user that copy information was dropped, so if they didn't
             # expect this outcome it can be fixed, but this is the correct
             # behavior in this circumstance.
 
             if crev:
                 self.ui.debug(" %s: copy %s:%s\n" % (fname, cfname, hex(crev)))
                 meta["copy"] = cfname
                 meta["copyrev"] = hex(crev)
                 fparent1, fparent2 = nullid, newfparent
             else:
                 self.ui.warn(_("warning: can't find ancestor for '%s' "
                                "copied from '%s'!\n") % (fname, cfname))
 
         elif fparent1 == nullid:
             fparent1, fparent2 = fparent2, nullid
         elif fparent2 != nullid:
             # is one parent an ancestor of the other?
             fparentancestors = flog.commonancestorsheads(fparent1, fparent2)
             if fparent1 in fparentancestors:
                 fparent1, fparent2 = fparent2, nullid
             elif fparent2 in fparentancestors:
                 fparent2 = nullid
 
         # is the file changed?
         text = fctx.data()
         if fparent2 != nullid or flog.cmp(fparent1, text) or meta:
             changelist.append(fname)
             return flog.add(text, meta, tr, linkrev, fparent1, fparent2)
         # are just the flags changed during merge?
         elif fname in manifest1 and manifest1.flags(fname) != fctx.flags():
             changelist.append(fname)
 
         return fparent1
 
     def checkcommitpatterns(self, wctx, vdirs, match, status, fail):
         """check for commit arguments that aren't commitable"""
         if match.isexact() or match.prefix():
             matched = set(status.modified + status.added + status.removed)
 
             for f in match.files():
                 f = self.dirstate.normalize(f)
                 if f == '.' or f in matched or f in wctx.substate:
                     continue
                 if f in status.deleted:
                     fail(f, _('file not found!'))
                 if f in vdirs: # visited directory
                     d = f + '/'
                     for mf in matched:
                         if mf.startswith(d):
                             break
                     else:
                         fail(f, _("no match under directory!"))
                 elif f not in self.dirstate:
                     fail(f, _("file not tracked!"))
 
     @unfilteredmethod
     def commit(self, text="", user=None, date=None, match=None, force=False,
                editor=False, extra=None):
         """Add a new revision to current repository.
 
         Revision information is gathered from the working directory,
         match can be used to filter the committed files. If editor is
         supplied, it is called to get a commit message.
         """
         if extra is None:
             extra = {}
 
         def fail(f, msg):
             raise error.Abort('%s: %s' % (f, msg))
 
         if not match:
             match = matchmod.always(self.root, '')
 
         if not force:
             vdirs = []
             match.explicitdir = vdirs.append
             match.bad = fail
 
         wlock = lock = tr = None
         try:
             wlock = self.wlock()
             lock = self.lock() # for recent changelog (see issue4368)
 
             wctx = self[None]
             merge = len(wctx.parents()) > 1
 
             if not force and merge and match.ispartial():
                 raise error.Abort(_('cannot partially commit a merge '
                                    '(do not specify files or patterns)'))
 
             status = self.status(match=match, clean=force)
             if force:
                 status.modified.extend(status.clean) # mq may commit clean files
 
             # check subrepos
             subs = []
             commitsubs = set()
             newstate = wctx.substate.copy()
             # only manage subrepos and .hgsubstate if .hgsub is present
             if '.hgsub' in wctx:
                 # we'll decide whether to track this ourselves, thanks
                 for c in status.modified, status.added, status.removed:
                     if '.hgsubstate' in c:
                         c.remove('.hgsubstate')
 
                 # compare current state to last committed state
                 # build new substate based on last committed state
                 oldstate = wctx.p1().substate
                 for s in sorted(newstate.keys()):
                     if not match(s):
                         # ignore working copy, use old state if present
                         if s in oldstate:
                             newstate[s] = oldstate[s]
                             continue
                         if not force:
                             raise error.Abort(
                                 _("commit with new subrepo %s excluded") % s)
                     dirtyreason = wctx.sub(s).dirtyreason(True)
                     if dirtyreason:
                         if not self.ui.configbool('ui', 'commitsubrepos'):
                             raise error.Abort(dirtyreason,
                                 hint=_("use --subrepos for recursive commit"))
                         subs.append(s)
                         commitsubs.add(s)
                     else:
                         bs = wctx.sub(s).basestate()
                         newstate[s] = (newstate[s][0], bs, newstate[s][2])
                         if oldstate.get(s, (None, None, None))[1] != bs:
                             subs.append(s)
 
                 # check for removed subrepos
                 for p in wctx.parents():
                     r = [s for s in p.substate if s not in newstate]
                     subs += [s for s in r if match(s)]
                 if subs:
                     if (not match('.hgsub') and
                         '.hgsub' in (wctx.modified() + wctx.added())):
                         raise error.Abort(
                             _("can't commit subrepos without .hgsub"))
                     status.modified.insert(0, '.hgsubstate')
 
             elif '.hgsub' in status.removed:
                 # clean up .hgsubstate when .hgsub is removed
                 if ('.hgsubstate' in wctx and
                     '.hgsubstate' not in (status.modified + status.added +
                                           status.removed)):
                     status.removed.insert(0, '.hgsubstate')
 
             # make sure all explicit patterns are matched
             if not force:
                 self.checkcommitpatterns(wctx, vdirs, match, status, fail)
 
             cctx = context.workingcommitctx(self, status,
                                             text, user, date, extra)
 
             # internal config: ui.allowemptycommit
             allowemptycommit = (wctx.branch() != wctx.p1().branch()
                                 or extra.get('close') or merge or cctx.files()
                                 or self.ui.configbool('ui', 'allowemptycommit'))
             if not allowemptycommit:
                 return None
 
             if merge and cctx.deleted():
                 raise error.Abort(_("cannot commit merge with missing files"))
 
             ms = mergemod.mergestate.read(self)
 
             if list(ms.unresolved()):
                 raise error.Abort(_("unresolved merge conflicts "
                                     "(see 'hg help resolve')"))
             if ms.mdstate() != 's' or list(ms.driverresolved()):
                 raise error.Abort(_('driver-resolved merge conflicts'),
                                   hint=_('run "hg resolve --all" to resolve'))
 
             if editor:
                 cctx._text = editor(self, cctx, subs)
             edited = (text != cctx._text)
 
             # Save commit message in case this transaction gets rolled back
             # (e.g. by a pretxncommit hook).  Leave the content alone on
             # the assumption that the user will use the same editor again.
             msgfn = self.savecommitmessage(cctx._text)
 
             # commit subs and write new state
             if subs:
                 for s in sorted(commitsubs):
                     sub = wctx.sub(s)
                     self.ui.status(_('committing subrepository %s\n') %
                         subrepo.subrelpath(sub))
                     sr = sub.commit(cctx._text, user, date)
                     newstate[s] = (newstate[s][0], sr)
                 subrepo.writestate(self, newstate)
 
             p1, p2 = self.dirstate.parents()
             hookp1, hookp2 = hex(p1), (p2 != nullid and hex(p2) or '')
             try:
                 self.hook("precommit", throw=True, parent1=hookp1,
                           parent2=hookp2)
                 tr = self.transaction('commit')
                 ret = self.commitctx(cctx, True)
             except: # re-raises
                 if edited:
                     self.ui.write(
                         _('note: commit message saved in %s\n') % msgfn)
                 raise
             # update bookmarks, dirstate and mergestate
             bookmarks.update(self, [p1, p2], ret)
             cctx.markcommitted(ret)
             ms.reset()
             tr.close()
 
         finally:
             lockmod.release(tr, lock, wlock)
 
         def commithook(node=hex(ret), parent1=hookp1, parent2=hookp2):
             # hack for command that use a temporary commit (eg: histedit)
             # temporary commit got stripped before hook release
             if self.changelog.hasnode(ret):
                 self.hook("commit", node=node, parent1=parent1,
                           parent2=parent2)
         self._afterlock(commithook)
         return ret
 
     @unfilteredmethod
     def commitctx(self, ctx, error=False):
         """Add a new revision to current repository.
         Revision information is passed via the context argument.
         """
 
         tr = None
         p1, p2 = ctx.p1(), ctx.p2()
         user = ctx.user()
 
         lock = self.lock()
         try:
             tr = self.transaction("commit")
             trp = weakref.proxy(tr)
 
             if ctx.files():
                 m1 = p1.manifest()
                 m2 = p2.manifest()
                 m = m1.copy()
 
                 # check in files
                 added = []
                 changed = []
                 removed = list(ctx.removed())
                 linkrev = len(self)
                 self.ui.note(_("committing files:\n"))
                 for f in sorted(ctx.modified() + ctx.added()):
                     self.ui.note(f + "\n")
                     try:
                         fctx = ctx[f]
                         if fctx is None:
                             removed.append(f)
                         else:
                             added.append(f)
                             m[f] = self._filecommit(fctx, m1, m2, linkrev,
                                                     trp, changed)
                             m.setflag(f, fctx.flags())
                     except OSError as inst:
                         self.ui.warn(_("trouble committing %s!\n") % f)
                         raise
                     except IOError as inst:
                         errcode = getattr(inst, 'errno', errno.ENOENT)
                         if error or errcode and errcode != errno.ENOENT:
                             self.ui.warn(_("trouble committing %s!\n") % f)
                         raise
 
                 # update manifest
                 self.ui.note(_("committing manifest\n"))
                 removed = [f for f in sorted(removed) if f in m1 or f in m2]
                 drop = [f for f in removed if f in m]
                 for f in drop:
                     del m[f]
                 mn = self.manifestlog.add(m, trp, linkrev,
                                           p1.manifestnode(), p2.manifestnode(),
                                           added, drop)
                 files = changed + removed
             else:
                 mn = p1.manifestnode()
                 files = []
 
             # update changelog
             self.ui.note(_("committing changelog\n"))
             self.changelog.delayupdate(tr)
             n = self.changelog.add(mn, files, ctx.description(),
                                    trp, p1.node(), p2.node(),
                                    user, ctx.date(), ctx.extra().copy())
             xp1, xp2 = p1.hex(), p2 and p2.hex() or ''
             self.hook('pretxncommit', throw=True, node=hex(n), parent1=xp1,
                       parent2=xp2)
             # set the new commit is proper phase
             targetphase = subrepo.newcommitphase(self.ui, ctx)
             if targetphase:
                 # retract boundary do not alter parent changeset.
                 # if a parent have higher the resulting phase will
                 # be compliant anyway
                 #
                 # if minimal phase was 0 we don't need to retract anything
                 phases.retractboundary(self, tr, targetphase, [n])
             tr.close()
             branchmap.updatecache(self.filtered('served'))
             return n
         finally:
             if tr:
                 tr.release()
             lock.release()
 
     @unfilteredmethod
     def destroying(self):
         '''Inform the repository that nodes are about to be destroyed.
         Intended for use by strip and rollback, so there's a common
         place for anything that has to be done before destroying history.
 
         This is mostly useful for saving state that is in memory and waiting
         to be flushed when the current lock is released. Because a call to
         destroyed is imminent, the repo will be invalidated causing those
         changes to stay in memory (waiting for the next unlock), or vanish
         completely.
         '''
         # When using the same lock to commit and strip, the phasecache is left
         # dirty after committing. Then when we strip, the repo is invalidated,
         # causing those changes to disappear.
         if '_phasecache' in vars(self):
             self._phasecache.write()
 
     @unfilteredmethod
     def destroyed(self):
         '''Inform the repository that nodes have been destroyed.
         Intended for use by strip and rollback, so there's a common
         place for anything that has to be done after destroying history.
         '''
         # When one tries to:
         # 1) destroy nodes thus calling this method (e.g. strip)
         # 2) use phasecache somewhere (e.g. commit)
         #
         # then 2) will fail because the phasecache contains nodes that were
         # removed. We can either remove phasecache from the filecache,
         # causing it to reload next time it is accessed, or simply filter
         # the removed nodes now and write the updated cache.
         self._phasecache.filterunknown(self)
         self._phasecache.write()
 
         # update the 'served' branch cache to help read only server process
         # Thanks to branchcache collaboration this is done from the nearest
         # filtered subset and it is expected to be fast.
         branchmap.updatecache(self.filtered('served'))
 
         # Ensure the persistent tag cache is updated.  Doing it now
         # means that the tag cache only has to worry about destroyed
         # heads immediately after a strip/rollback.  That in turn
         # guarantees that "cachetip == currenttip" (comparing both rev
         # and node) always means no nodes have been added or destroyed.
 
         # XXX this is suboptimal when qrefresh'ing: we strip the current
         # head, refresh the tag cache, then immediately add a new head.
         # But I think doing it this way is necessary for the "instant
         # tag cache retrieval" case to work.
         self.invalidate()
 
     def walk(self, match, node=None):
         '''
         walk recursively through the directory tree or a given
         changeset, finding all files matched by the match
         function
         '''
         return self[node].walk(match)
 
     def status(self, node1='.', node2=None, match=None,
                ignored=False, clean=False, unknown=False,
                listsubrepos=False):
         '''a convenience method that calls node1.status(node2)'''
         return self[node1].status(node2, match, ignored, clean, unknown,
                                   listsubrepos)
 
     def heads(self, start=None):
         heads = self.changelog.heads(start)
         # sort the output in rev descending order
         return sorted(heads, key=self.changelog.rev, reverse=True)
 
     def branchheads(self, branch=None, start=None, closed=False):
         '''return a (possibly filtered) list of heads for the given branch
 
         Heads are returned in topological order, from newest to oldest.
         If branch is None, use the dirstate branch.
         If start is not None, return only heads reachable from start.
         If closed is True, return heads that are marked as closed as well.
         '''
         if branch is None:
             branch = self[None].branch()
         branches = self.branchmap()
         if branch not in branches:
             return []
         # the cache returns heads ordered lowest to highest
         bheads = list(reversed(branches.branchheads(branch, closed=closed)))
         if start is not None:
             # filter out the heads that cannot be reached from startrev
             fbheads = set(self.changelog.nodesbetween([start], bheads)[2])
             bheads = [h for h in bheads if h in fbheads]
         return bheads
 
     def branches(self, nodes):
         if not nodes:
             nodes = [self.changelog.tip()]
         b = []
         for n in nodes:
             t = n
             while True:
                 p = self.changelog.parents(n)
                 if p[1] != nullid or p[0] == nullid:
                     b.append((t, n, p[0], p[1]))
                     break
                 n = p[0]
         return b
 
     def between(self, pairs):
         r = []
 
         for top, bottom in pairs:
             n, l, i = top, [], 0
             f = 1
 
             while n != bottom and n != nullid:
                 p = self.changelog.parents(n)[0]
                 if i == f:
                     l.append(n)
                     f = f * 2
                 n = p
                 i += 1
 
             r.append(l)
 
         return r
 
     def checkpush(self, pushop):
         """Extensions can override this function if additional checks have
         to be performed before pushing, or call it if they override push
         command.
         """
         pass
 
     @unfilteredpropertycache
     def prepushoutgoinghooks(self):
         """Return util.hooks consists of a pushop with repo, remote, outgoing
         methods, which are called before pushing changesets.
         """
         return util.hooks()
 
     def pushkey(self, namespace, key, old, new):
         try:
             tr = self.currenttransaction()
             hookargs = {}
             if tr is not None:
                 hookargs.update(tr.hookargs)
             hookargs['namespace'] = namespace
             hookargs['key'] = key
             hookargs['old'] = old
             hookargs['new'] = new
             self.hook('prepushkey', throw=True, **hookargs)
         except error.HookAbort as exc:
             self.ui.write_err(_("pushkey-abort: %s\n") % exc)
             if exc.hint:
                 self.ui.write_err(_("(%s)\n") % exc.hint)
             return False
         self.ui.debug('pushing key for "%s:%s"\n' % (namespace, key))
         ret = pushkey.push(self, namespace, key, old, new)
         def runhook():
             self.hook('pushkey', namespace=namespace, key=key, old=old, new=new,
                       ret=ret)
         self._afterlock(runhook)
         return ret
 
     def listkeys(self, namespace):
         self.hook('prelistkeys', throw=True, namespace=namespace)
         self.ui.debug('listing keys for "%s"\n' % namespace)
         values = pushkey.list(self, namespace)
         self.hook('listkeys', namespace=namespace, values=values)
         return values
 
     def debugwireargs(self, one, two, three=None, four=None, five=None):
         '''used to test argument passing over the wire'''
         return "%s %s %s %s %s" % (one, two, three, four, five)
 
     def savecommitmessage(self, text):
         fp = self.vfs('last-message.txt', 'wb')
         try:
             fp.write(text)
         finally:
             fp.close()
         return self.pathto(fp.name[len(self.root) + 1:])
 
 # used to avoid circular references so destructors work
 def aftertrans(files):
     renamefiles = [tuple(t) for t in files]
     def a():
         for vfs, src, dest in renamefiles:
             try:
                 vfs.rename(src, dest)
             except OSError: # journal file does not yet exist
                 pass
     return a
 
 def undoname(fn):
     base, name = os.path.split(fn)
     assert name.startswith('journal')
     return os.path.join(base, name.replace('journal', 'undo', 1))
 
 def instance(ui, path, create):
     return localrepository(ui, util.urllocalpath(path), create)
 
 def islocal(path):
     return True
 
 def newreporequirements(repo):
     """Determine the set of requirements for a new local repository.
 
     Extensions can wrap this function to specify custom requirements for
     new repositories.
     """
     ui = repo.ui
     requirements = set(['revlogv1'])
     if ui.configbool('format', 'usestore', True):
         requirements.add('store')
         if ui.configbool('format', 'usefncache', True):
             requirements.add('fncache')
             if ui.configbool('format', 'dotencode', True):
                 requirements.add('dotencode')
 
     if scmutil.gdinitconfig(ui):
         requirements.add('generaldelta')
     if ui.configbool('experimental', 'treemanifest', False):
         requirements.add('treemanifest')
     if ui.configbool('experimental', 'manifestv2', False):
         requirements.add('manifestv2')
 
     return requirements

mercurial/wireproto.py

@@ -1,957 +1,959
 # wireproto.py - generic wire protocol support functions
 #
 # Copyright 2005-2010 Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from __future__ import absolute_import
 
 import hashlib
 import itertools
 import os
 import sys
 import tempfile
 
 from .i18n import _
 from .node import (
     bin,
     hex,
 )
 
 from . import (
     bundle2,
     changegroup as changegroupmod,
     encoding,
     error,
     exchange,
     peer,
     pushkey as pushkeymod,
     streamclone,
     util,
 )
 
 urlerr = util.urlerr
 urlreq = util.urlreq
 
 bundle2required = _(
     'incompatible Mercurial client; bundle2 required\n'
     '(see https://www.mercurial-scm.org/wiki/IncompatibleClient)\n')
 
 class abstractserverproto(object):
     """abstract class that summarizes the protocol API
 
     Used as reference and documentation.
     """
 
     def getargs(self, args):
         """return the value for arguments in <args>
 
         returns a list of values (same order as <args>)"""
         raise NotImplementedError()
 
     def getfile(self, fp):
         """write the whole content of a file into a file like object
 
         The file is in the form::
 
             (<chunk-size>\n<chunk>)+0\n
 
         chunk size is the ascii version of the int.
         """
         raise NotImplementedError()
 
     def redirect(self):
         """may setup interception for stdout and stderr
 
         See also the `restore` method."""
         raise NotImplementedError()
 
     # If the `redirect` function does install interception, the `restore`
     # function MUST be defined. If interception is not used, this function
     # MUST NOT be defined.
     #
     # left commented here on purpose
     #
     #def restore(self):
     #    """reinstall previous stdout and stderr and return intercepted stdout
     #    """
     #    raise NotImplementedError()
 
     def groupchunks(self, fh):
         """Generator of chunks to send to the client.
 
         Some protocols may have compressed the contents.
         """
         raise NotImplementedError()
 
 class remotebatch(peer.batcher):
     '''batches the queued calls; uses as few roundtrips as possible'''
     def __init__(self, remote):
         '''remote must support _submitbatch(encbatch) and
         _submitone(op, encargs)'''
         peer.batcher.__init__(self)
         self.remote = remote
     def submit(self):
         req, rsp = [], []
         for name, args, opts, resref in self.calls:
             mtd = getattr(self.remote, name)
             batchablefn = getattr(mtd, 'batchable', None)
             if batchablefn is not None:
                 batchable = batchablefn(mtd.im_self, *args, **opts)
                 encargsorres, encresref = next(batchable)
                 if encresref:
                     req.append((name, encargsorres,))
                     rsp.append((batchable, encresref, resref,))
                 else:
                     resref.set(encargsorres)
             else:
                 if req:
                     self._submitreq(req, rsp)
                     req, rsp = [], []
                 resref.set(mtd(*args, **opts))
         if req:
             self._submitreq(req, rsp)
     def _submitreq(self, req, rsp):
         encresults = self.remote._submitbatch(req)
         for encres, r in zip(encresults, rsp):
             batchable, encresref, resref = r
             encresref.set(encres)
             resref.set(next(batchable))
 
 class remoteiterbatcher(peer.iterbatcher):
     def __init__(self, remote):
         super(remoteiterbatcher, self).__init__()
         self._remote = remote
 
     def __getattr__(self, name):
         if not getattr(self._remote, name, False):
             raise AttributeError(
                 'Attempted to iterbatch non-batchable call to %r' % name)
         return super(remoteiterbatcher, self).__getattr__(name)
 
     def submit(self):
         """Break the batch request into many patch calls and pipeline them.
 
         This is mostly valuable over http where request sizes can be
         limited, but can be used in other places as well.
         """
         req, rsp = [], []
         for name, args, opts, resref in self.calls:
             mtd = getattr(self._remote, name)
             batchable = mtd.batchable(mtd.im_self, *args, **opts)
             encargsorres, encresref = next(batchable)
             assert encresref
             req.append((name, encargsorres))
             rsp.append((batchable, encresref))
         if req:
             self._resultiter = self._remote._submitbatch(req)
         self._rsp = rsp
 
     def results(self):
         for (batchable, encresref), encres in itertools.izip(
                 self._rsp, self._resultiter):
             encresref.set(encres)
             yield next(batchable)
 
 # Forward a couple of names from peer to make wireproto interactions
 # slightly more sensible.
 batchable = peer.batchable
 future = peer.future
 
 # list of nodes encoding / decoding
 
 def decodelist(l, sep=' '):
     if l:
         return map(bin, l.split(sep))
     return []
 
 def encodelist(l, sep=' '):
     try:
         return sep.join(map(hex, l))
     except TypeError:
         raise
 
 # batched call argument encoding
 
 def escapearg(plain):
     return (plain
             .replace(':', ':c')
             .replace(',', ':o')
             .replace(';', ':s')
             .replace('=', ':e'))
 
 def unescapearg(escaped):
     return (escaped
             .replace(':e', '=')
             .replace(':s', ';')
             .replace(':o', ',')
             .replace(':c', ':'))
 
 def encodebatchcmds(req):
     """Return a ``cmds`` argument value for the ``batch`` command."""
     cmds = []
     for op, argsdict in req:
         # Old servers didn't properly unescape argument names. So prevent
         # the sending of argument names that may not be decoded properly by
         # servers.
         assert all(escapearg(k) == k for k in argsdict)
 
         args = ','.join('%s=%s' % (escapearg(k), escapearg(v))
                         for k, v in argsdict.iteritems())
         cmds.append('%s %s' % (op, args))
 
     return ';'.join(cmds)
 
 # mapping of options accepted by getbundle and their types
 #
 # Meant to be extended by extensions. It is extensions responsibility to ensure
 # such options are properly processed in exchange.getbundle.
 #
 # supported types are:
 #
 # :nodes: list of binary nodes
 # :csv:   list of comma-separated values
 # :scsv:  list of comma-separated values return as set
 # :plain: string with no transformation needed.
 gboptsmap = {'heads':  'nodes',
              'common': 'nodes',
              'obsmarkers': 'boolean',
              'bundlecaps': 'scsv',
              'listkeys': 'csv',
              'cg': 'boolean',
              'cbattempted': 'boolean'}
 
 # client side
 
 class wirepeer(peer.peerrepository):
     """Client-side interface for communicating with a peer repository.
 
     Methods commonly call wire protocol commands of the same name.
 
     See also httppeer.py and sshpeer.py for protocol-specific
     implementations of this interface.
     """
     def batch(self):
         if self.capable('batch'):
             return remotebatch(self)
         else:
             return peer.localbatch(self)
     def _submitbatch(self, req):
         """run batch request <req> on the server
 
         Returns an iterator of the raw responses from the server.
         """
         rsp = self._callstream("batch", cmds=encodebatchcmds(req))
         chunk = rsp.read(1024)
         work = [chunk]
         while chunk:
             while ';' not in chunk and chunk:
                 chunk = rsp.read(1024)
                 work.append(chunk)
             merged = ''.join(work)
             while ';' in merged:
                 one, merged = merged.split(';', 1)
                 yield unescapearg(one)
             chunk = rsp.read(1024)
             work = [merged, chunk]
         yield unescapearg(''.join(work))
 
     def _submitone(self, op, args):
         return self._call(op, **args)
 
     def iterbatch(self):
         return remoteiterbatcher(self)
 
     @batchable
     def lookup(self, key):
         self.requirecap('lookup', _('look up remote revision'))
         f = future()
         yield {'key': encoding.fromlocal(key)}, f
         d = f.value
         success, data = d[:-1].split(" ", 1)
         if int(success):
             yield bin(data)
         self._abort(error.RepoError(data))
 
     @batchable
     def heads(self):
         f = future()
         yield {}, f
         d = f.value
         try:
             yield decodelist(d[:-1])
         except ValueError:
             self._abort(error.ResponseError(_("unexpected response:"), d))
 
     @batchable
     def known(self, nodes):
         f = future()
         yield {'nodes': encodelist(nodes)}, f
         d = f.value
         try:
             yield [bool(int(b)) for b in d]
         except ValueError:
             self._abort(error.ResponseError(_("unexpected response:"), d))
 
     @batchable
     def branchmap(self):
         f = future()
         yield {}, f
         d = f.value
         try:
             branchmap = {}
             for branchpart in d.splitlines():
                 branchname, branchheads = branchpart.split(' ', 1)
                 branchname = encoding.tolocal(urlreq.unquote(branchname))
                 branchheads = decodelist(branchheads)
                 branchmap[branchname] = branchheads
             yield branchmap
         except TypeError:
             self._abort(error.ResponseError(_("unexpected response:"), d))
 
     def branches(self, nodes):
         n = encodelist(nodes)
         d = self._call("branches", nodes=n)
         try:
             br = [tuple(decodelist(b)) for b in d.splitlines()]
             return br
         except ValueError:
             self._abort(error.ResponseError(_("unexpected response:"), d))
 
     def between(self, pairs):
         batch = 8 # avoid giant requests
         r = []
         for i in xrange(0, len(pairs), batch):
             n = " ".join([encodelist(p, '-') for p in pairs[i:i + batch]])
             d = self._call("between", pairs=n)
             try:
                 r.extend(l and decodelist(l) or [] for l in d.splitlines())
             except ValueError:
                 self._abort(error.ResponseError(_("unexpected response:"), d))
         return r
 
     @batchable
     def pushkey(self, namespace, key, old, new):
         if not self.capable('pushkey'):
             yield False, None
         f = future()
         self.ui.debug('preparing pushkey for "%s:%s"\n' % (namespace, key))
         yield {'namespace': encoding.fromlocal(namespace),
                'key': encoding.fromlocal(key),
                'old': encoding.fromlocal(old),
                'new': encoding.fromlocal(new)}, f
         d = f.value
         d, output = d.split('\n', 1)
         try:
             d = bool(int(d))
         except ValueError:
             raise error.ResponseError(
                 _('push failed (unexpected response):'), d)
         for l in output.splitlines(True):
             self.ui.status(_('remote: '), l)
         yield d
 
     @batchable
     def listkeys(self, namespace):
         if not self.capable('pushkey'):
             yield {}, None
         f = future()
         self.ui.debug('preparing listkeys for "%s"\n' % namespace)
         yield {'namespace': encoding.fromlocal(namespace)}, f
         d = f.value
         self.ui.debug('received listkey for "%s": %i bytes\n'
                       % (namespace, len(d)))
         yield pushkeymod.decodekeys(d)
 
     def stream_out(self):
         return self._callstream('stream_out')
 
     def changegroup(self, nodes, kind):
         n = encodelist(nodes)
         f = self._callcompressable("changegroup", roots=n)
         return changegroupmod.cg1unpacker(f, 'UN')
 
     def changegroupsubset(self, bases, heads, kind):
         self.requirecap('changegroupsubset', _('look up remote changes'))
         bases = encodelist(bases)
         heads = encodelist(heads)
         f = self._callcompressable("changegroupsubset",
                                    bases=bases, heads=heads)
         return changegroupmod.cg1unpacker(f, 'UN')
 
     def getbundle(self, source, **kwargs):
         self.requirecap('getbundle', _('look up remote changes'))
         opts = {}
         bundlecaps = kwargs.get('bundlecaps')
         if bundlecaps is not None:
             kwargs['bundlecaps'] = sorted(bundlecaps)
         else:
             bundlecaps = () # kwargs could have it to None
         for key, value in kwargs.iteritems():
             if value is None:
                 continue
             keytype = gboptsmap.get(key)
             if keytype is None:
                 assert False, 'unexpected'
             elif keytype == 'nodes':
                 value = encodelist(value)
             elif keytype in ('csv', 'scsv'):
                 value = ','.join(value)
             elif keytype == 'boolean':
                 value = '%i' % bool(value)
             elif keytype != 'plain':
                 raise KeyError('unknown getbundle option type %s'
                                % keytype)
             opts[key] = value
         f = self._callcompressable("getbundle", **opts)
         if any((cap.startswith('HG2') for cap in bundlecaps)):
             return bundle2.getunbundler(self.ui, f)
         else:
             return changegroupmod.cg1unpacker(f, 'UN')
 
     def unbundle(self, cg, heads, url):
         '''Send cg (a readable file-like object representing the
         changegroup to push, typically a chunkbuffer object) to the
         remote server as a bundle.
 
         When pushing a bundle10 stream, return an integer indicating the
         result of the push (see localrepository.addchangegroup()).
 
         When pushing a bundle20 stream, return a bundle20 stream.
 
         `url` is the url the client thinks it's pushing to, which is
         visible to hooks.
         '''
 
         if heads != ['force'] and self.capable('unbundlehash'):
             heads = encodelist(['hashed',
                                 hashlib.sha1(''.join(sorted(heads))).digest()])
         else:
             heads = encodelist(heads)
 
         if util.safehasattr(cg, 'deltaheader'):
             # this a bundle10, do the old style call sequence
             ret, output = self._callpush("unbundle", cg, heads=heads)
             if ret == "":
                 raise error.ResponseError(
                     _('push failed:'), output)
             try:
                 ret = int(ret)
             except ValueError:
                 raise error.ResponseError(
                     _('push failed (unexpected response):'), ret)
 
             for l in output.splitlines(True):
                 self.ui.status(_('remote: '), l)
         else:
             # bundle2 push. Send a stream, fetch a stream.
             stream = self._calltwowaystream('unbundle', cg, heads=heads)
             ret = bundle2.getunbundler(self.ui, stream)
         return ret
 
     def debugwireargs(self, one, two, three=None, four=None, five=None):
         # don't pass optional arguments left at their default value
         opts = {}
         if three is not None:
             opts['three'] = three
         if four is not None:
             opts['four'] = four
         return self._call('debugwireargs', one=one, two=two, **opts)
 
     def _call(self, cmd, **args):
         """execute <cmd> on the server
 
         The command is expected to return a simple string.
 
         returns the server reply as a string."""
         raise NotImplementedError()
 
     def _callstream(self, cmd, **args):
         """execute <cmd> on the server
 
         The command is expected to return a stream. Note that if the
         command doesn't return a stream, _callstream behaves
         differently for ssh and http peers.
 
         returns the server reply as a file like object.
         """
         raise NotImplementedError()
 
     def _callcompressable(self, cmd, **args):
         """execute <cmd> on the server
 
         The command is expected to return a stream.
 
         The stream may have been compressed in some implementations. This
         function takes care of the decompression. This is the only difference
         with _callstream.
 
         returns the server reply as a file like object.
         """
         raise NotImplementedError()
 
     def _callpush(self, cmd, fp, **args):
         """execute a <cmd> on server
 
         The command is expected to be related to a push. Push has a special
         return method.
 
         returns the server reply as a (ret, output) tuple. ret is either
         empty (error) or a stringified int.
         """
         raise NotImplementedError()
 
     def _calltwowaystream(self, cmd, fp, **args):
         """execute <cmd> on server
 
         The command will send a stream to the server and get a stream in reply.
         """
         raise NotImplementedError()
 
     def _abort(self, exception):
         """clearly abort the wire protocol connection and raise the exception
         """
         raise NotImplementedError()
 
 # server side
 
 # wire protocol command can either return a string or one of these classes.
 class streamres(object):
     """wireproto reply: binary stream
 
     The call was successful and the result is a stream.
     Iterate on the `self.gen` attribute to retrieve chunks.
     """
     def __init__(self, gen):
         self.gen = gen
 
 class pushres(object):
     """wireproto reply: success with simple integer return
 
     The call was successful and returned an integer contained in `self.res`.
     """
     def __init__(self, res):
         self.res = res
 
 class pusherr(object):
     """wireproto reply: failure
 
     The call failed. The `self.res` attribute contains the error message.
     """
     def __init__(self, res):
         self.res = res
 
 class ooberror(object):
     """wireproto reply: failure of a batch of operation
 
     Something failed during a batch call. The error message is stored in
     `self.message`.
     """
     def __init__(self, message):
         self.message = message
 
 def getdispatchrepo(repo, proto, command):
     """Obtain the repo used for processing wire protocol commands.
 
     The intent of this function is to serve as a monkeypatch point for
     extensions that need commands to operate on different repo views under
     specialized circumstances.
     """
     return repo.filtered('served')
 
 def dispatch(repo, proto, command):
     repo = getdispatchrepo(repo, proto, command)
     func, spec = commands[command]
     args = proto.getargs(spec)
     return func(repo, proto, *args)
 
 def options(cmd, keys, others):
     opts = {}
     for k in keys:
         if k in others:
             opts[k] = others[k]
             del others[k]
     if others:
         sys.stderr.write("warning: %s ignored unexpected arguments %s\n"
                          % (cmd, ",".join(others)))
     return opts
 
 def bundle1allowed(repo, action):
     """Whether a bundle1 operation is allowed from the server.
 
     Priority is:
 
     1. server.bundle1gd.<action> (if generaldelta active)
     2. server.bundle1.<action>
     3. server.bundle1gd (if generaldelta active)
     4. server.bundle1
     """
     ui = repo.ui
     gd = 'generaldelta' in repo.requirements
 
     if gd:
         v = ui.configbool('server', 'bundle1gd.%s' % action, None)
         if v is not None:
             return v
 
     v = ui.configbool('server', 'bundle1.%s' % action, None)
     if v is not None:
         return v
 
     if gd:
         v = ui.configbool('server', 'bundle1gd', None)
         if v is not None:
             return v
 
     return ui.configbool('server', 'bundle1', True)
 
 # list of commands
 commands = {}
 
 def wireprotocommand(name, args=''):
     """decorator for wire protocol command"""
     def register(func):
         commands[name] = (func, args)
         return func
     return register
 
 @wireprotocommand('batch', 'cmds *')
 def batch(repo, proto, cmds, others):
     repo = repo.filtered("served")
     res = []
     for pair in cmds.split(';'):
         op, args = pair.split(' ', 1)
         vals = {}
         for a in args.split(','):
             if a:
                 n, v = a.split('=')
                 vals[unescapearg(n)] = unescapearg(v)
         func, spec = commands[op]
         if spec:
             keys = spec.split()
             data = {}
             for k in keys:
                 if k == '*':
                     star = {}
                     for key in vals.keys():
                         if key not in keys:
                             star[key] = vals[key]
                     data['*'] = star
                 else:
                     data[k] = vals[k]
             result = func(repo, proto, *[data[k] for k in keys])
         else:
             result = func(repo, proto)
         if isinstance(result, ooberror):
             return result
         res.append(escapearg(result))
     return ';'.join(res)
 
 @wireprotocommand('between', 'pairs')
 def between(repo, proto, pairs):
     pairs = [decodelist(p, '-') for p in pairs.split(" ")]
     r = []
     for b in repo.between(pairs):
         r.append(encodelist(b) + "\n")
     return "".join(r)
 
 @wireprotocommand('branchmap')
 def branchmap(repo, proto):
     branchmap = repo.branchmap()
     heads = []
     for branch, nodes in branchmap.iteritems():
         branchname = urlreq.quote(encoding.fromlocal(branch))
         branchnodes = encodelist(nodes)
         heads.append('%s %s' % (branchname, branchnodes))
     return '\n'.join(heads)
 
 @wireprotocommand('branches', 'nodes')
 def branches(repo, proto, nodes):
     nodes = decodelist(nodes)
     r = []
     for b in repo.branches(nodes):
         r.append(encodelist(b) + "\n")
     return "".join(r)
 
 @wireprotocommand('clonebundles', '')
 def clonebundles(repo, proto):
     """Server command for returning info for available bundles to seed clones.
 
     Clients will parse this response and determine what bundle to fetch.
 
     Extensions may wrap this command to filter or dynamically emit data
     depending on the request. e.g. you could advertise URLs for the closest
     data center given the client's IP address.
     """
     return repo.opener.tryread('clonebundles.manifest')
 
 wireprotocaps = ['lookup', 'changegroupsubset', 'branchmap', 'pushkey',
                  'known', 'getbundle', 'unbundlehash', 'batch']
 
 def _capabilities(repo, proto):
     """return a list of capabilities for a repo
 
     This function exists to allow extensions to easily wrap capabilities
     computation
 
     - returns a lists: easy to alter
     - change done here will be propagated to both `capabilities` and `hello`
       command without any other action needed.
     """
     # copy to prevent modification of the global list
     caps = list(wireprotocaps)
     if streamclone.allowservergeneration(repo.ui):
         if repo.ui.configbool('server', 'preferuncompressed', False):
             caps.append('stream-preferred')
         requiredformats = repo.requirements & repo.supportedformats
         # if our local revlogs are just revlogv1, add 'stream' cap
         if not requiredformats - set(('revlogv1',)):
             caps.append('stream')
         # otherwise, add 'streamreqs' detailing our local revlog format
         else:
             caps.append('streamreqs=%s' % ','.join(sorted(requiredformats)))
     if repo.ui.configbool('experimental', 'bundle2-advertise', True):
         capsblob = bundle2.encodecaps(bundle2.getrepocaps(repo))
         caps.append('bundle2=' + urlreq.quote(capsblob))
     caps.append('unbundle=%s' % ','.join(bundle2.bundlepriority))
     caps.append(
         'httpheader=%d' % repo.ui.configint('server', 'maxhttpheaderlen', 1024))
     if repo.ui.configbool('experimental', 'httppostargs', False):
         caps.append('httppostargs')
     return caps
 
 # If you are writing an extension and consider wrapping this function. Wrap
 # `_capabilities` instead.
 @wireprotocommand('capabilities')
 def capabilities(repo, proto):
     return ' '.join(_capabilities(repo, proto))
 
 @wireprotocommand('changegroup', 'roots')
 def changegroup(repo, proto, roots):
     nodes = decodelist(roots)
     cg = changegroupmod.changegroup(repo, nodes, 'serve')
     return streamres(proto.groupchunks(cg))
 
 @wireprotocommand('changegroupsubset', 'bases heads')
 def changegroupsubset(repo, proto, bases, heads):
     bases = decodelist(bases)
     heads = decodelist(heads)
     cg = changegroupmod.changegroupsubset(repo, bases, heads, 'serve')
     return streamres(proto.groupchunks(cg))
 
 @wireprotocommand('debugwireargs', 'one two *')
 def debugwireargs(repo, proto, one, two, others):
     # only accept optional args from the known set
     opts = options('debugwireargs', ['three', 'four'], others)
     return repo.debugwireargs(one, two, **opts)
 
 @wireprotocommand('getbundle', '*')
 def getbundle(repo, proto, others):
     opts = options('getbundle', gboptsmap.keys(), others)
     for k, v in opts.iteritems():
         keytype = gboptsmap[k]
         if keytype == 'nodes':
             opts[k] = decodelist(v)
         elif keytype == 'csv':
             opts[k] = list(v.split(','))
         elif keytype == 'scsv':
             opts[k] = set(v.split(','))
         elif keytype == 'boolean':
             # Client should serialize False as '0', which is a non-empty string
             # so it evaluates as a True bool.
             if v == '0':
                 opts[k] = False
             else:
                 opts[k] = bool(v)
         elif keytype != 'plain':
             raise KeyError('unknown getbundle option type %s'
                            % keytype)
 
     if not bundle1allowed(repo, 'pull'):
         if not exchange.bundle2requested(opts.get('bundlecaps')):
             return ooberror(bundle2required)
 
-    cg = exchange.getbundle(repo, 'serve', **opts)
-    return streamres(proto.groupchunks(cg))
+    chunks = exchange.getbundlechunks(repo, 'serve', **opts)
+    # TODO avoid util.chunkbuffer() here since it is adding overhead to
+    # what is fundamentally a generator proxying operation.
+    return streamres(proto.groupchunks(util.chunkbuffer(chunks)))
 
 @wireprotocommand('heads')
 def heads(repo, proto):
     h = repo.heads()
     return encodelist(h) + "\n"
 
 @wireprotocommand('hello')
 def hello(repo, proto):
     '''the hello command returns a set of lines describing various
     interesting things about the server, in an RFC822-like format.
     Currently the only one defined is "capabilities", which
     consists of a line in the form:
 
     capabilities: space separated list of tokens
     '''
     return "capabilities: %s\n" % (capabilities(repo, proto))
 
 @wireprotocommand('listkeys', 'namespace')
 def listkeys(repo, proto, namespace):
     d = repo.listkeys(encoding.tolocal(namespace)).items()
     return pushkeymod.encodekeys(d)
 
 @wireprotocommand('lookup', 'key')
 def lookup(repo, proto, key):
     try:
         k = encoding.tolocal(key)
         c = repo[k]
         r = c.hex()
         success = 1
     except Exception as inst:
         r = str(inst)
         success = 0
     return "%s %s\n" % (success, r)
 
 @wireprotocommand('known', 'nodes *')
 def known(repo, proto, nodes, others):
     return ''.join(b and "1" or "0" for b in repo.known(decodelist(nodes)))
 
 @wireprotocommand('pushkey', 'namespace key old new')
 def pushkey(repo, proto, namespace, key, old, new):
     # compatibility with pre-1.8 clients which were accidentally
     # sending raw binary nodes rather than utf-8-encoded hex
     if len(new) == 20 and new.encode('string-escape') != new:
         # looks like it could be a binary node
         try:
             new.decode('utf-8')
             new = encoding.tolocal(new) # but cleanly decodes as UTF-8
         except UnicodeDecodeError:
             pass # binary, leave unmodified
     else:
         new = encoding.tolocal(new) # normal path
 
     if util.safehasattr(proto, 'restore'):
 
         proto.redirect()
 
         try:
             r = repo.pushkey(encoding.tolocal(namespace), encoding.tolocal(key),
                              encoding.tolocal(old), new) or False
         except error.Abort:
             r = False
 
         output = proto.restore()
 
         return '%s\n%s' % (int(r), output)
 
     r = repo.pushkey(encoding.tolocal(namespace), encoding.tolocal(key),
                      encoding.tolocal(old), new)
     return '%s\n' % int(r)
 
 @wireprotocommand('stream_out')
 def stream(repo, proto):
     '''If the server supports streaming clone, it advertises the "stream"
     capability with a value representing the version and flags of the repo
     it is serving. Client checks to see if it understands the format.
     '''
     if not streamclone.allowservergeneration(repo.ui):
         return '1\n'
 
     def getstream(it):
         yield '0\n'
         for chunk in it:
             yield chunk
 
     try:
         # LockError may be raised before the first result is yielded. Don't
         # emit output until we're sure we got the lock successfully.
         it = streamclone.generatev1wireproto(repo)
         return streamres(getstream(it))
     except error.LockError:
         return '2\n'
 
 @wireprotocommand('unbundle', 'heads')
 def unbundle(repo, proto, heads):
     their_heads = decodelist(heads)
 
     try:
         proto.redirect()
 
         exchange.check_heads(repo, their_heads, 'preparing changes')
 
         # write bundle data to temporary file because it can be big
         fd, tempname = tempfile.mkstemp(prefix='hg-unbundle-')
         fp = os.fdopen(fd, 'wb+')
         r = 0
         try:
             proto.getfile(fp)
             fp.seek(0)
             gen = exchange.readbundle(repo.ui, fp, None)
             if (isinstance(gen, changegroupmod.cg1unpacker)
                 and not bundle1allowed(repo, 'push')):
                 return ooberror(bundle2required)
 
             r = exchange.unbundle(repo, gen, their_heads, 'serve',
                                   proto._client())
             if util.safehasattr(r, 'addpart'):
                 # The return looks streamable, we are in the bundle2 case and
                 # should return a stream.
                 return streamres(r.getchunks())
             return pushres(r)
 
         finally:
             fp.close()
             os.unlink(tempname)
 
     except (error.BundleValueError, error.Abort, error.PushRaced) as exc:
         # handle non-bundle2 case first
         if not getattr(exc, 'duringunbundle2', False):
             try:
                 raise
             except error.Abort:
                 # The old code we moved used sys.stderr directly.
                 # We did not change it to minimise code change.
                 # This need to be moved to something proper.
                 # Feel free to do it.
                 sys.stderr.write("abort: %s\n" % exc)
                 return pushres(0)
             except error.PushRaced:
                 return pusherr(str(exc))
 
         bundler = bundle2.bundle20(repo.ui)
         for out in getattr(exc, '_bundle2salvagedoutput', ()):
             bundler.addpart(out)
         try:
             try:
                 raise
             except error.PushkeyFailed as exc:
                 # check client caps
                 remotecaps = getattr(exc, '_replycaps', None)
                 if (remotecaps is not None
                         and 'pushkey' not in remotecaps.get('error', ())):
                     # no support remote side, fallback to Abort handler.
                     raise
                 part = bundler.newpart('error:pushkey')
                 part.addparam('in-reply-to', exc.partid)
                 if exc.namespace is not None:
                     part.addparam('namespace', exc.namespace, mandatory=False)
                 if exc.key is not None:
                     part.addparam('key', exc.key, mandatory=False)
                 if exc.new is not None:
                     part.addparam('new', exc.new, mandatory=False)
                 if exc.old is not None:
                     part.addparam('old', exc.old, mandatory=False)
                 if exc.ret is not None:
                     part.addparam('ret', exc.ret, mandatory=False)
         except error.BundleValueError as exc:
             errpart = bundler.newpart('error:unsupportedcontent')
             if exc.parttype is not None:
                 errpart.addparam('parttype', exc.parttype)
             if exc.params:
                 errpart.addparam('params', '\0'.join(exc.params))
         except error.Abort as exc:
             manargs = [('message', str(exc))]
             advargs = []
             if exc.hint is not None:
                 advargs.append(('hint', exc.hint))
             bundler.addpart(bundle2.bundlepart('error:abort',
                                                manargs, advargs))
         except error.PushRaced as exc:
             bundler.newpart('error:pushraced', [('message', str(exc))])
         return streamres(bundler.getchunks())

tests/test-getbundle.t

@@ -1,272 +1,272
 #require serve
 
 = Test the getbundle() protocol function =
 
 Create a test repository:
 
   $ hg init repo
   $ cd repo
   $ hg debugbuilddag -n -m '+2 :fork +5 :p1 *fork +6 :p2 /p1 :m1 +3' > /dev/null
   $ hg log -G --template '{node}\n'
   o  10c14a2cc935e1d8c31f9e98587dcf27fb08a6da
   |
   o  4801a72e5d88cb515b0c7e40fae34180f3f837f2
   |
   o  0b2f73f04880d9cb6a5cd8a757f0db0ad01e32c3
   |
   o    8365676dbab05860ce0d9110f2af51368b961bbd
   |\
   | o  5686dbbd9fc46cb806599c878d02fe1cb56b83d3
   | |
   | o  13c0170174366b441dc68e8e33757232fa744458
   | |
   | o  63476832d8ec6558cf9bbe3cbe0c757e5cf18043
   | |
   | o  700b7e19db54103633c4bf4a6a6b6d55f4d50c03
   | |
   | o  928b5f94cdb278bb536eba552de348a4e92ef24d
   | |
   | o  f34414c64173e0ecb61b25dc55e116dbbcc89bee
   | |
   | o  8931463777131cd73923e560b760061f2aa8a4bc
   | |
   o |  6621d79f61b23ec74cf4b69464343d9e0980ec8b
   | |
   o |  bac16991d12ff45f9dc43c52da1946dfadb83e80
   | |
   o |  ff42371d57168345fdf1a3aac66a51f6a45d41d2
   | |
   o |  d5f6e1ea452285324836a49d7d3c2a63cfed1d31
   | |
   o |  713346a995c363120712aed1aee7e04afd867638
   |/
   o  29a4d1f17bd3f0779ca0525bebb1cfb51067c738
   |
   o  7704483d56b2a7b5db54dcee7c62378ac629b348
   
   $ cd ..
 
 
 = Test locally =
 
 Get everything:
 
   $ hg debuggetbundle repo bundle
   $ hg debugbundle bundle
   7704483d56b2a7b5db54dcee7c62378ac629b348
   29a4d1f17bd3f0779ca0525bebb1cfb51067c738
   713346a995c363120712aed1aee7e04afd867638
   d5f6e1ea452285324836a49d7d3c2a63cfed1d31
   ff42371d57168345fdf1a3aac66a51f6a45d41d2
   bac16991d12ff45f9dc43c52da1946dfadb83e80
   6621d79f61b23ec74cf4b69464343d9e0980ec8b
   8931463777131cd73923e560b760061f2aa8a4bc
   f34414c64173e0ecb61b25dc55e116dbbcc89bee
   928b5f94cdb278bb536eba552de348a4e92ef24d
   700b7e19db54103633c4bf4a6a6b6d55f4d50c03
   63476832d8ec6558cf9bbe3cbe0c757e5cf18043
   13c0170174366b441dc68e8e33757232fa744458
   5686dbbd9fc46cb806599c878d02fe1cb56b83d3
   8365676dbab05860ce0d9110f2af51368b961bbd
   0b2f73f04880d9cb6a5cd8a757f0db0ad01e32c3
   4801a72e5d88cb515b0c7e40fae34180f3f837f2
   10c14a2cc935e1d8c31f9e98587dcf27fb08a6da
 
 Get part of linear run:
 
   $ hg debuggetbundle repo bundle -H 4801a72e5d88cb515b0c7e40fae34180f3f837f2 -C 8365676dbab05860ce0d9110f2af51368b961bbd
   $ hg debugbundle bundle
   0b2f73f04880d9cb6a5cd8a757f0db0ad01e32c3
   4801a72e5d88cb515b0c7e40fae34180f3f837f2
 
 Get missing branch and merge:
 
   $ hg debuggetbundle repo bundle -H 4801a72e5d88cb515b0c7e40fae34180f3f837f2 -C 13c0170174366b441dc68e8e33757232fa744458
   $ hg debugbundle bundle
   713346a995c363120712aed1aee7e04afd867638
   d5f6e1ea452285324836a49d7d3c2a63cfed1d31
   ff42371d57168345fdf1a3aac66a51f6a45d41d2
   bac16991d12ff45f9dc43c52da1946dfadb83e80
   6621d79f61b23ec74cf4b69464343d9e0980ec8b
   5686dbbd9fc46cb806599c878d02fe1cb56b83d3
   8365676dbab05860ce0d9110f2af51368b961bbd
   0b2f73f04880d9cb6a5cd8a757f0db0ad01e32c3
   4801a72e5d88cb515b0c7e40fae34180f3f837f2
 
 Get from only one head:
 
   $ hg debuggetbundle repo bundle -H 928b5f94cdb278bb536eba552de348a4e92ef24d -C 29a4d1f17bd3f0779ca0525bebb1cfb51067c738
   $ hg debugbundle bundle
   8931463777131cd73923e560b760061f2aa8a4bc
   f34414c64173e0ecb61b25dc55e116dbbcc89bee
   928b5f94cdb278bb536eba552de348a4e92ef24d
 
 Get parts of two branches:
 
   $ hg debuggetbundle repo bundle -H 13c0170174366b441dc68e8e33757232fa744458 -C 700b7e19db54103633c4bf4a6a6b6d55f4d50c03 -H bac16991d12ff45f9dc43c52da1946dfadb83e80 -C d5f6e1ea452285324836a49d7d3c2a63cfed1d31
   $ hg debugbundle bundle
   ff42371d57168345fdf1a3aac66a51f6a45d41d2
   bac16991d12ff45f9dc43c52da1946dfadb83e80
   63476832d8ec6558cf9bbe3cbe0c757e5cf18043
   13c0170174366b441dc68e8e33757232fa744458
 
 Check that we get all needed file changes:
 
   $ hg debugbundle bundle --all
   format: id, p1, p2, cset, delta base, len(delta)
   
   changelog
   ff42371d57168345fdf1a3aac66a51f6a45d41d2 d5f6e1ea452285324836a49d7d3c2a63cfed1d31 0000000000000000000000000000000000000000 ff42371d57168345fdf1a3aac66a51f6a45d41d2 d5f6e1ea452285324836a49d7d3c2a63cfed1d31 99
   bac16991d12ff45f9dc43c52da1946dfadb83e80 ff42371d57168345fdf1a3aac66a51f6a45d41d2 0000000000000000000000000000000000000000 bac16991d12ff45f9dc43c52da1946dfadb83e80 ff42371d57168345fdf1a3aac66a51f6a45d41d2 99
   63476832d8ec6558cf9bbe3cbe0c757e5cf18043 700b7e19db54103633c4bf4a6a6b6d55f4d50c03 0000000000000000000000000000000000000000 63476832d8ec6558cf9bbe3cbe0c757e5cf18043 bac16991d12ff45f9dc43c52da1946dfadb83e80 102
   13c0170174366b441dc68e8e33757232fa744458 63476832d8ec6558cf9bbe3cbe0c757e5cf18043 0000000000000000000000000000000000000000 13c0170174366b441dc68e8e33757232fa744458 63476832d8ec6558cf9bbe3cbe0c757e5cf18043 102
   
   manifest
   dac7984588fc4eea7acbf39693a9c1b06f5b175d 591f732a3faf1fb903815273f3c199a514a61ccb 0000000000000000000000000000000000000000 ff42371d57168345fdf1a3aac66a51f6a45d41d2 591f732a3faf1fb903815273f3c199a514a61ccb 113
   0772616e6b48a76afb6c1458e193cbb3dae2e4ff dac7984588fc4eea7acbf39693a9c1b06f5b175d 0000000000000000000000000000000000000000 bac16991d12ff45f9dc43c52da1946dfadb83e80 dac7984588fc4eea7acbf39693a9c1b06f5b175d 113
   eb498cd9af6c44108e43041e951ce829e29f6c80 bff2f4817ced57b386caf7c4e3e36a4bc9af7e93 0000000000000000000000000000000000000000 63476832d8ec6558cf9bbe3cbe0c757e5cf18043 0772616e6b48a76afb6c1458e193cbb3dae2e4ff 295
   b15709c071ddd2d93188508ba156196ab4f19620 eb498cd9af6c44108e43041e951ce829e29f6c80 0000000000000000000000000000000000000000 13c0170174366b441dc68e8e33757232fa744458 eb498cd9af6c44108e43041e951ce829e29f6c80 114
   
   mf
   4f73f97080266ab8e0c0561ca8d0da3eaf65b695 301ca08d026bb72cb4258a9d211bdf7ca0bcd810 0000000000000000000000000000000000000000 ff42371d57168345fdf1a3aac66a51f6a45d41d2 301ca08d026bb72cb4258a9d211bdf7ca0bcd810 17
   c7b583de053293870e145f45bd2d61643563fd06 4f73f97080266ab8e0c0561ca8d0da3eaf65b695 0000000000000000000000000000000000000000 bac16991d12ff45f9dc43c52da1946dfadb83e80 4f73f97080266ab8e0c0561ca8d0da3eaf65b695 18
   266ee3c0302a5a18f1cf96817ac79a51836179e9 edc0f6b8db80d68ae6aff2b19f7e5347ab68fa63 0000000000000000000000000000000000000000 63476832d8ec6558cf9bbe3cbe0c757e5cf18043 c7b583de053293870e145f45bd2d61643563fd06 149
   698c6a36220548cd3903ca7dada27c59aa500c52 266ee3c0302a5a18f1cf96817ac79a51836179e9 0000000000000000000000000000000000000000 13c0170174366b441dc68e8e33757232fa744458 266ee3c0302a5a18f1cf96817ac79a51836179e9 19
   
   nf11
   33fbc651630ffa7ccbebfe4eb91320a873e7291c 0000000000000000000000000000000000000000 0000000000000000000000000000000000000000 63476832d8ec6558cf9bbe3cbe0c757e5cf18043 0000000000000000000000000000000000000000 16
   
   nf12
   ddce0544363f037e9fb889faca058f52dc01c0a5 0000000000000000000000000000000000000000 0000000000000000000000000000000000000000 13c0170174366b441dc68e8e33757232fa744458 0000000000000000000000000000000000000000 16
   
   nf4
   3c1407305701051cbed9f9cb9a68bdfb5997c235 0000000000000000000000000000000000000000 0000000000000000000000000000000000000000 ff42371d57168345fdf1a3aac66a51f6a45d41d2 0000000000000000000000000000000000000000 15
   
   nf5
   0dbd89c185f53a1727c54cd1ce256482fa23968e 0000000000000000000000000000000000000000 0000000000000000000000000000000000000000 bac16991d12ff45f9dc43c52da1946dfadb83e80 0000000000000000000000000000000000000000 15
 
 Get branch and merge:
 
   $ hg debuggetbundle repo bundle -C 7704483d56b2a7b5db54dcee7c62378ac629b348 -H 0b2f73f04880d9cb6a5cd8a757f0db0ad01e32c3
   $ hg debugbundle bundle
   29a4d1f17bd3f0779ca0525bebb1cfb51067c738
   713346a995c363120712aed1aee7e04afd867638
   d5f6e1ea452285324836a49d7d3c2a63cfed1d31
   ff42371d57168345fdf1a3aac66a51f6a45d41d2
   bac16991d12ff45f9dc43c52da1946dfadb83e80
   6621d79f61b23ec74cf4b69464343d9e0980ec8b
   8931463777131cd73923e560b760061f2aa8a4bc
   f34414c64173e0ecb61b25dc55e116dbbcc89bee
   928b5f94cdb278bb536eba552de348a4e92ef24d
   700b7e19db54103633c4bf4a6a6b6d55f4d50c03
   63476832d8ec6558cf9bbe3cbe0c757e5cf18043
   13c0170174366b441dc68e8e33757232fa744458
   5686dbbd9fc46cb806599c878d02fe1cb56b83d3
   8365676dbab05860ce0d9110f2af51368b961bbd
   0b2f73f04880d9cb6a5cd8a757f0db0ad01e32c3
 
 = Test bundle2 =
 
   $ hg debuggetbundle repo bundle -t bundle2
   $ hg debugbundle bundle
   Stream params: {}
-  changegroup -- "sortdict([('version', '01'), ('nbchanges', '18')])"
+  changegroup -- "sortdict([('version', '01')])"
       7704483d56b2a7b5db54dcee7c62378ac629b348
       29a4d1f17bd3f0779ca0525bebb1cfb51067c738
       713346a995c363120712aed1aee7e04afd867638
       d5f6e1ea452285324836a49d7d3c2a63cfed1d31
       ff42371d57168345fdf1a3aac66a51f6a45d41d2
       bac16991d12ff45f9dc43c52da1946dfadb83e80
       6621d79f61b23ec74cf4b69464343d9e0980ec8b
       8931463777131cd73923e560b760061f2aa8a4bc
       f34414c64173e0ecb61b25dc55e116dbbcc89bee
       928b5f94cdb278bb536eba552de348a4e92ef24d
       700b7e19db54103633c4bf4a6a6b6d55f4d50c03
       63476832d8ec6558cf9bbe3cbe0c757e5cf18043
       13c0170174366b441dc68e8e33757232fa744458
       5686dbbd9fc46cb806599c878d02fe1cb56b83d3
       8365676dbab05860ce0d9110f2af51368b961bbd
       0b2f73f04880d9cb6a5cd8a757f0db0ad01e32c3
       4801a72e5d88cb515b0c7e40fae34180f3f837f2
       10c14a2cc935e1d8c31f9e98587dcf27fb08a6da
 = Test via HTTP =
 
 Get everything:
 
   $ hg serve -R repo -p $HGPORT -d --pid-file=hg.pid -E error.log -A access.log
   $ cat hg.pid >> $DAEMON_PIDS
   $ hg debuggetbundle http://localhost:$HGPORT/ bundle
   $ hg debugbundle bundle
   7704483d56b2a7b5db54dcee7c62378ac629b348
   29a4d1f17bd3f0779ca0525bebb1cfb51067c738
   713346a995c363120712aed1aee7e04afd867638
   d5f6e1ea452285324836a49d7d3c2a63cfed1d31
   ff42371d57168345fdf1a3aac66a51f6a45d41d2
   bac16991d12ff45f9dc43c52da1946dfadb83e80
   6621d79f61b23ec74cf4b69464343d9e0980ec8b
   8931463777131cd73923e560b760061f2aa8a4bc
   f34414c64173e0ecb61b25dc55e116dbbcc89bee
   928b5f94cdb278bb536eba552de348a4e92ef24d
   700b7e19db54103633c4bf4a6a6b6d55f4d50c03
   63476832d8ec6558cf9bbe3cbe0c757e5cf18043
   13c0170174366b441dc68e8e33757232fa744458
   5686dbbd9fc46cb806599c878d02fe1cb56b83d3
   8365676dbab05860ce0d9110f2af51368b961bbd
   0b2f73f04880d9cb6a5cd8a757f0db0ad01e32c3
   4801a72e5d88cb515b0c7e40fae34180f3f837f2
   10c14a2cc935e1d8c31f9e98587dcf27fb08a6da
 
 Get parts of two branches:
 
   $ hg debuggetbundle http://localhost:$HGPORT/ bundle -H 13c0170174366b441dc68e8e33757232fa744458 -C 700b7e19db54103633c4bf4a6a6b6d55f4d50c03 -H bac16991d12ff45f9dc43c52da1946dfadb83e80 -C d5f6e1ea452285324836a49d7d3c2a63cfed1d31
   $ hg debugbundle bundle
   ff42371d57168345fdf1a3aac66a51f6a45d41d2
   bac16991d12ff45f9dc43c52da1946dfadb83e80
   63476832d8ec6558cf9bbe3cbe0c757e5cf18043
   13c0170174366b441dc68e8e33757232fa744458
 
 Check that we get all needed file changes:
 
   $ hg debugbundle bundle --all
   format: id, p1, p2, cset, delta base, len(delta)
   
   changelog
   ff42371d57168345fdf1a3aac66a51f6a45d41d2 d5f6e1ea452285324836a49d7d3c2a63cfed1d31 0000000000000000000000000000000000000000 ff42371d57168345fdf1a3aac66a51f6a45d41d2 d5f6e1ea452285324836a49d7d3c2a63cfed1d31 99
   bac16991d12ff45f9dc43c52da1946dfadb83e80 ff42371d57168345fdf1a3aac66a51f6a45d41d2 0000000000000000000000000000000000000000 bac16991d12ff45f9dc43c52da1946dfadb83e80 ff42371d57168345fdf1a3aac66a51f6a45d41d2 99
   63476832d8ec6558cf9bbe3cbe0c757e5cf18043 700b7e19db54103633c4bf4a6a6b6d55f4d50c03 0000000000000000000000000000000000000000 63476832d8ec6558cf9bbe3cbe0c757e5cf18043 bac16991d12ff45f9dc43c52da1946dfadb83e80 102
   13c0170174366b441dc68e8e33757232fa744458 63476832d8ec6558cf9bbe3cbe0c757e5cf18043 0000000000000000000000000000000000000000 13c0170174366b441dc68e8e33757232fa744458 63476832d8ec6558cf9bbe3cbe0c757e5cf18043 102
   
   manifest
   dac7984588fc4eea7acbf39693a9c1b06f5b175d 591f732a3faf1fb903815273f3c199a514a61ccb 0000000000000000000000000000000000000000 ff42371d57168345fdf1a3aac66a51f6a45d41d2 591f732a3faf1fb903815273f3c199a514a61ccb 113
   0772616e6b48a76afb6c1458e193cbb3dae2e4ff dac7984588fc4eea7acbf39693a9c1b06f5b175d 0000000000000000000000000000000000000000 bac16991d12ff45f9dc43c52da1946dfadb83e80 dac7984588fc4eea7acbf39693a9c1b06f5b175d 113
   eb498cd9af6c44108e43041e951ce829e29f6c80 bff2f4817ced57b386caf7c4e3e36a4bc9af7e93 0000000000000000000000000000000000000000 63476832d8ec6558cf9bbe3cbe0c757e5cf18043 0772616e6b48a76afb6c1458e193cbb3dae2e4ff 295
   b15709c071ddd2d93188508ba156196ab4f19620 eb498cd9af6c44108e43041e951ce829e29f6c80 0000000000000000000000000000000000000000 13c0170174366b441dc68e8e33757232fa744458 eb498cd9af6c44108e43041e951ce829e29f6c80 114
   
   mf
   4f73f97080266ab8e0c0561ca8d0da3eaf65b695 301ca08d026bb72cb4258a9d211bdf7ca0bcd810 0000000000000000000000000000000000000000 ff42371d57168345fdf1a3aac66a51f6a45d41d2 301ca08d026bb72cb4258a9d211bdf7ca0bcd810 17
   c7b583de053293870e145f45bd2d61643563fd06 4f73f97080266ab8e0c0561ca8d0da3eaf65b695 0000000000000000000000000000000000000000 bac16991d12ff45f9dc43c52da1946dfadb83e80 4f73f97080266ab8e0c0561ca8d0da3eaf65b695 18
   266ee3c0302a5a18f1cf96817ac79a51836179e9 edc0f6b8db80d68ae6aff2b19f7e5347ab68fa63 0000000000000000000000000000000000000000 63476832d8ec6558cf9bbe3cbe0c757e5cf18043 c7b583de053293870e145f45bd2d61643563fd06 149
   698c6a36220548cd3903ca7dada27c59aa500c52 266ee3c0302a5a18f1cf96817ac79a51836179e9 0000000000000000000000000000000000000000 13c0170174366b441dc68e8e33757232fa744458 266ee3c0302a5a18f1cf96817ac79a51836179e9 19
   
   nf11
   33fbc651630ffa7ccbebfe4eb91320a873e7291c 0000000000000000000000000000000000000000 0000000000000000000000000000000000000000 63476832d8ec6558cf9bbe3cbe0c757e5cf18043 0000000000000000000000000000000000000000 16
   
   nf12
   ddce0544363f037e9fb889faca058f52dc01c0a5 0000000000000000000000000000000000000000 0000000000000000000000000000000000000000 13c0170174366b441dc68e8e33757232fa744458 0000000000000000000000000000000000000000 16
   
   nf4
   3c1407305701051cbed9f9cb9a68bdfb5997c235 0000000000000000000000000000000000000000 0000000000000000000000000000000000000000 ff42371d57168345fdf1a3aac66a51f6a45d41d2 0000000000000000000000000000000000000000 15
   
   nf5
   0dbd89c185f53a1727c54cd1ce256482fa23968e 0000000000000000000000000000000000000000 0000000000000000000000000000000000000000 bac16991d12ff45f9dc43c52da1946dfadb83e80 0000000000000000000000000000000000000000 15
 
 Verify we hit the HTTP server:
 
   $ cat access.log
   * - - [*] "GET /?cmd=capabilities HTTP/1.1" 200 - (glob)
   * - - [*] "GET /?cmd=getbundle HTTP/1.1" 200 - (glob)
   * - - [*] "GET /?cmd=capabilities HTTP/1.1" 200 - (glob)
   * - - [*] "GET /?cmd=getbundle HTTP/1.1" 200 - x-hgarg-1:common=700b7e19db54103633c4bf4a6a6b6d55f4d50c03+d5f6e1ea452285324836a49d7d3c2a63cfed1d31&heads=13c0170174366b441dc68e8e33757232fa744458+bac16991d12ff45f9dc43c52da1946dfadb83e80 (glob)
 
   $ cat error.log