comparison mercurial/changegroup.py @ 43076:2372284d9457

formatting: blacken the codebase

This is using my patch to black
(https://github.com/psf/black/pull/826) so we don't un-wrap collection
literals.

Done with:

  hg files 'set:**.py - mercurial/thirdparty/** - "contrib/python-zstandard/**"' | xargs black -S

# skip-blame mass-reformatting only

# no-check-commit reformats foo_bar functions

Differential Revision: https://phab.mercurial-scm.org/D6971

43075:57875cf423c9

    phases,
    pycompat,
    util,
)

from .interfaces import (
    repository,
)

_CHANGEGROUPV1_DELTA_HEADER = struct.Struct("20s20s20s20s")
_CHANGEGROUPV2_DELTA_HEADER = struct.Struct("20s20s20s20s20s")
_CHANGEGROUPV3_DELTA_HEADER = struct.Struct(">20s20s20s20s20sH")

LFS_REQUIREMENT = 'lfs'

readexactly = util.readexactly

def getchunk(stream):
    """return the next chunk from stream as a string"""
    d = readexactly(stream, 4)
    l = struct.unpack(">l", d)[0]

        if l:
            raise error.Abort(_("invalid chunk length %d") % l)
        return ""
    return readexactly(stream, l - 4)

def chunkheader(length):
    """return a changegroup chunk header (string)"""
    return struct.pack(">l", length + 4)

def closechunk():
    """return a changegroup chunk header (string) for a zero-length chunk"""
    return struct.pack(">l", 0)

def _fileheader(path):
    """Obtain a changegroup chunk header for a named path."""
    return chunkheader(len(path)) + path

def writechunks(ui, chunks, filename, vfs=None):
    """Write chunks to a file and return its filename.

    The stream is assumed to be a bundle file.

            if filename and vfs:
                vfs.unlink(cleanup)
            else:
                os.unlink(cleanup)

class cg1unpacker(object):
    """Unpacker for cg1 changegroup streams.

    A changegroup unpacker handles the framing of the revision data in
    the wire format. Most consumers will want to use the apply()

    consumers should prefer apply() instead.

    A few other public methods exist. Those are used only for
    bundlerepo and some debug commands - their use is discouraged.
    """
    deltaheader = _CHANGEGROUPV1_DELTA_HEADER
    deltaheadersize = deltaheader.size
    version = '01'
    _grouplistcount = 1 # One list of files after the manifests

    def __init__(self, fh, alg, extras=None):
        if alg is None:
            alg = 'UN'
        if alg not in util.compengines.supportedbundletypes:
            raise error.Abort(_('unknown stream compression type: %s')
                             % alg)
        if alg == 'BZ':
            alg = '_truncatedBZ'

        compengine = util.compengines.forbundletype(alg)
        self._stream = compengine.decompressorreader(fh)

    # These methods (compressed, read, seek, tell) all appear to only
    # be used by bundlerepo, but it's a little hard to tell.
    def compressed(self):
        return self._type is not None and self._type != 'UN'
    def read(self, l):
        return self._stream.read(l)
    def seek(self, pos):
        return self._stream.seek(pos)
    def tell(self):
        return self._stream.tell()
    def close(self):
        return self._stream.close()

    def _chunklength(self):
        d = readexactly(self._stream, 4)

                    break
                noentries = False
                yield chunkheader(len(chunk))
                pos = 0
                while pos < len(chunk):
                    next = pos + 2**20
                    yield chunk[pos:next]
                    pos = next
            yield closechunk()

    def _unpackmanifests(self, repo, revmap, trp, prog):

        deltas = self.deltaiter()
        repo.manifestlog.getstorage(b'').addgroup(deltas, revmap, trp)
        prog.complete()
        self.callback = None

    def apply(self, repo, tr, srctype, url, targetphase=phases.draft,
              expectedtotal=None):
        """Add the changegroup returned by source.read() to this repo.
        srctype is a string like 'push', 'pull', or 'unbundle'.  url is
        the URL of the repo where this changegroup is coming from.

        Return an integer summarizing the change to this repo:

        - more heads than before: 1+added heads (2..n)
        - fewer heads than before: -1-removed heads (-2..-n)
        - number of heads stays the same: 1
        """
        repo = repo.unfiltered()
        def csmap(x):
            repo.ui.debug("add changeset %s\n" % short(x))
            return len(cl)

        def revmap(x):

            # case we use the top level data. We overwrite the argument
            # because we need to use the top level value (if they exist)
            # in this function.
            srctype = tr.hookargs.setdefault('source', srctype)
            tr.hookargs.setdefault('url', url)
            repo.hook('prechangegroup',
                      throw=True, **pycompat.strkwargs(tr.hookargs))

            # write changelog data to temp files so concurrent readers
            # will not see an inconsistent view
            cl = repo.changelog
            cl.delayupdate(tr)

            trp = weakref.proxy(tr)
            # pull off the changeset group
            repo.ui.status(_("adding changesets\n"))
            clstart = len(cl)
            progress = repo.ui.makeprogress(_('changesets'), unit=_('chunks'),
                                            total=expectedtotal)
            self.callback = progress.increment

            efiles = set()
            def onchangelog(cl, node):
                efiles.update(cl.readfiles(node))

            self.changelogheader()
            deltas = self.deltaiter()
            cgnodes = cl.addgroup(deltas, csmap, trp, addrevisioncb=onchangelog)
            efiles = len(efiles)

            if not cgnodes:
                repo.ui.develwarn('applied empty changelog from changegroup',
                                  config='warn-empty-changegroup')
            clend = len(cl)
            changesets = clend - clstart
            progress.complete()
            self.callback = None

            # pull off the manifest group
            repo.ui.status(_("adding manifests\n"))
            # We know that we'll never have more manifests than we had
            # changesets.
            progress = repo.ui.makeprogress(_('manifests'), unit=_('chunks'),
                                            total=changesets)
            self._unpackmanifests(repo, revmap, trp, progress)

            needfiles = {}
            if repo.ui.configbool('server', 'validate'):
                cl = repo.changelog

                        needfiles.setdefault(f, set()).add(n)

            # process the files
            repo.ui.status(_("adding file changes\n"))
            newrevs, newfiles = _addchangegroupfiles(
                repo, self, revmap, trp, efiles, needfiles)

            # making sure the value exists
            tr.changes.setdefault('changegroup-count-changesets', 0)
            tr.changes.setdefault('changegroup-count-revisions', 0)
            tr.changes.setdefault('changegroup-count-files', 0)

                    hookargs = dict(tr.hookargs)
                else:
                    hookargs = dict(tr.hookargs)
                    hookargs['node'] = hex(cl.node(clstart))
                    hookargs['node_last'] = hex(cl.node(clend - 1))
                repo.hook('pretxnchangegroup',
                          throw=True, **pycompat.strkwargs(hookargs))

            added = [cl.node(r) for r in pycompat.xrange(clstart, clend)]
            phaseall = None
            if srctype in ('push', 'serve'):
                # Old servers can not push the boundary themselves.

                        args = hookargs.copy()
                        args['node'] = hex(n)
                        del args['node_last']
                        repo.hook("incoming", **pycompat.strkwargs(args))

                    newheads = [h for h in repo.heads()
                                if h not in oldheads]
                    repo.ui.log("incoming",
                                "%d incoming changes - new heads: %s\n",
                                len(added),
                                ', '.join([hex(c[:6]) for c in newheads]))

                tr.addpostclose('changegroup-runhooks-%020i' % clstart,
                                lambda tr: repo._afterlock(runhooks))
        finally:
            repo.ui.flush()
        # never return 0 here:
        if deltaheads < 0:
            ret = deltaheads - 1

        for chunkdata in iter(lambda: self.deltachunk(chain), {}):
            # Chunkdata: (node, p1, p2, cs, deltabase, delta, flags)
            yield chunkdata
            chain = chunkdata[0]

class cg2unpacker(cg1unpacker):
    """Unpacker for cg2 streams.

    cg2 streams add support for generaldelta, so the delta header
    format is slightly different. All other features about the data
    remain the same.
    """
    deltaheader = _CHANGEGROUPV2_DELTA_HEADER
    deltaheadersize = deltaheader.size
    version = '02'

    def _deltaheader(self, headertuple, prevnode):
        node, p1, p2, deltabase, cs = headertuple
        flags = 0
        return node, p1, p2, deltabase, cs, flags

class cg3unpacker(cg2unpacker):
    """Unpacker for cg3 streams.

    cg3 streams add support for exchanging treemanifests and revlog
    flags. It adds the revlog flags to the delta header and an empty chunk
    separating manifests and files.
    """
    deltaheader = _CHANGEGROUPV3_DELTA_HEADER
    deltaheadersize = deltaheader.size
    version = '03'
    _grouplistcount = 2 # One list of manifests and one list of files

    def _deltaheader(self, headertuple, prevnode):
        node, p1, p2, deltabase, cs, flags = headertuple
        return node, p1, p2, deltabase, cs, flags

            repo.ui.debug("adding %s revisions\n" % d)
            deltas = self.deltaiter()
            if not repo.manifestlog.getstorage(d).addgroup(deltas, revmap, trp):
                raise error.Abort(_("received dir revlog group is empty"))

class headerlessfixup(object):
    def __init__(self, fh, h):
        self._h = h
        self._fh = fh
    def read(self, n):
        if self._h:
            d, self._h = self._h[:n], self._h[n:]
            if len(d) < n:
                d += readexactly(self._fh, n - len(d))
            return d
        return readexactly(self._fh, n)

def _revisiondeltatochunks(delta, headerfn):
    """Serialize a revisiondelta to changegroup chunks."""

    # The captured revision delta may be encoded as a delta against
    # a base revision or as a full revision. The changegroup format

    elif delta.basenode == nullid:
        data = delta.revision
        prefix = mdiff.trivialdiffheader(len(data))
    else:
        data = delta.revision
        prefix = mdiff.replacediffheader(delta.baserevisionsize,
                                         len(data))

    meta = headerfn(delta)

    yield chunkheader(len(meta) + len(prefix) + len(data))
    yield meta
    if prefix:
        yield prefix
    yield data

def _sortnodesellipsis(store, nodes, cl, lookup):
    """Sort nodes for changegroup generation."""
    # Ellipses serving mode.
    #

    # changelog, so what we do is we sort the non-changelog histories
    # by the order in which they are used by the changelog.
    key = lambda n: cl.rev(lookup(n))
    return sorted(nodes, key=key)

def _resolvenarrowrevisioninfo(cl, store, ischangelog, rev, linkrev,
                               linknode, clrevtolocalrev, fullclnodes,
                               precomputedellipsis):
    linkparents = precomputedellipsis[linkrev]
    def local(clrev):
        """Turn a changelog revnum into a local revnum.

        The ellipsis dag is stored as revnums on the changelog,
        but when we're producing ellipsis entries for

            p = walk[0]
            walk = walk[1:]
            if p in clrevtolocalrev:
                return clrevtolocalrev[p]
            elif p in fullclnodes:
                walk.extend([pp for pp in cl.parentrevs(p)
                                if pp != nullrev])
            elif p in precomputedellipsis:
                walk.extend([pp for pp in precomputedellipsis[p]
                                if pp != nullrev])
            else:
                # In this case, we've got an ellipsis with parents
                # outside the current bundle (likely an
                # incremental pull). We "know" that we can use the
                # value of this same revlog at whatever revision

                        return i
                # We failed to resolve a parent for this node, so
                # we crash the changegroup construction.
                raise error.Abort(
                    'unable to resolve parent while packing %r %r'
                    ' for changeset %r' % (store.indexfile, rev, clrev))

        return nullrev

    if not linkparents or (
        store.parentrevs(rev) == (nullrev, nullrev)):
        p1, p2 = nullrev, nullrev
    elif len(linkparents) == 1:
        p1, = sorted(local(p) for p in linkparents)
        p2 = nullrev
    else:
        p1, p2 = sorted(local(p) for p in linkparents)

    p1node, p2node = store.node(p1), store.node(p2)

    return p1node, p2node, linknode

def deltagroup(repo, store, nodes, ischangelog, lookup, forcedeltaparentprev,
               topic=None,
               ellipses=False, clrevtolocalrev=None, fullclnodes=None,
               precomputedellipsis=None):
    """Calculate deltas for a set of revisions.

    Is a generator of ``revisiondelta`` instances.

    If topic is not None, progress detail will be generated using this

            else:
                # We could probably do this later and avoid the dict
                # holding state. But it likely doesn't matter.
                p1node, p2node, linknode = _resolvenarrowrevisioninfo(
                    cl, store, ischangelog, rev, linkrev, linknode,
                    clrevtolocalrev, fullclnodes, precomputedellipsis)

                adjustedparents[node] = (p1node, p2node)
                linknodes[node] = linknode

            filtered.append(node)

    # We expect the first pass to be fast, so we only engage the progress
    # meter for constructing the revision deltas.
    progress = None
    if topic is not None:
        progress = repo.ui.makeprogress(topic, unit=_('chunks'),
                                        total=len(nodes))

    configtarget = repo.ui.config('devel', 'bundle.delta')
    if configtarget not in ('', 'p1', 'full'):
        msg = _("""config "devel.bundle.delta" as unknown value: %s""")
        repo.ui.warn(msg % configtarget)

    revisions = store.emitrevisions(
        nodes,
        nodesorder=nodesorder,
        revisiondata=True,
        assumehaveparentrevisions=not ellipses,
        deltamode=deltamode)

    for i, revision in enumerate(revisions):
        if progress:
            progress.update(i + 1)

        yield revision

    if progress:
        progress.complete()

class cgpacker(object):
    def __init__(self, repo, oldmatcher, matcher, version,
                 builddeltaheader, manifestsend,
                 forcedeltaparentprev=False,
                 bundlecaps=None, ellipses=False,
                 shallow=False, ellipsisroots=None, fullnodes=None):
        """Given a source repo, construct a bundler.

        oldmatcher is a matcher that matches on files the client already has.
        These will not be included in the changegroup.

        if self._repo.ui.verbose and not self._repo.ui.debugflag:
            self._verbosenote = self._repo.ui.note
        else:
            self._verbosenote = lambda s: None

    def generate(self, commonrevs, clnodes, fastpathlinkrev, source,
                 changelog=True):
        """Yield a sequence of changegroup byte chunks.
        If changelog is False, changelog data won't be added to changegroup
        """

        repo = self._repo
        cl = repo.changelog

        self._verbosenote(_('uncompressed size of bundle content:\n'))
        size = 0

        clstate, deltas = self._generatechangelog(cl, clnodes,
                                                  generate=changelog)
        for delta in deltas:
            for chunk in _revisiondeltatochunks(delta,
                                                self._builddeltaheader):
                size += len(chunk)
                yield chunk

        close = closechunk()
        size += len(close)

        # Treemanifests don't work correctly with fastpathlinkrev
        # either, because we don't discover which directory nodes to
        # send along with files. This could probably be fixed.
        fastpathlinkrev = fastpathlinkrev and (
            'treemanifest' not in repo.requirements)

        fnodes = {}  # needed file nodes

        size = 0
        it = self.generatemanifests(
            commonrevs, clrevorder, fastpathlinkrev, manifests, fnodes, source,
            clstate['clrevtomanifestrev'])

        for tree, deltas in it:
            if tree:
                assert self.version == b'03'
                chunk = _fileheader(tree)

        self._verbosenote(_('%8.i (manifests)\n') % size)
        yield self._manifestsend

        mfdicts = None
        if self._ellipses and self._isshallow:
            mfdicts = [(self._repo.manifestlog[n].read(), lr)
                       for (n, lr) in manifests.iteritems()]

        manifests.clear()
        clrevs = set(cl.rev(x) for x in clnodes)

        it = self.generatefiles(changedfiles, commonrevs,
                                source, mfdicts, fastpathlinkrev,
                                fnodes, clrevs)

        for path, deltas in it:
            h = _fileheader(path)
            size = len(h)
            yield h

            if self._ellipses:
                # Only update manifests if x is going to be sent. Otherwise we
                # end up with bogus linkrevs specified for manifests and
                # we skip some manifest nodes that we should otherwise
                # have sent.
                if (x in self._fullclnodes
                    or cl.rev(x) in self._precomputedellipsis):

                    manifestnode = c.manifest
                    # Record the first changeset introducing this manifest
                    # version.
                    manifests.setdefault(manifestnode, x)
                    # Set this narrow-specific dict so we have the lowest
                    # manifest revnum to look up for this cl revnum. (Part of
                    # mapping changelog ellipsis parents to manifest ellipsis
                    # parents)
                    clrevtomanifestrev.setdefault(
                        cl.rev(x), mfl.rev(manifestnode))
                # We can't trust the changed files list in the changeset if the
                # client requested a shallow clone.
                if self._isshallow:
                    changedfiles.update(mfl[c.manifest].read().keys())
                else:

                changedfiles.update(c.files)

            return x

        gen = deltagroup(
            self._repo, cl, nodes, True, lookupcl,
            self._forcedeltaparentprev,
            ellipses=self._ellipses,
            topic=_('changesets'),
            clrevtolocalrev={},
            fullclnodes=self._fullclnodes,
            precomputedellipsis=self._precomputedellipsis)

        return state, gen

    def generatemanifests(self, commonrevs, clrevorder, fastpathlinkrev,
                          manifests, fnodes, source, clrevtolocalrev):
        """Returns an iterator of changegroup chunks containing manifests.

        `source` is unused here, but is used by extensions like remotefilelog to
        change what is sent based in pulls vs pushes, etc.
        """

                treemanifests to send.
                """
                clnode = nodes[x]
                mdata = mfl.get(tree, x).readfast(shallow=True)
                for p, n, fl in mdata.iterentries():
                    if fl == 't': # subdirectory manifest
                        subtree = tree + p + '/'
                        tmfclnodes = tmfnodes.setdefault(subtree, {})
                        tmfclnode = tmfclnodes.setdefault(n, clnode)
                        if clrevorder[clnode] < clrevorder[tmfclnode]:
                            tmfclnodes[n] = clnode

                        fclnodes = fnodes.setdefault(f, {})
                        fclnode = fclnodes.setdefault(n, clnode)
                        if clrevorder[clnode] < clrevorder[fclnode]:
                            fclnodes[n] = clnode
                return clnode
            return lookupmflinknode

        while tmfnodes:
            tree, nodes = tmfnodes.popitem()

                continue

            lookupfn = makelookupmflinknode(tree, nodes)

            deltas = deltagroup(
                self._repo, store, prunednodes, False, lookupfn,
                self._forcedeltaparentprev,
                ellipses=self._ellipses,
                topic=_('manifests'),
                clrevtolocalrev=clrevtolocalrev,
                fullclnodes=self._fullclnodes,
                precomputedellipsis=self._precomputedellipsis)

            if not self._oldmatcher.visitdir(store.tree[:-1]):
                yield tree, deltas
            else:
                # 'deltas' is a generator and we need to consume it even if

        # layer.
        frev, flr = store.rev, store.linkrev
        return [n for n in nodes if flr(frev(n)) not in commonrevs]

    # The 'source' parameter is useful for extensions
    def generatefiles(self, changedfiles, commonrevs, source,
                      mfdicts, fastpathlinkrev, fnodes, clrevs):
        changedfiles = [f for f in changedfiles
                        if self._matcher(f) and not self._oldmatcher(f)]

        if not fastpathlinkrev:
            def normallinknodes(unused, fname):
                return fnodes.get(fname, {})
        else:
            cln = self._repo.changelog.node

            def normallinknodes(store, fname):
                flinkrev = store.linkrev
                fnode = store.node
                revs = ((r, flinkrev(r)) for r in store)
                return dict((fnode(r), cln(lr))
                            for r, lr in revs if lr in clrevs)

        clrevtolocalrev = {}

        if self._isshallow:
            # In a shallow clone, the linknodes callback needs to also include

                        if fnode in links:
                            links[fnode] = min(links[fnode], lr, key=clrev)
                        elif fnode:
                            links[fnode] = lr
                return links
        else:
            linknodes = normallinknodes

        repo = self._repo
        progress = repo.ui.makeprogress(_('files'), unit=_('files'),
                                        total=len(changedfiles))
        for i, fname in enumerate(sorted(changedfiles)):
            filerevlog = repo.file(fname)
            if not filerevlog:
                raise error.Abort(_("empty or missing file data for %s") %
                                  fname)

            clrevtolocalrev.clear()

            linkrevnodes = linknodes(filerevlog, fname)
            # Lookup for filenodes, we collected the linkrev nodes above in the

            frev, flr = filerevlog.rev, filerevlog.linkrev
            # Skip sending any filenode we know the client already
            # has. This avoids over-sending files relatively
            # inexpensively, so it's not a problem if we under-filter
            # here.
            filenodes = [n for n in linkrevnodes
                         if flr(frev(n)) not in commonrevs]

            if not filenodes:
                continue

            progress.update(i + 1, item=fname)

            deltas = deltagroup(
                self._repo, filerevlog, filenodes, False, lookupfilelog,
                self._forcedeltaparentprev,
                ellipses=self._ellipses,
                clrevtolocalrev=clrevtolocalrev,
                fullclnodes=self._fullclnodes,
                precomputedellipsis=self._precomputedellipsis)

            yield fname, deltas

        progress.complete()

def _makecg1packer(repo, oldmatcher, matcher, bundlecaps,
                   ellipses=False, shallow=False, ellipsisroots=None,
                   fullnodes=None):
    builddeltaheader = lambda d: _CHANGEGROUPV1_DELTA_HEADER.pack(
        d.node, d.p1node, d.p2node, d.linknode)

    return cgpacker(repo, oldmatcher, matcher, b'01',
                    builddeltaheader=builddeltaheader,
                    manifestsend=b'',
                    forcedeltaparentprev=True,
                    bundlecaps=bundlecaps,
                    ellipses=ellipses,
                    shallow=shallow,
                    ellipsisroots=ellipsisroots,
                    fullnodes=fullnodes)

def _makecg2packer(repo, oldmatcher, matcher, bundlecaps,
                   ellipses=False, shallow=False, ellipsisroots=None,
                   fullnodes=None):
    builddeltaheader = lambda d: _CHANGEGROUPV2_DELTA_HEADER.pack(
        d.node, d.p1node, d.p2node, d.basenode, d.linknode)

    return cgpacker(repo, oldmatcher, matcher, b'02',
                    builddeltaheader=builddeltaheader,
                    manifestsend=b'',
                    bundlecaps=bundlecaps,
                    ellipses=ellipses,
                    shallow=shallow,
                    ellipsisroots=ellipsisroots,
                    fullnodes=fullnodes)

def _makecg3packer(repo, oldmatcher, matcher, bundlecaps,
                   ellipses=False, shallow=False, ellipsisroots=None,
                   fullnodes=None):
    builddeltaheader = lambda d: _CHANGEGROUPV3_DELTA_HEADER.pack(
        d.node, d.p1node, d.p2node, d.basenode, d.linknode, d.flags)

    return cgpacker(repo, oldmatcher, matcher, b'03',
                    builddeltaheader=builddeltaheader,
                    manifestsend=closechunk(),
                    bundlecaps=bundlecaps,
                    ellipses=ellipses,
                    shallow=shallow,
                    ellipsisroots=ellipsisroots,
                    fullnodes=fullnodes)

_packermap = {'01': (_makecg1packer, cg1unpacker),
             # cg2 adds support for exchanging generaldelta
             '02': (_makecg2packer, cg2unpacker),
             # cg3 adds support for exchanging revlog flags and treemanifests
             '03': (_makecg3packer, cg3unpacker),
}

def allsupportedversions(repo):
    versions = set(_packermap.keys())
    needv03 = False
    if (repo.ui.configbool('experimental', 'changegroup3') or
        repo.ui.configbool('experimental', 'treemanifest') or
        'treemanifest' in repo.requirements):
        # we keep version 03 because we need to to exchange treemanifest data
        #
        # we also keep vresion 01 and 02, because it is possible for repo to
        # contains both normal and tree manifest at the same time. so using
        # older version to pull data is viable

        needv03 = True
    if not needv03:
        versions.discard('03')
    return versions

# Changegroup versions that can be applied to the repo
def supportedincomingversions(repo):
    return allsupportedversions(repo)

# Changegroup versions that can be created from the repo
def supportedoutgoingversions(repo):
    versions = allsupportedversions(repo)
    if 'treemanifest' in repo.requirements:

        versions.discard('01')
        versions.discard('02')

    return versions

def localversion(repo):
    # Finds the best version to use for bundles that are meant to be used
    # locally, such as those from strip and shelve, and temporary bundles.
    return max(supportedoutgoingversions(repo))

def safeversion(repo):
    # Finds the smallest version that it's safe to assume clients of the repo
    # will support. For example, all hg versions that support generaldelta also
    # support changegroup 02.

    if 'generaldelta' in repo.requirements:
        versions.discard('01')
    assert versions
    return min(versions)

def getbundler(version, repo, bundlecaps=None, oldmatcher=None,
               matcher=None, ellipses=False, shallow=False,
               ellipsisroots=None, fullnodes=None):
    assert version in supportedoutgoingversions(repo)

    if matcher is None:
        matcher = matchmod.always()
    if oldmatcher is None:
        oldmatcher = matchmod.never()

    if version == '01' and not matcher.always():
        raise error.ProgrammingError('version 01 changegroups do not support '
                                     'sparse file matchers')

    if ellipses and version in (b'01', b'02'):
        raise error.Abort(
            _('ellipsis nodes require at least cg3 on client and server, '
              'but negotiated version %s') % version)

    # Requested files could include files not in the local store. So
    # filter those out.
    matcher = repo.narrowmatch(matcher)

    fn = _packermap[version][0]
    return fn(repo, oldmatcher, matcher, bundlecaps, ellipses=ellipses,
              shallow=shallow, ellipsisroots=ellipsisroots,
              fullnodes=fullnodes)

def getunbundler(version, fh, alg, extras=None):
    return _packermap[version][1](fh, alg, extras=extras)

def _changegroupinfo(repo, nodes, source):
    if repo.ui.verbose or source == 'bundle':
        repo.ui.status(_("%d changesets found\n") % len(nodes))
    if repo.ui.debugflag:
        repo.ui.debug("list of changesets:\n")
        for node in nodes:
            repo.ui.debug("%s\n" % hex(node))

def makechangegroup(repo, outgoing, version, source, fastpath=False,
                    bundlecaps=None):
    cgstream = makestream(repo, outgoing, version, source,
                          fastpath=fastpath, bundlecaps=bundlecaps)
    return getunbundler(version, util.chunkbuffer(cgstream), None,
                        {'clcount': len(outgoing.missing) })

def makestream(repo, outgoing, version, source, fastpath=False,
               bundlecaps=None, matcher=None):
    bundler = getbundler(version, repo, bundlecaps=bundlecaps,
                         matcher=matcher)

    repo = repo.unfiltered()
    commonrevs = outgoing.common
    csets = outgoing.missing
    heads = outgoing.missingheads
    # We go through the fast path if we get told to, or if all (unfiltered
    # heads have been requested (since we then know there all linkrevs will
    # be pulled by the client).
    heads.sort()
    fastpathlinkrev = fastpath or (
            repo.filtername is None and heads == sorted(repo.heads()))

    repo.hook('preoutgoing', throw=True, source=source)
    _changegroupinfo(repo, csets, source)
    return bundler.generate(commonrevs, csets, fastpathlinkrev, source)

def _addchangegroupfiles(repo, source, revmap, trp, expectedfiles, needfiles):
    revisions = 0
    files = 0
    progress = repo.ui.makeprogress(_('files'), unit=_('files'),
                                    total=expectedfiles)
    for chunkdata in iter(source.filelogheader, {}):
        files += 1
        f = chunkdata["filename"]
        repo.ui.debug("adding %s revisions\n" % f)
        progress.increment()

            for new in pycompat.xrange(o, len(fl)):
                n = fl.node(new)
                if n in needs:
                    needs.remove(n)
                else:
                    raise error.Abort(
                        _("received spurious file revlog entry"))
            if not needs:
                del needfiles[f]
    progress.complete()

    for f, needs in needfiles.iteritems():

        for n in needs:
            try:
                fl.rev(n)
            except error.LookupError:
                raise error.Abort(
                    _('missing file data for %s:%s - run hg verify') %
                    (f, hex(n)))

    return revisions, files

43076:2372284d9457

    phases,
    pycompat,
    util,
)

from .interfaces import repository

_CHANGEGROUPV1_DELTA_HEADER = struct.Struct("20s20s20s20s")
_CHANGEGROUPV2_DELTA_HEADER = struct.Struct("20s20s20s20s20s")
_CHANGEGROUPV3_DELTA_HEADER = struct.Struct(">20s20s20s20s20sH")

LFS_REQUIREMENT = 'lfs'

readexactly = util.readexactly


def getchunk(stream):
    """return the next chunk from stream as a string"""
    d = readexactly(stream, 4)
    l = struct.unpack(">l", d)[0]

        if l:
            raise error.Abort(_("invalid chunk length %d") % l)
        return ""
    return readexactly(stream, l - 4)


def chunkheader(length):
    """return a changegroup chunk header (string)"""
    return struct.pack(">l", length + 4)


def closechunk():
    """return a changegroup chunk header (string) for a zero-length chunk"""
    return struct.pack(">l", 0)


def _fileheader(path):
    """Obtain a changegroup chunk header for a named path."""
    return chunkheader(len(path)) + path


def writechunks(ui, chunks, filename, vfs=None):
    """Write chunks to a file and return its filename.

    The stream is assumed to be a bundle file.

            if filename and vfs:
                vfs.unlink(cleanup)
            else:
                os.unlink(cleanup)


class cg1unpacker(object):
    """Unpacker for cg1 changegroup streams.

    A changegroup unpacker handles the framing of the revision data in
    the wire format. Most consumers will want to use the apply()

    consumers should prefer apply() instead.

    A few other public methods exist. Those are used only for
    bundlerepo and some debug commands - their use is discouraged.
    """

    deltaheader = _CHANGEGROUPV1_DELTA_HEADER
    deltaheadersize = deltaheader.size
    version = '01'
    _grouplistcount = 1  # One list of files after the manifests

    def __init__(self, fh, alg, extras=None):
        if alg is None:
            alg = 'UN'
        if alg not in util.compengines.supportedbundletypes:
            raise error.Abort(_('unknown stream compression type: %s') % alg)
        if alg == 'BZ':
            alg = '_truncatedBZ'

        compengine = util.compengines.forbundletype(alg)
        self._stream = compengine.decompressorreader(fh)

    # These methods (compressed, read, seek, tell) all appear to only
    # be used by bundlerepo, but it's a little hard to tell.
    def compressed(self):
        return self._type is not None and self._type != 'UN'

    def read(self, l):
        return self._stream.read(l)

    def seek(self, pos):
        return self._stream.seek(pos)

    def tell(self):
        return self._stream.tell()

    def close(self):
        return self._stream.close()

    def _chunklength(self):
        d = readexactly(self._stream, 4)

                    break
                noentries = False
                yield chunkheader(len(chunk))
                pos = 0
                while pos < len(chunk):
                    next = pos + 2 ** 20
                    yield chunk[pos:next]
                    pos = next
            yield closechunk()

    def _unpackmanifests(self, repo, revmap, trp, prog):

        deltas = self.deltaiter()
        repo.manifestlog.getstorage(b'').addgroup(deltas, revmap, trp)
        prog.complete()
        self.callback = None

    def apply(
        self,
        repo,
        tr,
        srctype,
        url,
        targetphase=phases.draft,
        expectedtotal=None,
    ):
        """Add the changegroup returned by source.read() to this repo.
        srctype is a string like 'push', 'pull', or 'unbundle'.  url is
        the URL of the repo where this changegroup is coming from.

        Return an integer summarizing the change to this repo:

        - more heads than before: 1+added heads (2..n)
        - fewer heads than before: -1-removed heads (-2..-n)
        - number of heads stays the same: 1
        """
        repo = repo.unfiltered()

        def csmap(x):
            repo.ui.debug("add changeset %s\n" % short(x))
            return len(cl)

        def revmap(x):

            # case we use the top level data. We overwrite the argument
            # because we need to use the top level value (if they exist)
            # in this function.
            srctype = tr.hookargs.setdefault('source', srctype)
            tr.hookargs.setdefault('url', url)
            repo.hook(
                'prechangegroup', throw=True, **pycompat.strkwargs(tr.hookargs)
            )

            # write changelog data to temp files so concurrent readers
            # will not see an inconsistent view
            cl = repo.changelog
            cl.delayupdate(tr)

            trp = weakref.proxy(tr)
            # pull off the changeset group
            repo.ui.status(_("adding changesets\n"))
            clstart = len(cl)
            progress = repo.ui.makeprogress(
                _('changesets'), unit=_('chunks'), total=expectedtotal
            )
            self.callback = progress.increment

            efiles = set()

            def onchangelog(cl, node):
                efiles.update(cl.readfiles(node))

            self.changelogheader()
            deltas = self.deltaiter()
            cgnodes = cl.addgroup(deltas, csmap, trp, addrevisioncb=onchangelog)
            efiles = len(efiles)

            if not cgnodes:
                repo.ui.develwarn(
                    'applied empty changelog from changegroup',
                    config='warn-empty-changegroup',
                )
            clend = len(cl)
            changesets = clend - clstart
            progress.complete()
            self.callback = None

            # pull off the manifest group
            repo.ui.status(_("adding manifests\n"))
            # We know that we'll never have more manifests than we had
            # changesets.
            progress = repo.ui.makeprogress(
                _('manifests'), unit=_('chunks'), total=changesets
            )
            self._unpackmanifests(repo, revmap, trp, progress)

            needfiles = {}
            if repo.ui.configbool('server', 'validate'):
                cl = repo.changelog

                        needfiles.setdefault(f, set()).add(n)

            # process the files
            repo.ui.status(_("adding file changes\n"))
            newrevs, newfiles = _addchangegroupfiles(
                repo, self, revmap, trp, efiles, needfiles
            )

            # making sure the value exists
            tr.changes.setdefault('changegroup-count-changesets', 0)
            tr.changes.setdefault('changegroup-count-revisions', 0)
            tr.changes.setdefault('changegroup-count-files', 0)

                    hookargs = dict(tr.hookargs)
                else:
                    hookargs = dict(tr.hookargs)
                    hookargs['node'] = hex(cl.node(clstart))
                    hookargs['node_last'] = hex(cl.node(clend - 1))
                repo.hook(
                    'pretxnchangegroup',
                    throw=True,
                    **pycompat.strkwargs(hookargs)
                )

            added = [cl.node(r) for r in pycompat.xrange(clstart, clend)]
            phaseall = None
            if srctype in ('push', 'serve'):
                # Old servers can not push the boundary themselves.

                        args = hookargs.copy()
                        args['node'] = hex(n)
                        del args['node_last']
                        repo.hook("incoming", **pycompat.strkwargs(args))

                    newheads = [h for h in repo.heads() if h not in oldheads]
                    repo.ui.log(
                        "incoming",
                        "%d incoming changes - new heads: %s\n",
                        len(added),
                        ', '.join([hex(c[:6]) for c in newheads]),
                    )

                tr.addpostclose(
                    'changegroup-runhooks-%020i' % clstart,
                    lambda tr: repo._afterlock(runhooks),
                )
        finally:
            repo.ui.flush()
        # never return 0 here:
        if deltaheads < 0:
            ret = deltaheads - 1

        for chunkdata in iter(lambda: self.deltachunk(chain), {}):
            # Chunkdata: (node, p1, p2, cs, deltabase, delta, flags)
            yield chunkdata
            chain = chunkdata[0]


class cg2unpacker(cg1unpacker):
    """Unpacker for cg2 streams.

    cg2 streams add support for generaldelta, so the delta header
    format is slightly different. All other features about the data
    remain the same.
    """

    deltaheader = _CHANGEGROUPV2_DELTA_HEADER
    deltaheadersize = deltaheader.size
    version = '02'

    def _deltaheader(self, headertuple, prevnode):
        node, p1, p2, deltabase, cs = headertuple
        flags = 0
        return node, p1, p2, deltabase, cs, flags


class cg3unpacker(cg2unpacker):
    """Unpacker for cg3 streams.

    cg3 streams add support for exchanging treemanifests and revlog
    flags. It adds the revlog flags to the delta header and an empty chunk
    separating manifests and files.
    """

    deltaheader = _CHANGEGROUPV3_DELTA_HEADER
    deltaheadersize = deltaheader.size
    version = '03'
    _grouplistcount = 2  # One list of manifests and one list of files

    def _deltaheader(self, headertuple, prevnode):
        node, p1, p2, deltabase, cs, flags = headertuple
        return node, p1, p2, deltabase, cs, flags

            repo.ui.debug("adding %s revisions\n" % d)
            deltas = self.deltaiter()
            if not repo.manifestlog.getstorage(d).addgroup(deltas, revmap, trp):
                raise error.Abort(_("received dir revlog group is empty"))


class headerlessfixup(object):
    def __init__(self, fh, h):
        self._h = h
        self._fh = fh

    def read(self, n):
        if self._h:
            d, self._h = self._h[:n], self._h[n:]
            if len(d) < n:
                d += readexactly(self._fh, n - len(d))
            return d
        return readexactly(self._fh, n)


def _revisiondeltatochunks(delta, headerfn):
    """Serialize a revisiondelta to changegroup chunks."""

    # The captured revision delta may be encoded as a delta against
    # a base revision or as a full revision. The changegroup format

    elif delta.basenode == nullid:
        data = delta.revision
        prefix = mdiff.trivialdiffheader(len(data))
    else:
        data = delta.revision
        prefix = mdiff.replacediffheader(delta.baserevisionsize, len(data))

    meta = headerfn(delta)

    yield chunkheader(len(meta) + len(prefix) + len(data))
    yield meta
    if prefix:
        yield prefix
    yield data


def _sortnodesellipsis(store, nodes, cl, lookup):
    """Sort nodes for changegroup generation."""
    # Ellipses serving mode.
    #

    # changelog, so what we do is we sort the non-changelog histories
    # by the order in which they are used by the changelog.
    key = lambda n: cl.rev(lookup(n))
    return sorted(nodes, key=key)


def _resolvenarrowrevisioninfo(
    cl,
    store,
    ischangelog,
    rev,
    linkrev,
    linknode,
    clrevtolocalrev,
    fullclnodes,
    precomputedellipsis,
):
    linkparents = precomputedellipsis[linkrev]

    def local(clrev):
        """Turn a changelog revnum into a local revnum.

        The ellipsis dag is stored as revnums on the changelog,
        but when we're producing ellipsis entries for

            p = walk[0]
            walk = walk[1:]
            if p in clrevtolocalrev:
                return clrevtolocalrev[p]
            elif p in fullclnodes:
                walk.extend([pp for pp in cl.parentrevs(p) if pp != nullrev])
            elif p in precomputedellipsis:
                walk.extend(
                    [pp for pp in precomputedellipsis[p] if pp != nullrev]
                )
            else:
                # In this case, we've got an ellipsis with parents
                # outside the current bundle (likely an
                # incremental pull). We "know" that we can use the
                # value of this same revlog at whatever revision

                        return i
                # We failed to resolve a parent for this node, so
                # we crash the changegroup construction.
                raise error.Abort(
                    'unable to resolve parent while packing %r %r'
                    ' for changeset %r' % (store.indexfile, rev, clrev)
                )

        return nullrev

    if not linkparents or (store.parentrevs(rev) == (nullrev, nullrev)):
        p1, p2 = nullrev, nullrev
    elif len(linkparents) == 1:
        (p1,) = sorted(local(p) for p in linkparents)
        p2 = nullrev
    else:
        p1, p2 = sorted(local(p) for p in linkparents)

    p1node, p2node = store.node(p1), store.node(p2)

    return p1node, p2node, linknode


def deltagroup(
    repo,
    store,
    nodes,
    ischangelog,
    lookup,
    forcedeltaparentprev,
    topic=None,
    ellipses=False,
    clrevtolocalrev=None,
    fullclnodes=None,
    precomputedellipsis=None,
):
    """Calculate deltas for a set of revisions.

    Is a generator of ``revisiondelta`` instances.

    If topic is not None, progress detail will be generated using this

            else:
                # We could probably do this later and avoid the dict
                # holding state. But it likely doesn't matter.
                p1node, p2node, linknode = _resolvenarrowrevisioninfo(
                    cl,
                    store,
                    ischangelog,
                    rev,
                    linkrev,
                    linknode,
                    clrevtolocalrev,
                    fullclnodes,
                    precomputedellipsis,
                )

                adjustedparents[node] = (p1node, p2node)
                linknodes[node] = linknode

            filtered.append(node)

    # We expect the first pass to be fast, so we only engage the progress
    # meter for constructing the revision deltas.
    progress = None
    if topic is not None:
        progress = repo.ui.makeprogress(
            topic, unit=_('chunks'), total=len(nodes)
        )

    configtarget = repo.ui.config('devel', 'bundle.delta')
    if configtarget not in ('', 'p1', 'full'):
        msg = _("""config "devel.bundle.delta" as unknown value: %s""")
        repo.ui.warn(msg % configtarget)

    revisions = store.emitrevisions(
        nodes,
        nodesorder=nodesorder,
        revisiondata=True,
        assumehaveparentrevisions=not ellipses,
        deltamode=deltamode,
    )

    for i, revision in enumerate(revisions):
        if progress:
            progress.update(i + 1)

        yield revision

    if progress:
        progress.complete()


class cgpacker(object):
    def __init__(
        self,
        repo,
        oldmatcher,
        matcher,
        version,
        builddeltaheader,
        manifestsend,
        forcedeltaparentprev=False,
        bundlecaps=None,
        ellipses=False,
        shallow=False,
        ellipsisroots=None,
        fullnodes=None,
    ):
        """Given a source repo, construct a bundler.

        oldmatcher is a matcher that matches on files the client already has.
        These will not be included in the changegroup.

        if self._repo.ui.verbose and not self._repo.ui.debugflag:
            self._verbosenote = self._repo.ui.note
        else:
            self._verbosenote = lambda s: None

    def generate(
        self, commonrevs, clnodes, fastpathlinkrev, source, changelog=True
    ):
        """Yield a sequence of changegroup byte chunks.
        If changelog is False, changelog data won't be added to changegroup
        """

        repo = self._repo
        cl = repo.changelog

        self._verbosenote(_('uncompressed size of bundle content:\n'))
        size = 0

        clstate, deltas = self._generatechangelog(
            cl, clnodes, generate=changelog
        )
        for delta in deltas:
            for chunk in _revisiondeltatochunks(delta, self._builddeltaheader):
                size += len(chunk)
                yield chunk

        close = closechunk()
        size += len(close)

        # Treemanifests don't work correctly with fastpathlinkrev
        # either, because we don't discover which directory nodes to
        # send along with files. This could probably be fixed.
        fastpathlinkrev = fastpathlinkrev and (
            'treemanifest' not in repo.requirements
        )

        fnodes = {}  # needed file nodes

        size = 0
        it = self.generatemanifests(
            commonrevs,
            clrevorder,
            fastpathlinkrev,
            manifests,
            fnodes,
            source,
            clstate['clrevtomanifestrev'],
        )

        for tree, deltas in it:
            if tree:
                assert self.version == b'03'
                chunk = _fileheader(tree)

        self._verbosenote(_('%8.i (manifests)\n') % size)
        yield self._manifestsend

        mfdicts = None
        if self._ellipses and self._isshallow:
            mfdicts = [
                (self._repo.manifestlog[n].read(), lr)
                for (n, lr) in manifests.iteritems()
            ]

        manifests.clear()
        clrevs = set(cl.rev(x) for x in clnodes)

        it = self.generatefiles(
            changedfiles,
            commonrevs,
            source,
            mfdicts,
            fastpathlinkrev,
            fnodes,
            clrevs,
        )

        for path, deltas in it:
            h = _fileheader(path)
            size = len(h)
            yield h

            if self._ellipses:
                # Only update manifests if x is going to be sent. Otherwise we
                # end up with bogus linkrevs specified for manifests and
                # we skip some manifest nodes that we should otherwise
                # have sent.
                if (
                    x in self._fullclnodes
                    or cl.rev(x) in self._precomputedellipsis
                ):

                    manifestnode = c.manifest
                    # Record the first changeset introducing this manifest
                    # version.
                    manifests.setdefault(manifestnode, x)
                    # Set this narrow-specific dict so we have the lowest
                    # manifest revnum to look up for this cl revnum. (Part of
                    # mapping changelog ellipsis parents to manifest ellipsis
                    # parents)
                    clrevtomanifestrev.setdefault(
                        cl.rev(x), mfl.rev(manifestnode)
                    )
                # We can't trust the changed files list in the changeset if the
                # client requested a shallow clone.
                if self._isshallow:
                    changedfiles.update(mfl[c.manifest].read().keys())
                else:

                changedfiles.update(c.files)

            return x

        gen = deltagroup(
            self._repo,
            cl,
            nodes,
            True,
            lookupcl,
            self._forcedeltaparentprev,
            ellipses=self._ellipses,
            topic=_('changesets'),
            clrevtolocalrev={},
            fullclnodes=self._fullclnodes,
            precomputedellipsis=self._precomputedellipsis,
        )

        return state, gen

    def generatemanifests(
        self,
        commonrevs,
        clrevorder,
        fastpathlinkrev,
        manifests,
        fnodes,
        source,
        clrevtolocalrev,
    ):
        """Returns an iterator of changegroup chunks containing manifests.

        `source` is unused here, but is used by extensions like remotefilelog to
        change what is sent based in pulls vs pushes, etc.
        """

                treemanifests to send.
                """
                clnode = nodes[x]
                mdata = mfl.get(tree, x).readfast(shallow=True)
                for p, n, fl in mdata.iterentries():
                    if fl == 't':  # subdirectory manifest
                        subtree = tree + p + '/'
                        tmfclnodes = tmfnodes.setdefault(subtree, {})
                        tmfclnode = tmfclnodes.setdefault(n, clnode)
                        if clrevorder[clnode] < clrevorder[tmfclnode]:
                            tmfclnodes[n] = clnode

                        fclnodes = fnodes.setdefault(f, {})
                        fclnode = fclnodes.setdefault(n, clnode)
                        if clrevorder[clnode] < clrevorder[fclnode]:
                            fclnodes[n] = clnode
                return clnode

            return lookupmflinknode

        while tmfnodes:
            tree, nodes = tmfnodes.popitem()

                continue

            lookupfn = makelookupmflinknode(tree, nodes)

            deltas = deltagroup(
                self._repo,
                store,
                prunednodes,
                False,
                lookupfn,
                self._forcedeltaparentprev,
                ellipses=self._ellipses,
                topic=_('manifests'),
                clrevtolocalrev=clrevtolocalrev,
                fullclnodes=self._fullclnodes,
                precomputedellipsis=self._precomputedellipsis,
            )

            if not self._oldmatcher.visitdir(store.tree[:-1]):
                yield tree, deltas
            else:
                # 'deltas' is a generator and we need to consume it even if

        # layer.
        frev, flr = store.rev, store.linkrev
        return [n for n in nodes if flr(frev(n)) not in commonrevs]

    # The 'source' parameter is useful for extensions
    def generatefiles(
        self,
        changedfiles,
        commonrevs,
        source,
        mfdicts,
        fastpathlinkrev,
        fnodes,
        clrevs,
    ):
        changedfiles = [
            f
            for f in changedfiles
            if self._matcher(f) and not self._oldmatcher(f)
        ]

        if not fastpathlinkrev:

            def normallinknodes(unused, fname):
                return fnodes.get(fname, {})

        else:
            cln = self._repo.changelog.node

            def normallinknodes(store, fname):
                flinkrev = store.linkrev
                fnode = store.node
                revs = ((r, flinkrev(r)) for r in store)
                return dict(
                    (fnode(r), cln(lr)) for r, lr in revs if lr in clrevs
                )

        clrevtolocalrev = {}

        if self._isshallow:
            # In a shallow clone, the linknodes callback needs to also include

                        if fnode in links:
                            links[fnode] = min(links[fnode], lr, key=clrev)
                        elif fnode:
                            links[fnode] = lr
                return links

        else:
            linknodes = normallinknodes

        repo = self._repo
        progress = repo.ui.makeprogress(
            _('files'), unit=_('files'), total=len(changedfiles)
        )
        for i, fname in enumerate(sorted(changedfiles)):
            filerevlog = repo.file(fname)
            if not filerevlog:
                raise error.Abort(
                    _("empty or missing file data for %s") % fname
                )

            clrevtolocalrev.clear()

            linkrevnodes = linknodes(filerevlog, fname)
            # Lookup for filenodes, we collected the linkrev nodes above in the

            frev, flr = filerevlog.rev, filerevlog.linkrev
            # Skip sending any filenode we know the client already
            # has. This avoids over-sending files relatively
            # inexpensively, so it's not a problem if we under-filter
            # here.
            filenodes = [
                n for n in linkrevnodes if flr(frev(n)) not in commonrevs
            ]

            if not filenodes:
                continue

            progress.update(i + 1, item=fname)

            deltas = deltagroup(
                self._repo,
                filerevlog,
                filenodes,
                False,
                lookupfilelog,
                self._forcedeltaparentprev,
                ellipses=self._ellipses,
                clrevtolocalrev=clrevtolocalrev,
                fullclnodes=self._fullclnodes,
                precomputedellipsis=self._precomputedellipsis,
            )

            yield fname, deltas

        progress.complete()


def _makecg1packer(
    repo,
    oldmatcher,
    matcher,
    bundlecaps,
    ellipses=False,
    shallow=False,
    ellipsisroots=None,
    fullnodes=None,
):
    builddeltaheader = lambda d: _CHANGEGROUPV1_DELTA_HEADER.pack(
        d.node, d.p1node, d.p2node, d.linknode
    )

    return cgpacker(
        repo,
        oldmatcher,
        matcher,
        b'01',
        builddeltaheader=builddeltaheader,
        manifestsend=b'',
        forcedeltaparentprev=True,
        bundlecaps=bundlecaps,
        ellipses=ellipses,
        shallow=shallow,
        ellipsisroots=ellipsisroots,
        fullnodes=fullnodes,
    )


def _makecg2packer(
    repo,
    oldmatcher,
    matcher,
    bundlecaps,
    ellipses=False,
    shallow=False,
    ellipsisroots=None,
    fullnodes=None,
):
    builddeltaheader = lambda d: _CHANGEGROUPV2_DELTA_HEADER.pack(
        d.node, d.p1node, d.p2node, d.basenode, d.linknode
    )

    return cgpacker(
        repo,
        oldmatcher,
        matcher,
        b'02',
        builddeltaheader=builddeltaheader,
        manifestsend=b'',
        bundlecaps=bundlecaps,
        ellipses=ellipses,
        shallow=shallow,
        ellipsisroots=ellipsisroots,
        fullnodes=fullnodes,
    )


def _makecg3packer(
    repo,
    oldmatcher,
    matcher,
    bundlecaps,
    ellipses=False,
    shallow=False,
    ellipsisroots=None,
    fullnodes=None,
):
    builddeltaheader = lambda d: _CHANGEGROUPV3_DELTA_HEADER.pack(
        d.node, d.p1node, d.p2node, d.basenode, d.linknode, d.flags
    )

    return cgpacker(
        repo,
        oldmatcher,
        matcher,
        b'03',
        builddeltaheader=builddeltaheader,
        manifestsend=closechunk(),
        bundlecaps=bundlecaps,
        ellipses=ellipses,
        shallow=shallow,
        ellipsisroots=ellipsisroots,
        fullnodes=fullnodes,
    )


_packermap = {
    '01': (_makecg1packer, cg1unpacker),
    # cg2 adds support for exchanging generaldelta
    '02': (_makecg2packer, cg2unpacker),
    # cg3 adds support for exchanging revlog flags and treemanifests
    '03': (_makecg3packer, cg3unpacker),
}


def allsupportedversions(repo):
    versions = set(_packermap.keys())
    needv03 = False
    if (
        repo.ui.configbool('experimental', 'changegroup3')
        or repo.ui.configbool('experimental', 'treemanifest')
        or 'treemanifest' in repo.requirements
    ):
        # we keep version 03 because we need to to exchange treemanifest data
        #
        # we also keep vresion 01 and 02, because it is possible for repo to
        # contains both normal and tree manifest at the same time. so using
        # older version to pull data is viable

        needv03 = True
    if not needv03:
        versions.discard('03')
    return versions


# Changegroup versions that can be applied to the repo
def supportedincomingversions(repo):
    return allsupportedversions(repo)


# Changegroup versions that can be created from the repo
def supportedoutgoingversions(repo):
    versions = allsupportedversions(repo)
    if 'treemanifest' in repo.requirements:

        versions.discard('01')
        versions.discard('02')

    return versions


def localversion(repo):
    # Finds the best version to use for bundles that are meant to be used
    # locally, such as those from strip and shelve, and temporary bundles.
    return max(supportedoutgoingversions(repo))


def safeversion(repo):
    # Finds the smallest version that it's safe to assume clients of the repo
    # will support. For example, all hg versions that support generaldelta also
    # support changegroup 02.

    if 'generaldelta' in repo.requirements:
        versions.discard('01')
    assert versions
    return min(versions)


def getbundler(
    version,
    repo,
    bundlecaps=None,
    oldmatcher=None,
    matcher=None,
    ellipses=False,
    shallow=False,
    ellipsisroots=None,
    fullnodes=None,
):
    assert version in supportedoutgoingversions(repo)

    if matcher is None:
        matcher = matchmod.always()
    if oldmatcher is None:
        oldmatcher = matchmod.never()

    if version == '01' and not matcher.always():
        raise error.ProgrammingError(
            'version 01 changegroups do not support ' 'sparse file matchers'
        )

    if ellipses and version in (b'01', b'02'):
        raise error.Abort(
            _(
                'ellipsis nodes require at least cg3 on client and server, '
                'but negotiated version %s'
            )
            % version
        )

    # Requested files could include files not in the local store. So
    # filter those out.
    matcher = repo.narrowmatch(matcher)

    fn = _packermap[version][0]
    return fn(
        repo,
        oldmatcher,
        matcher,
        bundlecaps,
        ellipses=ellipses,
        shallow=shallow,
        ellipsisroots=ellipsisroots,
        fullnodes=fullnodes,
    )


def getunbundler(version, fh, alg, extras=None):
    return _packermap[version][1](fh, alg, extras=extras)


def _changegroupinfo(repo, nodes, source):
    if repo.ui.verbose or source == 'bundle':
        repo.ui.status(_("%d changesets found\n") % len(nodes))
    if repo.ui.debugflag:
        repo.ui.debug("list of changesets:\n")
        for node in nodes:
            repo.ui.debug("%s\n" % hex(node))


def makechangegroup(
    repo, outgoing, version, source, fastpath=False, bundlecaps=None
):
    cgstream = makestream(
        repo,
        outgoing,
        version,
        source,
        fastpath=fastpath,
        bundlecaps=bundlecaps,
    )
    return getunbundler(
        version,
        util.chunkbuffer(cgstream),
        None,
        {'clcount': len(outgoing.missing)},
    )


def makestream(
    repo,
    outgoing,
    version,
    source,
    fastpath=False,
    bundlecaps=None,
    matcher=None,
):
    bundler = getbundler(version, repo, bundlecaps=bundlecaps, matcher=matcher)

    repo = repo.unfiltered()
    commonrevs = outgoing.common
    csets = outgoing.missing
    heads = outgoing.missingheads
    # We go through the fast path if we get told to, or if all (unfiltered
    # heads have been requested (since we then know there all linkrevs will
    # be pulled by the client).
    heads.sort()
    fastpathlinkrev = fastpath or (
        repo.filtername is None and heads == sorted(repo.heads())
    )

    repo.hook('preoutgoing', throw=True, source=source)
    _changegroupinfo(repo, csets, source)
    return bundler.generate(commonrevs, csets, fastpathlinkrev, source)


def _addchangegroupfiles(repo, source, revmap, trp, expectedfiles, needfiles):
    revisions = 0
    files = 0
    progress = repo.ui.makeprogress(
        _('files'), unit=_('files'), total=expectedfiles
    )
    for chunkdata in iter(source.filelogheader, {}):
        files += 1
        f = chunkdata["filename"]
        repo.ui.debug("adding %s revisions\n" % f)
        progress.increment()

            for new in pycompat.xrange(o, len(fl)):
                n = fl.node(new)
                if n in needs:
                    needs.remove(n)
                else:
                    raise error.Abort(_("received spurious file revlog entry"))
            if not needs:
                del needfiles[f]
    progress.complete()

    for f, needs in needfiles.iteritems():

        for n in needs:
            try:
                fl.rev(n)
            except error.LookupError:
                raise error.Abort(
                    _('missing file data for %s:%s - run hg verify')
                    % (f, hex(n))
                )

    return revisions, files