author Yuya Nishihara <>
Tue, 05 Dec 2017 21:50:33 +0900
changeset 40931 c752fbe228fb5fb5338b4b90efcfed7c2046ef7b
parent 39897 d0db41af73c08bb55c7d6bb487bc94d6ae89c635
child 40422 308681291d116b50962b6df9323fd52e8deadc2b
child 41306 07fdac1d5c666cf00ddcd5358eba9cfd146d099a
permissions -rw-r--r--
repoview: extract a factory function of proxy class This makes sure that dynamically-created class objects are isolated from local binding of repo instances. The type cache is moved to module level as it isn't tied to each instance.

# - logic to computes, maintain and stores branchmap for local repo
# Copyright 2005-2007 Matt Mackall <>
# 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 struct

from .node import (
from . import (

calcsize = struct.calcsize
pack_into = struct.pack_into
unpack_from = struct.unpack_from

def _filename(repo):
    """name of a branchcache file for a given repo or repoview"""
    filename = "branch2"
    if repo.filtername:
        filename = '%s-%s' % (filename, repo.filtername)
    return filename

def read(repo):
        f = repo.cachevfs(_filename(repo))
        lines ='\n')
    except (IOError, OSError):
        return None

        cachekey = lines.pop(0).split(" ", 2)
        last, lrev = cachekey[:2]
        last, lrev = bin(last), int(lrev)
        filteredhash = None
        if len(cachekey) > 2:
            filteredhash = bin(cachekey[2])
        partial = branchcache(tipnode=last, tiprev=lrev,
        if not partial.validfor(repo):
            # invalidate the cache
            raise ValueError('tip differs')
        cl = repo.changelog
        for l in lines:
            if not l:
            node, state, label = l.split(" ", 2)
            if state not in 'oc':
                raise ValueError('invalid branch state')
            label = encoding.tolocal(label.strip())
            node = bin(node)
            if not cl.hasnode(node):
                raise ValueError('node %s does not exist' % hex(node))
            partial.setdefault(label, []).append(node)
            if state == 'c':
    except Exception as inst:
        if repo.ui.debugflag:
            msg = 'invalid branchheads cache'
            if repo.filtername is not None:
                msg += ' (%s)' % repo.filtername
            msg += ': %s\n'
            repo.ui.debug(msg % inst)
        partial = None
    return partial

### Nearest subset relation
# Nearest subset of filter X is a filter Y so that:
# * Y is included in X,
# * X - Y is as small as possible.
# This create and ordering used for branchmap purpose.
# the ordering may be partial
subsettable = {None: 'visible',
               'visible': 'served',
               'served': 'immutable',
               'immutable': 'base'}

def updatecache(repo):
    cl = repo.changelog
    filtername = repo.filtername
    partial = repo._branchcaches.get(filtername)

    revs = []
    if partial is None or not partial.validfor(repo):
        partial = read(repo)
        if partial is None:
            subsetname = subsettable.get(filtername)
            if subsetname is None:
                partial = branchcache()
                subset = repo.filtered(subsetname)
                partial = subset.branchmap().copy()
                extrarevs = subset.changelog.filteredrevs - cl.filteredrevs
                revs.extend(r for  r in extrarevs if r <= partial.tiprev)
    revs.extend(cl.revs(start=partial.tiprev + 1))
    if revs:
        partial.update(repo, revs)

    assert partial.validfor(repo), filtername
    repo._branchcaches[repo.filtername] = partial

def replacecache(repo, bm):
    """Replace the branchmap cache for a repo with a branch mapping.

    This is likely only called during clone with a branch map from a remote.
    rbheads = []
    closed = []
    for bheads in bm.itervalues():
        for h in bheads:
            r = repo.changelog.rev(h)
            b, c = repo.changelog.branchinfo(r)
            if c:

    if rbheads:
        rtiprev = max((int(repo.changelog.rev(node))
                for node in rbheads))
        cache = branchcache(bm,

        # Try to stick it as low as possible
        # filter above served are unlikely to be fetch from a clone
        for candidate in ('base', 'immutable', 'served'):
            rview = repo.filtered(candidate)
            if cache.validfor(rview):
                repo._branchcaches[candidate] = cache

class branchcache(dict):
    """A dict like object that hold branches heads cache.

    This cache is used to avoid costly computations to determine all the
    branch heads of a repo.

    The cache is serialized on disk in the following format:

    <tip hex node> <tip rev number> [optional filtered repo hex hash]
    <branch head hex node> <open/closed state> <branch name>
    <branch head hex node> <open/closed state> <branch name>

    The first line is used to check if the cache is still valid. If the
    branch cache is for a filtered repo view, an optional third hash is
    included that hashes the hashes of all filtered revisions.

    The open/closed state is represented by a single letter 'o' or 'c'.
    This field can be used to avoid changelog reads when determining if a
    branch head closes a branch or not.

    def __init__(self, entries=(), tipnode=nullid, tiprev=nullrev,
                 filteredhash=None, closednodes=None):
        super(branchcache, self).__init__(entries)
        self.tipnode = tipnode
        self.tiprev = tiprev
        self.filteredhash = filteredhash
        # closednodes is a set of nodes that close their branch. If the branch
        # cache has been updated, it may contain nodes that are no longer
        # heads.
        if closednodes is None:
            self._closednodes = set()
            self._closednodes = closednodes

    def validfor(self, repo):
        """Is the cache content valid regarding a repo

        - False when cached tipnode is unknown or if we detect a strip.
        - True when cache is up to date or a subset of current repo."""
            return ((self.tipnode == repo.changelog.node(self.tiprev))
                    and (self.filteredhash == \
                         scmutil.filteredhash(repo, self.tiprev)))
        except IndexError:
            return False

    def _branchtip(self, heads):
        '''Return tuple with last open head in heads and false,
        otherwise return last closed head and true.'''
        tip = heads[-1]
        closed = True
        for h in reversed(heads):
            if h not in self._closednodes:
                tip = h
                closed = False
        return tip, closed

    def branchtip(self, branch):
        '''Return the tipmost open head on branch head, otherwise return the
        tipmost closed head on branch.
        Raise KeyError for unknown branch.'''
        return self._branchtip(self[branch])[0]

    def iteropen(self, nodes):
        return (n for n in nodes if n not in self._closednodes)

    def branchheads(self, branch, closed=False):
        heads = self[branch]
        if not closed:
            heads = list(self.iteropen(heads))
        return heads

    def iterbranches(self):
        for bn, heads in self.iteritems():
            yield (bn, heads) + self._branchtip(heads)

    def copy(self):
        """return an deep copy of the branchcache object"""
        return branchcache(self, self.tipnode, self.tiprev, self.filteredhash,

    def write(self, repo):
            f = repo.cachevfs(_filename(repo), "w", atomictemp=True)
            cachekey = [hex(self.tipnode), '%d' % self.tiprev]
            if self.filteredhash is not None:
            f.write(" ".join(cachekey) + '\n')
            nodecount = 0
            for label, nodes in sorted(self.iteritems()):
                for node in nodes:
                    nodecount += 1
                    if node in self._closednodes:
                        state = 'c'
                        state = 'o'
                    f.write("%s %s %s\n" % (hex(node), state,
                        'wrote %s branch cache with %d labels and %d nodes\n',
                        repo.filtername, len(self), nodecount)
        except (IOError, OSError, error.Abort) as inst:
            # Abort may be raised by read only opener, so log and continue
            repo.ui.debug("couldn't write branch cache: %s\n" % inst)

    def update(self, repo, revgen):
        """Given a branchhead cache, self, that may have extra nodes or be
        missing heads, and a generator of nodes that are strictly a superset of
        heads missing, this function updates self to be correct.
        starttime = util.timer()
        cl = repo.changelog
        # collect new branch entries
        newbranches = {}
        getbranchinfo = repo.revbranchcache().branchinfo
        for r in revgen:
            branch, closesbranch = getbranchinfo(r)
            newbranches.setdefault(branch, []).append(r)
            if closesbranch:

        # fetch current topological heads to speed up filtering
        topoheads = set(cl.headrevs())

        # if older branchheads are reachable from new ones, they aren't
        # really branchheads. Note checking parents is insufficient:
        # 1 (branch a) -> 2 (branch b) -> 3 (branch a)
        for branch, newheadrevs in newbranches.iteritems():
            bheads = self.setdefault(branch, [])
            bheadset = set(cl.rev(node) for node in bheads)

            # This have been tested True on all internal usage of this function.
            # run it again in case of doubt
            # assert not (set(bheadrevs) & set(newheadrevs))

            # This prunes out two kinds of heads - heads that are superseded by
            # a head in newheadrevs, and newheadrevs that are not heads because
            # an existing head is their descendant.
            uncertain = bheadset - topoheads
            if uncertain:
                floorrev = min(uncertain)
                ancestors = set(cl.ancestors(newheadrevs, floorrev))
                bheadset -= ancestors
            bheadrevs = sorted(bheadset)
            self[branch] = [cl.node(rev) for rev in bheadrevs]
            tiprev = bheadrevs[-1]
            if tiprev > self.tiprev:
                self.tipnode = cl.node(tiprev)
                self.tiprev = tiprev

        if not self.validfor(repo):
            # cache key are not valid anymore
            self.tipnode = nullid
            self.tiprev = nullrev
            for heads in self.values():
                tiprev = max(cl.rev(node) for node in heads)
                if tiprev > self.tiprev:
                    self.tipnode = cl.node(tiprev)
                    self.tiprev = tiprev
        self.filteredhash = scmutil.filteredhash(repo, self.tiprev)

        duration = util.timer() - starttime
        repo.ui.log('branchcache', 'updated %s branch cache in %.4f seconds\n',
                    repo.filtername, duration)

# Revision branch info cache

_rbcversion = '-v1'
_rbcnames = 'rbc-names' + _rbcversion
_rbcrevs = 'rbc-revs' + _rbcversion
# [4 byte hash prefix][4 byte branch name number with sign bit indicating open]
_rbcrecfmt = '>4sI'
_rbcrecsize = calcsize(_rbcrecfmt)
_rbcnodelen = 4
_rbcbranchidxmask = 0x7fffffff
_rbccloseflag = 0x80000000

class revbranchcache(object):
    """Persistent cache, mapping from revision number to branch name and close.
    This is a low level cache, independent of filtering.

    Branch names are stored in rbc-names in internal encoding separated by 0.
    rbc-names is append-only, and each branch name is only stored once and will
    thus have a unique index.

    The branch info for each revision is stored in rbc-revs as constant size
    records. The whole file is read into memory, but it is only 'parsed' on
    demand. The file is usually append-only but will be truncated if repo
    modification is detected.
    The record for each revision contains the first 4 bytes of the
    corresponding node hash, and the record is only used if it still matches.
    Even a completely trashed rbc-revs fill thus still give the right result
    while converging towards full recovery ... assuming no incorrectly matching
    node hashes.
    The record also contains 4 bytes where 31 bits contains the index of the
    branch and the last bit indicate that it is a branch close commit.
    The usage pattern for rbc-revs is thus somewhat similar to 00changelog.i
    and will grow with it but be 1/8th of its size.

    def __init__(self, repo, readonly=True):
        assert repo.filtername is None
        self._repo = repo
        self._names = [] # branch names in local encoding with static index
        self._rbcrevs = bytearray()
        self._rbcsnameslen = 0 # length of names read at _rbcsnameslen
            bndata =
            self._rbcsnameslen = len(bndata) # for verification before writing
            if bndata:
                self._names = [encoding.tolocal(bn)
                               for bn in bndata.split('\0')]
        except (IOError, OSError):
            if readonly:
                # don't try to use cache - fall back to the slow path
                self.branchinfo = self._branchinfo

        if self._names:
                data =
                self._rbcrevs[:] = data
            except (IOError, OSError) as inst:
                repo.ui.debug("couldn't read revision branch cache: %s\n" %
        # remember number of good records on disk
        self._rbcrevslen = min(len(self._rbcrevs) // _rbcrecsize,
        if self._rbcrevslen == 0:
            self._names = []
        self._rbcnamescount = len(self._names) # number of names read at
                                               # _rbcsnameslen
        self._namesreverse = dict((b, r) for r, b in enumerate(self._names))

    def _clear(self):
        self._rbcsnameslen = 0
        del self._names[:]
        self._rbcnamescount = 0
        self._rbcrevslen = len(self._repo.changelog)
        self._rbcrevs = bytearray(self._rbcrevslen * _rbcrecsize)

    def branchinfo(self, rev):
        """Return branch name and close flag for rev, using and updating
        persistent cache."""
        changelog = self._repo.changelog
        rbcrevidx = rev * _rbcrecsize

        # avoid negative index, is fast without cache
        if rev == nullrev:
            return changelog.branchinfo(rev)

        # if requested rev isn't allocated, grow and cache the rev info
        if len(self._rbcrevs) < rbcrevidx + _rbcrecsize:
            return self._branchinfo(rev)

        # fast path: extract data from cache, use it if node is matching
        reponode = changelog.node(rev)[:_rbcnodelen]
        cachenode, branchidx = unpack_from(
            _rbcrecfmt, util.buffer(self._rbcrevs), rbcrevidx)
        close = bool(branchidx & _rbccloseflag)
        if close:
            branchidx &= _rbcbranchidxmask
        if cachenode == '\0\0\0\0':
        elif cachenode == reponode:
                return self._names[branchidx], close
            except IndexError:
                # recover from invalid reference to unknown branch
                self._repo.ui.debug("referenced branch names not found"
                    " - rebuilding revision branch cache from scratch\n")
            # rev/node map has changed, invalidate the cache from here up
            self._repo.ui.debug("history modification detected - truncating "
                "revision branch cache to revision %d\n" % rev)
            truncate = rbcrevidx + _rbcrecsize
            del self._rbcrevs[truncate:]
            self._rbcrevslen = min(self._rbcrevslen, truncate)

        # fall back to slow path and make sure it will be written to disk
        return self._branchinfo(rev)

    def _branchinfo(self, rev):
        """Retrieve branch info from changelog and update _rbcrevs"""
        changelog = self._repo.changelog
        b, close = changelog.branchinfo(rev)
        if b in self._namesreverse:
            branchidx = self._namesreverse[b]
            branchidx = len(self._names)
            self._namesreverse[b] = branchidx
        reponode = changelog.node(rev)
        if close:
            branchidx |= _rbccloseflag
        self._setcachedata(rev, reponode, branchidx)
        return b, close

    def _setcachedata(self, rev, node, branchidx):
        """Writes the node's branch data to the in-memory cache data."""
        if rev == nullrev:
        rbcrevidx = rev * _rbcrecsize
        if len(self._rbcrevs) < rbcrevidx + _rbcrecsize:
            self._rbcrevs.extend('\0' *
                                 (len(self._repo.changelog) * _rbcrecsize -
        pack_into(_rbcrecfmt, self._rbcrevs, rbcrevidx, node, branchidx)
        self._rbcrevslen = min(self._rbcrevslen, rev)

        tr = self._repo.currenttransaction()
        if tr:
            tr.addfinalize('write-revbranchcache', self.write)

    def write(self, tr=None):
        """Save branch cache if it is dirty."""
        repo = self._repo
        wlock = None
        step = ''
            if self._rbcnamescount < len(self._names):
                step = ' names'
                wlock = repo.wlock(wait=False)
                if self._rbcnamescount != 0:
                    f =, 'ab')
                    if f.tell() == self._rbcsnameslen:
                        repo.ui.debug("%s changed - rewriting it\n" % _rbcnames)
                        self._rbcnamescount = 0
                        self._rbcrevslen = 0
                if self._rbcnamescount == 0:
                    # before rewriting names, make sure references are removed
                    repo.cachevfs.unlinkpath(_rbcrevs, ignoremissing=True)
                    f =, 'wb')
                                  for b in self._names[self._rbcnamescount:]))
                self._rbcsnameslen = f.tell()
                self._rbcnamescount = len(self._names)

            start = self._rbcrevslen * _rbcrecsize
            if start != len(self._rbcrevs):
                step = ''
                if wlock is None:
                    wlock = repo.wlock(wait=False)
                revs = min(len(repo.changelog),
                           len(self._rbcrevs) // _rbcrecsize)
                f =, 'ab')
                if f.tell() != start:
                    repo.ui.debug("truncating cache/%s to %d\n"
                                  % (_rbcrevs, start))
                    if f.tell() != start:
                        start = 0
                end = revs * _rbcrecsize
                self._rbcrevslen = revs
        except (IOError, OSError, error.Abort, error.LockError) as inst:
            repo.ui.debug("couldn't write revision branch cache%s: %s\n"
                          % (step, inst))
            if wlock is not None: