def bundle(args):
'''create a mercurial bundle'''
bundle_commits = list((c, p) for c, t, p in GitHgHelper.rev_list(
'--topo-order', '--full-history', '--parents', '--reverse', *args.rev))
if bundle_commits:
# TODO: better UX. For instance, this will fail with an exception when
# the parent commit doesn't have mercurial metadata.
GRAFT = {
None: False,
'false': False,
'true': True,
}
try:
graft = Git.config('cinnabar.graft', values=GRAFT)
except InvalidConfig as e:
logging.error(e.message)
return 1
store = PushStore(graft=graft)
if args.version == 1:
b2caps = {}
elif args.version == 2:
b2caps = {
'HG20': (),
'changegroup': ('01', '02'),
}
with open(args.path, 'wb') as fh:
if not b2caps:
fh.write('HG10UN')
for data in create_bundle(store, bundle_commits, b2caps):
fh.write(data)
store.close(rollback=True)
def local_bases():
h = chain(heads(), (w for w in what if w))
for c, t, p in GitHgHelper.rev_list('--topo-order', '--full-history',
'--boundary', *h):
if c[0] != '-':
continue
yield store.hg_changeset(c[1:])
for w in what:
rev = store.hg_changeset(w)
if rev:
yield rev
def local_bases():
h = chain(heads(), (w for w in what if w))
for c, t, p in GitHgHelper.rev_list('--topo-order', '--full-history',
'--boundary', *h):
if c[0] != '-':
continue
yield store.hg_changeset(c[1:])
for w in what:
rev = store.hg_changeset(w)
if rev:
yield rev
def push(repo, store, what, repo_heads, repo_branches, dry_run=False):
def heads():
for sha1 in store.heads(repo_branches):
yield '^%s' % store.changeset_ref(sha1)
def local_bases():
h = chain(heads(), (w for w in what if w))
for c, t, p in GitHgHelper.rev_list('--topo-order', '--full-history',
'--boundary', *h):
if c[0] != '-':
continue
yield store.hg_changeset(c[1:])
for w in what:
rev = store.hg_changeset(w)
if rev:
yield rev
common = findcommon(repo, store, set(local_bases()))
logging.info('common: %s', common)
def revs():
for sha1 in common:
yield '^%s' % store.changeset_ref(sha1)
revs = chain(revs(), (w for w in what if w))
push_commits = list((c, p) for c, t, p in GitHgHelper.rev_list(
'--topo-order', '--full-history', '--parents', '--reverse', *revs))
pushed = False
if push_commits:
has_root = any(len(p) == 40 for p in push_commits)
force = all(v[1] for v in what.values())
if has_root and repo_heads:
if not force:
raise Exception('Cannot push a new root')
else:
logging.warn('Pushing a new root')
if force:
repo_heads = ['force']
else:
if not repo_heads:
repo_heads = [NULL_NODE_ID]
repo_heads = [unhexlify(h) for h in repo_heads]
if push_commits and not dry_run:
if repo.local():
repo.local().ui.setconfig('server', 'validate', True)
b2caps = bundle2caps(repo) if unbundle20 else {}
logging.getLogger('bundle2').debug('%r', b2caps)
if b2caps:
b2caps['replycaps'] = encodecaps({'error': ['abort']})
cg = create_bundle(store, push_commits, b2caps)
if not isinstance(repo, HelperRepo):
cg = util.chunkbuffer(cg)
if not b2caps:
cg = cg1unpacker(cg, 'UN')
reply = repo.unbundle(cg, repo_heads, '')
if unbundle20 and isinstance(reply, unbundle20):
parts = iter(reply.iterparts())
for part in parts:
logging.getLogger('bundle2').debug('part: %s', part.type)
logging.getLogger('bundle2').debug('params: %r', part.params)
if part.type == 'output':
sys.stderr.write(part.read())
elif part.type == 'reply:changegroup':
# TODO: should check params['in-reply-to']
reply = int(part.params['return'])
elif part.type == 'error:abort':
raise error.Abort(part.params['message'],
hint=part.params.get('hint'))
else:
logging.getLogger('bundle2').warning(
'ignoring bundle2 part: %s', part.type)
pushed = reply != 0
return gitdag(push_commits) if pushed or dry_run else ()
def findcommon(repo, store, hgheads):
logger = logging.getLogger('findcommon')
logger.debug(hgheads)
if not hgheads:
return set()
sample_size = 100
sample = _sample(hgheads, sample_size)
known = repo.known(unhexlify(h) for h in sample)
known = set(h for h, k in izip(sample, known) if k)
logger.info('initial sample size: %d', len(sample))
if len(known) == len(hgheads):
logger.info('all heads known')
return hgheads
git_heads = set(store.changeset_ref(h) for h in hgheads)
git_known = set(store.changeset_ref(h) for h in known)
if logger.isEnabledFor(logging.DEBUG):
logger.debug('known (sub)set: (%d) %s', len(known), sorted(git_known))
args = ['--topo-order', '--full-history', '--parents']
def revs():
for h in git_known:
yield '^%s' % h
for h in git_heads:
if h not in git_known:
yield h
args.extend(revs())
revs = ((c, parents) for c, t, parents in GitHgHelper.rev_list(*args))
dag = gitdag(chain(revs, ((k, ()) for k in git_known)))
dag.tag_nodes_and_parents(git_known, 'known')
def log_dag(tag):
if not logger.isEnabledFor(logging.DEBUG):
return
logger.debug('%s dag size: %d', tag,
sum(1 for n in dag.iternodes(tag)))
heads = sorted(dag.heads(tag))
logger.debug('%s dag heads: (%d) %s', tag, len(heads), heads)
roots = sorted(dag.roots(tag))
logger.debug('%s dag roots: (%d) %s', tag, len(roots), roots)
log_dag('unknown')
log_dag('known')
while True:
unknown = set(chain(dag.heads(), dag.roots()))
if not unknown:
break
sample = set(_sample(unknown, sample_size))
if len(sample) < sample_size:
sample |= set(_sample(set(dag.iternodes()),
sample_size - len(sample)))
sample = list(sample)
hg_sample = [store.hg_changeset(h) for h in sample]
known = repo.known(unhexlify(h) for h in hg_sample)
unknown = set(h for h, k in izip(sample, known) if not k)
known = set(h for h, k in izip(sample, known) if k)
logger.info('next sample size: %d', len(sample))
if logger.isEnabledFor(logging.DEBUG):
logger.debug('known (sub)set: (%d) %s', len(known), sorted(known))
logger.debug('unknown (sub)set: (%d) %s', len(unknown),
sorted(unknown))
dag.tag_nodes_and_parents(known, 'known')
dag.tag_nodes_and_children(unknown, 'unknown')
log_dag('unknown')
log_dag('known')
return [store.hg_changeset(h) for h in dag.heads('known')]
def fsck_quick(force=False):
status = FsckStatus()
store = GitHgStore()
metadata_commit = Git.resolve_ref('refs/cinnabar/metadata')
if not metadata_commit:
status.info('There does not seem to be any git-cinnabar metadata.\n'
'Is this a git-cinnabar clone?')
return 1
broken_metadata = Git.resolve_ref('refs/cinnabar/broken')
checked_metadata = Git.resolve_ref('refs/cinnabar/checked')
if checked_metadata == broken_metadata:
checked_metadata = None
if metadata_commit == checked_metadata and not force:
status.info('The git-cinnabar metadata was already checked and is '
'presumably clean.\n'
'Try `--force` if you want to check anyways.')
return 0
elif force:
checked_metadata = None
commit = GitCommit(metadata_commit)
if commit.body != 'files-meta unified-manifests-v2':
status.info(
'The git-cinnabar metadata is incompatible with this version.\n'
'Please use the git-cinnabar version it was used with last.\n')
return 1
if len(commit.parents) > 6 or len(commit.parents) < 5:
status.report('The git-cinnabar metadata seems to be corrupted in '
'unexpected ways.\n')
return 1
changesets, manifests, hg2git, git2hg, files_meta = commit.parents[:5]
commit = GitCommit(changesets)
heads = OrderedDict(
(node, branch)
for node, _, branch in (d.partition(' ')
for d in commit.body.splitlines()))
if len(heads) != len(commit.parents):
status.report('The git-cinnabar metadata seems to be corrupted in '
'unexpected ways.\n')
return 1
manifest_nodes = []
parents = None
fix_changeset_heads = False
def get_checked_metadata(num):
if not checked_metadata:
return None
commit = Git.resolve_ref('{}^{}'.format(checked_metadata, num))
if commit:
return GitCommit(commit)
checked_commit = get_checked_metadata(1)
# TODO: Check that the recorded heads are actually dag heads.
for c, changeset_node in progress_iter(
'Checking {} changeset heads',
((c, node) for c, node in izip(commit.parents, heads)
if not checked_commit or c not in checked_commit.parents)):
gitsha1 = GitHgHelper.hg2git(changeset_node)
if gitsha1 == NULL_NODE_ID:
status.report('Missing hg2git metadata for changeset %s' %
changeset_node)
continue
if gitsha1 != c:
if parents is None:
parents = set(commit.parents)
if gitsha1 not in parents:
status.report('Inconsistent metadata:\n'
' Head metadata says changeset %s maps to %s\n'
' but hg2git metadata says it maps to %s' %
(changeset_node, c, gitsha1))
continue
fix_changeset_heads = True
changeset = store._changeset(c, include_parents=True)
if not changeset:
status.report('Missing git2hg metadata for git commit %s' % c)
continue
if changeset.node != changeset_node:
if changeset.node not in heads:
status.report(
'Inconsistent metadata:\n'
' Head metadata says %s maps to changeset %s\n'
' but git2hg metadata says it maps to changeset %s' %
(c, changeset_node, changeset.node))
continue
fix_changeset_heads = True
if changeset.node != changeset.sha1:
status.report('Sha1 mismatch for changeset %s' % changeset.node)
continue
changeset_branch = changeset.branch or 'default'
if heads[changeset.node] != changeset_branch:
status.report(
'Inconsistent metadata:\n'
' Head metadata says changeset %s is in branch %s\n'
' but git2hg metadata says it is in branch %s' %
(changeset.node, heads[changeset.node], changeset_branch))
continue
manifest_nodes.append(changeset.manifest)
if status('broken'):
return 1
# Rebuilding manifests benefits from limiting the difference with
# the last rebuilt manifest. Similarly, building the list of unique
# files in all manifests benefits from that too.
# Unfortunately, the manifest heads are not ordered in a topological
# relevant matter, and the differences between two consecutive manifests
# can be much larger than they could be. The consequence is spending a
# large amount of time rebuilding the manifests and gathering the files
# list. It's actually faster to attempt to reorder them according to
# some heuristics first, such that the differences are smaller.
# Here, we use the depth from the root node(s) to reorder the manifests.
# This doesn't give the most optimal ordering, but it's already much
# faster. On a clone of multiple mozilla-* repositories with > 1400 heads,
# it's close to an order of magnitude difference on the "Checking
# manifests" loop.
depths = {}
roots = {}
manifest_queue = []
revs = []
revs.append('{}^@'.format(manifests))
if checked_metadata:
revs.append('^{}^2^@'.format(checked_metadata))
for m, _, parents in progress_iter(
'Loading {} manifests',
GitHgHelper.rev_list('--topo-order', '--reverse', '--full-history',
*revs)):
manifest_queue.append((m, parents))
if parents:
depth = {}
for p in parents:
for root, num in depths.get(p, {}).iteritems():
if root in depth:
depth[root] = max(depth[root], num + 1)
else:
depth[root] = num + 1
if depth:
depths[m] = depth
del depth
continue
depths[m] = {m: 0}
roots[m] = parents
if status('broken'):
return 1
# TODO: check that all manifest_nodes gathered above are available in the
# manifests dag, and that the dag heads are the recorded heads.
manifests_commit = GitCommit(manifests)
checked_commit = get_checked_metadata(2)
depths = [([depths[p].get(r, 0) for r in roots], p)
for p in manifests_commit.parents
if not checked_commit or p not in checked_commit.parents]
manifests_commit_parents = [p for _, p in sorted(depths)]
previous = None
all_interesting = set()
for m in progress_iter('Checking {} manifest heads',
manifests_commit_parents):
c = GitCommit(m)
if not SHA1_RE.match(c.body):
status.report('Invalid manifest metadata in git commit %s' % m)
continue
gitsha1 = GitHgHelper.hg2git(c.body)
if gitsha1 == NULL_NODE_ID:
status.report('Missing hg2git metadata for manifest %s' % c.body)
continue
if not GitHgHelper.check_manifest(c.body):
status.report('Sha1 mismatch for manifest %s' % c.body)
files = {}
if previous:
for _, _, before, after, d, path in GitHgHelper.diff_tree(
previous, m):
if d in 'AM' and before != after and \
(path, after) not in all_interesting:
files[path] = after
else:
for _, t, sha1, path in GitHgHelper.ls_tree(m, recursive=True):
if (path, sha1) not in all_interesting:
files[path] = sha1
all_interesting.update(files.iteritems())
previous = m
if status('broken'):
return 1
# Don't check files that were already there in the previously checked
# manifests.
previous = None
for parents in roots.itervalues():
for p in parents:
if previous:
for _, _, before, after, d, path in GitHgHelper.diff_tree(
previous, p):
if d in 'AM' and before != after:
all_interesting.discard((path, after))
else:
for _, t, sha1, path in GitHgHelper.ls_tree(p, recursive=True):
all_interesting.discard((path, sha1))
previous = p
progress = Progress('Checking {} files')
while all_interesting and manifest_queue:
(m, parents) = manifest_queue.pop()
changes = get_changes(m, parents, all=True)
for path, hg_file, hg_fileparents in changes:
if hg_fileparents[1:] == (hg_file, ):
continue
elif hg_fileparents[:1] == (hg_file, ):
continue
# Reaching here means the file received a modification compared
# to its parents. If it's a file we're going to check below,
# it means we don't need to check its parents if somehow they were
# going to be checked. If it's not a file we're going to check
# below, it's because it's either a file we weren't interested in
# in the first place, or it's the parent of a file we have checked.
# Either way, we aren't interested in the parents.
for p in hg_fileparents:
all_interesting.discard((path, p))
if (path, hg_file) not in all_interesting:
continue
all_interesting.remove((path, hg_file))
if not GitHgHelper.check_file(hg_file, *hg_fileparents):
p = store.manifest_path(path)
status.report('Sha1 mismatch for file %s\n'
' revision %s' % (p, hg_file))
print_parents = ' '.join(p for p in hg_fileparents
if p != NULL_NODE_ID)
if print_parents:
status.report(' with parent%s %s' %
('s' if len(print_parents) > 41 else '',
print_parents))
progress.progress()
progress.finish()
if all_interesting:
status.info('Could not find the following files:')
for path, sha1 in sorted(all_interesting):
p = store.manifest_path(path)
status.info(' %s %s' % (sha1, path))
status.info('This might be a bug in `git cinnabar fsck`. Please open '
'an issue, with the message above, on\n'
'https://github.com/glandium/git-cinnabar/issues')
return 1
if status('broken'):
status.info('Your git-cinnabar repository appears to be corrupted.\n'
'Please open an issue, with the information above, on\n'
'https://github.com/glandium/git-cinnabar/issues')
Git.update_ref('refs/cinnabar/broken', metadata_commit)
if checked_metadata:
status.info(
'\nThen please try to run `git cinnabar rollback --fsck` to '
'restore last known state, and to update from the mercurial '
'repository.')
else:
status.info('\nThen please try to run `git cinnabar reclone`.')
status.info(
'\nPlease note this may affect the commit sha1s of mercurial '
'changesets, and may require to rebase your local branches.')
status.info(
'\nAlternatively, you may start afresh with a new clone. In any '
'case, please keep this corrupted repository around for further '
'debugging.')
return 1
refresh = []
if fix_changeset_heads:
status.fix('Fixing changeset heads metadata order.')
refresh.append('refs/cinnabar/changesets')
interval_expired('fsck', 0)
store.close(refresh=refresh)
GitHgHelper._helper = False
metadata_commit = Git.resolve_ref('refs/cinnabar/metadata')
Git.update_ref('refs/cinnabar/checked', metadata_commit)
return 0
def fsck(args):
'''check cinnabar metadata consistency'''
if not args.commit and not args.full:
return fsck_quick(args.force)
status = FsckStatus()
store = GitHgStore()
if args.full and args.commit:
logging.error('Cannot pass both --full and a commit')
return 1
if args.commit:
commits = set()
all_git_commits = {}
for c in args.commit:
cs = store.hg_changeset(c)
if cs:
commits.add(c)
c = cs.node
commit = GitHgHelper.hg2git(c)
if commit == NULL_NODE_ID and not cs:
status.info('Unknown commit or changeset: %s' % c)
return 1
if not cs:
cs = store.hg_changeset(commit)
commits.add(commit)
all_git_commits = GitHgHelper.rev_list('--no-walk=unsorted', *commits)
else:
all_refs = dict(
(ref, sha1) for sha1, ref in Git.for_each_ref('refs/cinnabar'))
if 'refs/cinnabar/metadata' in all_refs:
git_heads = '%s^^@' % all_refs['refs/cinnabar/metadata']
else:
assert False
all_git_commits = GitHgHelper.rev_list('--topo-order',
'--full-history', '--reverse',
git_heads)
dag = gitdag()
GitHgHelper.reset_heads('manifests')
full_file_check = FileFindParents.logger.isEnabledFor(logging.DEBUG)
for node, tree, parents in progress_iter('Checking {} changesets',
all_git_commits):
node = store._replace.get(node, node)
hg_node = store.hg_changeset(node)
if not hg_node:
status.report('Missing note for git commit: ' + node)
continue
GitHgHelper.seen('git2hg', node)
changeset_data = store.changeset(hg_node)
changeset = changeset_data.node
GitHgHelper.seen('hg2git', changeset)
changeset_ref = store.changeset_ref(changeset)
if not changeset_ref:
status.report('Missing changeset in hg2git branch: %s' % changeset)
continue
elif str(changeset_ref) != node:
status.report('Commit mismatch for changeset %s\n'
' hg2git: %s\n commit: %s' %
(changeset, changeset_ref, node))
hg_changeset = store.changeset(changeset, include_parents=True)
if hg_changeset.node != hg_changeset.sha1:
status.report('Sha1 mismatch for changeset %s' % changeset)
dag.add(hg_changeset.node,
(hg_changeset.parent1, hg_changeset.parent2),
changeset_data.branch or 'default')
raw_changeset = Changeset.from_git_commit(node)
patcher = ChangesetPatcher.from_diff(raw_changeset, changeset_data)
if patcher != store.read_changeset_data(node):
status.fix('Adjusted changeset metadata for %s' % changeset)
GitHgHelper.set('changeset', changeset, NULL_NODE_ID)
GitHgHelper.set('changeset', changeset, node)
GitHgHelper.put_blob(patcher, want_sha1=False)
GitHgHelper.set('changeset-metadata', changeset, NULL_NODE_ID)
GitHgHelper.set('changeset-metadata', changeset, ':1')
manifest = changeset_data.manifest
if GitHgHelper.seen('hg2git', manifest) or manifest == NULL_NODE_ID:
continue
manifest_ref = store.manifest_ref(manifest)
if not manifest_ref:
status.report('Missing manifest in hg2git branch: %s' % manifest)
parents = tuple(
store.changeset(p).manifest for p in hg_changeset.parents)
git_parents = tuple(
store.manifest_ref(p) for p in parents if p != NULL_NODE_ID)
# This doesn't change the value but makes the helper track the manifest
# dag.
GitHgHelper.set('manifest', manifest, manifest_ref)
if not GitHgHelper.check_manifest(manifest):
status.report('Sha1 mismatch for manifest %s' % manifest)
manifest_commit_parents = GitCommit(manifest_ref).parents
if sorted(manifest_commit_parents) != sorted(git_parents):
# TODO: better error
status.report(
'%s(%s) %s != %s' %
(manifest, manifest_ref, manifest_commit_parents, git_parents))
# TODO: check that manifest content matches changeset content
changes = get_changes(manifest_ref, git_parents)
for path, hg_file, hg_fileparents in changes:
if hg_file != NULL_NODE_ID and (hg_file == HG_EMPTY_FILE
or GitHgHelper.seen(
'hg2git', hg_file)):
if full_file_check:
file = store.file(hg_file, hg_fileparents, git_parents,
store.manifest_path(path))
valid = file.node == file.sha1
else:
valid = GitHgHelper.check_file(hg_file, *hg_fileparents)
if not valid:
status.report('Sha1 mismatch for file %s in manifest %s' %
(hg_file, manifest_ref))
if not args.commit and not status('broken'):
store_manifest_heads = set(store._manifest_heads_orig)
manifest_heads = set(GitHgHelper.heads('manifests'))
if store_manifest_heads != manifest_heads:
def iter_manifests(a, b):
for h in a - b:
yield h
for h in b:
yield '^%s' % h
for m, t, p in GitHgHelper.rev_list(
'--topo-order', '--full-history', '--reverse',
*iter_manifests(manifest_heads, store_manifest_heads)):
status.fix('Missing manifest commit in manifest branch: %s' %
m)
for m, t, p in GitHgHelper.rev_list(
'--topo-order', '--full-history', '--reverse',
*iter_manifests(store_manifest_heads, manifest_heads)):
status.fix('Removing metadata commit %s with no corresponding '
'changeset' % (m))
for h in store_manifest_heads - manifest_heads:
if GitHgHelper.seen('hg2git', store.hg_manifest(h)):
status.fix('Removing non-head reference to %s in manifests'
' metadata.' % h)
dangling = ()
if not args.commit and not status('broken'):
dangling = GitHgHelper.dangling('hg2git')
for obj in dangling:
status.fix('Removing dangling metadata for ' + obj)
# Theoretically, we should figure out if they are files, manifests
# or changesets and set the right variable accordingly, but in
# practice, it makes no difference. Reevaluate when GitHgStore.close
# is modified, though.
GitHgHelper.set('file', obj, NULL_NODE_ID)
GitHgHelper.set('file-meta', obj, NULL_NODE_ID)
if not args.commit and not status('broken'):
dangling = GitHgHelper.dangling('git2hg')
for c in dangling:
status.fix('Removing dangling note for commit ' + c)
GitHgHelper.set('changeset-metadata', c, NULL_NODE_ID)
if status('broken'):
status.info(
'Your git-cinnabar repository appears to be corrupted. There\n'
'are known issues in older revisions that have been fixed.\n'
'Please try running the following command to reset:\n'
' git cinnabar reclone\n\n'
'Please note this command may change the commit sha1s. Your\n'
'local branches will however stay untouched.\n'
'Please report any corruption that fsck would detect after a\n'
'reclone.')
if not args.commit:
status.info('Checking head references...')
computed_heads = defaultdict(set)
for branch, head in dag.all_heads():
computed_heads[branch].add(head)
for branch in sorted(dag.tags()):
stored_heads = store.heads({branch})
for head in computed_heads[branch] - stored_heads:
status.fix('Adding missing head %s in branch %s' %
(head, branch))
store.add_head(head)
for head in stored_heads - computed_heads[branch]:
status.fix('Removing non-head reference to %s in branch %s' %
(head, branch))
del store._hgheads[head]
metadata_commit = Git.resolve_ref('refs/cinnabar/metadata')
if status('broken'):
Git.update_ref('refs/cinnabar/broken', metadata_commit)
return 1
if args.full:
Git.update_ref('refs/cinnabar/checked', metadata_commit)
interval_expired('fsck', 0)
store.close()
if status('fixed'):
return 2
return 0
def push(self, *refspecs):
try:
default = 'never' if self._graft else 'phase'
values = {
None: default,
'': default,
'never': 'never',
'phase': 'phase',
'always': 'always',
}
data = Git.config('cinnabar.data', self._remote.name,
values=values)
except InvalidConfig as e:
logging.error(e.message)
return 1
pushes = {s.lstrip('+'): (d, s.startswith('+'))
for s, d in (r.split(':', 1) for r in refspecs)}
if not self._repo.capable('unbundle'):
for source, (dest, force) in pushes.iteritems():
self._helper.write(
'error %s Remote does not support the "unbundle" '
'capability\n' % dest)
self._helper.write('\n')
self._helper.flush()
else:
repo_heads = self._branchmap.heads()
PushStore.adopt(self._store, self._graft)
pushed = push(self._repo, self._store, pushes, repo_heads,
self._branchmap.names(), self._dry_run)
status = {}
for source, (dest, _) in pushes.iteritems():
if dest.startswith('refs/tags/'):
if source:
status[dest] = 'Pushing tags is unsupported'
else:
status[dest] = \
'Deleting remote tags is unsupported'
continue
if not dest.startswith(('refs/heads/bookmarks/',
'hg/bookmarks/')):
if source:
status[dest] = bool(len(pushed))
else:
status[dest] = \
'Deleting remote branches is unsupported'
continue
name = unquote(dest[21:])
if source:
source = self._store.hg_changeset(Git.resolve_ref(source))\
or ''
status[dest] = self._repo.pushkey(
'bookmarks', name, self._bookmarks.get(name, ''), source)
for source, (dest, force) in pushes.iteritems():
if status[dest] is True:
self._helper.write('ok %s\n' % dest)
elif status[dest]:
self._helper.write('error %s %s\n' % (dest, status[dest]))
else:
self._helper.write('error %s nothing changed on remote\n'
% dest)
self._helper.write('\n')
self._helper.flush()
if not pushed or self._dry_run:
data = False
elif data == 'always':
data = True
elif data == 'phase':
phases = self._repo.listkeys('phases')
drafts = {}
if not phases.get('publishing', False):
drafts = set(p for p, is_draft in phases.iteritems()
if int(is_draft))
if not drafts:
data = True
else:
def draft_commits():
for d in drafts:
c = self._store.changeset_ref(d)
if c:
yield '^%s^@' % c
for h in pushed.heads():
yield h
args = ['--ancestry-path', '--topo-order']
args.extend(draft_commits())
pushed_drafts = tuple(
c for c, t, p in GitHgHelper.rev_list(*args))
# Theoretically, we could have commits with no
# metadata that the remote declares are public, while
# the rest of our push is in a draft state. That is
# however so unlikely that it's not worth the effort
# to support partial metadata storage.
data = not bool(pushed_drafts)
elif data == 'never':
data = False
self._store.close(rollback=not data)
def push(repo, store, what, repo_heads, repo_branches, dry_run=False):
def heads():
for sha1 in store.heads(repo_branches):
yield b'^%s' % store.changeset_ref(sha1)
def local_bases():
h = chain(heads(), (w for w, _, _ in what if w))
for c, t, p in GitHgHelper.rev_list(b'--topo-order', b'--full-history',
b'--boundary', *h):
if c[:1] != b'-':
continue
yield store.hg_changeset(c[1:])
for w, _, _ in what:
if w:
rev = store.hg_changeset(w)
if rev:
yield rev
common = findcommon(repo, store, set(local_bases()))
logging.info('common: %s', common)
def revs():
for sha1 in common:
yield b'^%s' % store.changeset_ref(sha1)
revs = chain(revs(), (w for w, _, _ in what if w))
push_commits = list((c, p) for c, t, p in GitHgHelper.rev_list(
b'--topo-order', b'--full-history', b'--parents', b'--reverse', *revs))
pushed = False
if push_commits:
has_root = any(not p for (c, p) in push_commits)
force = all(v for _, _, v in what)
if has_root and repo_heads:
if not force:
raise Exception('Cannot push a new root')
else:
logging.warn('Pushing a new root')
if force:
repo_heads = [b'force']
else:
if not repo_heads:
repo_heads = [NULL_NODE_ID]
repo_heads = [unhexlify(h) for h in repo_heads]
if push_commits and not dry_run:
if repo.local():
repo.local().ui.setconfig(b'server', b'validate', True)
if unbundle20:
b2caps = repo.capable(b'bundle2') or {}
else:
b2caps = {}
if b2caps:
b2caps = decodecaps(unquote_to_bytes(b2caps))
logging.getLogger('bundle2').debug('%r', b2caps)
if b2caps:
b2caps[b'replycaps'] = encodecaps({b'error': [b'abort']})
cg = create_bundle(store, push_commits, b2caps)
if not isinstance(repo, HelperRepo):
cg = chunkbuffer(cg)
if not b2caps:
cg = cg1unpacker(cg, b'UN')
reply = repo.unbundle(cg, repo_heads, b'')
if unbundle20 and isinstance(reply, unbundle20):
parts = iter(reply.iterparts())
for part in parts:
logging.getLogger('bundle2').debug('part: %s', part.type)
logging.getLogger('bundle2').debug('params: %r', part.params)
if part.type == b'output':
sys.stderr.write(fsdecode(part.read()))
elif part.type == b'reply:changegroup':
# TODO: should check params['in-reply-to']
reply = int(part.params[b'return'])
elif part.type == b'error:abort':
message = part.params[b'message'].decode('utf-8')
hint = part.params.get(b'hint')
if hint:
message += '\n\n' + hint.decode('utf-8')
raise Exception(message)
else:
logging.getLogger(b'bundle2').warning(
'ignoring bundle2 part: %s', part.type)
pushed = reply != 0
return gitdag(push_commits) if pushed or dry_run else ()
def fsck_quick():
status = FsckStatus()
store = GitHgStore()
metadata_commit = Git.resolve_ref('refs/cinnabar/metadata')
if not metadata_commit:
status.info(
'There does not seem to be any git-cinnabar metadata.\n'
'Is this a git-cinnabar clone?'
)
return 1
commit = GitCommit(metadata_commit)
if commit.body != 'files-meta unified-manifests-v2':
status.info(
'The git-cinnabar metadata is incompatible with this version.\n'
'Please use the git-cinnabar version it was used with last.\n'
)
return 1
if len(commit.parents) > 6 or len(commit.parents) < 5:
status.report('The git-cinnabar metadata seems to be corrupted in '
'unexpected ways.\n')
return 1
changesets, manifests, hg2git, git2hg, files_meta = commit.parents[:5]
commit = GitCommit(changesets)
heads = OrderedDict(
(node, branch)
for node, _, branch in (d.partition(' ')
for d in commit.body.splitlines()))
if len(heads) != len(commit.parents):
status.report('The git-cinnabar metadata seems to be corrupted in '
'unexpected ways.\n')
return 1
manifest_nodes = []
parents = None
fix_changeset_heads = False
# TODO: Check that the recorded heads are actually dag heads.
for c, changeset_node in progress_iter(
'Checking {} changeset heads', izip(commit.parents, heads)):
gitsha1 = GitHgHelper.hg2git(changeset_node)
if gitsha1 == NULL_NODE_ID:
status.report('Missing hg2git metadata for changeset %s'
% changeset_node)
continue
if gitsha1 != c:
if parents is None:
parents = set(commit.parents)
if gitsha1 not in parents:
status.report(
'Inconsistent metadata:\n'
' Head metadata says changeset %s maps to %s\n'
' but hg2git metadata says it maps to %s'
% (changeset_node, c, gitsha1))
continue
fix_changeset_heads = True
changeset = store._changeset(c, include_parents=True)
if not changeset:
status.report('Missing git2hg metadata for git commit %s' % c)
continue
if changeset.node != changeset_node:
if changeset.node not in heads:
status.report(
'Inconsistent metadata:\n'
' Head metadata says %s maps to changeset %s\n'
' but git2hg metadata says it maps to changeset %s'
% (c, changeset_node, changeset.node))
continue
fix_changeset_heads = True
if changeset.node != changeset.sha1:
status.report('Sha1 mismatch for changeset %s' % changeset.node)
continue
changeset_branch = changeset.branch or 'default'
if heads[changeset.node] != changeset_branch:
status.report(
'Inconsistent metadata:\n'
' Head metadata says changeset %s is in branch %s\n'
' but git2hg metadata says it is in branch %s'
% (changeset.node, heads[changeset.node], changeset_branch))
continue
manifest_nodes.append(changeset.manifest)
if status('broken'):
return 1
# Rebuilding manifests benefits from limiting the difference with
# the last rebuilt manifest. Similarly, building the list of unique
# files in all manifests benefits from that too.
# Unfortunately, the manifest heads are not ordered in a topological
# relevant matter, and the differences between two consecutive manifests
# can be much larger than they could be. The consequence is spending a
# large amount of time rebuilding the manifests and gathering the files
# list. It's actually faster to attempt to reorder them according to
# some heuristics first, such that the differences are smaller.
# Here, we use the depth from the root node(s) to reorder the manifests.
# This doesn't give the most optimal ordering, but it's already much
# faster. On a clone of multiple mozilla-* repositories with > 1400 heads,
# it's close to an order of magnitude difference on the "Checking
# manifests" loop.
depths = {}
roots = []
manifest_queue = []
for m, _, parents in progress_iter(
'Loading {} manifests', GitHgHelper.rev_list(
'--topo-order', '--reverse', '--full-history', '%s^@'
% manifests)):
manifest_queue.append((m, parents))
if parents:
depth = {}
for p in parents:
for root, num in depths[p].iteritems():
if root in depth:
depth[root] = max(depth[root], num + 1)
else:
depth[root] = num + 1
depths[m] = depth
del depth
else:
depths[m] = {m: 0}
roots.append(m)
if status('broken'):
return 1
# TODO: check that all manifest_nodes gathered above are available in the
# manifests dag, and that the dag heads are the recorded heads.
manifests_commit = GitCommit(manifests)
depths = [
[depths[p].get(r, 0) for r in roots]
for p in manifests_commit.parents
]
manifests_commit_parents = [
p for _, p in sorted(zip(depths, manifests_commit.parents))
]
previous = None
all_interesting = set()
for m in progress_iter('Checking {} manifest heads',
manifests_commit_parents):
c = GitCommit(m)
if not SHA1_RE.match(c.body):
status.report('Invalid manifest metadata in git commit %s' % m)
continue
gitsha1 = GitHgHelper.hg2git(c.body)
if gitsha1 == NULL_NODE_ID:
status.report('Missing hg2git metadata for manifest %s' % c.body)
continue
if not GitHgHelper.check_manifest(c.body):
status.report('Sha1 mismatch for manifest %s' % c.body)
files = {}
if previous:
for _, _, before, after, d, path in GitHgHelper.diff_tree(
previous, m):
if d in 'AM' and before != after and \
(path, after) not in all_interesting:
files[path] = after
else:
for _, t, sha1, path in GitHgHelper.ls_tree(m, recursive=True):
if (path, sha1) not in all_interesting:
files[path] = sha1
all_interesting.update(files.iteritems())
previous = m
if status('broken'):
return 1
progress = Progress('Checking {} files')
while all_interesting and manifest_queue:
(m, parents) = manifest_queue.pop()
changes = get_changes(m, parents, all=True)
for path, hg_file, hg_fileparents in changes:
if hg_fileparents[1:] == (hg_file,):
continue
elif hg_fileparents[:1] == (hg_file,):
continue
# Reaching here means the file received a modification compared
# to its parents. If it's a file we're going to check below,
# it means we don't need to check its parents if somehow they were
# going to be checked. If it's not a file we're going to check
# below, it's because it's either a file we weren't interested in
# in the first place, or it's the parent of a file we have checked.
# Either way, we aren't interested in the parents.
for p in hg_fileparents:
all_interesting.discard((path, p))
if (path, hg_file) not in all_interesting:
continue
all_interesting.remove((path, hg_file))
if not GitHgHelper.check_file(hg_file, *hg_fileparents):
p = store.manifest_path(path)
status.report(
'Sha1 mismatch for file %s\n'
' revision %s' % (p, hg_file))
print_parents = ' '.join(p for p in hg_fileparents
if p != NULL_NODE_ID)
if print_parents:
status.report(' with parent%s %s' % (
's' if len(print_parents) > 41 else '',
print_parents))
progress.progress()
progress.finish()
if all_interesting:
status.info('Could not find the following files:')
for path, sha1 in sorted(all_interesting):
p = store.manifest_path(path)
status.info(' %s %s' % (sha1, path))
status.info(
'This might be a bug in `git cinnabar fsck`. Please open '
'an issue, with the message above, on\n'
'https://github.com/glandium/git-cinnabar/issues')
return 1
if status('broken'):
status.info(
'Your git-cinnabar repository appears to be corrupted.\n'
'Please open an issue, with the information above, on\n'
'https://github.com/glandium/git-cinnabar/issues')
Git.update_ref('refs/cinnabar/broken', metadata_commit)
if Git.resolve_ref('refs/cinnabar/checked'):
status.info(
'\nThen please try to run `git cinnabar rollback --fsck` to '
'restore last known state, and to update from the mercurial '
'repository.')
else:
status.info('\nThen please try to run `git cinnabar reclone`.')
status.info(
'\nPlease note this may affect the commit sha1s of mercurial '
'changesets, and may require to rebase your local branches.')
status.info(
'\nAlternatively, you may start afresh with a new clone. In any '
'case, please keep this corrupted repository around for further '
'debugging.')
return 1
refresh = []
if fix_changeset_heads:
status.fix('Fixing changeset heads metadata order.')
refresh.append('refs/cinnabar/changesets')
interval_expired('fsck', 0)
store.close(refresh=refresh)
GitHgHelper._helper = False
metadata_commit = Git.resolve_ref('refs/cinnabar/metadata')
Git.update_ref('refs/cinnabar/checked', metadata_commit)
return 0
def fsck(args):
'''check cinnabar metadata consistency'''
if not args.commit and not args.full:
return fsck_quick()
status = FsckStatus()
store = GitHgStore()
if args.full and args.commit:
logging.error('Cannot pass both --full and a commit')
return 1
if args.commit:
commits = set()
all_git_commits = {}
for c in args.commit:
cs = store.hg_changeset(c)
if cs:
commits.add(c)
c = cs.node
commit = GitHgHelper.hg2git(c)
if commit == NULL_NODE_ID and not cs:
status.info('Unknown commit or changeset: %s' % c)
return 1
if not cs:
cs = store.hg_changeset(commit)
commits.add(commit)
all_git_commits = GitHgHelper.rev_list('--no-walk=unsorted', *commits)
else:
all_refs = dict((ref, sha1)
for sha1, ref in Git.for_each_ref('refs/cinnabar'))
if 'refs/cinnabar/metadata' in all_refs:
git_heads = '%s^^@' % all_refs['refs/cinnabar/metadata']
else:
assert False
all_git_commits = GitHgHelper.rev_list(
'--topo-order', '--full-history', '--reverse', git_heads)
dag = gitdag()
GitHgHelper.reset_heads('manifests')
full_file_check = FileFindParents.logger.isEnabledFor(logging.DEBUG)
for node, tree, parents in progress_iter('Checking {} changesets',
all_git_commits):
node = store._replace.get(node, node)
hg_node = store.hg_changeset(node)
if not hg_node:
status.report('Missing note for git commit: ' + node)
continue
GitHgHelper.seen('git2hg', node)
changeset_data = store.changeset(hg_node)
changeset = changeset_data.node
GitHgHelper.seen('hg2git', changeset)
changeset_ref = store.changeset_ref(changeset)
if not changeset_ref:
status.report('Missing changeset in hg2git branch: %s' % changeset)
continue
elif str(changeset_ref) != node:
status.report('Commit mismatch for changeset %s\n'
' hg2git: %s\n commit: %s'
% (changeset, changeset_ref, node))
hg_changeset = store.changeset(changeset, include_parents=True)
if hg_changeset.node != hg_changeset.sha1:
status.report('Sha1 mismatch for changeset %s' % changeset)
dag.add(hg_changeset.node,
(hg_changeset.parent1, hg_changeset.parent2),
changeset_data.branch or 'default')
raw_changeset = Changeset.from_git_commit(node)
patcher = ChangesetPatcher.from_diff(raw_changeset, changeset_data)
if patcher != store.read_changeset_data(node):
status.fix('Adjusted changeset metadata for %s' % changeset)
GitHgHelper.set('changeset', changeset, NULL_NODE_ID)
GitHgHelper.set('changeset', changeset, node)
GitHgHelper.put_blob(patcher, want_sha1=False)
GitHgHelper.set('changeset-metadata', changeset, NULL_NODE_ID)
GitHgHelper.set('changeset-metadata', changeset, ':1')
manifest = changeset_data.manifest
if GitHgHelper.seen('hg2git', manifest) or manifest == NULL_NODE_ID:
continue
manifest_ref = store.manifest_ref(manifest)
if not manifest_ref:
status.report('Missing manifest in hg2git branch: %s' % manifest)
parents = tuple(
store.changeset(p).manifest
for p in hg_changeset.parents
)
git_parents = tuple(store.manifest_ref(p) for p in parents
if p != NULL_NODE_ID)
# This doesn't change the value but makes the helper track the manifest
# dag.
GitHgHelper.set('manifest', manifest, manifest_ref)
if not GitHgHelper.check_manifest(manifest):
status.report('Sha1 mismatch for manifest %s' % manifest)
manifest_commit_parents = GitCommit(manifest_ref).parents
if sorted(manifest_commit_parents) != sorted(git_parents):
# TODO: better error
status.report('%s(%s) %s != %s' % (manifest, manifest_ref,
manifest_commit_parents,
git_parents))
# TODO: check that manifest content matches changeset content
changes = get_changes(manifest_ref, git_parents)
for path, hg_file, hg_fileparents in changes:
if hg_file != NULL_NODE_ID and (hg_file == HG_EMPTY_FILE or
GitHgHelper.seen('hg2git',
hg_file)):
if full_file_check:
file = store.file(hg_file, hg_fileparents, git_parents,
store.manifest_path(path))
valid = file.node == file.sha1
else:
valid = GitHgHelper.check_file(hg_file,
*hg_fileparents)
if not valid:
status.report(
'Sha1 mismatch for file %s in manifest %s'
% (hg_file, manifest_ref))
if not args.commit and not status('broken'):
store_manifest_heads = set(store._manifest_heads_orig)
manifest_heads = set(GitHgHelper.heads('manifests'))
if store_manifest_heads != manifest_heads:
def iter_manifests(a, b):
for h in a - b:
yield h
for h in b:
yield '^%s' % h
for m, t, p in GitHgHelper.rev_list(
'--topo-order', '--full-history', '--reverse',
*iter_manifests(manifest_heads, store_manifest_heads)):
status.fix('Missing manifest commit in manifest branch: %s'
% m)
for m, t, p in GitHgHelper.rev_list(
'--topo-order', '--full-history', '--reverse',
*iter_manifests(store_manifest_heads, manifest_heads)):
status.fix('Removing metadata commit %s with no corresponding '
'changeset' % (m))
for h in store_manifest_heads - manifest_heads:
if GitHgHelper.seen('hg2git', store.hg_manifest(h)):
status.fix('Removing non-head reference to %s in manifests'
' metadata.' % h)
dangling = ()
if not args.commit and not status('broken'):
dangling = GitHgHelper.dangling('hg2git')
for obj in dangling:
status.fix('Removing dangling metadata for ' + obj)
# Theoretically, we should figure out if they are files, manifests
# or changesets and set the right variable accordingly, but in
# practice, it makes no difference. Reevaluate when GitHgStore.close
# is modified, though.
GitHgHelper.set('file', obj, NULL_NODE_ID)
GitHgHelper.set('file-meta', obj, NULL_NODE_ID)
if not args.commit and not status('broken'):
dangling = GitHgHelper.dangling('git2hg')
for c in dangling:
status.fix('Removing dangling note for commit ' + c)
GitHgHelper.set('changeset-metadata', c, NULL_NODE_ID)
if status('broken'):
status.info(
'Your git-cinnabar repository appears to be corrupted. There\n'
'are known issues in older revisions that have been fixed.\n'
'Please try running the following command to reset:\n'
' git cinnabar reclone\n\n'
'Please note this command may change the commit sha1s. Your\n'
'local branches will however stay untouched.\n'
'Please report any corruption that fsck would detect after a\n'
'reclone.')
if not args.commit:
status.info('Checking head references...')
computed_heads = defaultdict(set)
for branch, head in dag.all_heads():
computed_heads[branch].add(head)
for branch in sorted(dag.tags()):
stored_heads = store.heads({branch})
for head in computed_heads[branch] - stored_heads:
status.fix('Adding missing head %s in branch %s' %
(head, branch))
store.add_head(head)
for head in stored_heads - computed_heads[branch]:
status.fix('Removing non-head reference to %s in branch %s' %
(head, branch))
del store._hgheads[head]
metadata_commit = Git.resolve_ref('refs/cinnabar/metadata')
if status('broken'):
Git.update_ref('refs/cinnabar/broken', metadata_commit)
return 1
if args.full:
Git.update_ref('refs/cinnabar/checked', metadata_commit)
interval_expired('fsck', 0)
store.close()
if status('fixed'):
return 2
return 0
def create_hg_manifest(self, commit, parents):
manifest = GeneratedManifestInfo(NULL_NODE_ID)
changeset_files = []
if parents:
parent_changeset = self.changeset(self.hg_changeset(parents[0]))
parent_manifest = self.manifest(parent_changeset.manifest)
parent_node = parent_manifest.node
if len(parents) == 2:
parent2_changeset = self.changeset(self.hg_changeset(parents[1]))
parent2_manifest = self.manifest(parent2_changeset.manifest)
parent2_node = parent2_manifest.node
if parent_node == parent2_node:
parents = parents[:1]
if not parents:
for line in Git.ls_tree(commit, recursive=True):
mode, typ, sha1, path = line
node = self.create_file(sha1)
manifest.add(path, node, self.ATTR[mode], modified=True)
changeset_files.append(path)
manifest.parents = []
manifest.delta_node = NULL_NODE_ID
return manifest, changeset_files
elif len(parents) == 2:
if not experiment('merge'):
raise Exception('Pushing merges is not supported yet')
if not self._merge_warn:
logging.warning('Pushing merges is experimental.')
logging.warning('This may irremediably push bad state to the '
'mercurial server!')
self._merge_warn = 1
git_manifests = (self.manifest_ref(parent_node),
self.manifest_ref(parent2_node))
# TODO: this would benefit from less git queries
file_dags = {}
for m, tree, mparents in GitHgHelper.rev_list(
b'--parents', b'--topo-order',
b'--full-history', b'--reverse',
b'%s...%s' % git_manifests):
for p in mparents:
for path, sha1_after, sha1_before in manifest_diff(p, m):
path = GitHgStore.manifest_path(path)
if path not in file_dags:
file_dags[path] = gitdag()
dag = file_dags[path]
if sha1_before == NULL_NODE_ID:
dag.add(sha1_after, ())
else:
dag.add(sha1_after, (sha1_before,))
files = [(p, mode, sha1) for mode, _, sha1, p in
Git.ls_tree(commit, recursive=True)]
manifests = sorted_merge(parent_manifest, parent2_manifest,
key=lambda i: i.path, non_key=lambda i: i)
for line in sorted_merge(files, manifests):
path, f, m = line
if not m:
m = (None, None)
manifest_line_p1, manifest_line_p2 = m
if not f: # File was removed
if manifest_line_p1:
manifest.removed.add(path)
changeset_files.append(path)
continue
mode, sha1 = f
attr = self.ATTR[mode]
if manifest_line_p1 and not manifest_line_p2:
file_parents = (manifest_line_p1.sha1,)
elif manifest_line_p2 and not manifest_line_p1:
file_parents = (manifest_line_p2.sha1,)
elif not manifest_line_p1 and not manifest_line_p2:
file_parents = ()
changeset_files.append(path)
elif manifest_line_p1.sha1 == manifest_line_p2.sha1:
file_parents = (manifest_line_p1.sha1,)
else:
if self._merge_warn == 1:
logging.warning('This may take a while...')
self._merge_warn = 2
file_parents = ()
dag = file_dags.pop(path)
if dag:
dag.tag_nodes_and_parents(
(manifest_line_p1.sha1,), 'a')
if dag._tags.get(manifest_line_p2.sha1) == 'a':
file_parents = (manifest_line_p1.sha1,)
else:
dag._tags.clear()
dag.tag_nodes_and_parents(
(manifest_line_p2.sha1,), 'b')
if dag._tags.get(manifest_line_p1.sha1) == 'b':
file_parents = (manifest_line_p2.sha1,)
if not file_parents:
file_parents = (manifest_line_p1.sha1,
manifest_line_p2.sha1)
assert file_parents is not None
f = self._create_file_internal(sha1, *file_parents)
file_parents = tuple(p for p in (f.parent1, f.parent2)
if p != NULL_NODE_ID)
merged = len(file_parents) == 2
if not merged and file_parents:
if self.git_file_ref(file_parents[0]) == sha1:
node = file_parents[0]
else:
merged = True
if merged or not file_parents:
node = self._store_file_internal(f)
elif file_parents:
node = file_parents[0]
attr_change = (manifest_line_p1 and
manifest_line_p1.attr != attr)
manifest.add(path, node, attr, modified=merged or attr_change)
if merged or attr_change:
changeset_files.append(path)
if manifest.raw_data == parent_manifest.raw_data:
return parent_manifest, []
manifest.parents = (parent_node, parent2_node)
return manifest, changeset_files
def process_diff(diff):
for (mode_before, mode_after, sha1_before, sha1_after, status,
path) in diff:
if status[:1] == b'R':
yield status[1:], (
b'000000', sha1_before, NULL_NODE_ID, b'D')
yield path, (mode_after, sha1_before, sha1_after,
status)
git_diff = sorted(
l for l in process_diff(GitHgHelper.diff_tree(
parents[0], commit, detect_copy=True))
)
if not git_diff:
return parent_manifest, []
parent_lines = OrderedDict((l.path, l) for l in parent_manifest)
items = manifest.items
for line in sorted_merge(iteritems(parent_lines), git_diff,
non_key=lambda i: i[1]):
path, manifest_line, change = line
if not change:
items.append(manifest_line)
continue
mode_after, sha1_before, sha1_after, status = change
path2 = status[1:]
status = status[:1]
attr = self.ATTR.get(mode_after)
if status == b'D':
manifest.removed.add(path)
changeset_files.append(path)
continue
if status in b'MT':
if sha1_before == sha1_after:
node = manifest_line.sha1
else:
node = self.create_file(sha1_after, manifest_line.sha1)
elif status in b'RC':
if sha1_after != EMPTY_BLOB:
node = self.create_copy(
(path2, parent_lines[path2].sha1), sha1_after,
path=path)
else:
node = self.create_file(sha1_after)
else:
assert status == b'A'
node = self.create_file(sha1_after)
manifest.add(path, node, attr, modified=True)
changeset_files.append(path)
manifest.parents = (parent_node,)
manifest.delta_node = parent_node
return manifest, changeset_files
def push(repo, store, what, repo_heads, repo_branches, dry_run=False):
def heads():
for sha1 in store.heads(repo_branches):
yield '^%s' % store.changeset_ref(sha1)
def local_bases():
h = chain(heads(), (w for w in what if w))
for c, t, p in GitHgHelper.rev_list('--topo-order', '--full-history',
'--boundary', *h):
if c[0] != '-':
continue
yield store.hg_changeset(c[1:])
for w in what:
rev = store.hg_changeset(w)
if rev:
yield rev
common = findcommon(repo, store, set(local_bases()))
logging.info('common: %s', common)
def revs():
for sha1 in common:
yield '^%s' % store.changeset_ref(sha1)
revs = chain(revs(), (w for w in what if w))
push_commits = list((c, p) for c, t, p in GitHgHelper.rev_list(
'--topo-order', '--full-history', '--parents', '--reverse', *revs))
pushed = False
if push_commits:
has_root = any(len(p) == 40 for p in push_commits)
force = all(v[1] for v in what.values())
if has_root and repo_heads:
if not force:
raise Exception('Cannot push a new root')
else:
logging.warn('Pushing a new root')
if force:
repo_heads = ['force']
else:
if not repo_heads:
repo_heads = [NULL_NODE_ID]
repo_heads = [unhexlify(h) for h in repo_heads]
if push_commits and not dry_run:
if repo.local():
repo.local().ui.setconfig('server', 'validate', True)
if unbundle20:
b2caps = repo.capable('bundle2') or {}
else:
b2caps = {}
if b2caps:
b2caps = decodecaps(urllib.unquote(b2caps))
logging.getLogger('bundle2').debug('%r', b2caps)
if b2caps:
b2caps['replycaps'] = encodecaps({'error': ['abort']})
cg = create_bundle(store, push_commits, b2caps)
if not isinstance(repo, HelperRepo):
cg = chunkbuffer(cg)
if not b2caps:
cg = cg1unpacker(cg, 'UN')
reply = repo.unbundle(cg, repo_heads, '')
if unbundle20 and isinstance(reply, unbundle20):
parts = iter(reply.iterparts())
for part in parts:
logging.getLogger('bundle2').debug('part: %s', part.type)
logging.getLogger('bundle2').debug('params: %r', part.params)
if part.type == 'output':
sys.stderr.write(part.read())
elif part.type == 'reply:changegroup':
# TODO: should check params['in-reply-to']
reply = int(part.params['return'])
elif part.type == 'error:abort':
raise error.Abort(part.params['message'],
hint=part.params.get('hint'))
else:
logging.getLogger('bundle2').warning(
'ignoring bundle2 part: %s', part.type)
pushed = reply != 0
return gitdag(push_commits) if pushed or dry_run else ()
def findcommon(repo, store, hgheads):
logger = logging.getLogger('findcommon')
logger.debug(hgheads)
if not hgheads:
logger.info('no requests')
return set()
sample_size = 100
sample = _sample(hgheads, sample_size)
requests = 1
known = repo.known(unhexlify(h) for h in sample)
known = set(h for h, k in izip(sample, known) if k)
logger.debug('initial sample size: %d', len(sample))
if len(known) == len(hgheads):
logger.debug('all heads known')
logger.info('1 request')
return hgheads
git_heads = set(store.changeset_ref(h) for h in hgheads)
git_known = set(store.changeset_ref(h) for h in known)
if logger.isEnabledFor(logging.DEBUG):
logger.debug('known (sub)set: (%d) %s', len(known), sorted(git_known))
args = ['--topo-order', '--full-history', '--parents']
def revs():
for h in git_known:
yield '^%s' % h
for h in git_heads:
if h not in git_known:
yield h
args.extend(revs())
revs = ((c, parents) for c, t, parents in GitHgHelper.rev_list(*args))
dag = gitdag(chain(revs, ((k, ()) for k in git_known)))
dag.tag_nodes_and_parents(git_known, 'known')
def log_dag(tag):
if not logger.isEnabledFor(logging.DEBUG):
return
logger.debug('%s dag size: %d', tag,
sum(1 for n in dag.iternodes(tag)))
heads = sorted(dag.heads(tag))
logger.debug('%s dag heads: (%d) %s', tag, len(heads), heads)
roots = sorted(dag.roots(tag))
logger.debug('%s dag roots: (%d) %s', tag, len(roots), roots)
log_dag('unknown')
log_dag('known')
while True:
unknown = set(chain(dag.heads(), dag.roots()))
if not unknown:
break
sample = set(_sample(unknown, sample_size))
if len(sample) < sample_size:
sample |= set(_sample(set(dag.iternodes()),
sample_size - len(sample)))
sample = list(sample)
hg_sample = [store.hg_changeset(h) for h in sample]
requests += 1
known = repo.known(unhexlify(h) for h in hg_sample)
unknown = set(h for h, k in izip(sample, known) if not k)
known = set(h for h, k in izip(sample, known) if k)
logger.debug('next sample size: %d', len(sample))
if logger.isEnabledFor(logging.DEBUG):
logger.debug('known (sub)set: (%d) %s', len(known), sorted(known))
logger.debug('unknown (sub)set: (%d) %s', len(unknown),
sorted(unknown))
dag.tag_nodes_and_parents(known, 'known')
dag.tag_nodes_and_children(unknown, 'unknown')
log_dag('unknown')
log_dag('known')
logger.info('%d requests', requests)
return [store.hg_changeset(h) for h in dag.heads('known')]