def __init__(self, reader, length=None):
self._reader = reader
self._length = length
self._read = 0
self._pos = 0
self._buf = ''
self._progress = Progress(' {}%' if self._length else ' {} bytes')
class ReaderProgress(object):
def __init__(self, reader, length=None):
self._reader = reader
self._length = length
self._read = 0
self._pos = 0
self._buf = ''
self._progress = Progress(' {}%' if self._length else ' {} bytes')
def read(self, length):
# See comment above tell
if self._pos < self._read:
assert self._read - self._pos <= 8
assert length <= len(self._buf)
data = self._buf[:length]
self._buf = self._buf[length:]
self._pos += length
else:
assert self._read == self._pos
data = self._reader.read(length)
self._read += len(data)
self._pos = self._read
# Keep the last 8 bytes we read for GzipFile
self._buf = data[-8:]
self.progress()
return data
def progress(self):
if self._length:
count = self._read * 100 / self._length
else:
count = self._read
self._progress.progress(count)
def finish(self):
self._progress.finish()
# GzipFile wants to seek to the end of the file and back, so we add
# enough tell/seek support to make it happy. It also rewinds 8 bytes
# for the CRC, so we also handle that.
def tell(self):
return self._pos
def seek(self, pos, how=os.SEEK_SET):
if how == os.SEEK_END:
self._pos = self._length + pos
elif how == os.SEEK_SET:
self._pos = pos
elif how == os.SEEK_CUR:
self._pos += pos
else:
raise NotImplementedError()
return self._pos
class ReaderProgress(object):
def __init__(self, reader, length=None):
self._reader = reader
self._length = length
self._read = 0
self._pos = 0
self._buf = ''
self._progress = Progress(' {}%' if self._length else ' {} bytes')
def read(self, length):
# See comment above tell
if self._pos < self._read:
assert self._read - self._pos <= 8
assert length <= len(self._buf)
data = self._buf[:length]
self._buf = self._buf[length:]
self._pos += length
else:
assert self._read == self._pos
data = self._reader.read(length)
self._read += len(data)
self._pos = self._read
# Keep the last 8 bytes we read for GzipFile
self._buf = data[-8:]
self.progress()
return data
def progress(self):
if self._length:
count = self._read * 100 // self._length
else:
count = self._read
self._progress.progress(count)
def finish(self):
self._progress.finish()
# GzipFile wants to seek to the end of the file and back, so we add
# enough tell/seek support to make it happy. It also rewinds 8 bytes
# for the CRC, so we also handle that.
def tell(self):
return self._pos
def seek(self, pos, how=os.SEEK_SET):
if how == os.SEEK_END:
self._pos = self._length + pos
elif how == os.SEEK_SET:
self._pos = pos
elif how == os.SEEK_CUR:
self._pos += pos
else:
raise NotImplementedError()
return self._pos
def __init__(self, reader, length=None):
self._reader = reader
self._length = length
self._read = 0
self._pos = 0
self._buf = ''
self._progress = Progress(' {}%' if self._length else ' {} bytes')
class ReaderProgress(object):
def __init__(self, reader, length=None):
self._reader = reader
self._length = length
self._read = 0
self._progress = Progress(' {}%' if self._length else ' {} bytes')
def read(self, length):
data = self._reader.read(length)
self._read += len(data)
if self._length:
count = self._read * 100 // self._length
else:
count = self._read
self._progress.progress(count)
return data
def finish(self):
self._progress.finish()
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_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
