mirror of
https://github.com/facebook/sapling.git
synced 2024-10-07 15:27:13 +03:00
100b68ca2e
Marker are now written as soon as possible but within a transaction. Using a transaction ensure a proper behavior on error and rollback compatibility. Flush logic are not necessary anymore and are dropped from lock release. With this changeset, the obsstore is open, written and closed for every single added marker. This is expected to be highly inefficient and batched write should be implemented "quickly". Another issue is that every flush of the file will invalidate the obsstore filecache and trigger a full re instantiation of the repo.obsstore attribute (including, reading and parsing entry). This is also expected to be highly inefficient and proper filecache operation should be implemented "quickly" too. A side benefit of the filecache issue is that repo.obsstore object is properly invalidated on transaction abortion.
282 lines
8.7 KiB
Python
282 lines
8.7 KiB
Python
# obsolete.py - obsolete markers handling
|
|
#
|
|
# Copyright 2012 Pierre-Yves David <pierre-yves.david@ens-lyon.org>
|
|
# Logilab SA <contact@logilab.fr>
|
|
#
|
|
# This software may be used and distributed according to the terms of the
|
|
# GNU General Public License version 2 or any later version.
|
|
|
|
"""Obsolete markers handling
|
|
|
|
An obsolete marker maps an old changeset to a list of new
|
|
changesets. If the list of new changesets is empty, the old changeset
|
|
is said to be "killed". Otherwise, the old changeset is being
|
|
"replaced" by the new changesets.
|
|
|
|
Obsolete markers can be used to record and distribute changeset graph
|
|
transformations performed by history rewriting operations, and help
|
|
building new tools to reconciliate conflicting rewriting actions. To
|
|
facilitate conflicts resolution, markers include various annotations
|
|
besides old and news changeset identifiers, such as creation date or
|
|
author name.
|
|
|
|
|
|
Format
|
|
------
|
|
|
|
Markers are stored in an append-only file stored in
|
|
'.hg/store/obsstore'.
|
|
|
|
The file starts with a version header:
|
|
|
|
- 1 unsigned byte: version number, starting at zero.
|
|
|
|
|
|
The header is followed by the markers. Each marker is made of:
|
|
|
|
- 1 unsigned byte: number of new changesets "R", could be zero.
|
|
|
|
- 1 unsigned 32-bits integer: metadata size "M" in bytes.
|
|
|
|
- 1 byte: a bit field. It is reserved for flags used in obsolete
|
|
markers common operations, to avoid repeated decoding of metadata
|
|
entries.
|
|
|
|
- 20 bytes: obsoleted changeset identifier.
|
|
|
|
- N*20 bytes: new changesets identifiers.
|
|
|
|
- M bytes: metadata as a sequence of nul-terminated strings. Each
|
|
string contains a key and a value, separated by a color ':', without
|
|
additional encoding. Keys cannot contain '\0' or ':' and values
|
|
cannot contain '\0'.
|
|
"""
|
|
import struct
|
|
from mercurial import util, base85
|
|
from i18n import _
|
|
|
|
_pack = struct.pack
|
|
_unpack = struct.unpack
|
|
|
|
|
|
|
|
# data used for parsing and writing
|
|
_fmversion = 0
|
|
_fmfixed = '>BIB20s'
|
|
_fmnode = '20s'
|
|
_fmfsize = struct.calcsize(_fmfixed)
|
|
_fnodesize = struct.calcsize(_fmnode)
|
|
|
|
def _readmarkers(data):
|
|
"""Read and enumerate markers from raw data"""
|
|
off = 0
|
|
diskversion = _unpack('>B', data[off:off + 1])[0]
|
|
off += 1
|
|
if diskversion != _fmversion:
|
|
raise util.Abort(_('parsing obsolete marker: unknown version %r')
|
|
% diskversion)
|
|
|
|
# Loop on markers
|
|
l = len(data)
|
|
while off + _fmfsize <= l:
|
|
# read fixed part
|
|
cur = data[off:off + _fmfsize]
|
|
off += _fmfsize
|
|
nbsuc, mdsize, flags, pre = _unpack(_fmfixed, cur)
|
|
# read replacement
|
|
sucs = ()
|
|
if nbsuc:
|
|
s = (_fnodesize * nbsuc)
|
|
cur = data[off:off + s]
|
|
sucs = _unpack(_fmnode * nbsuc, cur)
|
|
off += s
|
|
# read metadata
|
|
# (metadata will be decoded on demand)
|
|
metadata = data[off:off + mdsize]
|
|
if len(metadata) != mdsize:
|
|
raise util.Abort(_('parsing obsolete marker: metadata is too '
|
|
'short, %d bytes expected, got %d')
|
|
% (len(metadata), mdsize))
|
|
off += mdsize
|
|
yield (pre, sucs, flags, metadata)
|
|
|
|
def encodemeta(meta):
|
|
"""Return encoded metadata string to string mapping.
|
|
|
|
Assume no ':' in key and no '\0' in both key and value."""
|
|
for key, value in meta.iteritems():
|
|
if ':' in key or '\0' in key:
|
|
raise ValueError("':' and '\0' are forbidden in metadata key'")
|
|
if '\0' in value:
|
|
raise ValueError("':' are forbidden in metadata value'")
|
|
return '\0'.join(['%s:%s' % (k, meta[k]) for k in sorted(meta)])
|
|
|
|
def decodemeta(data):
|
|
"""Return string to string dictionary from encoded version."""
|
|
d = {}
|
|
for l in data.split('\0'):
|
|
if l:
|
|
key, value = l.split(':')
|
|
d[key] = value
|
|
return d
|
|
|
|
class marker(object):
|
|
"""Wrap obsolete marker raw data"""
|
|
|
|
def __init__(self, repo, data):
|
|
# the repo argument will be used to create changectx in later version
|
|
self._repo = repo
|
|
self._data = data
|
|
self._decodedmeta = None
|
|
|
|
def precnode(self):
|
|
"""Precursor changeset node identifier"""
|
|
return self._data[0]
|
|
|
|
def succnodes(self):
|
|
"""List of successor changesets node identifiers"""
|
|
return self._data[1]
|
|
|
|
def metadata(self):
|
|
"""Decoded metadata dictionary"""
|
|
if self._decodedmeta is None:
|
|
self._decodedmeta = decodemeta(self._data[3])
|
|
return self._decodedmeta
|
|
|
|
def date(self):
|
|
"""Creation date as (unixtime, offset)"""
|
|
parts = self.metadata()['date'].split(' ')
|
|
return (float(parts[0]), int(parts[1]))
|
|
|
|
class obsstore(object):
|
|
"""Store obsolete markers
|
|
|
|
Markers can be accessed with two mappings:
|
|
- precursors: old -> set(new)
|
|
- successors: new -> set(old)
|
|
"""
|
|
|
|
def __init__(self, sopener):
|
|
self._all = []
|
|
# new markers to serialize
|
|
self.precursors = {}
|
|
self.successors = {}
|
|
self.sopener = sopener
|
|
data = sopener.tryread('obsstore')
|
|
if data:
|
|
for marker in _readmarkers(data):
|
|
self._load(marker)
|
|
|
|
def __iter__(self):
|
|
return iter(self._all)
|
|
|
|
def __nonzero__(self):
|
|
return bool(self._all)
|
|
|
|
def create(self, transaction, prec, succs=(), flag=0, metadata=None):
|
|
"""obsolete: add a new obsolete marker
|
|
|
|
* ensuring it is hashable
|
|
* check mandatory metadata
|
|
* encode metadata
|
|
"""
|
|
if metadata is None:
|
|
metadata = {}
|
|
if len(prec) != 20:
|
|
raise ValueError(prec)
|
|
for succ in succs:
|
|
if len(succ) != 20:
|
|
raise ValueError(succ)
|
|
marker = (str(prec), tuple(succs), int(flag), encodemeta(metadata))
|
|
self.add(transaction, marker)
|
|
|
|
def add(self, transaction, marker):
|
|
"""Add a new marker to the store"""
|
|
if marker not in self._all:
|
|
f = self.sopener('obsstore', 'ab')
|
|
try:
|
|
offset = f.tell()
|
|
transaction.add('obsstore', offset)
|
|
if offset == 0:
|
|
# new file add version header
|
|
f.write(_pack('>B', _fmversion))
|
|
_writemarkers(f.write, [marker])
|
|
finally:
|
|
# XXX: f.close() == filecache invalidation == obsstore rebuilt.
|
|
# call 'filecacheentry.refresh()' here
|
|
f.close()
|
|
self._load(marker)
|
|
|
|
def mergemarkers(self, transation, data):
|
|
other = _readmarkers(data)
|
|
local = set(self._all)
|
|
new = [m for m in other if m not in local]
|
|
for marker in new:
|
|
# XXX: N marker == N x (open, write, close)
|
|
# we should write them all at once
|
|
self.add(transation, marker)
|
|
|
|
def _load(self, marker):
|
|
self._all.append(marker)
|
|
pre, sucs = marker[:2]
|
|
self.precursors.setdefault(pre, set()).add(marker)
|
|
for suc in sucs:
|
|
self.successors.setdefault(suc, set()).add(marker)
|
|
|
|
def _writemarkers(write, markers):
|
|
# Kept separate from flushmarkers(), it will be reused for
|
|
# markers exchange.
|
|
for marker in markers:
|
|
pre, sucs, flags, metadata = marker
|
|
nbsuc = len(sucs)
|
|
format = _fmfixed + (_fmnode * nbsuc)
|
|
data = [nbsuc, len(metadata), flags, pre]
|
|
data.extend(sucs)
|
|
write(_pack(format, *data))
|
|
write(metadata)
|
|
|
|
def listmarkers(repo):
|
|
"""List markers over pushkey"""
|
|
if not repo.obsstore:
|
|
return {}
|
|
data = [_pack('>B', _fmversion)]
|
|
_writemarkers(data.append, repo.obsstore)
|
|
return {'dump': base85.b85encode(''.join(data))}
|
|
|
|
def pushmarker(repo, key, old, new):
|
|
"""Push markers over pushkey"""
|
|
if key != 'dump':
|
|
repo.ui.warn(_('unknown key: %r') % key)
|
|
return 0
|
|
if old:
|
|
repo.ui.warn(_('unexpected old value') % key)
|
|
return 0
|
|
data = base85.b85decode(new)
|
|
lock = repo.lock()
|
|
try:
|
|
tr = repo.transaction('pushkey: obsolete markers')
|
|
try:
|
|
repo.obsstore.mergemarkers(tr, data)
|
|
tr.close()
|
|
return 1
|
|
finally:
|
|
tr.release()
|
|
finally:
|
|
lock.release()
|
|
|
|
def allmarkers(repo):
|
|
"""all obsolete markers known in a repository"""
|
|
for markerdata in repo.obsstore:
|
|
yield marker(repo, markerdata)
|
|
|
|
def precursormarkers(ctx):
|
|
"""obsolete marker making this changeset obsolete"""
|
|
for data in ctx._repo.obsstore.precursors.get(ctx.node(), ()):
|
|
yield marker(ctx._repo, data)
|
|
|
|
def successormarkers(ctx):
|
|
"""obsolete marker marking this changeset as a successors"""
|
|
for data in ctx._repo.obsstore.successors.get(ctx.node(), ()):
|
|
yield marker(ctx._repo, data)
|
|
|