##// END OF EJS Templates
bdiff: replace hash algorithm...
bdiff: replace hash algorithm This patch replaces lyhash with the hash algorithm used by diffutils. The algorithm has its origins in Git commit 2e9d1410, which is all the way back from 1992. The license header in the code at that revision in GPL v2. I have not performed an extensive analysis of the distribution (and therefore buckets) of hash output. However, `hg perfbdiff` gives some clear wins. I'd like to think that if it is good enough for diffutils it is good enough for us? From the mozilla-unified repository: $ perfbdiff -m 3041e4d59df2 ! wall 0.053271 comb 0.060000 user 0.060000 sys 0.000000 (best of 100) ! wall 0.035827 comb 0.040000 user 0.040000 sys 0.000000 (best of 100) $ perfbdiff 0e9928989e9c --alldata --count 100 ! wall 6.204277 comb 6.200000 user 6.200000 sys 0.000000 (best of 3) ! wall 4.309710 comb 4.300000 user 4.300000 sys 0.000000 (best of 3) From the hg repo: $ perfbdiff 35000 --alldata --count 1000 ! wall 0.660358 comb 0.660000 user 0.660000 sys 0.000000 (best of 15) ! wall 0.534092 comb 0.530000 user 0.530000 sys 0.000000 (best of 19) Looking at the generated assembly and statistical profiler output from the kernel level, I believe there is room to make this function even faster. Namely, we're still consuming data character by character instead of at the word level. This translates to more loop iterations and more instructions. At this juncture though, the real performance killer is that we're hashing every line. We should get a significant speedup if we change the algorithm to find the longest prefix, longest suffix, treat those as single "lines" and then only do the line splitting and hashing on the parts that are different. That will require a lot of C code, however. I'm optimistic this approach could result in a ~2x speedup.

File last commit:

r28368:b9296b33 default
r30318:e1d6aa0e default
Show More
darcs.py
221 lines | 8.2 KiB | text/x-python | PythonLexer
# darcs.py - darcs support for the convert extension
#
# Copyright 2007-2009 Matt Mackall <mpm@selenic.com> and others
#
# 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 errno
import os
import re
import shutil
import tempfile
from mercurial.i18n import _
from mercurial import (
error,
util,
)
from . import common
NoRepo = common.NoRepo
# The naming drift of ElementTree is fun!
try:
import xml.etree.cElementTree.ElementTree as ElementTree
import xml.etree.cElementTree.XMLParser as XMLParser
except ImportError:
try:
import xml.etree.ElementTree.ElementTree as ElementTree
import xml.etree.ElementTree.XMLParser as XMLParser
except ImportError:
try:
import elementtree.cElementTree.ElementTree as ElementTree
import elementtree.cElementTree.XMLParser as XMLParser
except ImportError:
try:
import elementtree.ElementTree.ElementTree as ElementTree
import elementtree.ElementTree.XMLParser as XMLParser
except ImportError:
pass
class darcs_source(common.converter_source, common.commandline):
def __init__(self, ui, path, revs=None):
common.converter_source.__init__(self, ui, path, revs=revs)
common.commandline.__init__(self, ui, 'darcs')
# check for _darcs, ElementTree so that we can easily skip
# test-convert-darcs if ElementTree is not around
if not os.path.exists(os.path.join(path, '_darcs')):
raise NoRepo(_("%s does not look like a darcs repository") % path)
common.checktool('darcs')
version = self.run0('--version').splitlines()[0].strip()
if version < '2.1':
raise error.Abort(_('darcs version 2.1 or newer needed (found %r)')
% version)
if "ElementTree" not in globals():
raise error.Abort(_("Python ElementTree module is not available"))
self.path = os.path.realpath(path)
self.lastrev = None
self.changes = {}
self.parents = {}
self.tags = {}
# Check darcs repository format
format = self.format()
if format:
if format in ('darcs-1.0', 'hashed'):
raise NoRepo(_("%s repository format is unsupported, "
"please upgrade") % format)
else:
self.ui.warn(_('failed to detect repository format!'))
def before(self):
self.tmppath = tempfile.mkdtemp(
prefix='convert-' + os.path.basename(self.path) + '-')
output, status = self.run('init', repodir=self.tmppath)
self.checkexit(status)
tree = self.xml('changes', xml_output=True, summary=True,
repodir=self.path)
tagname = None
child = None
for elt in tree.findall('patch'):
node = elt.get('hash')
name = elt.findtext('name', '')
if name.startswith('TAG '):
tagname = name[4:].strip()
elif tagname is not None:
self.tags[tagname] = node
tagname = None
self.changes[node] = elt
self.parents[child] = [node]
child = node
self.parents[child] = []
def after(self):
self.ui.debug('cleaning up %s\n' % self.tmppath)
shutil.rmtree(self.tmppath, ignore_errors=True)
def recode(self, s, encoding=None):
if isinstance(s, unicode):
# XMLParser returns unicode objects for anything it can't
# encode into ASCII. We convert them back to str to get
# recode's normal conversion behavior.
s = s.encode('latin-1')
return super(darcs_source, self).recode(s, encoding)
def xml(self, cmd, **kwargs):
# NOTE: darcs is currently encoding agnostic and will print
# patch metadata byte-for-byte, even in the XML changelog.
etree = ElementTree()
# While we are decoding the XML as latin-1 to be as liberal as
# possible, etree will still raise an exception if any
# non-printable characters are in the XML changelog.
parser = XMLParser(encoding='latin-1')
p = self._run(cmd, **kwargs)
etree.parse(p.stdout, parser=parser)
p.wait()
self.checkexit(p.returncode)
return etree.getroot()
def format(self):
output, status = self.run('show', 'repo', no_files=True,
repodir=self.path)
self.checkexit(status)
m = re.search(r'^\s*Format:\s*(.*)$', output, re.MULTILINE)
if not m:
return None
return ','.join(sorted(f.strip() for f in m.group(1).split(',')))
def manifest(self):
man = []
output, status = self.run('show', 'files', no_directories=True,
repodir=self.tmppath)
self.checkexit(status)
for line in output.split('\n'):
path = line[2:]
if path:
man.append(path)
return man
def getheads(self):
return self.parents[None]
def getcommit(self, rev):
elt = self.changes[rev]
date = util.strdate(elt.get('local_date'), '%a %b %d %H:%M:%S %Z %Y')
desc = elt.findtext('name') + '\n' + elt.findtext('comment', '')
# etree can return unicode objects for name, comment, and author,
# so recode() is used to ensure str objects are emitted.
return common.commit(author=self.recode(elt.get('author')),
date=util.datestr(date, '%Y-%m-%d %H:%M:%S %1%2'),
desc=self.recode(desc).strip(),
parents=self.parents[rev])
def pull(self, rev):
output, status = self.run('pull', self.path, all=True,
match='hash %s' % rev,
no_test=True, no_posthook=True,
external_merge='/bin/false',
repodir=self.tmppath)
if status:
if output.find('We have conflicts in') == -1:
self.checkexit(status, output)
output, status = self.run('revert', all=True, repodir=self.tmppath)
self.checkexit(status, output)
def getchanges(self, rev, full):
if full:
raise error.Abort(_("convert from darcs does not support --full"))
copies = {}
changes = []
man = None
for elt in self.changes[rev].find('summary').getchildren():
if elt.tag in ('add_directory', 'remove_directory'):
continue
if elt.tag == 'move':
if man is None:
man = self.manifest()
source, dest = elt.get('from'), elt.get('to')
if source in man:
# File move
changes.append((source, rev))
changes.append((dest, rev))
copies[dest] = source
else:
# Directory move, deduce file moves from manifest
source = source + '/'
for f in man:
if not f.startswith(source):
continue
fdest = dest + '/' + f[len(source):]
changes.append((f, rev))
changes.append((fdest, rev))
copies[fdest] = f
else:
changes.append((elt.text.strip(), rev))
self.pull(rev)
self.lastrev = rev
return sorted(changes), copies, set()
def getfile(self, name, rev):
if rev != self.lastrev:
raise error.Abort(_('internal calling inconsistency'))
path = os.path.join(self.tmppath, name)
try:
data = util.readfile(path)
mode = os.lstat(path).st_mode
except IOError as inst:
if inst.errno == errno.ENOENT:
return None, None
raise
mode = (mode & 0o111) and 'x' or ''
return data, mode
def gettags(self):
return self.tags