##// 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:

r29205:a0939666 default
r30318:e1d6aa0e default
Show More
git.py
399 lines | 13.7 KiB | text/x-python | PythonLexer
# git.py - git support for the convert extension
#
# Copyright 2005-2009 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 os
from mercurial.i18n import _
from mercurial import (
config,
error,
node as nodemod,
)
from . import (
common,
)
class submodule(object):
def __init__(self, path, node, url):
self.path = path
self.node = node
self.url = url
def hgsub(self):
return "%s = [git]%s" % (self.path, self.url)
def hgsubstate(self):
return "%s %s" % (self.node, self.path)
class convert_git(common.converter_source, common.commandline):
# Windows does not support GIT_DIR= construct while other systems
# cannot remove environment variable. Just assume none have
# both issues.
def _gitcmd(self, cmd, *args, **kwargs):
return cmd('--git-dir=%s' % self.path, *args, **kwargs)
def gitrun0(self, *args, **kwargs):
return self._gitcmd(self.run0, *args, **kwargs)
def gitrun(self, *args, **kwargs):
return self._gitcmd(self.run, *args, **kwargs)
def gitrunlines0(self, *args, **kwargs):
return self._gitcmd(self.runlines0, *args, **kwargs)
def gitrunlines(self, *args, **kwargs):
return self._gitcmd(self.runlines, *args, **kwargs)
def gitpipe(self, *args, **kwargs):
return self._gitcmd(self._run3, *args, **kwargs)
def __init__(self, ui, path, revs=None):
super(convert_git, self).__init__(ui, path, revs=revs)
common.commandline.__init__(self, ui, 'git')
# Pass an absolute path to git to prevent from ever being interpreted
# as a URL
path = os.path.abspath(path)
if os.path.isdir(path + "/.git"):
path += "/.git"
if not os.path.exists(path + "/objects"):
raise common.NoRepo(_("%s does not look like a Git repository") %
path)
# The default value (50) is based on the default for 'git diff'.
similarity = ui.configint('convert', 'git.similarity', default=50)
if similarity < 0 or similarity > 100:
raise error.Abort(_('similarity must be between 0 and 100'))
if similarity > 0:
self.simopt = ['-C%d%%' % similarity]
findcopiesharder = ui.configbool('convert', 'git.findcopiesharder',
False)
if findcopiesharder:
self.simopt.append('--find-copies-harder')
else:
self.simopt = []
common.checktool('git', 'git')
self.path = path
self.submodules = []
self.catfilepipe = self.gitpipe('cat-file', '--batch')
def after(self):
for f in self.catfilepipe:
f.close()
def getheads(self):
if not self.revs:
output, status = self.gitrun('rev-parse', '--branches', '--remotes')
heads = output.splitlines()
if status:
raise error.Abort(_('cannot retrieve git heads'))
else:
heads = []
for rev in self.revs:
rawhead, ret = self.gitrun('rev-parse', '--verify', rev)
heads.append(rawhead[:-1])
if ret:
raise error.Abort(_('cannot retrieve git head "%s"') % rev)
return heads
def catfile(self, rev, type):
if rev == nodemod.nullhex:
raise IOError
self.catfilepipe[0].write(rev+'\n')
self.catfilepipe[0].flush()
info = self.catfilepipe[1].readline().split()
if info[1] != type:
raise error.Abort(_('cannot read %r object at %s') % (type, rev))
size = int(info[2])
data = self.catfilepipe[1].read(size)
if len(data) < size:
raise error.Abort(_('cannot read %r object at %s: unexpected size')
% (type, rev))
# read the trailing newline
self.catfilepipe[1].read(1)
return data
def getfile(self, name, rev):
if rev == nodemod.nullhex:
return None, None
if name == '.hgsub':
data = '\n'.join([m.hgsub() for m in self.submoditer()])
mode = ''
elif name == '.hgsubstate':
data = '\n'.join([m.hgsubstate() for m in self.submoditer()])
mode = ''
else:
data = self.catfile(rev, "blob")
mode = self.modecache[(name, rev)]
return data, mode
def submoditer(self):
null = nodemod.nullhex
for m in sorted(self.submodules, key=lambda p: p.path):
if m.node != null:
yield m
def parsegitmodules(self, content):
"""Parse the formatted .gitmodules file, example file format:
[submodule "sub"]\n
\tpath = sub\n
\turl = git://giturl\n
"""
self.submodules = []
c = config.config()
# Each item in .gitmodules starts with whitespace that cant be parsed
c.parse('.gitmodules', '\n'.join(line.strip() for line in
content.split('\n')))
for sec in c.sections():
s = c[sec]
if 'url' in s and 'path' in s:
self.submodules.append(submodule(s['path'], '', s['url']))
def retrievegitmodules(self, version):
modules, ret = self.gitrun('show', '%s:%s' % (version, '.gitmodules'))
if ret:
# This can happen if a file is in the repo that has permissions
# 160000, but there is no .gitmodules file.
self.ui.warn(_("warning: cannot read submodules config file in "
"%s\n") % version)
return
try:
self.parsegitmodules(modules)
except error.ParseError:
self.ui.warn(_("warning: unable to parse .gitmodules in %s\n")
% version)
return
for m in self.submodules:
node, ret = self.gitrun('rev-parse', '%s:%s' % (version, m.path))
if ret:
continue
m.node = node.strip()
def getchanges(self, version, full):
if full:
raise error.Abort(_("convert from git does not support --full"))
self.modecache = {}
cmd = ['diff-tree','-z', '--root', '-m', '-r'] + self.simopt + [version]
output, status = self.gitrun(*cmd)
if status:
raise error.Abort(_('cannot read changes in %s') % version)
changes = []
copies = {}
seen = set()
entry = None
subexists = [False]
subdeleted = [False]
difftree = output.split('\x00')
lcount = len(difftree)
i = 0
skipsubmodules = self.ui.configbool('convert', 'git.skipsubmodules',
False)
def add(entry, f, isdest):
seen.add(f)
h = entry[3]
p = (entry[1] == "100755")
s = (entry[1] == "120000")
renamesource = (not isdest and entry[4][0] == 'R')
if f == '.gitmodules':
if skipsubmodules:
return
subexists[0] = True
if entry[4] == 'D' or renamesource:
subdeleted[0] = True
changes.append(('.hgsub', nodemod.nullhex))
else:
changes.append(('.hgsub', ''))
elif entry[1] == '160000' or entry[0] == ':160000':
if not skipsubmodules:
subexists[0] = True
else:
if renamesource:
h = nodemod.nullhex
self.modecache[(f, h)] = (p and "x") or (s and "l") or ""
changes.append((f, h))
while i < lcount:
l = difftree[i]
i += 1
if not entry:
if not l.startswith(':'):
continue
entry = l.split()
continue
f = l
if entry[4][0] == 'C':
copysrc = f
copydest = difftree[i]
i += 1
f = copydest
copies[copydest] = copysrc
if f not in seen:
add(entry, f, False)
# A file can be copied multiple times, or modified and copied
# simultaneously. So f can be repeated even if fdest isn't.
if entry[4][0] == 'R':
# rename: next line is the destination
fdest = difftree[i]
i += 1
if fdest not in seen:
add(entry, fdest, True)
# .gitmodules isn't imported at all, so it being copied to
# and fro doesn't really make sense
if f != '.gitmodules' and fdest != '.gitmodules':
copies[fdest] = f
entry = None
if subexists[0]:
if subdeleted[0]:
changes.append(('.hgsubstate', nodemod.nullhex))
else:
self.retrievegitmodules(version)
changes.append(('.hgsubstate', ''))
return (changes, copies, set())
def getcommit(self, version):
c = self.catfile(version, "commit") # read the commit hash
end = c.find("\n\n")
message = c[end + 2:]
message = self.recode(message)
l = c[:end].splitlines()
parents = []
author = committer = None
for e in l[1:]:
n, v = e.split(" ", 1)
if n == "author":
p = v.split()
tm, tz = p[-2:]
author = " ".join(p[:-2])
if author[0] == "<": author = author[1:-1]
author = self.recode(author)
if n == "committer":
p = v.split()
tm, tz = p[-2:]
committer = " ".join(p[:-2])
if committer[0] == "<": committer = committer[1:-1]
committer = self.recode(committer)
if n == "parent":
parents.append(v)
if committer and committer != author:
message += "\ncommitter: %s\n" % committer
tzs, tzh, tzm = tz[-5:-4] + "1", tz[-4:-2], tz[-2:]
tz = -int(tzs) * (int(tzh) * 3600 + int(tzm))
date = tm + " " + str(tz)
c = common.commit(parents=parents, date=date, author=author,
desc=message,
rev=version)
return c
def numcommits(self):
output, ret = self.gitrunlines('rev-list', '--all')
if ret:
raise error.Abort(_('cannot retrieve number of commits in %s') \
% self.path)
return len(output)
def gettags(self):
tags = {}
alltags = {}
output, status = self.gitrunlines('ls-remote', '--tags', self.path)
if status:
raise error.Abort(_('cannot read tags from %s') % self.path)
prefix = 'refs/tags/'
# Build complete list of tags, both annotated and bare ones
for line in output:
line = line.strip()
if line.startswith("error:") or line.startswith("fatal:"):
raise error.Abort(_('cannot read tags from %s') % self.path)
node, tag = line.split(None, 1)
if not tag.startswith(prefix):
continue
alltags[tag[len(prefix):]] = node
# Filter out tag objects for annotated tag refs
for tag in alltags:
if tag.endswith('^{}'):
tags[tag[:-3]] = alltags[tag]
else:
if tag + '^{}' in alltags:
continue
else:
tags[tag] = alltags[tag]
return tags
def getchangedfiles(self, version, i):
changes = []
if i is None:
output, status = self.gitrunlines('diff-tree', '--root', '-m',
'-r', version)
if status:
raise error.Abort(_('cannot read changes in %s') % version)
for l in output:
if "\t" not in l:
continue
m, f = l[:-1].split("\t")
changes.append(f)
else:
output, status = self.gitrunlines('diff-tree', '--name-only',
'--root', '-r', version,
'%s^%s' % (version, i + 1), '--')
if status:
raise error.Abort(_('cannot read changes in %s') % version)
changes = [f.rstrip('\n') for f in output]
return changes
def getbookmarks(self):
bookmarks = {}
# Handle local and remote branches
remoteprefix = self.ui.config('convert', 'git.remoteprefix', 'remote')
reftypes = [
# (git prefix, hg prefix)
('refs/remotes/origin/', remoteprefix + '/'),
('refs/heads/', '')
]
exclude = set([
'refs/remotes/origin/HEAD',
])
try:
output, status = self.gitrunlines('show-ref')
for line in output:
line = line.strip()
rev, name = line.split(None, 1)
# Process each type of branch
for gitprefix, hgprefix in reftypes:
if not name.startswith(gitprefix) or name in exclude:
continue
name = '%s%s' % (hgprefix, name[len(gitprefix):])
bookmarks[name] = rev
except Exception:
pass
return bookmarks
def checkrevformat(self, revstr, mapname='splicemap'):
""" git revision string is a 40 byte hex """
self.checkhexformat(revstr, mapname)