mirror of
https://github.com/mnauw/git-remote-hg.git
synced 2025-10-26 06:06:06 +01:00
git-hg-helper: repurpose marks subcommand to gc subcommand
This commit is contained in:
Mark Nauwelaerts
116
git-hg-helper
116
git-hg-helper
@@ -13,6 +13,7 @@ import subprocess
|
||||
import argparse
|
||||
import textwrap
|
||||
import logging
|
||||
import threading
|
||||
|
||||
# thanks go to git-remote-helper for some helper functions
|
||||
|
||||
@@ -59,9 +60,7 @@ class GitHgRepo:
|
||||
def identity(self):
|
||||
return '[%s|%s]' % (os.getcwd(), self.topdir)
|
||||
|
||||
# run a git cmd in repo dir, captures stdout and stderr by default
|
||||
# override in kwargs if otherwise desired
|
||||
def run_cmd(self, args, check=False, **kwargs):
|
||||
def start_cmd(self, args, **kwargs):
|
||||
cmd = ['git'] + args
|
||||
popen_options = { 'cwd': self.topdir,
|
||||
'stdout': subprocess.PIPE, 'stderr': subprocess.PIPE }
|
||||
@@ -69,6 +68,12 @@ class GitHgRepo:
|
||||
log('%s running cmd %s with options %s', self.identity(),
|
||||
cmd, popen_options)
|
||||
process = subprocess.Popen(cmd, **popen_options)
|
||||
return process
|
||||
|
||||
# run a git cmd in repo dir, captures stdout and stderr by default
|
||||
# override in kwargs if otherwise desired
|
||||
def run_cmd(self, args, check=False, **kwargs):
|
||||
process = self.start_cmd(args, **kwargs)
|
||||
output = process.communicate()[0]
|
||||
if check and process.returncode != 0:
|
||||
die('command failed: %s', ' '.join(cmd))
|
||||
@@ -362,7 +367,7 @@ class GitRevCommand(SubCommand):
|
||||
print gitcommit
|
||||
|
||||
|
||||
class MarksCommand(SubCommand):
|
||||
class GcCommand(SubCommand):
|
||||
|
||||
def argumentparser(self):
|
||||
usage = '%%(prog)s %s [options] <remote>...' % (self.subcommand)
|
||||
@@ -370,31 +375,30 @@ class MarksCommand(SubCommand):
|
||||
formatter_class=argparse.RawDescriptionHelpFormatter)
|
||||
p.add_argument('-n', '--dry-run', action='store_true',
|
||||
help='do not actually update any metadata files')
|
||||
p.add_argument('--keep', metavar='REVID',
|
||||
help='only retain ancestors of REVID (including) in \
|
||||
hg tracking metadata, akin to hg\'s strip')
|
||||
p.epilog = textwrap.dedent("""\
|
||||
Performs checks on <remote>'s marks files and ensures these are consistent
|
||||
Performs cleanup on <remote>'s marks files and ensures these are consistent
|
||||
(never affecting or touching any git repository objects or history).
|
||||
The marks files are considered consistent if they "join"
|
||||
on the :mark number (with no dangling hg or git commit id on either side).
|
||||
on the :mark number (along with a valid git commit id).
|
||||
|
||||
While fetching from an hg remote usually results in a sane state, there
|
||||
are some cases where that might not suffice (or not even succeed in the first
|
||||
place). Also, fetching will only ever add tracked metadata marks, whereas
|
||||
sometimes forgetting about some state might be required for consistent state
|
||||
recovery (e.g. a strip performed on a remote Mercurial repo). Executing this
|
||||
command should allow a subsequent fetch to succeed and restore a sane state
|
||||
quickly. In particular, making good use of --keep following a strip allows
|
||||
a subsequent fetch to recover quickly without extensive history processing.
|
||||
|
||||
Furthermore, since git-fast-import (used during fetch) also dumps
|
||||
non-commit SHA-1 in the marks file, the latter can become pretty large.
|
||||
It will reduce in size by either performing a push (git-fast-export only
|
||||
dumps commit objects to marks file) or by running this helper command.
|
||||
This command can be useful in following scenarios:
|
||||
* following a git gc command;
|
||||
this could prune objects and lead to (then) invalid commit ids in marks
|
||||
(in which case git-fast-export or git-fast-import would complain bitterly).
|
||||
Such pruning is more likely to happen with remote hg repos with multiple heads.
|
||||
* cleaning marks-git of a fetch-only remote;
|
||||
git-fast-import (used during fetch) also dumps non-commit SHA-1 in the marks file,
|
||||
so the latter can become pretty large. It will reduce in size either by a push
|
||||
(git-fast-export only dumps commit objects) or by running this helper command.
|
||||
""")
|
||||
return p
|
||||
|
||||
def print_commits(self, gm, dest):
|
||||
for c in gm.marks.keys():
|
||||
dest.write(c + '\n')
|
||||
dest.flush()
|
||||
dest.close()
|
||||
|
||||
def do(self, options, args):
|
||||
mydir = os.path.dirname(__file__)
|
||||
import imp
|
||||
@@ -407,31 +411,37 @@ class MarksCommand(SubCommand):
|
||||
if not remote in hg_repos:
|
||||
self.usage('%s is not a valid hg remote' % (remote))
|
||||
hgpath = os.path.join(self.githgrepo.gitdir, 'hg', remote)
|
||||
print "Loading hg marks ..."
|
||||
hgm = remotehg.Marks(os.path.join(hgpath, 'marks-hg'), None)
|
||||
print "Loading git marks ..."
|
||||
gm = GitMarks(os.path.join(hgpath, 'marks-git'))
|
||||
repo = hg.repository(ui.ui(), hg_repos[remote])
|
||||
ctx = ctxrev = None
|
||||
if options.keep != None:
|
||||
strip = options.keep
|
||||
if strip in repo and repo[strip]:
|
||||
ctx = repo[strip]
|
||||
ctxrev = ctx.rev()
|
||||
if not ctxrev >= 0:
|
||||
self.usage('revision %s not found in repository %s' % (strip, repo.root))
|
||||
# git-gc may have dropped unreachable commits
|
||||
# (in particular due to multiple hg head cases)
|
||||
# need to drop those so git-fast-export or git-fast-import does not complain
|
||||
print "Performing garbage collection on git commits ..."
|
||||
process = self.githgrepo.start_cmd(['cat-file', '--batch-check'], \
|
||||
stdin=subprocess.PIPE)
|
||||
thread = threading.Thread(target=self.print_commits, args=(gm, process.stdin))
|
||||
thread.start()
|
||||
git_marks = set({})
|
||||
for l in process.stdout:
|
||||
sp = l.strip().split(' ', 2)
|
||||
if sp[1] == 'commit':
|
||||
git_marks.add(gm.from_rev(sp[0]))
|
||||
thread.join()
|
||||
# reduce down to marks that are common to both
|
||||
common_marks = set(hgm.rev_marks.keys()).intersection(gm.rev_marks.keys())
|
||||
print "Computing marks intersection ..."
|
||||
common_marks = set(hgm.rev_marks.keys()).intersection(git_marks)
|
||||
hg_rev_marks = {}
|
||||
git_rev_marks = {}
|
||||
for m in common_marks:
|
||||
rev = hgm.rev_marks[m]
|
||||
# also check if still around in repo
|
||||
if rev in repo and \
|
||||
not (ctxrev != None and repo[rev].rev() > ctxrev):
|
||||
hg_rev_marks[m] = hgm.rev_marks[m]
|
||||
git_rev_marks[m] = gm.rev_marks[m]
|
||||
hg_rev_marks[m] = hgm.rev_marks[m]
|
||||
git_rev_marks[m] = gm.rev_marks[m]
|
||||
# common marks will not not include any refs/notes/hg
|
||||
# let's not discard those casually, though they are not vital
|
||||
revlist = subprocess.Popen(['git', 'rev-list', 'refs/notes/hg'], stdout=subprocess.PIPE)
|
||||
print "Including notes commits ..."
|
||||
revlist = self.githgrepo.start_cmd(['rev-list', 'refs/notes/hg'])
|
||||
for l in revlist.stdout.readlines():
|
||||
c = l.strip()
|
||||
m = gm.marks.get(c, 0)
|
||||
@@ -445,38 +455,18 @@ class MarksCommand(SubCommand):
|
||||
print "Trimmed hg marks from #%d down to #%d" % (len(hgm.rev_marks), len(hg_rev_marks))
|
||||
if len(gm.rev_marks) != len(git_rev_marks):
|
||||
print "Trimmed git marks from #%d down to #%d" % (len(gm.rev_marks), len(git_rev_marks))
|
||||
# make hg tips as consistent as possible
|
||||
for b in hgm.tips:
|
||||
tip = hgm.tips[b]
|
||||
if tip not in repo or not repo[tip]:
|
||||
if not ctx:
|
||||
print "Could not determine safe value for tip of %s (rev %s)" % (b, tip)
|
||||
print "Please use --strip to provide fallback value."
|
||||
return
|
||||
else:
|
||||
# basically only the revision number is used by a fetch
|
||||
tip = ctx.hex()
|
||||
else:
|
||||
hgrevs = set(hg_rev_marks.values())
|
||||
while True:
|
||||
if tip in hgrevs:
|
||||
break
|
||||
parent = repo[tip].parents()[0].hex()
|
||||
if parent == tip:
|
||||
break
|
||||
tip = parent
|
||||
if hgm.tips[b] != tip:
|
||||
hgm.tips[b] = tip
|
||||
print "Updated tip of %s to %s" % (b, tip)
|
||||
# marks-hg tips irrelevant nowadays
|
||||
# now update and store
|
||||
if not options.dry_run:
|
||||
# hg marks
|
||||
print "Writing hg marks ..."
|
||||
hgm.rev_marks = hg_rev_marks
|
||||
hgm.marks = {}
|
||||
for mark, rev in hg_rev_marks.iteritems():
|
||||
hgm.marks[rev] = mark
|
||||
hgm.store()
|
||||
# git marks
|
||||
print "Writing git marks ..."
|
||||
gm.rev_marks = git_rev_marks
|
||||
gm.marks = {}
|
||||
for mark, rev in git_rev_marks.iteritems():
|
||||
@@ -847,7 +837,7 @@ def get_subcommands():
|
||||
'hg-rev': HgRevCommand,
|
||||
'git-rev': GitRevCommand,
|
||||
'repo': RepoCommand,
|
||||
'marks': MarksCommand,
|
||||
'gc': GcCommand,
|
||||
'sub': SubRepoCommand,
|
||||
'help' : HelpCommand
|
||||
}
|
||||
@@ -868,7 +858,7 @@ def do_usage():
|
||||
|
||||
hg-rev \t show hg revision corresponding to a git revision
|
||||
git-rev \t find git revision corresponding to a hg revision
|
||||
marks \t perform maintenance on repo tracking marks
|
||||
gc \t perform maintenance and consistency cleanup on repo tracking marks
|
||||
sub \t manage subrepos
|
||||
repo \t show local hg repo backing a remote
|
||||
|
||||
|
||||
Reference in New Issue
Block a user