Changeset - b61824e61e68
[Not reviewed]
beta
0 1 0
Marcin Kuzminski - 13 years ago 2012-11-29 19:29:33
marcin@python-works.com
Don't cache dulwich Repos, in pararell multithreaded evniroment dulwich pack file openers
can break badly. We need to create Repo() objects always for each call, even when it's
taken from cache.
Fixed issue with dulwich filedescriptor leak ref #573
1 file changed with 20 insertions and 13 deletions:
0 comments (0 inline, 0 general)
rhodecode/lib/vcs/backends/git/repository.py
Show inline comments
 
@@ -9,117 +9,124 @@
 
    :copyright: (c) 2010-2011 by Marcin Kuzminski, Lukasz Balcerzak.
 
"""
 

	
 
import os
 
import re
 
import time
 
import posixpath
 
import logging
 
import traceback
 
import urllib
 
import urllib2
 
from dulwich.repo import Repo, NotGitRepository
 
from dulwich.objects import Tag
 
from string import Template
 
from subprocess import Popen, PIPE
 
from rhodecode.lib.vcs.backends.base import BaseRepository
 
from rhodecode.lib.vcs.exceptions import BranchDoesNotExistError
 
from rhodecode.lib.vcs.exceptions import ChangesetDoesNotExistError
 
from rhodecode.lib.vcs.exceptions import EmptyRepositoryError
 
from rhodecode.lib.vcs.exceptions import RepositoryError
 
from rhodecode.lib.vcs.exceptions import TagAlreadyExistError
 
from rhodecode.lib.vcs.exceptions import TagDoesNotExistError
 
from rhodecode.lib.vcs.utils import safe_unicode, makedate, date_fromtimestamp
 
from rhodecode.lib.vcs.utils.lazy import LazyProperty
 
from rhodecode.lib.vcs.utils.ordered_dict import OrderedDict
 
from rhodecode.lib.vcs.utils.paths import abspath
 
from rhodecode.lib.vcs.utils.paths import get_user_home
 
from .workdir import GitWorkdir
 
from .changeset import GitChangeset
 
from .inmemory import GitInMemoryChangeset
 
from .config import ConfigFile
 
from rhodecode.lib import subprocessio
 

	
 

	
 
log = logging.getLogger(__name__)
 

	
 

	
 
class GitRepository(BaseRepository):
 
    """
 
    Git repository backend.
 
    """
 
    DEFAULT_BRANCH_NAME = 'master'
 
    scm = 'git'
 

	
 
    def __init__(self, repo_path, create=False, src_url=None,
 
                 update_after_clone=False, bare=False):
 

	
 
        self.path = abspath(repo_path)
 
        self._repo = self._get_repo(create, src_url, update_after_clone, bare)
 
        repo = self._get_repo(create, src_url, update_after_clone, bare)
 
        self.bare = repo.bare
 

	
 
        self._config_files = [
 
            bare and abspath(self.path, 'config') or abspath(self.path, '.git',
 
                'config'),
 
            abspath(get_user_home(), '.gitconfig'),
 
        ]
 

	
 
    @property
 
    def _repo(self):
 
        repo = Repo(self.path)
 
        #temporary set that to now at later we will move it to constructor
 
        baseui = None
 
        if baseui is None:
 
            from mercurial.ui import ui
 
            baseui = ui()
 
        # patch the instance of GitRepo with an "FAKE" ui object to add
 
        # compatibility layer with Mercurial
 
        setattr(self._repo, 'ui', baseui)
 

	
 
        try:
 
            self.head = self._repo.head()
 
        except KeyError:
 
            self.head = None
 
        setattr(repo, 'ui', baseui)
 
        return repo
 

	
 
        self._config_files = [
 
            bare and abspath(self.path, 'config') or abspath(self.path, '.git',
 
                'config'),
 
            abspath(get_user_home(), '.gitconfig'),
 
        ]
 
        self.bare = self._repo.bare
 
    @property
 
    def head(self):
 
        try:
 
            return self._repo.head()
 
        except KeyError:
 
            return None
 

	
 
    @LazyProperty
 
    def revisions(self):
 
        """
 
        Returns list of revisions' ids, in ascending order.  Being lazy
 
        attribute allows external tools to inject shas from cache.
 
        """
 
        return self._get_all_revisions()
 

	
 
    def run_git_command(self, cmd):
 
        """
 
        Runs given ``cmd`` as git command and returns tuple
 
        (returncode, stdout, stderr).
 

	
 
        .. note::
 
           This method exists only until log/blame functionality is implemented
 
           at Dulwich (see https://bugs.launchpad.net/bugs/645142). Parsing
 
           os command's output is road to hell...
 

	
 
        :param cmd: git command to be executed
 
        """
 

	
 
        _copts = ['-c', 'core.quotepath=false', ]
 
        _str_cmd = False
 
        if isinstance(cmd, basestring):
 
            cmd = [cmd]
 
            _str_cmd = True
 

	
 
        gitenv = os.environ
 
        # need to clean fix GIT_DIR !
 
        if 'GIT_DIR' in gitenv:
 
            del gitenv['GIT_DIR']
 
        gitenv['GIT_CONFIG_NOGLOBAL'] = '1'
 

	
 
        cmd = ['git'] + _copts + cmd
 
        if _str_cmd:
 
            cmd = ' '.join(cmd)
 
        try:
 
            opts = dict(
 
                env=gitenv,
 
                shell=False,
 
            )
 
            if os.path.isdir(self.path):
 
                opts['cwd'] = self.path
 
            p = subprocessio.SubprocessIOChunker(cmd, **opts)
 
        except (EnvironmentError, OSError), err:
 
            log.error(traceback.format_exc())
 
            raise RepositoryError("Couldn't run git command (%s).\n"
0 comments (0 inline, 0 general)