Changeset - 0e2d450feb03
[Not reviewed]
default
0 8 0
Mads Kiilerich - 10 years ago 2015-06-09 22:53:24
madski@unity3d.com
git: run external commands as list of strings so we really get correct quoting (Issue #135)

a6dfd14d4b20 from
https://bitbucket.org/conservancy/kallithea/pull-request/17/add-quotes-to-repo-urls-for-git-backend
fixed that issue but did not make it "safe". The vcs git backend still used
command strings but tried to quote them correctly ... but that approach is
almost impossible to get right.

Instead, pass a string list all the way to the subprocess module and let it do
the quoting. This also makes some of the code more simple.
8 files changed with 57 insertions and 91 deletions:
0 comments (0 inline, 0 general)
kallithea/controllers/compare.py
Show inline comments
 
@@ -85,103 +85,103 @@ class CompareController(BaseRepoControll
 
                # rev numbers from hgrepo can be used in other_repo - org_rev ancestors cannot
 

	
 
            #no remote compare do it on the same repository
 
            else:
 
                hgrepo = other_repo._repo
 

	
 
            ancestors = hgrepo.revs("ancestor(id(%s), id(%s))", org_rev, other_rev)
 
            if ancestors:
 
                # FIXME: picks arbitrary ancestor - but there is usually only one
 
                try:
 
                    ancestor = hgrepo[ancestors.first()].hex()
 
                except AttributeError:
 
                    # removed in hg 3.2
 
                    ancestor = hgrepo[ancestors[0]].hex()
 

	
 
            other_revs = hgrepo.revs("ancestors(id(%s)) and not ancestors(id(%s)) and not id(%s)",
 
                                     other_rev, org_rev, org_rev)
 
            other_changesets = [other_repo.get_changeset(rev) for rev in other_revs]
 
            org_revs = hgrepo.revs("ancestors(id(%s)) and not ancestors(id(%s)) and not id(%s)",
 
                                   org_rev, other_rev, other_rev)
 

	
 
            org_changesets = [org_repo.get_changeset(hgrepo[rev].hex()) for rev in org_revs]
 

	
 
        elif alias == 'git':
 
            if org_repo != other_repo:
 
                from dulwich.repo import Repo
 
                from dulwich.client import SubprocessGitClient
 

	
 
                gitrepo = Repo(org_repo.path)
 
                SubprocessGitClient(thin_packs=False).fetch(other_repo.path, gitrepo)
 

	
 
                gitrepo_remote = Repo(other_repo.path)
 
                SubprocessGitClient(thin_packs=False).fetch(org_repo.path, gitrepo_remote)
 

	
 
                revs = []
 
                for x in gitrepo_remote.get_walker(include=[other_rev],
 
                                                   exclude=[org_rev]):
 
                    revs.append(x.commit.id)
 

	
 
                other_changesets = [other_repo.get_changeset(rev) for rev in reversed(revs)]
 
                if other_changesets:
 
                    ancestor = other_changesets[0].parents[0].raw_id
 
                else:
 
                    # no changesets from other repo, ancestor is the other_rev
 
                    ancestor = other_rev
 

	
 
            else:
 
                so, se = org_repo.run_git_command(
 
                    'log --reverse --pretty="format: %%H" -s %s..%s'
 
                        % (org_rev, other_rev)
 
                    ['log', '--reverse', '--pretty=format:%H',
 
                     '-s', '%s..%s' % (org_rev, other_rev)]
 
                )
 
                other_changesets = [org_repo.get_changeset(cs)
 
                              for cs in re.findall(r'[0-9a-fA-F]{40}', so)]
 
                so, se = org_repo.run_git_command(
 
                    'merge-base %s %s' % (org_rev, other_rev)
 
                    ['merge-base', org_rev, other_rev]
 
                )
 
                ancestor = re.findall(r'[0-9a-fA-F]{40}', so)[0]
 
            org_changesets = []
 

	
 
        else:
 
            raise Exception('Bad alias only git and hg is allowed')
 

	
 
        return other_changesets, org_changesets, ancestor
 

	
 
    @LoginRequired()
 
    @HasRepoPermissionAnyDecorator('repository.read', 'repository.write',
 
                                   'repository.admin')
 
    def index(self, repo_name):
 
        c.compare_home = True
 
        org_repo = c.db_repo.repo_name
 
        other_repo = request.GET.get('other_repo', org_repo)
 
        c.a_repo = Repository.get_by_repo_name(org_repo)
 
        c.cs_repo = Repository.get_by_repo_name(other_repo)
 
        c.a_ref_name = c.cs_ref_name = _('Select changeset')
 
        return render('compare/compare_diff.html')
 

	
 
    @LoginRequired()
 
    @HasRepoPermissionAnyDecorator('repository.read', 'repository.write',
 
                                   'repository.admin')
 
    def compare(self, repo_name, org_ref_type, org_ref_name, other_ref_type, other_ref_name):
 
        org_repo = c.db_repo.repo_name
 
        other_repo = request.GET.get('other_repo', org_repo)
 
        # If merge is True:
 
        #   Show what org would get if merged with other:
 
        #   List changesets that are ancestors of other but not of org.
 
        #   New changesets in org is thus ignored.
 
        #   Diff will be from common ancestor, and merges of org to other will thus be ignored.
 
        # If merge is False:
 
        #   Make a raw diff from org to other, no matter if related or not.
 
        #   Changesets in one and not in the other will be ignored
 
        merge = bool(request.GET.get('merge'))
 
        # fulldiff disables cut_off_limit
 
        c.fulldiff = request.GET.get('fulldiff')
 
        # partial uses compare_cs.html template directly
 
        partial = request.environ.get('HTTP_X_PARTIAL_XHR')
 
        # as_form puts hidden input field with changeset revisions
 
        c.as_form = partial and request.GET.get('as_form')
 
        # swap url for compare_diff page - never partial and never as_form
 
        c.swap_url = h.url('compare_url',
 
            repo_name=other_repo,
 
            org_ref_type=other_ref_type, org_ref_name=other_ref_name,
 
            other_repo=org_repo,
 
            other_ref_type=org_ref_type, other_ref_name=org_ref_name,
kallithea/lib/hooks.py
Show inline comments
 
@@ -402,69 +402,69 @@ def handle_git_receive(repo_path, revs, 
 
    engine = engine_from_config(conf, 'sqlalchemy.db1.')
 
    init_model(engine)
 

	
 
    baseui = make_ui('db')
 
    # fix if it's not a bare repo
 
    if repo_path.endswith(os.sep + '.git'):
 
        repo_path = repo_path[:-5]
 

	
 
    repo = Repository.get_by_full_path(repo_path)
 
    if not repo:
 
        raise OSError('Repository %s not found in database'
 
                      % (safe_str(repo_path)))
 

	
 
    _hooks = dict(baseui.configitems('hooks')) or {}
 

	
 
    if hook_type == 'pre':
 
        repo = repo.scm_instance
 
    else:
 
        #post push shouldn't use the cached instance never
 
        repo = repo.scm_instance_no_cache()
 

	
 
    if hook_type == 'pre':
 
        pre_push(baseui, repo)
 

	
 
    # if push hook is enabled via web interface
 
    elif hook_type == 'post' and _hooks.get(Ui.HOOK_PUSH):
 
        rev_data = []
 
        for l in revs:
 
            old_rev, new_rev, ref = l.split(' ')
 
            _ref_data = ref.split('/')
 
            if _ref_data[1] in ['tags', 'heads']:
 
                rev_data.append({'old_rev': old_rev,
 
                                 'new_rev': new_rev,
 
                                 'ref': ref,
 
                                 'type': _ref_data[1],
 
                                 'name': _ref_data[2].strip()})
 

	
 
        git_revs = []
 

	
 
        for push_ref in rev_data:
 
            _type = push_ref['type']
 
            if _type == 'heads':
 
                if push_ref['old_rev'] == EmptyChangeset().raw_id:
 
                    # update the symbolic ref if we push new repo
 
                    if repo.is_empty():
 
                        repo._repo.refs.set_symbolic_ref('HEAD',
 
                                            'refs/heads/%s' % push_ref['name'])
 

	
 
                    cmd = "for-each-ref --format='%(refname)' 'refs/heads/*'"
 
                    cmd = ['for-each-ref', '--format=%(refname)','refs/heads/*']
 
                    heads = repo.run_git_command(cmd)[0]
 
                    cmd = ['log', push_ref['new_rev'],
 
                           '--reverse', '--pretty=format:%H', '--not']
 
                    heads = heads.replace(push_ref['ref'], '')
 
                    heads = ' '.join(map(lambda c: c.strip('\n').strip(),
 
                                         heads.splitlines()))
 
                    cmd = (('log %(new_rev)s' % push_ref) +
 
                           ' --reverse --pretty=format:"%H" --not ' + heads)
 
                    for l in heads.splitlines():
 
                        cmd.append(l.strip())
 
                    git_revs += repo.run_git_command(cmd)[0].splitlines()
 

	
 
                elif push_ref['new_rev'] == EmptyChangeset().raw_id:
 
                    #delete branch case
 
                    git_revs += ['delete_branch=>%s' % push_ref['name']]
 
                else:
 
                    cmd = (('log %(old_rev)s..%(new_rev)s' % push_ref) +
 
                           ' --reverse --pretty=format:"%H"')
 
                    cmd = ['log', '%(old_rev)s..%(new_rev)s' % push_ref,
 
                           '--reverse', '--pretty=format:%H']
 
                    git_revs += repo.run_git_command(cmd)[0].splitlines()
 

	
 
            elif _type == 'tags':
 
                git_revs += ['tag=>%s' % push_ref['name']]
 

	
 
        log_push_action(baseui, repo, _git_revs=git_revs)
kallithea/lib/middleware/pygrack.py
Show inline comments
 
@@ -37,147 +37,142 @@ class FileWrapper(object):
 
            self.fd, self.content_length, self.content_length - self.remain
 
        )
 

	
 

	
 
class GitRepository(object):
 
    git_folder_signature = set(['config', 'head', 'info', 'objects', 'refs'])
 
    commands = ['git-upload-pack', 'git-receive-pack']
 

	
 
    def __init__(self, repo_name, content_path, extras):
 
        files = set([f.lower() for f in os.listdir(content_path)])
 
        if  not (self.git_folder_signature.intersection(files)
 
                == self.git_folder_signature):
 
            raise OSError('%s missing git signature' % content_path)
 
        self.content_path = content_path
 
        self.valid_accepts = ['application/x-%s-result' %
 
                              c for c in self.commands]
 
        self.repo_name = repo_name
 
        self.extras = extras
 

	
 
    def _get_fixedpath(self, path):
 
        """
 
        Small fix for repo_path
 

	
 
        :param path:
 
        """
 
        return path.split(self.repo_name, 1)[-1].strip('/')
 

	
 
    def inforefs(self, request, environ):
 
        """
 
        WSGI Response producer for HTTP GET Git Smart
 
        HTTP /info/refs request.
 
        """
 

	
 
        git_command = request.GET.get('service')
 
        if git_command not in self.commands:
 
            log.debug('command %s not allowed' % git_command)
 
            return exc.HTTPMethodNotAllowed()
 

	
 
        # note to self:
 
        # please, resist the urge to add '\n' to git capture and increment
 
        # line count by 1.
 
        # The code in Git client not only does NOT need '\n', but actually
 
        # blows up if you sprinkle "flush" (0000) as "0001\n".
 
        # It reads binary, per number of bytes specified.
 
        # if you do add '\n' as part of data, count it.
 
        server_advert = '# service=%s' % git_command
 
        packet_len = str(hex(len(server_advert) + 4)[2:].rjust(4, '0')).lower()
 
        _git_path = kallithea.CONFIG.get('git_path', 'git')
 
        cmd = [_git_path, git_command[4:],
 
               '--stateless-rpc', '--advertise-refs', self.content_path]
 
        log.debug('handling cmd %s', cmd)
 
        try:
 
            out = subprocessio.SubprocessIOChunker(
 
                r'%s %s --stateless-rpc --advertise-refs "%s"' % (
 
                    _git_path, git_command[4:], self.content_path),
 
                starting_values=[
 
                    packet_len + server_advert + '0000'
 
                ]
 
            out = subprocessio.SubprocessIOChunker(cmd,
 
                starting_values=[packet_len + server_advert + '0000']
 
            )
 
        except EnvironmentError, e:
 
            log.error(traceback.format_exc())
 
            raise exc.HTTPExpectationFailed()
 
        resp = Response()
 
        resp.content_type = 'application/x-%s-advertisement' % str(git_command)
 
        resp.charset = None
 
        resp.app_iter = out
 
        return resp
 

	
 
    def backend(self, request, environ):
 
        """
 
        WSGI Response producer for HTTP POST Git Smart HTTP requests.
 
        Reads commands and data from HTTP POST's body.
 
        returns an iterator obj with contents of git command's
 
        response to stdout
 
        """
 
        _git_path = kallithea.CONFIG.get('git_path', 'git')
 
        git_command = self._get_fixedpath(request.path_info)
 
        if git_command not in self.commands:
 
            log.debug('command %s not allowed' % git_command)
 
            return exc.HTTPMethodNotAllowed()
 

	
 
        if 'CONTENT_LENGTH' in environ:
 
            inputstream = FileWrapper(environ['wsgi.input'],
 
                                      request.content_length)
 
        else:
 
            inputstream = environ['wsgi.input']
 

	
 
        try:
 
            gitenv = os.environ
 
        gitenv = dict(os.environ)
 
            # forget all configs
 
            gitenv['GIT_CONFIG_NOGLOBAL'] = '1'
 
            opts = dict(
 
                env=gitenv,
 
                cwd=self.content_path,
 
            )
 
            cmd = r'%s %s --stateless-rpc "%s"' % (_git_path, git_command[4:],
 
                                                   self.content_path),
 
            log.debug('handling cmd %s' % cmd)
 
        cmd = [_git_path, git_command[4:], '--stateless-rpc', self.content_path]
 
        log.debug('handling cmd %s', cmd)
 
        try:
 
            out = subprocessio.SubprocessIOChunker(
 
                cmd,
 
                inputstream=inputstream,
 
                **opts
 
                env=gitenv,
 
                cwd=self.content_path,
 
            )
 
        except EnvironmentError, e:
 
            log.error(traceback.format_exc())
 
            raise exc.HTTPExpectationFailed()
 

	
 
        if git_command in [u'git-receive-pack']:
 
            # updating refs manually after each push.
 
            # Needed for pre-1.7.0.4 git clients using regular HTTP mode.
 
            from kallithea.lib.vcs import get_repo
 
            from dulwich.server import update_server_info
 
            repo = get_repo(self.content_path)
 
            if repo:
 
                update_server_info(repo._repo)
 

	
 
        resp = Response()
 
        resp.content_type = 'application/x-%s-result' % git_command.encode('utf8')
 
        resp.charset = None
 
        resp.app_iter = out
 
        return resp
 

	
 
    def __call__(self, environ, start_response):
 
        request = Request(environ)
 
        _path = self._get_fixedpath(request.path_info)
 
        if _path.startswith('info/refs'):
 
            app = self.inforefs
 
        elif [a for a in self.valid_accepts if a in request.accept]:
 
            app = self.backend
 
        try:
 
            resp = app(request, environ)
 
        except exc.HTTPException, e:
 
            resp = e
 
            log.error(traceback.format_exc())
 
        except Exception, e:
 
            log.error(traceback.format_exc())
 
            resp = exc.HTTPInternalServerError()
 
        return resp(environ, start_response)
 

	
 

	
 
class GitDirectory(object):
 

	
 
    def __init__(self, repo_root, repo_name, extras):
 
        repo_location = os.path.join(repo_root, repo_name)
 
        if not os.path.isdir(repo_location):
 
            raise OSError(repo_location)
 

	
 
        self.content_path = repo_location
 
        self.repo_name = repo_name
 
        self.repo_location = repo_location
kallithea/lib/utils.py
Show inline comments
 
@@ -754,97 +754,97 @@ class BasePasterCommand(Command):
 

	
 
    def update_parser(self):
 
        """
 
        Abstract method.  Allows for the class's parser to be updated
 
        before the superclass's `run` method is called.  Necessary to
 
        allow options/arguments to be passed through to the underlying
 
        celery command.
 
        """
 
        raise NotImplementedError("Abstract Method.")
 

	
 
    def bootstrap_config(self, conf):
 
        """
 
        Loads the pylons configuration.
 
        """
 
        from pylons import config as pylonsconfig
 

	
 
        self.path_to_ini_file = os.path.realpath(conf)
 
        conf = paste.deploy.appconfig('config:' + self.path_to_ini_file)
 
        pylonsconfig.init_app(conf.global_conf, conf.local_conf)
 

	
 
    def _init_session(self):
 
        """
 
        Inits SqlAlchemy Session
 
        """
 
        logging.config.fileConfig(self.path_to_ini_file)
 
        from pylons import config
 
        from kallithea.model import init_model
 
        from kallithea.lib.utils2 import engine_from_config
 

	
 
        #get to remove repos !!
 
        add_cache(config)
 
        engine = engine_from_config(config, 'sqlalchemy.db1.')
 
        init_model(engine)
 

	
 

	
 
def check_git_version():
 
    """
 
    Checks what version of git is installed in system, and issues a warning
 
    if it's too old for Kallithea to work properly.
 
    """
 
    from kallithea import BACKENDS
 
    from kallithea.lib.vcs.backends.git.repository import GitRepository
 
    from kallithea.lib.vcs.conf import settings
 
    from distutils.version import StrictVersion
 

	
 
    if 'git' not in BACKENDS:
 
        return None
 

	
 
    stdout, stderr = GitRepository._run_git_command('--version', _bare=True,
 
    stdout, stderr = GitRepository._run_git_command(['--version'], _bare=True,
 
                                                    _safe=True)
 

	
 
    m = re.search("\d+.\d+.\d+", stdout)
 
    if m:
 
        ver = StrictVersion(m.group(0))
 
    else:
 
        ver = StrictVersion('0.0.0')
 

	
 
    req_ver = StrictVersion('1.7.4')
 

	
 
    log.debug('Git executable: "%s" version %s detected: %s'
 
              % (settings.GIT_EXECUTABLE_PATH, ver, stdout))
 
    if stderr:
 
        log.warning('Error detecting git version: %r' % stderr)
 
    elif ver < req_ver:
 
        log.warning('Kallithea detected git version %s, which is too old '
 
                    'for the system to function properly. '
 
                    'Please upgrade to version %s or later.' % (ver, req_ver))
 
    return ver
 

	
 

	
 
@decorator.decorator
 
def jsonify(func, *args, **kwargs):
 
    """Action decorator that formats output for JSON
 

	
 
    Given a function that will return content, this decorator will turn
 
    the result into JSON, with a content-type of 'application/json' and
 
    output it.
 

	
 
    """
 
    from pylons.decorators.util import get_pylons
 
    from kallithea.lib.compat import json
 
    pylons = get_pylons(args)
 
    pylons.response.headers['Content-Type'] = 'application/json; charset=utf-8'
 
    data = func(*args, **kwargs)
 
    if isinstance(data, (list, tuple)):
 
        msg = "JSON responses with Array envelopes are susceptible to " \
 
              "cross-site data leak attacks, see " \
 
              "http://wiki.pylonshq.com/display/pylonsfaq/Warnings"
 
        warnings.warn(msg, Warning, 2)
 
        log.warning(msg)
 
    log.debug("Returning JSON wrapped action output")
 
    return json.dumps(data, encoding='utf-8')
 

	
 

	
 
def conditional_cache(region, prefix, condition, func):
 
    """
 

	
kallithea/lib/vcs/backends/git/changeset.py
Show inline comments
 
@@ -144,97 +144,97 @@ class GitChangeset(BaseChangeset):
 
                    raise ChangesetError('%s have not been found' % curdir)
 

	
 
                # cache all items from the given traversed tree
 
                for item, stat, id in tree.iteritems():
 
                    if curdir:
 
                        name = '/'.join((curdir, item))
 
                    else:
 
                        name = item
 
                    self._paths[name] = id
 
                    self._stat_modes[name] = stat
 
            if not path in self._paths:
 
                raise NodeDoesNotExistError("There is no file nor directory "
 
                    "at the given path '%s' at revision %s"
 
                    % (path, safe_str(self.short_id)))
 
        return self._paths[path]
 

	
 
    def _get_kind(self, path):
 
        obj = self.repository._repo[self._get_id_for_path(path)]
 
        if isinstance(obj, objects.Blob):
 
            return NodeKind.FILE
 
        elif isinstance(obj, objects.Tree):
 
            return NodeKind.DIR
 

	
 
    def _get_filectx(self, path):
 
        path = self._fix_path(path)
 
        if self._get_kind(path) != NodeKind.FILE:
 
            raise ChangesetError("File does not exist for revision %s at "
 
                " '%s'" % (self.raw_id, path))
 
        return path
 

	
 
    def _get_file_nodes(self):
 
        return chain(*(t[2] for t in self.walk()))
 

	
 
    @LazyProperty
 
    def parents(self):
 
        """
 
        Returns list of parents changesets.
 
        """
 
        return [self.repository.get_changeset(parent)
 
                for parent in self._commit.parents]
 

	
 
    @LazyProperty
 
    def children(self):
 
        """
 
        Returns list of children changesets.
 
        """
 
        rev_filter = settings.GIT_REV_FILTER
 
        so, se = self.repository.run_git_command(
 
            "rev-list %s --children" % (rev_filter)
 
            ['rev-list', rev_filter, '--children']
 
        )
 

	
 
        children = []
 
        pat = re.compile(r'^%s' % self.raw_id)
 
        for l in so.splitlines():
 
            if pat.match(l):
 
                childs = l.split(' ')[1:]
 
                children.extend(childs)
 
        return [self.repository.get_changeset(cs) for cs in children]
 

	
 
    def next(self, branch=None):
 
        if branch and self.branch != branch:
 
            raise VCSError('Branch option used on changeset not belonging '
 
                           'to that branch')
 

	
 
        cs = self
 
        while True:
 
            try:
 
                next_ = cs.revision + 1
 
                next_rev = cs.repository.revisions[next_]
 
            except IndexError:
 
                raise ChangesetDoesNotExistError
 
            cs = cs.repository.get_changeset(next_rev)
 

	
 
            if not branch or branch == cs.branch:
 
                return cs
 

	
 
    def prev(self, branch=None):
 
        if branch and self.branch != branch:
 
            raise VCSError('Branch option used on changeset not belonging '
 
                           'to that branch')
 

	
 
        cs = self
 
        while True:
 
            try:
 
                prev_ = cs.revision - 1
 
                if prev_ < 0:
 
                    raise IndexError
 
                prev_rev = cs.repository.revisions[prev_]
 
            except IndexError:
 
                raise ChangesetDoesNotExistError
 
            cs = cs.repository.get_changeset(prev_rev)
 

	
 
            if not branch or branch == cs.branch:
 
                return cs
 

	
 
    def diff(self, ignore_whitespace=True, context=3):
 
        rev1 = self.parents[0] if self.parents else self.repository.EMPTY_CHANGESET
 
@@ -243,130 +243,130 @@ class GitChangeset(BaseChangeset):
 
                                    ignore_whitespace=ignore_whitespace,
 
                                    context=context))
 

	
 
    def get_file_mode(self, path):
 
        """
 
        Returns stat mode of the file at the given ``path``.
 
        """
 
        # ensure path is traversed
 
        path = safe_str(path)
 
        self._get_id_for_path(path)
 
        return self._stat_modes[path]
 

	
 
    def get_file_content(self, path):
 
        """
 
        Returns content of the file at given ``path``.
 
        """
 
        id = self._get_id_for_path(path)
 
        blob = self.repository._repo[id]
 
        return blob.as_pretty_string()
 

	
 
    def get_file_size(self, path):
 
        """
 
        Returns size of the file at given ``path``.
 
        """
 
        id = self._get_id_for_path(path)
 
        blob = self.repository._repo[id]
 
        return blob.raw_length()
 

	
 
    def get_file_changeset(self, path):
 
        """
 
        Returns last commit of the file at the given ``path``.
 
        """
 
        return self.get_file_history(path, limit=1)[0]
 

	
 
    def get_file_history(self, path, limit=None):
 
        """
 
        Returns history of file as reversed list of ``Changeset`` objects for
 
        which file at given ``path`` has been modified.
 

	
 
        TODO: This function now uses os underlying 'git' and 'grep' commands
 
        which is generally not good. Should be replaced with algorithm
 
        iterating commits.
 
        """
 
        self._get_filectx(path)
 
        cs_id = safe_str(self.id)
 
        f_path = safe_str(path)
 

	
 
        if limit:
 
            cmd = 'log -n %s --pretty="format: %%H" -s %s -- "%s"' % (
 
                      safe_int(limit, 0), cs_id, f_path)
 
            cmd = ['log', '-n', str(safe_int(limit, 0)),
 
                   '--pretty=format:%H', '-s', cs_id, '--', f_path]
 

	
 
        else:
 
            cmd = 'log --pretty="format: %%H" -s %s -- "%s"' % (
 
                      cs_id, f_path)
 
            cmd = ['log',
 
                   '--pretty=format:%H', '-s', cs_id, '--', f_path]
 
        so, se = self.repository.run_git_command(cmd)
 
        ids = re.findall(r'[0-9a-fA-F]{40}', so)
 
        return [self.repository.get_changeset(sha) for sha in ids]
 

	
 
    def get_file_history_2(self, path):
 
        """
 
        Returns history of file as reversed list of ``Changeset`` objects for
 
        which file at given ``path`` has been modified.
 

	
 
        """
 
        self._get_filectx(path)
 
        from dulwich.walk import Walker
 
        include = [self.id]
 
        walker = Walker(self.repository._repo.object_store, include,
 
                        paths=[path], max_entries=1)
 
        return [self.repository.get_changeset(sha)
 
                for sha in (x.commit.id for x in walker)]
 

	
 
    def get_file_annotate(self, path):
 
        """
 
        Returns a generator of four element tuples with
 
            lineno, sha, changeset lazy loader and line
 

	
 
        TODO: This function now uses os underlying 'git' command which is
 
        generally not good. Should be replaced with algorithm iterating
 
        commits.
 
        """
 
        cmd = 'blame -l --root -r %s -- "%s"' % (self.id, path)
 
        cmd = ['blame', '-l', '--root', '-r', self.id, '--', path]
 
        # -l     ==> outputs long shas (and we need all 40 characters)
 
        # --root ==> doesn't put '^' character for boundaries
 
        # -r sha ==> blames for the given revision
 
        so, se = self.repository.run_git_command(cmd)
 

	
 
        for i, blame_line in enumerate(so.split('\n')[:-1]):
 
            ln_no = i + 1
 
            sha, line = re.split(r' ', blame_line, 1)
 
            yield (ln_no, sha, lambda: self.repository.get_changeset(sha), line)
 

	
 
    def fill_archive(self, stream=None, kind='tgz', prefix=None,
 
                     subrepos=False):
 
        """
 
        Fills up given stream.
 

	
 
        :param stream: file like object.
 
        :param kind: one of following: ``zip``, ``tgz`` or ``tbz2``.
 
            Default: ``tgz``.
 
        :param prefix: name of root directory in archive.
 
            Default is repository name and changeset's raw_id joined with dash
 
            (``repo-tip.<KIND>``).
 
        :param subrepos: include subrepos in this archive.
 

	
 
        :raise ImproperArchiveTypeError: If given kind is wrong.
 
        :raise VcsError: If given stream is None
 

	
 
        """
 
        allowed_kinds = settings.ARCHIVE_SPECS.keys()
 
        if kind not in allowed_kinds:
 
            raise ImproperArchiveTypeError('Archive kind not supported use one'
 
                'of %s', allowed_kinds)
 

	
 
        if prefix is None:
 
            prefix = '%s-%s' % (self.repository.name, self.short_id)
 
        elif prefix.startswith('/'):
 
            raise VCSError("Prefix cannot start with leading slash")
 
        elif prefix.strip() == '':
 
            raise VCSError("Prefix cannot be empty")
 

	
 
        if kind == 'zip':
 
            frmt = 'zip'
 
        else:
 
            frmt = 'tar'
 
        _git_path = settings.GIT_EXECUTABLE_PATH
 
        cmd = '%s archive --format=%s --prefix=%s/ %s' % (_git_path,
 
                                                frmt, prefix, self.raw_id)
 
        if kind == 'tgz':
 
            cmd += ' | gzip -9'
 
@@ -426,98 +426,98 @@ class GitChangeset(BaseChangeset):
 

	
 
    def get_node(self, path):
 
        if isinstance(path, unicode):
 
            path = path.encode('utf-8')
 
        path = self._fix_path(path)
 
        if not path in self.nodes:
 
            try:
 
                id_ = self._get_id_for_path(path)
 
            except ChangesetError:
 
                raise NodeDoesNotExistError("Cannot find one of parents' "
 
                    "directories for a given path: %s" % path)
 

	
 
            _GL = lambda m: m and objects.S_ISGITLINK(m)
 
            if _GL(self._stat_modes.get(path)):
 
                node = SubModuleNode(path, url=None, changeset=id_,
 
                                     alias=self.repository.alias)
 
            else:
 
                obj = self.repository._repo.get_object(id_)
 

	
 
                if isinstance(obj, objects.Tree):
 
                    if path == '':
 
                        node = RootNode(changeset=self)
 
                    else:
 
                        node = DirNode(path, changeset=self)
 
                    node._tree = obj
 
                elif isinstance(obj, objects.Blob):
 
                    node = FileNode(path, changeset=self)
 
                    node._blob = obj
 
                else:
 
                    raise NodeDoesNotExistError("There is no file nor directory "
 
                        "at the given path '%s' at revision %s"
 
                        % (path, self.short_id))
 
            # cache node
 
            self.nodes[path] = node
 
        return self.nodes[path]
 

	
 
    @LazyProperty
 
    def affected_files(self):
 
        """
 
        Gets a fast accessible file changes for given changeset
 
        """
 
        added, modified, deleted = self._changes_cache
 
        return list(added.union(modified).union(deleted))
 

	
 
    @LazyProperty
 
    def _diff_name_status(self):
 
        output = []
 
        for parent in self.parents:
 
            cmd = 'diff --name-status %s %s --encoding=utf8' % (parent.raw_id,
 
                                                                self.raw_id)
 
            cmd = ['diff', '--name-status', parent.raw_id, self.raw_id,
 
                   '--encoding=utf8']
 
            so, se = self.repository.run_git_command(cmd)
 
            output.append(so.strip())
 
        return '\n'.join(output)
 

	
 
    @LazyProperty
 
    def _changes_cache(self):
 
        added = set()
 
        modified = set()
 
        deleted = set()
 
        _r = self.repository._repo
 

	
 
        parents = self.parents
 
        if not self.parents:
 
            parents = [EmptyChangeset()]
 
        for parent in parents:
 
            if isinstance(parent, EmptyChangeset):
 
                oid = None
 
            else:
 
                oid = _r[parent.raw_id].tree
 
            changes = _r.object_store.tree_changes(oid, _r[self.raw_id].tree)
 
            for (oldpath, newpath), (_, _), (_, _) in changes:
 
                if newpath and oldpath:
 
                    modified.add(newpath)
 
                elif newpath and not oldpath:
 
                    added.add(newpath)
 
                elif not newpath and oldpath:
 
                    deleted.add(oldpath)
 
        return added, modified, deleted
 

	
 
    def _get_paths_for_status(self, status):
 
        """
 
        Returns sorted list of paths for given ``status``.
 

	
 
        :param status: one of: *added*, *modified* or *deleted*
 
        """
 
        added, modified, deleted = self._changes_cache
 
        return sorted({
 
            'added': list(added),
 
            'modified': list(modified),
 
            'deleted': list(deleted)}[status]
 
        )
 

	
 
    @LazyProperty
 
    def added(self):
 
        """
 
        Returns list of added ``FileNode`` objects.
 
        """
 
        if not self.parents:
kallithea/lib/vcs/backends/git/repository.py
Show inline comments
 
# -*- coding: utf-8 -*-
 
"""
 
    vcs.backends.git.repository
 
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

	
 
    Git repository implementation.
 

	
 
    :created_on: Apr 8, 2010
 
    :copyright: (c) 2010-2011 by Marcin Kuzminski, Lukasz Balcerzak.
 
"""
 

	
 
import os
 
import re
 
import time
 
import errno
 
import urllib
 
import urllib2
 
import logging
 
import posixpath
 
import string
 
import sys
 
if sys.platform == "win32":
 
    from subprocess import list2cmdline
 
    def quote(s):
 
        return list2cmdline([s])
 
else:
 
    try:
 
        # Python <=2.7
 
        from pipes import quote
 
    except ImportError:
 
        # Python 3.3+
 
        from shlex import quote
 

	
 
from dulwich.objects import Tag
 
from dulwich.repo import Repo, NotGitRepository
 
from dulwich.config import ConfigFile
 

	
 
from kallithea.lib.vcs import subprocessio
 
from kallithea.lib.vcs.backends.base import BaseRepository, CollectionGenerator
 
from kallithea.lib.vcs.conf import settings
 

	
 
from kallithea.lib.vcs.exceptions import (
 
    BranchDoesNotExistError, ChangesetDoesNotExistError, EmptyRepositoryError,
 
    RepositoryError, TagAlreadyExistError, TagDoesNotExistError
 
)
 
from kallithea.lib.vcs.utils import safe_unicode, makedate, date_fromtimestamp
 
from kallithea.lib.vcs.utils.lazy import LazyProperty
 
from kallithea.lib.vcs.utils.ordered_dict import OrderedDict
 
from kallithea.lib.vcs.utils.paths import abspath, get_user_home
 

	
 
from kallithea.lib.vcs.utils.hgcompat import (
 
    hg_url, httpbasicauthhandler, httpdigestauthhandler
 
)
 

	
 
from .changeset import GitChangeset
 
from .inmemory import GitInMemoryChangeset
 
from .workdir import GitWorkdir
 

	
 
SHA_PATTERN = re.compile(r'^[[0-9a-fA-F]{12}|[0-9a-fA-F]{40}]$')
 

	
 
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)
 
        repo = self._get_repo(create, src_url, update_after_clone, bare)
 
        self.bare = repo.bare
 

	
 
    @property
 
    def _config_files(self):
 
        return [
 
@@ -90,116 +78,111 @@ class GitRepository(BaseRepository):
 
    @property
 
    def head(self):
 
        try:
 
            return self._repo.head()
 
        except KeyError:
 
            return None
 

	
 
    @property
 
    def _empty(self):
 
        """
 
        Checks if repository is empty ie. without any changesets
 
        """
 

	
 
        try:
 
            self.revisions[0]
 
        except (KeyError, IndexError):
 
            return True
 
        return False
 

	
 
    @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()
 

	
 
    @classmethod
 
    def _run_git_command(cls, cmd, **opts):
 
        """
 
        Runs given ``cmd`` as git command and returns tuple
 
        (stdout, stderr).
 

	
 
        :param cmd: git command to be executed
 
        :param opts: env options to pass into Subprocess command
 
        """
 

	
 
        if '_bare' in opts:
 
            _copts = []
 
            del opts['_bare']
 
        else:
 
            _copts = ['-c', 'core.quotepath=false', ]
 
        safe_call = False
 
        if '_safe' in opts:
 
            #no exc on failure
 
            del opts['_safe']
 
            safe_call = True
 

	
 
        _str_cmd = False
 
        if isinstance(cmd, basestring):
 
            cmd = [cmd]
 
            _str_cmd = True
 
        assert isinstance(cmd, list), cmd
 

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

	
 
        _git_path = settings.GIT_EXECUTABLE_PATH
 
        cmd = [_git_path] + _copts + cmd
 
        if _str_cmd:
 
            cmd = ' '.join(cmd)
 

	
 
        try:
 
            _opts = dict(
 
                env=gitenv,
 
                shell=True,
 
                shell=False,
 
            )
 
            _opts.update(opts)
 
            p = subprocessio.SubprocessIOChunker(cmd, **_opts)
 
        except (EnvironmentError, OSError), err:
 
            tb_err = ("Couldn't run git command (%s).\n"
 
                      "Original error was:%s\n" % (cmd, err))
 
            log.error(tb_err)
 
            if safe_call:
 
                return '', err
 
            else:
 
                raise RepositoryError(tb_err)
 

	
 
        return ''.join(p.output), ''.join(p.error)
 

	
 
    def run_git_command(self, cmd):
 
        opts = {}
 
        if os.path.isdir(self.path):
 
            opts['cwd'] = self.path
 
        return self._run_git_command(cmd, **opts)
 

	
 
    @classmethod
 
    def _check_url(cls, url):
 
        """
 
        Function will check given url and try to verify if it's a valid
 
        link. Sometimes it may happened that git will issue basic
 
        auth request that can cause whole API to hang when used from python
 
        or other external calls.
 

	
 
        On failures it'll raise urllib2.HTTPError, exception is also thrown
 
        when the return code is non 200
 
        """
 

	
 
        # check first if it's not an local url
 
        if os.path.isdir(url) or url.startswith('file:'):
 
            return True
 

	
 
        if '+' in url[:url.find('://')]:
 
            url = url[url.find('+') + 1:]
 

	
 
        handlers = []
 
        url_obj = hg_url(url)
 
        test_uri, authinfo = url_obj.authinfo()
 
        url_obj.passwd = '*****'
 
        cleaned_uri = str(url_obj)
 

	
 
        if not test_uri.endswith('info/refs'):
 
            test_uri = test_uri.rstrip('/') + '/info/refs'
 

	
 
@@ -223,97 +206,97 @@ class GitRepository(BaseRepository):
 
            resp = o.open(req)
 
            if resp.code != 200:
 
                raise Exception('Return Code is not 200')
 
        except Exception, e:
 
            # means it cannot be cloned
 
            raise urllib2.URLError("[%s] org_exc: %s" % (cleaned_uri, e))
 

	
 
        # now detect if it's proper git repo
 
        gitdata = resp.read()
 
        if not 'service=git-upload-pack' in gitdata:
 
            raise urllib2.URLError(
 
                "url [%s] does not look like an git" % (cleaned_uri))
 

	
 
        return True
 

	
 
    def _get_repo(self, create, src_url=None, update_after_clone=False,
 
                  bare=False):
 
        if create and os.path.exists(self.path):
 
            raise RepositoryError("Location already exist")
 
        if src_url and not create:
 
            raise RepositoryError("Create should be set to True if src_url is "
 
                                  "given (clone operation creates repository)")
 
        try:
 
            if create and src_url:
 
                GitRepository._check_url(src_url)
 
                self.clone(src_url, update_after_clone, bare)
 
                return Repo(self.path)
 
            elif create:
 
                os.makedirs(self.path)
 
                if bare:
 
                    return Repo.init_bare(self.path)
 
                else:
 
                    return Repo.init(self.path)
 
            else:
 
                return self._repo
 
        except (NotGitRepository, OSError), err:
 
            raise RepositoryError(err)
 

	
 
    def _get_all_revisions(self):
 
        # we must check if this repo is not empty, since later command
 
        # fails if it is. And it's cheaper to ask than throw the subprocess
 
        # errors
 
        try:
 
            self._repo.head()
 
        except KeyError:
 
            return []
 

	
 
        rev_filter = settings.GIT_REV_FILTER
 
        cmd = 'rev-list %s --reverse --date-order' % (rev_filter)
 
        cmd = ['rev-list', rev_filter, '--reverse', '--date-order']
 
        try:
 
            so, se = self.run_git_command(cmd)
 
        except RepositoryError:
 
            # Can be raised for empty repositories
 
            return []
 
        return so.splitlines()
 

	
 
    def _get_all_revisions2(self):
 
        #alternate implementation using dulwich
 
        includes = [x[1][0] for x in self._parsed_refs.iteritems()
 
                    if x[1][1] != 'T']
 
        return [c.commit.id for c in self._repo.get_walker(include=includes)]
 

	
 
    def _get_revision(self, revision):
 
        """
 
        For git backend we always return integer here. This way we ensure
 
        that changeset's revision attribute would become integer.
 
        """
 

	
 
        is_null = lambda o: len(o) == revision.count('0')
 

	
 
        if self._empty:
 
            raise EmptyRepositoryError("There are no changesets yet")
 

	
 
        if revision in (None, '', 'tip', 'HEAD', 'head', -1):
 
            return self.revisions[-1]
 

	
 
        is_bstr = isinstance(revision, (str, unicode))
 
        if ((is_bstr and revision.isdigit() and len(revision) < 12)
 
            or isinstance(revision, int) or is_null(revision)):
 
            try:
 
                revision = self.revisions[int(revision)]
 
            except IndexError:
 
                msg = ("Revision %s does not exist for %s" % (revision, self))
 
                raise ChangesetDoesNotExistError(msg)
 

	
 
        elif is_bstr:
 
            # get by branch/tag name
 
            _ref_revision = self._parsed_refs.get(revision)
 
            if _ref_revision:  # and _ref_revision[1] in ['H', 'RH', 'T']:
 
                return _ref_revision[0]
 

	
 
            _tags_shas = self.tags.values()
 
            # maybe it's a tag ? we don't have them in self.revisions
 
            if revision in _tags_shas:
 
                return _tags_shas[_tags_shas.index(revision)]
 

	
 
            elif not SHA_PATTERN.match(revision) or revision not in self.revisions:
 
@@ -506,235 +489,225 @@ class GitRepository(BaseRepository):
 
                        heads[n] = val
 

	
 
        return heads if reverse else dict((y, x) for x, y in heads.iteritems())
 

	
 
    def get_changeset(self, revision=None):
 
        """
 
        Returns ``GitChangeset`` object representing commit from git repository
 
        at the given revision or head (most recent commit) if None given.
 
        """
 
        if isinstance(revision, GitChangeset):
 
            return revision
 
        revision = self._get_revision(revision)
 
        changeset = GitChangeset(repository=self, revision=revision)
 
        return changeset
 

	
 
    def get_changesets(self, start=None, end=None, start_date=None,
 
           end_date=None, branch_name=None, reverse=False):
 
        """
 
        Returns iterator of ``GitChangeset`` objects from start to end (both
 
        are inclusive), in ascending date order (unless ``reverse`` is set).
 

	
 
        :param start: changeset ID, as str; first returned changeset
 
        :param end: changeset ID, as str; last returned changeset
 
        :param start_date: if specified, changesets with commit date less than
 
          ``start_date`` would be filtered out from returned set
 
        :param end_date: if specified, changesets with commit date greater than
 
          ``end_date`` would be filtered out from returned set
 
        :param branch_name: if specified, changesets not reachable from given
 
          branch would be filtered out from returned set
 
        :param reverse: if ``True``, returned generator would be reversed
 
          (meaning that returned changesets would have descending date order)
 

	
 
        :raise BranchDoesNotExistError: If given ``branch_name`` does not
 
            exist.
 
        :raise ChangesetDoesNotExistError: If changeset for given ``start`` or
 
          ``end`` could not be found.
 

	
 
        """
 
        if branch_name and branch_name not in self.branches:
 
            raise BranchDoesNotExistError("Branch '%s' not found" \
 
                                          % branch_name)
 
        # actually we should check now if it's not an empty repo to not spaw
 
        # subprocess commands
 
        if self._empty:
 
            raise EmptyRepositoryError("There are no changesets yet")
 

	
 
        # %H at format means (full) commit hash, initial hashes are retrieved
 
        # in ascending date order
 
        cmd_template = 'log --date-order --reverse --pretty=format:"%H"'
 
        cmd_params = {}
 
        cmd = ['log', '--date-order', '--reverse', '--pretty=format:%H']
 
        if start_date:
 
            cmd_template += ' --since "$since"'
 
            cmd_params['since'] = start_date.strftime('%m/%d/%y %H:%M:%S')
 
            cmd += ['--since', start_date.strftime('%m/%d/%y %H:%M:%S')]
 
        if end_date:
 
            cmd_template += ' --until "$until"'
 
            cmd_params['until'] = end_date.strftime('%m/%d/%y %H:%M:%S')
 
            cmd += ['--until', end_date.strftime('%m/%d/%y %H:%M:%S')]
 
        if branch_name:
 
            cmd_template += ' $branch_name'
 
            cmd_params['branch_name'] = branch_name
 
            cmd.append(branch_name)
 
        else:
 
            rev_filter = settings.GIT_REV_FILTER
 
            cmd_template += ' %s' % (rev_filter)
 
            cmd.append(settings.GIT_REV_FILTER)
 

	
 
        cmd = string.Template(cmd_template).safe_substitute(**cmd_params)
 
        revs = self.run_git_command(cmd)[0].splitlines()
 
        start_pos = 0
 
        end_pos = len(revs)
 
        if start:
 
            _start = self._get_revision(start)
 
            try:
 
                start_pos = revs.index(_start)
 
            except ValueError:
 
                pass
 

	
 
        if end is not None:
 
            _end = self._get_revision(end)
 
            try:
 
                end_pos = revs.index(_end)
 
            except ValueError:
 
                pass
 

	
 
        if None not in [start, end] and start_pos > end_pos:
 
            raise RepositoryError('start cannot be after end')
 

	
 
        if end_pos is not None:
 
            end_pos += 1
 

	
 
        revs = revs[start_pos:end_pos]
 
        if reverse:
 
            revs = reversed(revs)
 
        return CollectionGenerator(self, revs)
 

	
 
    def get_diff(self, rev1, rev2, path=None, ignore_whitespace=False,
 
                 context=3):
 
        """
 
        Returns (git like) *diff*, as plain text. Shows changes introduced by
 
        ``rev2`` since ``rev1``.
 

	
 
        :param rev1: Entry point from which diff is shown. Can be
 
          ``self.EMPTY_CHANGESET`` - in this case, patch showing all
 
          the changes since empty state of the repository until ``rev2``
 
        :param rev2: Until which revision changes should be shown.
 
        :param ignore_whitespace: If set to ``True``, would not show whitespace
 
          changes. Defaults to ``False``.
 
        :param context: How many lines before/after changed lines should be
 
          shown. Defaults to ``3``.
 
        """
 
        flags = ['-U%s' % context, '--full-index', '--binary', '-p', '-M', '--abbrev=40']
 
        if ignore_whitespace:
 
            flags.append('-w')
 

	
 
        if hasattr(rev1, 'raw_id'):
 
            rev1 = getattr(rev1, 'raw_id')
 

	
 
        if hasattr(rev2, 'raw_id'):
 
            rev2 = getattr(rev2, 'raw_id')
 

	
 
        if rev1 == self.EMPTY_CHANGESET:
 
            rev2 = self.get_changeset(rev2).raw_id
 
            cmd = ' '.join(['show'] + flags + [rev2])
 
            cmd = ['show'] + flags + [rev2]
 
        else:
 
            rev1 = self.get_changeset(rev1).raw_id
 
            rev2 = self.get_changeset(rev2).raw_id
 
            cmd = ' '.join(['diff'] + flags + [rev1, rev2])
 
            cmd = ['diff'] + flags + [rev1, rev2]
 

	
 
        if path:
 
            cmd += ' -- "%s"' % path
 
            cmd += ['--', path]
 

	
 
        stdout, stderr = self.run_git_command(cmd)
 
        # TODO: don't ignore stderr
 
        # If we used 'show' command, strip first few lines (until actual diff
 
        # starts)
 
        if rev1 == self.EMPTY_CHANGESET:
 
            parts = stdout.split('\ndiff ', 1)
 
            if len(parts) > 1:
 
                stdout = 'diff ' + parts[1]
 
        return stdout
 

	
 
    @LazyProperty
 
    def in_memory_changeset(self):
 
        """
 
        Returns ``GitInMemoryChangeset`` object for this repository.
 
        """
 
        return GitInMemoryChangeset(self)
 

	
 
    def clone(self, url, update_after_clone=True, bare=False):
 
        """
 
        Tries to clone changes from external location.
 

	
 
        :param update_after_clone: If set to ``False``, git won't checkout
 
          working directory
 
        :param bare: If set to ``True``, repository would be cloned into
 
          *bare* git repository (no working directory at all).
 
        """
 
        url = self._get_url(url)
 
        cmd = ['clone', '-q']
 
        if bare:
 
            cmd.append('--bare')
 
        elif not update_after_clone:
 
            cmd.append('--no-checkout')
 
        cmd += ['--', quote(url), quote(self.path)]
 
        cmd = ' '.join(cmd)
 
        cmd += ['--', url, self.path]
 
        # If error occurs run_git_command raises RepositoryError already
 
        self.run_git_command(cmd)
 

	
 
    def pull(self, url):
 
        """
 
        Tries to pull changes from external location.
 
        """
 
        url = self._get_url(url)
 
        cmd = ['pull', "--ff-only", quote(url)]
 
        cmd = ' '.join(cmd)
 
        cmd = ['pull', '--ff-only', url]
 
        # If error occurs run_git_command raises RepositoryError already
 
        self.run_git_command(cmd)
 

	
 
    def fetch(self, url):
 
        """
 
        Tries to pull changes from external location.
 
        """
 
        url = self._get_url(url)
 
        so, se = self.run_git_command('ls-remote -h %s' % quote(url))
 
        refs = []
 
        so, se = self.run_git_command(['ls-remote', '-h', url])
 
        cmd = ['fetch', url, '--']
 
        for line in (x for x in so.splitlines()):
 
            sha, ref = line.split('\t')
 
            refs.append(ref)
 
        refs = ' '.join(('+%s:%s' % (r, r) for r in refs))
 
        cmd = '''fetch %s -- %s''' % (quote(url), refs)
 
            cmd.append('+%s:%s' % (ref, ref))
 
        self.run_git_command(cmd)
 

	
 
    def _update_server_info(self):
 
        """
 
        runs gits update-server-info command in this repo instance
 
        """
 
        from dulwich.server import update_server_info
 
        try:
 
            update_server_info(self._repo)
 
        except OSError, e:
 
            if e.errno != errno.ENOENT:
 
                raise
 
            # Workaround for dulwich crashing on for example its own dulwich/tests/data/repos/simple_merge.git/info/refs.lock
 
            log.error('Ignoring error running update-server-info: %s', e)
 

	
 
    @LazyProperty
 
    def workdir(self):
 
        """
 
        Returns ``Workdir`` instance for this repository.
 
        """
 
        return GitWorkdir(self)
 

	
 
    def get_config_value(self, section, name, config_file=None):
 
        """
 
        Returns configuration value for a given [``section``] and ``name``.
 

	
 
        :param section: Section we want to retrieve value from
 
        :param name: Name of configuration we want to retrieve
 
        :param config_file: A path to file which should be used to retrieve
 
          configuration from (might also be a list of file paths)
 
        """
 
        if config_file is None:
 
            config_file = []
 
        elif isinstance(config_file, basestring):
 
            config_file = [config_file]
 

	
 
        def gen_configs():
 
            for path in config_file + self._config_files:
 
                try:
 
                    yield ConfigFile.from_path(path)
 
                except (IOError, OSError, ValueError):
 
                    continue
 

	
 
        for config in gen_configs():
 
            try:
 
                return config.get(section, name)
 
            except KeyError:
 
                continue
kallithea/lib/vcs/subprocessio.py
Show inline comments
 
@@ -297,101 +297,99 @@ class SubprocessIOChunker(object):
 
       does not block the parallel inpipe reading occurring parallel thread.)
 

	
 
    The purpose of the object is to allow us to wrap subprocess interactions into
 
    an iterable that can be passed to a WSGI server as the application's return
 
    value. Because of stream-processing-ability, WSGI does not have to read ALL
 
    of the subprocess's output and buffer it, before handing it to WSGI server for
 
    HTTP response. Instead, the class initializer reads just a bit of the stream
 
    to figure out if error occurred or likely to occur and if not, just hands the
 
    further iteration over subprocess output to the server for completion of HTTP
 
    response.
 

	
 
    The real or perceived subprocess error is trapped and raised as one of
 
    EnvironmentError family of exceptions
 

	
 
    Example usage:
 
    #    try:
 
    #        answer = SubprocessIOChunker(
 
    #            cmd,
 
    #            input,
 
    #            buffer_size = 65536,
 
    #            chunk_size = 4096
 
    #            )
 
    #    except (EnvironmentError) as e:
 
    #        print str(e)
 
    #        raise e
 
    #
 
    #    return answer
 

	
 

	
 
    """
 

	
 
    def __init__(self, cmd, inputstream=None, buffer_size=65536,
 
                 chunk_size=4096, starting_values=[], **kwargs):
 
        """
 
        Initializes SubprocessIOChunker
 

	
 
        :param cmd: A Subprocess.Popen style "cmd". Can be string or array of strings
 
        :param inputstream: (Default: None) A file-like, string, or file pointer.
 
        :param buffer_size: (Default: 65536) A size of total buffer per stream in bytes.
 
        :param chunk_size: (Default: 4096) A max size of a chunk. Actual chunk may be smaller.
 
        :param starting_values: (Default: []) An array of strings to put in front of output que.
 
        """
 

	
 
        if inputstream:
 
            input_streamer = StreamFeeder(inputstream)
 
            input_streamer.start()
 
            inputstream = input_streamer.output
 

	
 
        _shell = kwargs.get('shell', True)
 
        if isinstance(cmd, (list, tuple)):
 
            cmd = ' '.join(cmd)
 
        # Note: fragile cmd mangling has been removed for use in Kallithea
 
        assert isinstance(cmd, list), cmd
 

	
 
        kwargs['shell'] = _shell
 
        _p = subprocess.Popen(cmd, bufsize=-1,
 
                              stdin=inputstream,
 
                              stdout=subprocess.PIPE,
 
                              stderr=subprocess.PIPE,
 
                              **kwargs)
 

	
 
        bg_out = BufferedGenerator(_p.stdout, buffer_size, chunk_size,
 
                                   starting_values)
 
        bg_err = BufferedGenerator(_p.stderr, 16000, 1, bottomless=True)
 

	
 
        while not bg_out.done_reading and not bg_out.reading_paused and not bg_err.length:
 
            # doing this until we reach either end of file, or end of buffer.
 
            bg_out.data_added_event.wait(1)
 
            bg_out.data_added_event.clear()
 

	
 
        # at this point it's still ambiguous if we are done reading or just full buffer.
 
        # Either way, if error (returned by ended process, or implied based on
 
        # presence of stuff in stderr output) we error out.
 
        # Else, we are happy.
 
        _returncode = _p.poll()
 
        if _returncode or (_returncode is None and bg_err.length):
 
            try:
 
                _p.terminate()
 
            except Exception:
 
                pass
 
            bg_out.stop()
 
            out = ''.join(bg_out)
 
            bg_err.stop()
 
            err = ''.join(bg_err)
 
            if (err.strip() == 'fatal: The remote end hung up unexpectedly' and
 
                out.startswith('0034shallow ')):
 
                # hack inspired by https://github.com/schacon/grack/pull/7
 
                bg_out = iter([out])
 
                _p = None
 
            elif err:
 
                raise EnvironmentError(
 
                    "Subprocess exited due to an error:\n" + err)
 
            else:
 
                raise EnvironmentError(
 
                    "Subprocess exited with non 0 ret code:%s" % _returncode)
 
        self.process = _p
 
        self.output = bg_out
 
        self.error = bg_err
 
        self.inputstream = inputstream
 

	
 
    def __iter__(self):
 
        return self
 

	
kallithea/tests/vcs/test_git.py
Show inline comments
 
@@ -637,122 +637,122 @@ class GitSpecificTest(unittest.TestCase)
 
            changeset.added
 

	
 
    def test_error_is_raised_for_removed_if_diff_name_status_is_wrong(self):
 
        repo = mock.MagicMock()
 
        changeset = GitChangeset(repo, 'foobar')
 
        changeset._diff_name_status = 'foobar'
 
        with self.assertRaises(VCSError):
 
            changeset.added
 

	
 

	
 
class GitSpecificWithRepoTest(_BackendTestMixin, unittest.TestCase):
 
    backend_alias = 'git'
 

	
 
    @classmethod
 
    def _get_commits(cls):
 
        return [
 
            {
 
                'message': 'Initial',
 
                'author': 'Joe Doe <joe.doe@example.com>',
 
                'date': datetime.datetime(2010, 1, 1, 20),
 
                'added': [
 
                    FileNode('foobar/static/js/admin/base.js', content='base'),
 
                    FileNode('foobar/static/admin', content='admin',
 
                        mode=0120000), # this is a link
 
                    FileNode('foo', content='foo'),
 
                ],
 
            },
 
            {
 
                'message': 'Second',
 
                'author': 'Joe Doe <joe.doe@example.com>',
 
                'date': datetime.datetime(2010, 1, 1, 22),
 
                'added': [
 
                    FileNode('foo2', content='foo2'),
 
                ],
 
            },
 
        ]
 

	
 
    def test_paths_slow_traversing(self):
 
        cs = self.repo.get_changeset()
 
        self.assertEqual(cs.get_node('foobar').get_node('static').get_node('js')
 
            .get_node('admin').get_node('base.js').content, 'base')
 

	
 
    def test_paths_fast_traversing(self):
 
        cs = self.repo.get_changeset()
 
        self.assertEqual(cs.get_node('foobar/static/js/admin/base.js').content,
 
            'base')
 

	
 
    def test_workdir_get_branch(self):
 
        self.repo.run_git_command('checkout -b production')
 
        self.repo.run_git_command(['checkout', '-b', 'production'])
 
        # Regression test: one of following would fail if we don't check
 
        # .git/HEAD file
 
        self.repo.run_git_command('checkout production')
 
        self.repo.run_git_command(['checkout', 'production'])
 
        self.assertEqual(self.repo.workdir.get_branch(), 'production')
 
        self.repo.run_git_command('checkout master')
 
        self.repo.run_git_command(['checkout', 'master'])
 
        self.assertEqual(self.repo.workdir.get_branch(), 'master')
 

	
 
    def test_get_diff_runs_git_command_with_hashes(self):
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo.get_diff(0, 1)
 
        self.repo.run_git_command.assert_called_once_with(
 
          'diff -U%s --full-index --binary -p -M --abbrev=40 %s %s' %
 
            (3, self.repo._get_revision(0), self.repo._get_revision(1)))
 
            ['diff', '-U3', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(0), self.repo._get_revision(1)])
 

	
 
    def test_get_diff_runs_git_command_with_str_hashes(self):
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo.get_diff(self.repo.EMPTY_CHANGESET, 1)
 
        self.repo.run_git_command.assert_called_once_with(
 
            'show -U%s --full-index --binary -p -M --abbrev=40 %s' %
 
            (3, self.repo._get_revision(1)))
 
            ['show', '-U3', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(1)])
 

	
 
    def test_get_diff_runs_git_command_with_path_if_its_given(self):
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo.get_diff(0, 1, 'foo')
 
        self.repo.run_git_command.assert_called_once_with(
 
          'diff -U%s --full-index --binary -p -M --abbrev=40 %s %s -- "foo"'
 
            % (3, self.repo._get_revision(0), self.repo._get_revision(1)))
 
            ['diff', '-U3', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
 

	
 

	
 
class GitRegressionTest(_BackendTestMixin, unittest.TestCase):
 
    backend_alias = 'git'
 

	
 
    @classmethod
 
    def _get_commits(cls):
 
        return [
 
            {
 
                'message': 'Initial',
 
                'author': 'Joe Doe <joe.doe@example.com>',
 
                'date': datetime.datetime(2010, 1, 1, 20),
 
                'added': [
 
                    FileNode('bot/__init__.py', content='base'),
 
                    FileNode('bot/templates/404.html', content='base'),
 
                    FileNode('bot/templates/500.html', content='base'),
 
                ],
 
            },
 
            {
 
                'message': 'Second',
 
                'author': 'Joe Doe <joe.doe@example.com>',
 
                'date': datetime.datetime(2010, 1, 1, 22),
 
                'added': [
 
                    FileNode('bot/build/migrations/1.py', content='foo2'),
 
                    FileNode('bot/build/migrations/2.py', content='foo2'),
 
                    FileNode('bot/build/static/templates/f.html', content='foo2'),
 
                    FileNode('bot/build/static/templates/f1.html', content='foo2'),
 
                    FileNode('bot/build/templates/err.html', content='foo2'),
 
                    FileNode('bot/build/templates/err2.html', content='foo2'),
 
                ],
 
            },
 
        ]
 

	
 
    def test_similar_paths(self):
 
        cs = self.repo.get_changeset()
 
        paths = lambda *n:[x.path for x in n]
 
        self.assertEqual(paths(*cs.get_nodes('bot')), ['bot/build', 'bot/templates', 'bot/__init__.py'])
 
        self.assertEqual(paths(*cs.get_nodes('bot/build')), ['bot/build/migrations', 'bot/build/static', 'bot/build/templates'])
 
        self.assertEqual(paths(*cs.get_nodes('bot/build/static')), ['bot/build/static/templates'])
 
        # this get_nodes below causes troubles !
 
        self.assertEqual(paths(*cs.get_nodes('bot/build/static/templates')), ['bot/build/static/templates/f.html', 'bot/build/static/templates/f1.html'])
 
        self.assertEqual(paths(*cs.get_nodes('bot/build/templates')), ['bot/build/templates/err.html', 'bot/build/templates/err2.html'])
 
        self.assertEqual(paths(*cs.get_nodes('bot/templates/')), ['bot/templates/404.html', 'bot/templates/500.html'])
 

	
 
if __name__ == '__main__':
 
    unittest.main()
0 comments (0 inline, 0 general)