Changeset - 55d2b08d9c44
[Not reviewed]
stable
0 4 0
Branko Majic (branko) - 8 years ago 2018-02-09 18:12:19
branko@majic.rs
vcs: sanitize diff context values (Issue #306)

- Updated Git repository implementation to ensure context falls within
0 to 2**31-1 range (inclusive) when fetching a diff.
- Added tests for Git repositories for checking passed-in negative and
overflowing contexts (for the --unified option).
- Updated Mercurial repository implementation to ensure context is not
negative when fetching a diff.
- Added tests for Mercurial repositories for checking passed-in
negative context (for the --unified option).
4 files changed with 91 insertions and 2 deletions:
0 comments (0 inline, 0 general)
kallithea/lib/vcs/backends/git/repository.py
Show inline comments
 
@@ -398,337 +398,359 @@ class GitRepository(BaseRepository):
 
        if not self.revisions:
 
            return {}
 

	
 
        sortkey = lambda ctx: ctx[0]
 
        _tags = [(x[0], x[1][0])
 
                 for x in self._parsed_refs.iteritems() if x[1][1] == 'T']
 
        return OrderedDict(sorted(_tags, key=sortkey, reverse=True))
 

	
 
    def tag(self, name, user, revision=None, message=None, date=None,
 
            **kwargs):
 
        """
 
        Creates and returns a tag for the given ``revision``.
 

	
 
        :param name: name for new tag
 
        :param user: full username, i.e.: "Joe Doe <joe.doe@example.com>"
 
        :param revision: changeset id for which new tag would be created
 
        :param message: message of the tag's commit
 
        :param date: date of tag's commit
 

	
 
        :raises TagAlreadyExistError: if tag with same name already exists
 
        """
 
        if name in self.tags:
 
            raise TagAlreadyExistError("Tag %s already exists" % name)
 
        changeset = self.get_changeset(revision)
 
        message = message or "Added tag %s for commit %s" % (name,
 
            changeset.raw_id)
 
        self._repo.refs["refs/tags/%s" % name] = changeset._commit.id
 

	
 
        self._parsed_refs = self._get_parsed_refs()
 
        self.tags = self._get_tags()
 
        return changeset
 

	
 
    def remove_tag(self, name, user, message=None, date=None):
 
        """
 
        Removes tag with the given ``name``.
 

	
 
        :param name: name of the tag to be removed
 
        :param user: full username, i.e.: "Joe Doe <joe.doe@example.com>"
 
        :param message: message of the tag's removal commit
 
        :param date: date of tag's removal commit
 

	
 
        :raises TagDoesNotExistError: if tag with given name does not exists
 
        """
 
        if name not in self.tags:
 
            raise TagDoesNotExistError("Tag %s does not exist" % name)
 
        tagpath = posixpath.join(self._repo.refs.path, 'refs', 'tags', name)
 
        try:
 
            os.remove(tagpath)
 
            self._parsed_refs = self._get_parsed_refs()
 
            self.tags = self._get_tags()
 
        except OSError as e:
 
            raise RepositoryError(e.strerror)
 

	
 
    @LazyProperty
 
    def bookmarks(self):
 
        """
 
        Gets bookmarks for this repository
 
        """
 
        return {}
 

	
 
    @LazyProperty
 
    def _parsed_refs(self):
 
        return self._get_parsed_refs()
 

	
 
    def _get_parsed_refs(self):
 
        # cache the property
 
        _repo = self._repo
 
        refs = _repo.get_refs()
 
        keys = [('refs/heads/', 'H'),
 
                ('refs/remotes/origin/', 'RH'),
 
                ('refs/tags/', 'T')]
 
        _refs = {}
 
        for ref, sha in refs.iteritems():
 
            for k, type_ in keys:
 
                if ref.startswith(k):
 
                    _key = ref[len(k):]
 
                    if type_ == 'T':
 
                        obj = _repo.get_object(sha)
 
                        if isinstance(obj, Tag):
 
                            sha = _repo.get_object(sha).object[1]
 
                    _refs[_key] = [sha, type_]
 
                    break
 
        return _refs
 

	
 
    def _heads(self, reverse=False):
 
        refs = self._repo.get_refs()
 
        heads = {}
 

	
 
        for key, val in refs.items():
 
            for ref_key in ['refs/heads/', 'refs/remotes/origin/']:
 
                if key.startswith(ref_key):
 
                    n = key[len(ref_key):]
 
                    if n not in ['HEAD']:
 
                        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 = ['log', '--date-order', '--reverse', '--pretty=format:%H']
 
        if start_date:
 
            cmd += ['--since', start_date.strftime('%m/%d/%y %H:%M:%S')]
 
        if end_date:
 
            cmd += ['--until', end_date.strftime('%m/%d/%y %H:%M:%S')]
 
        if branch_name:
 
            cmd.append(branch_name)
 
        else:
 
            cmd.append(settings.GIT_REV_FILTER)
 

	
 
        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``.
 
          shown. Defaults to ``3``. Due to limitations in Git, if
 
          value passed-in is greater than ``2**31-1``
 
          (``2147483647``), it will be set to ``2147483647``
 
          instead. If negative value is passed-in, it will be set to
 
          ``0`` instead.
 
        """
 

	
 
        # Git internally uses a signed long int for storing context
 
        # size (number of lines to show before and after the
 
        # differences). This can result in integer overflow, so we
 
        # ensure the requested context is smaller by one than the
 
        # number that would cause the overflow. It is highly unlikely
 
        # that a single file will contain that many lines, so this
 
        # kind of change should not cause any realistic consequences.
 
        overflowed_long_int = 2**31
 

	
 
        if context >= overflowed_long_int:
 
            context = overflowed_long_int-1
 

	
 
        # Negative context values make no sense, and will result in
 
        # errors. Ensure this does not happen.
 
        if context < 0:
 
            context = 0
 

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

	
 
        if 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 += ['--', 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', 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', url])
 
        cmd = ['fetch', url, '--']
 
        for line in (x for x in so.splitlines()):
 
            sha, ref = line.split('\t')
 
            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 as e:
 
            if e.errno not in [errno.ENOENT, errno.EROFS]:
 
                raise
 
            # Workaround for dulwich crashing on for example its own dulwich/tests/data/repos/simple_merge.git/info/refs.lock
 
            log.error('Ignoring %s running update-server-info: %s', type(e).__name__, 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
 
        return None
 

	
 
    def get_user_name(self, config_file=None):
 
        """
 
        Returns user's name from global configuration file.
 

	
 
        :param config_file: A path to file which should be used to retrieve
 
          configuration from (might also be a list of file paths)
 
        """
 
        return self.get_config_value('user', 'name', config_file)
 

	
 
    def get_user_email(self, config_file=None):
 
        """
 
        Returns user's email from global configuration file.
 

	
 
        :param config_file: A path to file which should be used to retrieve
 
          configuration from (might also be a list of file paths)
 
        """
 
        return self.get_config_value('user', 'email', config_file)
kallithea/lib/vcs/backends/hg/repository.py
Show inline comments
 
@@ -55,386 +55,393 @@ class MercurialRepository(BaseRepository
 
        ``repo_path``.
 

	
 
        :param repo_path: local path of the repository
 
        :param create=False: if set to True, would try to create repository if
 
           it does not exist rather than raising exception
 
        :param baseui=None: user data
 
        :param src_url=None: would try to clone repository from given location
 
        :param update_after_clone=False: sets update of working copy after
 
          making a clone
 
        """
 

	
 
        if not isinstance(repo_path, str):
 
            raise VCSError('Mercurial backend requires repository path to '
 
                           'be instance of <str> got %s instead' %
 
                           type(repo_path))
 

	
 
        self.path = abspath(repo_path)
 
        self.baseui = baseui or ui.ui()
 
        # We've set path and ui, now we can set _repo itself
 
        self._repo = self._get_repo(create, src_url, update_after_clone)
 

	
 
    @property
 
    def _empty(self):
 
        """
 
        Checks if repository is empty ie. without any changesets
 
        """
 
        # TODO: Following raises errors when using InMemoryChangeset...
 
        # return len(self._repo.changelog) == 0
 
        return len(self.revisions) == 0
 

	
 
    @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()
 

	
 
    @LazyProperty
 
    def name(self):
 
        return os.path.basename(self.path)
 

	
 
    @LazyProperty
 
    def branches(self):
 
        return self._get_branches()
 

	
 
    @LazyProperty
 
    def closed_branches(self):
 
        return self._get_branches(normal=False, closed=True)
 

	
 
    @LazyProperty
 
    def allbranches(self):
 
        """
 
        List all branches, including closed branches.
 
        """
 
        return self._get_branches(closed=True)
 

	
 
    def _get_branches(self, normal=True, closed=False):
 
        """
 
        Gets branches for this repository
 
        Returns only not closed branches by default
 

	
 
        :param closed: return also closed branches for mercurial
 
        :param normal: return also normal branches
 
        """
 

	
 
        if self._empty:
 
            return {}
 

	
 
        bt = OrderedDict()
 
        for bn, _heads, tip, isclosed in sorted(self._repo.branchmap().iterbranches()):
 
            if isclosed:
 
                if closed:
 
                    bt[safe_unicode(bn)] = hex(tip)
 
            else:
 
                if normal:
 
                    bt[safe_unicode(bn)] = hex(tip)
 

	
 
        return bt
 

	
 
    @LazyProperty
 
    def tags(self):
 
        """
 
        Gets tags for this repository
 
        """
 
        return self._get_tags()
 

	
 
    def _get_tags(self):
 
        if self._empty:
 
            return {}
 

	
 
        sortkey = lambda ctx: ctx[0]  # sort by name
 
        _tags = [(safe_unicode(n), hex(h),) for n, h in
 
                 self._repo.tags().items()]
 

	
 
        return OrderedDict(sorted(_tags, key=sortkey, reverse=True))
 

	
 
    def tag(self, name, user, revision=None, message=None, date=None,
 
            **kwargs):
 
        """
 
        Creates and returns a tag for the given ``revision``.
 

	
 
        :param name: name for new tag
 
        :param user: full username, i.e.: "Joe Doe <joe.doe@example.com>"
 
        :param revision: changeset id for which new tag would be created
 
        :param message: message of the tag's commit
 
        :param date: date of tag's commit
 

	
 
        :raises TagAlreadyExistError: if tag with same name already exists
 
        """
 
        if name in self.tags:
 
            raise TagAlreadyExistError("Tag %s already exists" % name)
 
        changeset = self.get_changeset(revision)
 
        local = kwargs.setdefault('local', False)
 

	
 
        if message is None:
 
            message = "Added tag %s for changeset %s" % (name,
 
                changeset.short_id)
 

	
 
        if date is None:
 
            date = datetime.datetime.now().ctime()
 

	
 
        try:
 
            self._repo.tag(name, changeset._ctx.node(), message, local, user,
 
                date)
 
        except Abort as e:
 
            raise RepositoryError(e.message)
 

	
 
        # Reinitialize tags
 
        self.tags = self._get_tags()
 
        tag_id = self.tags[name]
 

	
 
        return self.get_changeset(revision=tag_id)
 

	
 
    def remove_tag(self, name, user, message=None, date=None):
 
        """
 
        Removes tag with the given ``name``.
 

	
 
        :param name: name of the tag to be removed
 
        :param user: full username, i.e.: "Joe Doe <joe.doe@example.com>"
 
        :param message: message of the tag's removal commit
 
        :param date: date of tag's removal commit
 

	
 
        :raises TagDoesNotExistError: if tag with given name does not exists
 
        """
 
        if name not in self.tags:
 
            raise TagDoesNotExistError("Tag %s does not exist" % name)
 
        if message is None:
 
            message = "Removed tag %s" % name
 
        if date is None:
 
            date = datetime.datetime.now().ctime()
 
        local = False
 

	
 
        try:
 
            self._repo.tag(name, nullid, message, local, user, date)
 
            self.tags = self._get_tags()
 
        except Abort as e:
 
            raise RepositoryError(e.message)
 

	
 
    @LazyProperty
 
    def bookmarks(self):
 
        """
 
        Gets bookmarks for this repository
 
        """
 
        return self._get_bookmarks()
 

	
 
    def _get_bookmarks(self):
 
        if self._empty:
 
            return {}
 

	
 
        sortkey = lambda ctx: ctx[0]  # sort by name
 
        _bookmarks = [(safe_unicode(n), hex(h),) for n, h in
 
                 self._repo._bookmarks.items()]
 
        return OrderedDict(sorted(_bookmarks, key=sortkey, reverse=True))
 

	
 
    def _get_all_revisions(self):
 

	
 
        return [self._repo[x].hex() for x in self._repo.filtered('visible').changelog.revs()]
 

	
 
    def get_diff(self, rev1, rev2, path='', 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``.
 
          shown. Defaults to ``3``. If negative value is passed-in, it will be
 
          set to ``0`` instead.
 
        """
 

	
 
        # Negative context values make no sense, and will result in
 
        # errors. Ensure this does not happen.
 
        if context < 0:
 
            context = 0
 

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

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

	
 
        # Check if given revisions are present at repository (may raise
 
        # ChangesetDoesNotExistError)
 
        if rev1 != self.EMPTY_CHANGESET:
 
            self.get_changeset(rev1)
 
        self.get_changeset(rev2)
 
        if path:
 
            file_filter = match(self.path, '', [path])
 
        else:
 
            file_filter = None
 

	
 
        return ''.join(patch.diff(self._repo, rev1, rev2, match=file_filter,
 
                          opts=diffopts(git=True,
 
                                        showfunc=True,
 
                                        ignorews=ignore_whitespace,
 
                                        context=context)))
 

	
 
    @classmethod
 
    def _check_url(cls, url, repoui=None):
 
        """
 
        Function will check given url and try to verify if it's a valid
 
        link. Sometimes it may happened that mercurial 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 url.startswith('ssh:'):
 
            # in case of invalid uri or authentication issues, sshpeer will
 
            # throw an exception.
 
            sshpeer(repoui or ui.ui(), url).lookup('tip')
 
            return True
 

	
 
        url_prefix = None
 
        if '+' in url[:url.find('://')]:
 
            url_prefix, url = url.split('+', 1)
 

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

	
 
        if authinfo:
 
            #create a password manager
 
            passmgr = urllib2.HTTPPasswordMgrWithDefaultRealm()
 
            passmgr.add_password(*authinfo)
 

	
 
            handlers.extend((httpbasicauthhandler(passmgr),
 
                             httpdigestauthhandler(passmgr)))
 

	
 
        o = urllib2.build_opener(*handlers)
 
        o.addheaders = [('Content-Type', 'application/mercurial-0.1'),
 
                        ('Accept', 'application/mercurial-0.1')]
 

	
 
        q = {"cmd": 'between'}
 
        q.update({'pairs': "%s-%s" % ('0' * 40, '0' * 40)})
 
        qs = '?%s' % urllib.urlencode(q)
 
        cu = "%s%s" % (test_uri, qs)
 
        req = urllib2.Request(cu, None, {})
 

	
 
        try:
 
            resp = o.open(req)
 
            if resp.code != 200:
 
                raise Exception('Return Code is not 200')
 
        except Exception as e:
 
            # means it cannot be cloned
 
            raise urllib2.URLError("[%s] org_exc: %s" % (cleaned_uri, e))
 

	
 
        if not url_prefix: # skip svn+http://... (and git+... too)
 
            # now check if it's a proper hg repo
 
            try:
 
                httppeer(repoui or ui.ui(), url).lookup('tip')
 
            except Exception as e:
 
                raise urllib2.URLError(
 
                    "url [%s] does not look like an hg repo org_exc: %s"
 
                    % (cleaned_uri, e))
 

	
 
        return True
 

	
 
    def _get_repo(self, create, src_url=None, update_after_clone=False):
 
        """
 
        Function will check for mercurial repository in given path and return
 
        a localrepo object. If there is no repository in that path it will
 
        raise an exception unless ``create`` parameter is set to True - in
 
        that case repository would be created and returned.
 
        If ``src_url`` is given, would try to clone repository from the
 
        location at given clone_point. Additionally it'll make update to
 
        working copy accordingly to ``update_after_clone`` flag
 
        """
 

	
 
        try:
 
            if src_url:
 
                url = str(self._get_url(src_url))
 
                opts = {}
 
                if not update_after_clone:
 
                    opts.update({'noupdate': True})
 
                MercurialRepository._check_url(url, self.baseui)
 
                clone(self.baseui, url, self.path, **opts)
 

	
 
                # Don't try to create if we've already cloned repo
 
                create = False
 
            return localrepository(self.baseui, self.path, create=create)
 
        except (Abort, RepoError) as err:
 
            if create:
 
                msg = "Cannot create repository at %s. Original error was %s"\
 
                    % (self.path, err)
 
            else:
 
                msg = "Not valid repository at %s. Original error was %s"\
 
                    % (self.path, err)
 
            raise RepositoryError(msg)
 

	
 
    @LazyProperty
 
    def in_memory_changeset(self):
 
        return MercurialInMemoryChangeset(self)
 

	
 
    @LazyProperty
 
    def description(self):
 
        undefined_description = u'unknown'
 
        _desc = self._repo.ui.config('web', 'description', None, untrusted=True)
 
        return safe_unicode(_desc or undefined_description)
 

	
 
    @LazyProperty
 
    def contact(self):
 
        undefined_contact = u'Unknown'
 
        return safe_unicode(get_contact(self._repo.ui.config)
 
                            or undefined_contact)
 

	
 
    @LazyProperty
 
    def last_change(self):
 
        """
 
        Returns last change made on this repository as datetime object
 
        """
 
        return date_fromtimestamp(self._get_mtime(), makedate()[1])
 

	
 
    def _get_mtime(self):
 
        try:
 
            return time.mktime(self.get_changeset().date.timetuple())
 
        except RepositoryError:
 
            #fallback to filesystem
 
            cl_path = os.path.join(self.path, '.hg', "00changelog.i")
 
            st_path = os.path.join(self.path, '.hg', "store")
 
            if os.path.exists(cl_path):
 
                return os.stat(cl_path).st_mtime
 
            else:
 
                return os.stat(st_path).st_mtime
 

	
 
    def _get_revision(self, revision):
 
        """
 
        Gets an ID revision given as str. This will always return a fill
 
        40 char revision number
 

	
 
        :param revision: str or int or None
 
        """
 
        if isinstance(revision, unicode):
 
            revision = safe_str(revision)
 

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

	
 
        if revision in [-1, 'tip', None]:
 
            revision = 'tip'
 

	
 
        try:
 
            revision = hex(self._repo.lookup(revision))
 
        except (LookupError, ):
 
            msg = ("Ambiguous identifier `%s` for %s" % (revision, self))
 
            raise ChangesetDoesNotExistError(msg)
 
        except (IndexError, ValueError, RepoLookupError, TypeError):
 
            msg = ("Revision %s does not exist for %s" % (revision, self))
 
            raise ChangesetDoesNotExistError(msg)
 

	
 
        return revision
 

	
 
    def get_ref_revision(self, ref_type, ref_name):
 
        """
 
        Returns revision number for the given reference.
 
        """
 
        ref_name = safe_str(ref_name)
 
        if ref_type == 'rev' and not ref_name.strip('0'):
 
            return self.EMPTY_CHANGESET
 
        # lookup up the exact node id
kallithea/tests/vcs/test_git.py
Show inline comments
 
@@ -538,313 +538,353 @@ class GitChangesetTest(unittest.TestCase
 
            for rev, data in revision_dict.items():
 
                cs = self.repo.get_changeset(rev)
 

	
 
                l1_1 = [x[1] for x in cs.get_file_annotate(fname)]
 
                l1_2 = [x[2]().raw_id for x in cs.get_file_annotate(fname)]
 
                self.assertEqual(l1_1, l1_2)
 
                l1 = l1_1
 
                l2 = files[fname][rev]['changesets']
 
                self.assertTrue(l1 == l2 , "The lists of revision for %s@rev %s"
 
                                "from annotation list should match each other, "
 
                                "got \n%s \nvs \n%s " % (fname, rev, l1, l2))
 

	
 
    def test_files_state(self):
 
        """
 
        Tests state of FileNodes.
 
        """
 
        node = self.repo\
 
            .get_changeset('e6ea6d16e2f26250124a1f4b4fe37a912f9d86a0')\
 
            .get_node('vcs/utils/diffs.py')
 
        self.assertTrue(node.state, NodeState.ADDED)
 
        self.assertTrue(node.added)
 
        self.assertFalse(node.changed)
 
        self.assertFalse(node.not_changed)
 
        self.assertFalse(node.removed)
 

	
 
        node = self.repo\
 
            .get_changeset('33fa3223355104431402a888fa77a4e9956feb3e')\
 
            .get_node('.hgignore')
 
        self.assertTrue(node.state, NodeState.CHANGED)
 
        self.assertFalse(node.added)
 
        self.assertTrue(node.changed)
 
        self.assertFalse(node.not_changed)
 
        self.assertFalse(node.removed)
 

	
 
        node = self.repo\
 
            .get_changeset('e29b67bd158580fc90fc5e9111240b90e6e86064')\
 
            .get_node('setup.py')
 
        self.assertTrue(node.state, NodeState.NOT_CHANGED)
 
        self.assertFalse(node.added)
 
        self.assertFalse(node.changed)
 
        self.assertTrue(node.not_changed)
 
        self.assertFalse(node.removed)
 

	
 
        # If node has REMOVED state then trying to fetch it would raise
 
        # ChangesetError exception
 
        chset = self.repo.get_changeset(
 
            'fa6600f6848800641328adbf7811fd2372c02ab2')
 
        path = 'vcs/backends/BaseRepository.py'
 
        self.assertRaises(NodeDoesNotExistError, chset.get_node, path)
 
        # but it would be one of ``removed`` (changeset's attribute)
 
        self.assertTrue(path in [rf.path for rf in chset.removed])
 

	
 
        chset = self.repo.get_changeset(
 
            '54386793436c938cff89326944d4c2702340037d')
 
        changed = ['setup.py', 'tests/test_nodes.py', 'vcs/backends/hg.py',
 
            'vcs/nodes.py']
 
        self.assertEqual(set(changed), set([f.path for f in chset.changed]))
 

	
 
    def test_commit_message_is_unicode(self):
 
        for cs in self.repo:
 
            self.assertEqual(type(cs.message), unicode)
 

	
 
    def test_changeset_author_is_unicode(self):
 
        for cs in self.repo:
 
            self.assertEqual(type(cs.author), unicode)
 

	
 
    def test_repo_files_content_is_unicode(self):
 
        changeset = self.repo.get_changeset()
 
        for node in changeset.get_node('/'):
 
            if node.is_file():
 
                self.assertEqual(type(node.content), unicode)
 

	
 
    def test_wrong_path(self):
 
        # There is 'setup.py' in the root dir but not there:
 
        path = 'foo/bar/setup.py'
 
        tip = self.repo.get_changeset()
 
        self.assertRaises(VCSError, tip.get_node, path)
 

	
 
    def test_author_email(self):
 
        self.assertEqual('marcin@python-blog.com',
 
          self.repo.get_changeset('c1214f7e79e02fc37156ff215cd71275450cffc3')\
 
          .author_email)
 
        self.assertEqual('lukasz.balcerzak@python-center.pl',
 
          self.repo.get_changeset('ff7ca51e58c505fec0dd2491de52c622bb7a806b')\
 
          .author_email)
 
        self.assertEqual('none@none',
 
          self.repo.get_changeset('8430a588b43b5d6da365400117c89400326e7992')\
 
          .author_email)
 

	
 
    def test_author_username(self):
 
        self.assertEqual('Marcin Kuzminski',
 
          self.repo.get_changeset('c1214f7e79e02fc37156ff215cd71275450cffc3')\
 
          .author_name)
 
        self.assertEqual('Lukasz Balcerzak',
 
          self.repo.get_changeset('ff7ca51e58c505fec0dd2491de52c622bb7a806b')\
 
          .author_name)
 
        self.assertEqual('marcink',
 
          self.repo.get_changeset('8430a588b43b5d6da365400117c89400326e7992')\
 
          .author_name)
 

	
 

	
 
class GitSpecificTest(unittest.TestCase):
 

	
 
    def test_error_is_raised_for_added_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
 

	
 
    def test_error_is_raised_for_changed_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
 

	
 
    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'])
 
        # Regression test: one of following would fail if we don't check
 
        # .git/HEAD file
 
        self.repo.run_git_command(['checkout', 'production'])
 
        self.assertEqual(self.repo.workdir.get_branch(), 'production')
 
        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', '-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', '-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', '-U3', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
 

	
 
    def test_get_diff_does_not_sanitize_valid_context(self):
 
        almost_overflowed_long_int = 2**31-1
 

	
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo.get_diff(0, 1, 'foo', context=almost_overflowed_long_int)
 
        self.repo.run_git_command.assert_called_once_with(
 
            ['diff', '-U' + str(almost_overflowed_long_int), '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
 

	
 
    def test_get_diff_sanitizes_overflowing_context(self):
 
        overflowed_long_int = 2**31
 
        sanitized_overflowed_long_int = overflowed_long_int-1
 

	
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo.get_diff(0, 1, 'foo', context=overflowed_long_int)
 

	
 
        self.repo.run_git_command.assert_called_once_with(
 
            ['diff', '-U' + str(sanitized_overflowed_long_int), '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
 

	
 
    def test_get_diff_does_not_sanitize_zero_context(self):
 
        zero_context = 0
 

	
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo.get_diff(0, 1, 'foo', context=zero_context)
 

	
 
        self.repo.run_git_command.assert_called_once_with(
 
            ['diff', '-U' + str(zero_context), '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
 

	
 
    def test_get_diff_sanitizes_negative_context(self):
 
        negative_context = -10
 

	
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo.get_diff(0, 1, 'foo', context=negative_context)
 

	
 
        self.repo.run_git_command.assert_called_once_with(
 
            ['diff', '-U0', '--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'])
 

	
 

	
 
class GitHooksTest(unittest.TestCase):
 
    """
 
    Tests related to hook functionality of Git repositories.
 
    """
 

	
 
    def setUp(self):
 
        # For each run we want a fresh repo.
 
        self.repo_directory = get_new_dir("githookrepo")
 
        self.repo = GitRepository(self.repo_directory, create=True)
 

	
 
        # Create a dictionary where keys are hook names, and values are paths to
 
        # them. Deduplicates code in tests a bit.
 
        self.hook_directory = self.repo.get_hook_location()
 
        self.kallithea_hooks = {h: os.path.join(self.hook_directory, h) for h in ("pre-receive", "post-receive")}
 

	
 
    def test_hooks_created_if_missing(self):
 
        """
 
        Tests if hooks are installed in repository if they are missing.
 
        """
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            if os.path.exists(hook_path):
 
                os.remove(hook_path)
 

	
 
        ScmModel().install_git_hooks(repo=self.repo)
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            self.assertTrue(os.path.exists(hook_path))
 

	
 
    def test_kallithea_hooks_updated(self):
 
        """
 
        Tests if hooks are updated if they are Kallithea hooks already.
 
        """
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            with open(hook_path, "w") as f:
 
                f.write("KALLITHEA_HOOK_VER=0.0.0\nJUST_BOGUS")
 

	
 
        ScmModel().install_git_hooks(repo=self.repo)
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            with open(hook_path) as f:
 
                self.assertNotIn("JUST_BOGUS", f.read())
 

	
 
    def test_custom_hooks_untouched(self):
 
        """
 
        Tests if hooks are left untouched if they are not Kallithea hooks.
 
        """
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            with open(hook_path, "w") as f:
 
                f.write("#!/bin/bash\n#CUSTOM_HOOK")
 

	
 
        ScmModel().install_git_hooks(repo=self.repo)
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            with open(hook_path) as f:
 
                self.assertIn("CUSTOM_HOOK", f.read())
 

	
 
    def test_custom_hooks_forced_update(self):
 
        """
 
        Tests if hooks are forcefully updated even though they are custom hooks.
 
        """
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            with open(hook_path, "w") as f:
 
                f.write("#!/bin/bash\n#CUSTOM_HOOK")
 

	
 
        ScmModel().install_git_hooks(repo=self.repo, force_create=True)
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            with open(hook_path) as f:
 
                self.assertIn("KALLITHEA_HOOK_VER", f.read())
 

	
 

	
 
if __name__ == '__main__':
 
    unittest.main()
kallithea/tests/vcs/test_hg.py
Show inline comments
 

	
 
import os
 

	
 
import mock
 

	
 
from kallithea.lib.vcs.backends.hg import MercurialRepository, MercurialChangeset
 
from kallithea.lib.vcs.exceptions import RepositoryError, VCSError, NodeDoesNotExistError
 
from kallithea.lib.vcs.nodes import NodeKind, NodeState
 
from kallithea.tests.vcs.conf import TEST_HG_REPO, TEST_HG_REPO_CLONE, \
 
    TEST_HG_REPO_PULL
 
from kallithea.lib.vcs.utils.compat import unittest
 

	
 

	
 
class MercurialRepositoryTest(unittest.TestCase):
 

	
 
    def __check_for_existing_repo(self):
 
        if os.path.exists(TEST_HG_REPO_CLONE):
 
            self.fail('Cannot test mercurial clone repo as location %s already '
 
                      'exists. You should manually remove it first.'
 
                      % TEST_HG_REPO_CLONE)
 

	
 
    def setUp(self):
 
        self.repo = MercurialRepository(TEST_HG_REPO)
 

	
 
    def test_wrong_repo_path(self):
 
        wrong_repo_path = '/tmp/errorrepo'
 
        self.assertRaises(RepositoryError, MercurialRepository, wrong_repo_path)
 

	
 
    def test_unicode_path_repo(self):
 
        self.assertRaises(VCSError,lambda:MercurialRepository(u'iShouldFail'))
 

	
 
    def test_repo_clone(self):
 
        self.__check_for_existing_repo()
 
        repo = MercurialRepository(TEST_HG_REPO)
 
        repo_clone = MercurialRepository(TEST_HG_REPO_CLONE,
 
            src_url=TEST_HG_REPO, update_after_clone=True)
 
        self.assertEqual(len(repo.revisions), len(repo_clone.revisions))
 
        # Checking hashes of changesets should be enough
 
        for changeset in repo.get_changesets():
 
            raw_id = changeset.raw_id
 
            self.assertEqual(raw_id, repo_clone.get_changeset(raw_id).raw_id)
 

	
 
    def test_repo_clone_with_update(self):
 
        repo = MercurialRepository(TEST_HG_REPO)
 
        repo_clone = MercurialRepository(TEST_HG_REPO_CLONE + '_w_update',
 
            src_url=TEST_HG_REPO, update_after_clone=True)
 
        self.assertEqual(len(repo.revisions), len(repo_clone.revisions))
 

	
 
        #check if current workdir was updated
 
        self.assertEqual(os.path.isfile(os.path.join(TEST_HG_REPO_CLONE \
 
                                                    + '_w_update',
 
                                                    'MANIFEST.in')), True,)
 

	
 
    def test_repo_clone_without_update(self):
 
        repo = MercurialRepository(TEST_HG_REPO)
 
        repo_clone = MercurialRepository(TEST_HG_REPO_CLONE + '_wo_update',
 
            src_url=TEST_HG_REPO, update_after_clone=False)
 
        self.assertEqual(len(repo.revisions), len(repo_clone.revisions))
 
        self.assertEqual(os.path.isfile(os.path.join(TEST_HG_REPO_CLONE \
 
                                                    + '_wo_update',
 
                                                    'MANIFEST.in')), False,)
 

	
 
    def test_pull(self):
 
        if os.path.exists(TEST_HG_REPO_PULL):
 
            self.fail('Cannot test mercurial pull command as location %s '
 
                      'already exists. You should manually remove it first'
 
                      % TEST_HG_REPO_PULL)
 
        repo_new = MercurialRepository(TEST_HG_REPO_PULL, create=True)
 
        self.assertTrue(len(self.repo.revisions) > len(repo_new.revisions))
 

	
 
        repo_new.pull(self.repo.path)
 
        repo_new = MercurialRepository(TEST_HG_REPO_PULL)
 
        self.assertTrue(len(self.repo.revisions) == len(repo_new.revisions))
 

	
 
    def test_revisions(self):
 
        # there are 21 revisions at bitbucket now
 
        # so we can assume they would be available from now on
 
        subset = set(['b986218ba1c9b0d6a259fac9b050b1724ed8e545',
 
                 '3d8f361e72ab303da48d799ff1ac40d5ac37c67e',
 
                 '6cba7170863a2411822803fa77a0a264f1310b35',
 
                 '56349e29c2af3ac913b28bde9a2c6154436e615b',
 
                 '2dda4e345facb0ccff1a191052dd1606dba6781d',
 
                 '6fff84722075f1607a30f436523403845f84cd9e',
 
                 '7d4bc8ec6be56c0f10425afb40b6fc315a4c25e7',
 
                 '3803844fdbd3b711175fc3da9bdacfcd6d29a6fb',
 
                 'dc5d2c0661b61928834a785d3e64a3f80d3aad9c',
 
                 'be90031137367893f1c406e0a8683010fd115b79',
 
                 'db8e58be770518cbb2b1cdfa69146e47cd481481',
 
                 '84478366594b424af694a6c784cb991a16b87c21',
 
                 '17f8e105dddb9f339600389c6dc7175d395a535c',
 
                 '20a662e756499bde3095ffc9bc0643d1def2d0eb',
 
                 '2e319b85e70a707bba0beff866d9f9de032aa4f9',
 
                 '786facd2c61deb9cf91e9534735124fb8fc11842',
 
                 '94593d2128d38210a2fcd1aabff6dda0d6d9edf8',
 
                 'aa6a0de05b7612707db567078e130a6cd114a9a7',
 
                 'eada5a770da98ab0dd7325e29d00e0714f228d09'
 
                ])
 
        self.assertTrue(subset.issubset(set(self.repo.revisions)))
 

	
 

	
 
        # check if we have the proper order of revisions
 
        org = ['b986218ba1c9b0d6a259fac9b050b1724ed8e545',
 
                '3d8f361e72ab303da48d799ff1ac40d5ac37c67e',
 
                '6cba7170863a2411822803fa77a0a264f1310b35',
 
                '56349e29c2af3ac913b28bde9a2c6154436e615b',
 
                '2dda4e345facb0ccff1a191052dd1606dba6781d',
 
                '6fff84722075f1607a30f436523403845f84cd9e',
 
                '7d4bc8ec6be56c0f10425afb40b6fc315a4c25e7',
 
                '3803844fdbd3b711175fc3da9bdacfcd6d29a6fb',
 
                'dc5d2c0661b61928834a785d3e64a3f80d3aad9c',
 
                'be90031137367893f1c406e0a8683010fd115b79',
 
                'db8e58be770518cbb2b1cdfa69146e47cd481481',
 
                '84478366594b424af694a6c784cb991a16b87c21',
 
                '17f8e105dddb9f339600389c6dc7175d395a535c',
 
                '20a662e756499bde3095ffc9bc0643d1def2d0eb',
 
                '2e319b85e70a707bba0beff866d9f9de032aa4f9',
 
                '786facd2c61deb9cf91e9534735124fb8fc11842',
 
                '94593d2128d38210a2fcd1aabff6dda0d6d9edf8',
 
                'aa6a0de05b7612707db567078e130a6cd114a9a7',
 
                'eada5a770da98ab0dd7325e29d00e0714f228d09',
 
                '2c1885c735575ca478bf9e17b0029dca68824458',
 
                'd9bcd465040bf869799b09ad732c04e0eea99fe9',
 
                '469e9c847fe1f6f7a697b8b25b4bc5b48780c1a7',
 
                '4fb8326d78e5120da2c7468dcf7098997be385da',
 
                '62b4a097164940bd66030c4db51687f3ec035eed',
 
                '536c1a19428381cfea92ac44985304f6a8049569',
 
                '965e8ab3c44b070cdaa5bf727ddef0ada980ecc4',
 
                '9bb326a04ae5d98d437dece54be04f830cf1edd9',
 
                'f8940bcb890a98c4702319fbe36db75ea309b475',
 
                'ff5ab059786ebc7411e559a2cc309dfae3625a3b',
 
                '6b6ad5f82ad5bb6190037671bd254bd4e1f4bf08',
 
                'ee87846a61c12153b51543bf860e1026c6d3dcba', ]
 
        self.assertEqual(org, self.repo.revisions[:31])
 

	
 
    def test_iter_slice(self):
 
        sliced = list(self.repo[:10])
 
        itered = list(self.repo)[:10]
 
        self.assertEqual(sliced, itered)
 

	
 
    def test_slicing(self):
 
        #4 1 5 10 95
 
        for sfrom, sto, size in [(0, 4, 4), (1, 2, 1), (10, 15, 5),
 
                                 (10, 20, 10), (5, 100, 95)]:
 
            revs = list(self.repo[sfrom:sto])
 
            self.assertEqual(len(revs), size)
 
            self.assertEqual(revs[0], self.repo.get_changeset(sfrom))
 
            self.assertEqual(revs[-1], self.repo.get_changeset(sto - 1))
 

	
 
    def test_branches(self):
 
        # TODO: Need more tests here
 

	
 
        #active branches
 
        self.assertTrue('default' in self.repo.branches)
 
        self.assertTrue('stable' in self.repo.branches)
 

	
 
        # closed
 
        self.assertTrue('git' in self.repo._get_branches(closed=True))
 
        self.assertTrue('web' in self.repo._get_branches(closed=True))
 

	
 
        for name, id in self.repo.branches.items():
 
            self.assertTrue(isinstance(
 
                self.repo.get_changeset(id), MercurialChangeset))
 

	
 
    def test_tip_in_tags(self):
 
        # tip is always a tag
 
        self.assertIn('tip', self.repo.tags)
 

	
 
    def test_tip_changeset_in_tags(self):
 
        tip = self.repo.get_changeset()
 
        self.assertEqual(self.repo.tags['tip'], tip.raw_id)
 

	
 
    def test_initial_changeset(self):
 

	
 
        init_chset = self.repo.get_changeset(0)
 
        self.assertEqual(init_chset.message, 'initial import')
 
        self.assertEqual(init_chset.author,
 
            'Marcin Kuzminski <marcin@python-blog.com>')
 
        self.assertEqual(sorted(init_chset._file_paths),
 
            sorted([
 
                'vcs/__init__.py',
 
                'vcs/backends/BaseRepository.py',
 
                'vcs/backends/__init__.py',
 
            ])
 
        )
 
        self.assertEqual(sorted(init_chset._dir_paths),
 
            sorted(['', 'vcs', 'vcs/backends']))
 

	
 
        self.assertRaises(NodeDoesNotExistError, init_chset.get_node, path='foobar')
 

	
 
        node = init_chset.get_node('vcs/')
 
        self.assertTrue(hasattr(node, 'kind'))
 
        self.assertEqual(node.kind, NodeKind.DIR)
 

	
 
        node = init_chset.get_node('vcs')
 
        self.assertTrue(hasattr(node, 'kind'))
 
        self.assertEqual(node.kind, NodeKind.DIR)
 

	
 
        node = init_chset.get_node('vcs/__init__.py')
 
        self.assertTrue(hasattr(node, 'kind'))
 
        self.assertEqual(node.kind, NodeKind.FILE)
 

	
 
    def test_not_existing_changeset(self):
 
        #rawid
 
        self.assertRaises(RepositoryError, self.repo.get_changeset,
 
            'abcd' * 10)
 
        #shortid
 
        self.assertRaises(RepositoryError, self.repo.get_changeset,
 
            'erro' * 4)
 
        #numeric
 
        self.assertRaises(RepositoryError, self.repo.get_changeset,
 
            self.repo.count() + 1)
 

	
 

	
 
        # Small chance we ever get to this one
 
        revision = pow(2, 30)
 
        self.assertRaises(RepositoryError, self.repo.get_changeset, revision)
 

	
 
    def test_changeset10(self):
 

	
 
        chset10 = self.repo.get_changeset(10)
 
        README = """===
 
VCS
 
===
 

	
 
Various Version Control System management abstraction layer for Python.
 

	
 
Introduction
 
------------
 

	
 
TODO: To be written...
 

	
 
"""
 
        node = chset10.get_node('README.rst')
 
        self.assertEqual(node.kind, NodeKind.FILE)
 
        self.assertEqual(node.content, README)
 

	
 
    @mock.patch('kallithea.lib.vcs.backends.hg.repository.diffopts')
 
    def test_get_diff_does_not_sanitize_zero_context(self, mock_diffopts):
 
        zero_context = 0
 

	
 
        self.repo.get_diff(0, 1, 'foo', context=zero_context)
 

	
 
        mock_diffopts.assert_called_once_with(git=True, showfunc=True, ignorews=False, context=zero_context)
 

	
 
    @mock.patch('kallithea.lib.vcs.backends.hg.repository.diffopts')
 
    def test_get_diff_sanitizes_negative_context(self, mock_diffopts):
 
        negative_context = -10
 
        zero_context = 0
 

	
 
        self.repo.get_diff(0, 1, 'foo', context=negative_context)
 

	
 
        mock_diffopts.assert_called_once_with(git=True, showfunc=True, ignorews=False, context=zero_context)
 

	
 

	
 
class MercurialChangesetTest(unittest.TestCase):
 

	
 
    def setUp(self):
 
        self.repo = MercurialRepository(TEST_HG_REPO)
 

	
 
    def _test_equality(self, changeset):
 
        revision = changeset.revision
 
        self.assertEqual(changeset, self.repo.get_changeset(revision))
 

	
 
    def test_equality(self):
 
        self.setUp()
 
        revs = [0, 10, 20]
 
        changesets = [self.repo.get_changeset(rev) for rev in revs]
 
        for changeset in changesets:
 
            self._test_equality(changeset)
 

	
 
    def test_default_changeset(self):
 
        tip = self.repo.get_changeset('tip')
 
        self.assertEqual(tip, self.repo.get_changeset())
 
        self.assertEqual(tip, self.repo.get_changeset(revision=None))
 
        self.assertEqual(tip, list(self.repo[-1:])[0])
 

	
 
    def test_root_node(self):
 
        tip = self.repo.get_changeset('tip')
 
        self.assertTrue(tip.root is tip.get_node(''))
 

	
 
    def test_lazy_fetch(self):
 
        """
 
        Test if changeset's nodes expands and are cached as we walk through
 
        the revision. This test is somewhat hard to write as order of tests
 
        is a key here. Written by running command after command in a shell.
 
        """
 
        self.setUp()
 
        chset = self.repo.get_changeset(45)
 
        self.assertTrue(len(chset.nodes) == 0)
 
        root = chset.root
 
        self.assertTrue(len(chset.nodes) == 1)
 
        self.assertTrue(len(root.nodes) == 8)
 
        # accessing root.nodes updates chset.nodes
 
        self.assertTrue(len(chset.nodes) == 9)
 

	
 
        docs = root.get_node('docs')
 
        # we haven't yet accessed anything new as docs dir was already cached
 
        self.assertTrue(len(chset.nodes) == 9)
 
        self.assertTrue(len(docs.nodes) == 8)
 
        # accessing docs.nodes updates chset.nodes
 
        self.assertTrue(len(chset.nodes) == 17)
 

	
 
        self.assertTrue(docs is chset.get_node('docs'))
 
        self.assertTrue(docs is root.nodes[0])
 
        self.assertTrue(docs is root.dirs[0])
 
        self.assertTrue(docs is chset.get_node('docs'))
 

	
 
    def test_nodes_with_changeset(self):
 
        self.setUp()
 
        chset = self.repo.get_changeset(45)
 
        root = chset.root
 
        docs = root.get_node('docs')
 
        self.assertTrue(docs is chset.get_node('docs'))
 
        api = docs.get_node('api')
 
        self.assertTrue(api is chset.get_node('docs/api'))
 
        index = api.get_node('index.rst')
 
        self.assertTrue(index is chset.get_node('docs/api/index.rst'))
 
        self.assertTrue(index is chset.get_node('docs')\
 
            .get_node('api')\
 
            .get_node('index.rst'))
 

	
 
    def test_branch_and_tags(self):
 
        chset0 = self.repo.get_changeset(0)
 
        self.assertEqual(chset0.branch, 'default')
 
        self.assertEqual(chset0.tags, [])
 

	
 
        chset10 = self.repo.get_changeset(10)
 
        self.assertEqual(chset10.branch, 'default')
 
        self.assertEqual(chset10.tags, [])
 

	
 
        chset44 = self.repo.get_changeset(44)
 
        self.assertEqual(chset44.branch, 'web')
 

	
 
        tip = self.repo.get_changeset('tip')
 
        self.assertTrue('tip' in tip.tags)
 

	
 
    def _test_file_size(self, revision, path, size):
 
        node = self.repo.get_changeset(revision).get_node(path)
 
        self.assertTrue(node.is_file())
 
        self.assertEqual(node.size, size)
 

	
 
    def test_file_size(self):
 
        to_check = (
 
            (10, 'setup.py', 1068),
 
            (20, 'setup.py', 1106),
 
            (60, 'setup.py', 1074),
 

	
 
            (10, 'vcs/backends/base.py', 2921),
 
            (20, 'vcs/backends/base.py', 3936),
 
            (60, 'vcs/backends/base.py', 6189),
 
        )
 
        for revision, path, size in to_check:
 
            self._test_file_size(revision, path, size)
 

	
 
    def _test_dir_size(self, revision, path, size):
 
        node = self.repo.get_changeset(revision).get_node(path)
 
        self.assertFalse(node.is_file())
 
        self.assertEqual(node.size, size)
 

	
 
    def test_dir_size(self):
 
        to_check = (
 
            ('96507bd11ecc', '/', 682421),
 
            ('a53d9201d4bc', '/', 682410),
 
            ('90243de06161', '/', 682006),
 
        )
 
        for revision, path, size in to_check:
 
            self._test_dir_size(revision, path, size)
 

	
 
    def test_repo_size(self):
 
        self.assertEqual(self.repo.size, 682421)
 

	
 
    def test_file_history(self):
 
        # we can only check if those revisions are present in the history
 
        # as we cannot update this test every time file is changed
 
        files = {
 
            'setup.py': [7, 18, 45, 46, 47, 69, 77],
 
            'vcs/nodes.py': [7, 8, 24, 26, 30, 45, 47, 49, 56, 57, 58, 59, 60,
 
                61, 73, 76],
 
            'vcs/backends/hg.py': [4, 5, 6, 11, 12, 13, 14, 15, 16, 21, 22, 23,
 
                26, 27, 28, 30, 31, 33, 35, 36, 37, 38, 39, 40, 41, 44, 45, 47,
 
                48, 49, 53, 54, 55, 58, 60, 61, 67, 68, 69, 70, 73, 77, 78, 79,
 
                82],
 
        }
 
        for path, revs in files.items():
 
            tip = self.repo.get_changeset(revs[-1])
 
            node = tip.get_node(path)
 
            node_revs = [chset.revision for chset in node.history]
 
            self.assertTrue(set(revs).issubset(set(node_revs)),
 
                "We assumed that %s is subset of revisions for which file %s "
 
                "has been changed, and history of that node returned: %s"
 
                % (revs, path, node_revs))
 

	
 
    def test_file_annotate(self):
 
        files = {
 
                 'vcs/backends/__init__.py':
 
                  {89: {'lines_no': 31,
 
                        'changesets': [32, 32, 61, 32, 32, 37, 32, 32, 32, 44,
 
                                       37, 37, 37, 37, 45, 37, 44, 37, 37, 37,
 
                                       32, 32, 32, 32, 37, 32, 37, 37, 32,
 
                                       32, 32]},
 
                   20: {'lines_no': 1,
 
                        'changesets': [4]},
 
                   55: {'lines_no': 31,
 
                        'changesets': [32, 32, 45, 32, 32, 37, 32, 32, 32, 44,
 
                                       37, 37, 37, 37, 45, 37, 44, 37, 37, 37,
 
                                       32, 32, 32, 32, 37, 32, 37, 37, 32,
 
                                       32, 32]}},
 
                 'vcs/exceptions.py':
 
                 {89: {'lines_no': 18,
 
                       'changesets': [16, 16, 16, 16, 16, 16, 16, 16, 16, 16,
 
                                      16, 16, 17, 16, 16, 18, 18, 18]},
 
                  20: {'lines_no': 18,
 
                       'changesets': [16, 16, 16, 16, 16, 16, 16, 16, 16, 16,
 
                                      16, 16, 17, 16, 16, 18, 18, 18]},
 
                  55: {'lines_no': 18, 'changesets': [16, 16, 16, 16, 16, 16,
 
                                                      16, 16, 16, 16, 16, 16,
 
                                                      17, 16, 16, 18, 18, 18]}},
 
                 'MANIFEST.in': {89: {'lines_no': 5,
 
                                      'changesets': [7, 7, 7, 71, 71]},
 
                                 20: {'lines_no': 3,
 
                                      'changesets': [7, 7, 7]},
 
                                 55: {'lines_no': 3,
 
                                     'changesets': [7, 7, 7]}}}
 

	
 
        for fname, revision_dict in files.items():
 
            for rev, data in revision_dict.items():
 
                cs = self.repo.get_changeset(rev)
 
                l1_1 = [x[1] for x in cs.get_file_annotate(fname)]
 
                l1_2 = [x[2]().raw_id for x in cs.get_file_annotate(fname)]
 
                self.assertEqual(l1_1, l1_2)
 
                l1 = l1_2 = [x[2]().revision for x in cs.get_file_annotate(fname)]
 
                l2 = files[fname][rev]['changesets']
 
                self.assertTrue(l1 == l2 , "The lists of revision for %s@rev%s"
 
                                "from annotation list should match each other,"
 
                                "got \n%s \nvs \n%s " % (fname, rev, l1, l2))
 

	
 
    def test_changeset_state(self):
 
        """
 
        Tests which files have been added/changed/removed at particular revision
 
        """
 

	
 
        # rev 46ad32a4f974:
 
        # hg st --rev 46ad32a4f974
 
        #    changed: 13
 
        #    added:   20
0 comments (0 inline, 0 general)