# This program is distributed in the hope that it will be useful,

# but WITHOUT ANY WARRANTY; without even the implied warranty of

# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the

# GNU General Public License for more details.

#

# You should have received a copy of the GNU General Public License

# along with this program.  If not, see <http://www.gnu.org/licenses/>.

"""

kallithea.lib.auth_modules.auth_crowd

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Kallithea authentication plugin for Atlassian CROWD

This file was forked by the Kallithea project in July 2014.

Original author and date, and relevant copyright and licensing information is below:

:created_on: Created on Nov 17, 2012

:author: marcink

:copyright: (c) 2013 RhodeCode GmbH, and others.

:license: GPLv3, see LICENSE.md for more details.

"""

import base64

import logging

import urllib2

from kallithea.lib import auth_modules

from kallithea.lib.compat import json, formatted_json, hybrid_property

from kallithea.model.db import User

log = logging.getLogger(__name__)

class CrowdServer(object):

    def __init__(self, *args, **kwargs):

        """

        Create a new CrowdServer object that points to IP/Address 'host',

        on the given port, and using the given method (https/http). user and

        passwd can be set here or with set_credentials. If unspecified,

        "version" defaults to "latest".

        example::

            cserver = CrowdServer(host="127.0.0.1",

                                  port="8095",

                                  user="some_app",

                                  passwd="some_passwd",

                                  version="1")

        """

            kwargs["port"] = "8095"

        self._logger = kwargs.get("logger", logging.getLogger(__name__))

        self._uri = "%s://%s:%s/crowd" % (kwargs.get("method", "http"),

                                    kwargs.get("host", "127.0.0.1"),

                                    kwargs.get("port", "8095"))

        self.set_credentials(kwargs.get("user", ""),

                             kwargs.get("passwd", ""))

        self._version = kwargs.get("version", "latest")

        self._url_list = None

        self._appname = "crowd"

    def set_credentials(self, user, passwd):

        self.user = user

        self.passwd = passwd

        self._make_opener()

    def _make_opener(self):

        mgr = urllib2.HTTPPasswordMgrWithDefaultRealm()

        mgr.add_password(None, self._uri, self.user, self.passwd)

        handler = urllib2.HTTPBasicAuthHandler(mgr)

        self.opener = urllib2.build_opener(handler)

    def _request(self, url, body=None, headers=None,

                 method=None, noformat=False,

                 empty_response_ok=False):

        _headers = {"Content-type": "application/json",

                    "Accept": "application/json"}

        if self.user and self.passwd:

            authstring = base64.b64encode("%s:%s" % (self.user, self.passwd))

            _headers["Authorization"] = "Basic %s" % authstring

        if headers:

            _headers.update(headers)

        log.debug("Sent crowd: \n%s",

                  formatted_json({"url": url, "body": body,

                                           "headers": _headers}))

        req = urllib2.Request(url, body, _headers)

        if method:

            req.get_method = lambda: method

        global msg

        msg = ""

        try:

            rdoc = self.opener.open(req)

            msg = "".join(rdoc.readlines())

            if not msg and empty_response_ok:

                rval = {}

                rval["status"] = True

                rval["error"] = "Response body was empty"

                        start_id = results[0]['raw_id']

                        last_rev = repo.get_changeset(revision=start_id).revision

                    # there are new changesets to index or a new repo to index

                    if last_rev == 0 or num_of_revs > last_rev + 1:

                        # delete the docs in the index for the previous

                        # last changeset(s)

                        for hit in results:

                            q = qp.parse(u"last:t AND %s AND raw_id:%s" %

                                            (repo_name, hit['raw_id']))

                            writer.delete_by_query(q)

                        # index from the previous last changeset + all new ones

                        indexed_total += self.index_changesets(writer,

                                                repo_name, repo, start_id)

                        writer_is_dirty = True

                log.debug('indexed %s changesets for repo %s',

                          indexed_total, repo_name

                )

            finally:

                if writer_is_dirty:

                    log.debug('>> COMMITING CHANGES TO CHANGESET INDEX<<')

                    writer.commit(merge=True)

                    log.debug('>>> FINISHED REBUILDING CHANGESET INDEX <<<')

                else:

                    log.debug('>> NOTHING TO COMMIT TO CHANGESET INDEX<<')

    def update_file_index(self):

        log.debug((u'STARTING INCREMENTAL INDEXING UPDATE FOR EXTENSIONS %s '

                   'AND REPOS %s') % (INDEX_EXTENSIONS, self.repo_paths.keys()))

        idx = open_dir(self.index_location, indexname=self.indexname)

        # The set of all paths in the index

        indexed_paths = set()

        # The set of all paths we need to re-index

        to_index = set()

        writer = idx.writer()

        writer_is_dirty = False

        try:

            with idx.reader() as reader:

                # Loop over the stored fields in the index

                for fields in reader.all_stored_fields():

                    indexed_path = fields['path']

                    indexed_repo_path = fields['repository']

                    indexed_paths.add(indexed_path)

                        continue

                    repo = self.repo_paths[indexed_repo_path]

                    try:

                        node = self.get_node(repo, indexed_path)

                        # Check if this file was changed since it was indexed

                        indexed_time = fields['modtime']

                        mtime = self.get_node_mtime(node)

                        if mtime > indexed_time:

                            # The file has changed, delete it and add it to

                            # the list of files to reindex

                            log.debug(

                                'adding to reindex list %s mtime: %s vs %s',

                                    indexed_path, mtime, indexed_time

                            )

                            writer.delete_by_term('fileid', indexed_path)

                            writer_is_dirty = True

                            to_index.add(indexed_path)

                    except (ChangesetError, NodeDoesNotExistError):

                        # This file was deleted since it was indexed

                        log.debug('removing from index %s', indexed_path)

                        writer.delete_by_term('path', indexed_path)

                        writer_is_dirty = True

            # Loop over the files in the filesystem

            # Assume we have a function that gathers the filenames of the

            # documents to be indexed

            ri_cnt_total = 0  # indexed

            riwc_cnt_total = 0  # indexed with content

            for repo_name, repo in self.repo_paths.items():

                # skip indexing if there aren't any revisions

                if len(repo) < 1:

                    continue

                ri_cnt = 0   # indexed

                riwc_cnt = 0  # indexed with content

                for path in self.get_paths(repo):

                    path = safe_unicode(path)

                    if path in to_index or path not in indexed_paths:

                        # This is either a file that's changed, or a new file

                        # that wasn't indexed before. So index it!

                        i, iwc = self.add_doc(writer, path, repo, repo_name)

                        writer_is_dirty = True

                        log.debug('re indexing %s', path)

                        ri_cnt += i

                        ri_cnt_total += 1

    def author(self):

        return safe_unicode(getattr(self._commit, self._author_property))

    @LazyProperty

    def date(self):

        return date_fromtimestamp(getattr(self._commit, self._date_property),

                                  getattr(self._commit, self._date_tz_property))

    @LazyProperty

    def _timestamp(self):

        return getattr(self._commit, self._date_property)

    @LazyProperty

    def status(self):

        """

        Returns modified, added, removed, deleted files for current changeset

        """

        return self.changed, self.added, self.removed

    @LazyProperty

    def tags(self):

        _tags = []

        for tname, tsha in self.repository.tags.iteritems():

            if tsha == self.raw_id:

                _tags.append(tname)

        return _tags

    @LazyProperty

    def branch(self):

        heads = self.repository._heads(reverse=False)

        ref = heads.get(self.raw_id)

        if ref:

            return safe_unicode(ref)

    def _fix_path(self, path):

        """

        Paths are stored without trailing slash so we need to get rid off it if

        needed.

        """

        if path.endswith('/'):

            path = path.rstrip('/')

        return path

    def _get_id_for_path(self, path):

        path = safe_str(path)

        # FIXME: Please, spare a couple of minutes and make those codes cleaner;

            path = path.strip('/')

            # set root tree

            tree = self.repository._repo[self._tree_id]

            if path == '':

                self._paths[''] = tree.id

                return tree.id

            splitted = path.split('/')

            dirs, name = splitted[:-1], splitted[-1]

            curdir = ''

            # initially extract things from root dir

            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

            for dir in dirs:

                if curdir:

                    curdir = '/'.join((curdir, dir))

                else:

                    curdir = dir

                dir_id = None

                for item, stat, id in tree.iteritems():

                    if dir == item:

                        dir_id = id

                if dir_id:

                    # Update tree

                    tree = self.repository._repo[dir_id]

                    if not isinstance(tree, objects.Tree):

                        raise ChangesetError('%s is not a directory' % curdir)

                else:

                    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

                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', 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]

            cmd += ' | bzip2 -9'

        if stream is None:

            raise VCSError('You need to pass in a valid stream for filling'

                           ' with archival data')

        popen = Popen(cmd, stdout=PIPE, stderr=PIPE, shell=True,

                      cwd=self.repository.path)

        buffer_size = 1024 * 8

        chunk = popen.stdout.read(buffer_size)

        while chunk:

            stream.write(chunk)

            chunk = popen.stdout.read(buffer_size)

        # Make sure all descriptors would be read

        popen.communicate()

    def get_nodes(self, path):

        if self._get_kind(path) != NodeKind.DIR:

            raise ChangesetError("Directory does not exist for revision %s at "

                " '%s'" % (self.revision, path))

        path = self._fix_path(path)

        id = self._get_id_for_path(path)

        tree = self.repository._repo[id]

        dirnodes = []

        filenodes = []

        als = self.repository.alias

        for name, stat, id in tree.iteritems():

            if objects.S_ISGITLINK(stat):

                dirnodes.append(SubModuleNode(name, url=None, changeset=id,

                                              alias=als))

                continue

            obj = self.repository._repo.get_object(id)

            if path != '':

                obj_path = '/'.join((path, name))

            else:

                obj_path = name

            if obj_path not in self._stat_modes:

                self._stat_modes[obj_path] = stat

            if isinstance(obj, objects.Tree):

                dirnodes.append(DirNode(obj_path, changeset=self))

            elif isinstance(obj, objects.Blob):

                filenodes.append(FileNode(obj_path, changeset=self, mode=stat))

            else:

                raise ChangesetError("Requested object should be Tree "

                                     "or Blob, is %r" % type(obj))

        nodes = dirnodes + filenodes

        for node in nodes:

                self.nodes[node.path] = node

        nodes.sort()

        return nodes

    def get_node(self, path):

        if isinstance(path, unicode):

            path = path.encode('utf-8')

        path = self._fix_path(path)

            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', parent.raw_id, self.raw_id,

                   '--encoding=utf8']

            so, se = self.repository.run_git_command(cmd)

            output.append(so.strip())

        return '\n'.join(output)

        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('git://'):

            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'

        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 = [('User-Agent', 'git/1.7.8.0')]  # fake some git

        q = {"service": 'git-upload-pack'}

        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))

        # now detect if it's proper git repo

        gitdata = resp.read()

            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 Repo(self.path)

        except (NotGitRepository, OSError) as 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', 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

        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:

                msg = ("Revision %s does not exist for %s" % (revision, self))

                raise ChangesetDoesNotExistError(msg)

        # Ensure we return full id

        if not SHA_PATTERN.match(str(revision)):

            raise ChangesetDoesNotExistError("Given revision %s not recognized"

                % revision)

        return revision

    def get_ref_revision(self, ref_type, ref_name):

        """

        Returns ``MercurialChangeset`` object representing repository's

        changeset at the given ``revision``.

        """

        return self._get_revision(ref_name)

    def _get_archives(self, archive_name='tip'):

        for i in [('zip', '.zip'), ('gz', '.tar.gz'), ('bz2', '.tar.bz2')]:

                yield {"type": i[0], "extension": i[1], "node": archive_name}

    def _get_url(self, url):

        """

        Returns normalized url. If schema is not given, would fall to

        filesystem (``file:///``) schema.

        """

        url = safe_str(url)

            url = ':///'.join(('file', url))

        return url

    def get_hook_location(self):

        """

        returns absolute path to location where hooks are stored

        """

        loc = os.path.join(self.path, 'hooks')

        if not self.bare:

            loc = os.path.join(self.path, '.git', 'hooks')

        return loc

    @LazyProperty

    def name(self):

        return os.path.basename(self.path)

    @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:

            idx_loc = '' if self.bare else '.git'

            # fallback to filesystem

            in_path = os.path.join(self.path, idx_loc, "index")

            he_path = os.path.join(self.path, idx_loc, "HEAD")

            if os.path.exists(in_path):

                return os.stat(in_path).st_mtime

            else:

                return os.stat(he_path).st_mtime

    @LazyProperty

    def description(self):

        undefined_description = u'unknown'

        _desc = self._repo.get_description()

        return safe_unicode(_desc or undefined_description)

    @LazyProperty

    def contact(self):

        undefined_contact = u'Unknown'

        return undefined_contact

    @property

        elif prefix.strip() == '':

            raise VCSError("Prefix cannot be empty")

        archival.archive(self.repository._repo, stream, self.raw_id,

                         kind, prefix=prefix, subrepos=subrepos)

    def get_nodes(self, path):

        """

        Returns combined ``DirNode`` and ``FileNode`` objects list representing

        state of changeset at the given ``path``. If node at the given ``path``

        is not instance of ``DirNode``, ChangesetError would be raised.

        """

        if self._get_kind(path) != NodeKind.DIR:

            raise ChangesetError("Directory does not exist for revision %s at "

                " '%s'" % (self.revision, path))

        path = self._fix_path(path)

        filenodes = [FileNode(f, changeset=self) for f in self._file_paths

            if os.path.dirname(f) == path]

        dirs = path == '' and '' or [d for d in self._dir_paths

            if d and posixpath.dirname(d) == path]

        dirnodes = [DirNode(d, changeset=self) for d in dirs

            if os.path.dirname(d) == path]

        als = self.repository.alias

        for k, vals in self._extract_submodules().iteritems():

            #vals = url,rev,type

            loc = vals[0]

            cs = vals[1]

            dirnodes.append(SubModuleNode(k, url=loc, changeset=cs,

                                          alias=als))

        nodes = dirnodes + filenodes

        # cache nodes

        for node in nodes:

            self.nodes[node.path] = node

        nodes.sort()

        return nodes

    def get_node(self, path):

        """

        Returns ``Node`` object from the given ``path``. If there is no node at

        the given ``path``, ``ChangesetError`` would be raised.

        """

        path = self._fix_path(path)

            if path in self._file_paths:

                node = FileNode(path, changeset=self)

            elif path in self._dir_paths or path in self._dir_paths:

                if path == '':

                    node = RootNode(changeset=self)

                else:

                    node = DirNode(path, changeset=self)

            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

        """

        return self._ctx.files()

    @property

    def added(self):

        """

        Returns list of added ``FileNode`` objects.

        """

        return AddedFileNodesGenerator([n for n in self.status[1]], self)

    @property

    def changed(self):

        """

        Returns list of modified ``FileNode`` objects.

        """

        return ChangedFileNodesGenerator([n for n in self.status[0]], self)

    @property

    def removed(self):

        """

        Returns list of removed ``FileNode`` objects.

        """

        return RemovedFileNodesGenerator([n for n in self.status[2]], self)

    @LazyProperty

    def extra(self):

        return self._ctx.extra()

        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

        _revset_predicates = {

                'branch': 'branch',

                'book': 'bookmark',

                'tag': 'tag',

                'rev': 'id',

            }

        # avoid expensive branch(x) iteration over whole repo

        rev_spec = "%%s & %s(%%s)" % _revset_predicates[ref_type]

        try:

            revs = self._repo.revs(rev_spec, ref_name, ref_name)

        except LookupError:

            msg = ("Ambiguous identifier %s:%s for %s" % (ref_type, ref_name, self.name))

            raise ChangesetDoesNotExistError(msg)

        except RepoLookupError:

            msg = ("Revision %s:%s does not exist for %s" % (ref_type, ref_name, self.name))

            raise ChangesetDoesNotExistError(msg)

        if revs:

            try:

                revision = revs.last()

            except AttributeError:

                # removed in hg 3.2

                revision = revs[-1]

        else:

            # TODO: just report 'not found'?

            revision = ref_name

        return self._get_revision(revision)

    def _get_archives(self, archive_name='tip'):

        allowed = self.baseui.configlist("web", "allow_archive",

                                         untrusted=True)

        for i in [('zip', '.zip'), ('gz', '.tar.gz'), ('bz2', '.tar.bz2')]:

            if i[0] in allowed or self._repo.ui.configbool("web",

                                                           "allow" + i[0],

                                                           untrusted=True):

                yield {"type": i[0], "extension": i[1], "node": archive_name}

    def _get_url(self, url):

        """

        Returns normalized url. If schema is not given, would fall

        to filesystem

        (``file:///``) schema.

        """

        url = safe_str(url)

            url = "file:" + urllib.pathname2url(url)

        return url

    def get_hook_location(self):

        """

        returns absolute path to location where hooks are stored

        """

        return os.path.join(self.path, '.hg', '.hgrc')

    def get_changeset(self, revision=None):

        """

        Returns ``MercurialChangeset`` object representing repository's

        changeset at the given ``revision``.

        """

        revision = self._get_revision(revision)

        changeset = MercurialChangeset(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 ``MercurialChangeset`` objects from start to end

        (both are inclusive)

        :param start: None, str, int or mercurial lookup format

        :param end:  None, str, int or mercurial lookup format

        :param start_date:

        :param end_date:

        :param branch_name:

        :param reversed: return changesets in reversed order

        """

        start_raw_id = self._get_revision(start)

        start_pos = self.revisions.index(start_raw_id) if start else None

        end_raw_id = self._get_revision(end)

        end_pos = self.revisions.index(end_raw_id) if end else None

        if None not in [start, end] and start_pos > end_pos:

            raise RepositoryError("Start revision '%s' cannot be "

                                  "after end revision '%s'" % (start, end))

        if branch_name and branch_name not in self.allbranches.keys():

            msg = ("Branch %s not found in %s" % (branch_name, self))

            raise BranchDoesNotExistError(msg)

        if end_pos is not None:

            end_pos += 1

        # filter branches

        filter_ = []

        import chardet

        encoding = chardet.detect(unicode_)['encoding']

        if encoding is None:

            raise UnicodeEncodeError()