#

# 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.indexers.daemon

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

A daemon will read from task table and run tasks

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: Jan 26, 2010

:author: marcink

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

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

"""

import logging

import os

import sys

import traceback

from os.path import dirname

from shutil import rmtree

from time import mktime

from whoosh.index import create_in, exists_in, open_dir

from whoosh.qparser import QueryParser

from kallithea.lib.conf import INDEX_EXTENSIONS, INDEX_FILENAMES

from kallithea.lib.indexers import CHGSET_IDX_NAME, CHGSETS_SCHEMA, IDX_NAME, SCHEMA

from kallithea.lib.utils2 import safe_str

from kallithea.lib.vcs.exceptions import ChangesetDoesNotExistError, ChangesetError, NodeDoesNotExistError, RepositoryError

from kallithea.model import db

from kallithea.model.scm import ScmModel

# Add location of top level folder to sys.path

project_path = dirname(dirname(dirname(dirname(os.path.realpath(__file__)))))

sys.path.append(project_path)

log = logging.getLogger('whoosh_indexer')

class WhooshIndexingDaemon(object):

    """

    Daemon for atomic indexing jobs

    """

    def __init__(self, indexname=IDX_NAME, index_location=None,

                 repo_location=None, repo_list=None,

                 repo_update_list=None):

        self.indexname = indexname

        self.index_location = index_location

        if not index_location:

            raise Exception('You have to provide index location')

        self.repo_location = repo_location

        if not repo_location:

            raise Exception('You have to provide repositories location')

        self.repo_paths = ScmModel().repo_scan(self.repo_location)

        # filter repo list

        if repo_list:

            repo_list = set(repo_list)

            self.filtered_repo_paths = {}

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

                if repo_name in repo_list:

                    self.filtered_repo_paths[repo_name] = repo

            self.repo_paths = self.filtered_repo_paths

        # filter update repo list

        self.filtered_repo_update_paths = {}

        if repo_update_list:

            self.filtered_repo_update_paths = {}

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

                if repo_name in repo_update_list:

                    self.filtered_repo_update_paths[repo_name] = repo

            self.repo_paths = self.filtered_repo_update_paths

        self.initial = True

        if not os.path.isdir(self.index_location):

            os.makedirs(self.index_location)

            log.info('Cannot run incremental index since it does not '

                     'yet exist - running full build')

        elif not exists_in(self.index_location, IDX_NAME):

            log.info('Running full index build, as the file content '

                     'index does not exist')

        elif not exists_in(self.index_location, CHGSET_IDX_NAME):

            log.info('Running full index build, as the changeset '

                     'index does not exist')

        else:

            self.initial = False

    def _get_index_revision(self, repo):

        db_repo = db.Repository.get_by_repo_name(repo.name)

        landing_rev = 'tip'

        if db_repo:

            _rev_type, _rev = db_repo.landing_rev

            landing_rev = _rev

        return landing_rev

    def _get_index_changeset(self, repo, index_rev=None):

        if not index_rev:

            index_rev = self._get_index_revision(repo)

        cs = repo.get_changeset(index_rev)

        return cs

    def get_paths(self, repo):

        """

        recursive walk in root dir and return a set of all path in that dir

        based on repository walk function

        """

        index_paths_ = set()

        try:

            cs = self._get_index_changeset(repo)

            for _topnode, _dirs, files in cs.walk('/'):

                for f in files:

                    index_paths_.add(os.path.join(repo.path, f.path))

        except RepositoryError:

            log.debug(traceback.format_exc())

            pass

        return index_paths_

    def get_node(self, repo, path, index_rev=None):

        """

        gets a filenode based on given full path.

        :param repo: scm repo instance

        :param path: full path including root location

        :return: FileNode

        """

        # FIXME: paths should be normalized ... or even better: don't include repo.path

        assert path.startswith(repo.path)

        assert path[len(repo.path)] in (os.path.sep, os.path.altsep)

        node_path = path[len(repo.path) + 1:]

        cs = self._get_index_changeset(repo, index_rev=index_rev)

        node = cs.get_node(node_path)

        return node

    def is_indexable_node(self, node):

        """

        Just index the content of chosen files, skipping binary files

        """

        return (node.extension in INDEX_EXTENSIONS or node.name in INDEX_FILENAMES) and \

               not node.is_binary

    def get_node_mtime(self, node):

        return mktime(node.last_changeset.date.timetuple())

    def add_doc(self, writer, path, repo, repo_name, index_rev=None):

        """

        Adding doc to writer this function itself fetches data from

        the instance of vcs backend

        """

        try:

            node = self.get_node(repo, path, index_rev)

        except (ChangesetError, NodeDoesNotExistError):

            log.debug("    >> %s - not found in %s %s", path, repo, index_rev)

            return 0, 0

        indexed = indexed_w_content = 0

        if self.is_indexable_node(node):

            bytes_content = node.content

            if b'\0' in bytes_content:

                log.warning('    >> %s - no text content', path)

                u_content = ''

            else:

                log.debug('    >> %s', path)

                u_content = safe_str(bytes_content)

                indexed_w_content += 1

        else:

            log.debug('    >> %s - not indexable', path)

            # just index file name without it's content

            u_content = ''

            indexed += 1

        writer.add_document(

            fileid=path,

            repository_rawname=repo_name,

            repository=repo_name,

            path=path,

            content=u_content,

            modtime=self.get_node_mtime(node),

            extension=node.extension

        )

        return indexed, indexed_w_content

    def index_changesets(self, writer, repo_name, repo, start_rev=None):

        """

        Add all changeset in the vcs repo starting at start_rev

        to the index writer

        :param writer: the whoosh index writer to add to

        :param repo_name: name of the repository from whence the

          changeset originates including the repository group

        :param repo: the vcs repository instance to index changesets for,

          the presumption is the repo has changesets to index

        :param start_rev=None: the full sha id to start indexing from

          if start_rev is None then index from the first changeset in

          the repo

        """

        if start_rev is None:

            start_rev = repo[0].raw_id

        log.debug('Indexing changesets in %s, starting at rev %s',

                  repo_name, start_rev)

        indexed = 0

        cs_iter = repo.get_changesets(start=start_rev)

        total = len(cs_iter)

        for cs in cs_iter:

            indexed += 1

            log.debug('    >> %s %s/%s', cs, indexed, total)

            writer.add_document(

                raw_id=cs.raw_id,

                date=cs._timestamp,

                repository_rawname=repo_name,

                repository=repo_name,

                author=cs.author,

                message=cs.message,

                last=cs.last,

                added=' '.join(node.path for node in cs.added).lower(),

                removed=' '.join(node.path for node in cs.removed).lower(),

                changed=' '.join(node.path for node in cs.changed).lower(),

                parents=' '.join(cs.raw_id for cs in cs.parents),

            )

        return indexed

    def index_files(self, file_idx_writer, repo_name, repo):

        """

        Index files for given repo_name

        :param file_idx_writer: the whoosh index writer to add to

        :param repo_name: name of the repository we're indexing

        :param repo: instance of vcs repo

        """

        i_cnt = iwc_cnt = 0

        log.debug('Building file index for %s @revision:%s', repo_name,

                                                self._get_index_revision(repo))

        index_rev = self._get_index_revision(repo)

        for idx_path in self.get_paths(repo):

            i, iwc = self.add_doc(file_idx_writer, idx_path, repo, repo_name, index_rev)

            i_cnt += i

            iwc_cnt += iwc

        log.debug('added %s files %s with content for repo %s',

                  i_cnt + iwc_cnt, iwc_cnt, repo.path)

        return i_cnt, iwc_cnt

    def update_changeset_index(self):

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

        with idx.searcher() as searcher:

            writer = idx.writer()

            writer_is_dirty = False

            try:

                indexed_total = 0

                repo_name = None

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

                    log.debug('Updating changeset index for repo %s', repo_name)

                    # skip indexing if there aren't any revs in the repo

                    num_of_revs = len(repo)

                    if num_of_revs < 1:

                        continue

                    qp = QueryParser('repository', schema=CHGSETS_SCHEMA)

                    q = qp.parse("last:t AND %s" % repo_name)

                    results = searcher.search(q)

                    # default to scanning the entire repo

                    last_rev = 0

                    start_id = None

                    if len(results) > 0:

                        # assuming that there is only one result, if not this

                        # may require a full re-index.

                        start_id = results[0]['raw_id']

                        try:

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

                        except ChangesetDoesNotExistError:

                            log.error('previous last revision %s not found - indexing from scratch', start_id)

                            start_id = None

                    # 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("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('STARTING INCREMENTAL INDEXING UPDATE FOR EXTENSIONS %s '

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

        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)

                    if indexed_repo_path not in self.filtered_repo_update_paths:

                        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 sorted(self.repo_paths.items()):

                log.debug('Updating file index for repo %s', repo_name)

                # 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):

                    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

                        ri_cnt += i

                        ri_cnt_total += 1

                        riwc_cnt += iwc

                        riwc_cnt_total += iwc

                log.debug('added %s files %s with content for repo %s',

                             ri_cnt + riwc_cnt, riwc_cnt, repo.path

                )

            log.debug('indexed %s files in total and %s with content',

                        ri_cnt_total, riwc_cnt_total

            )

        finally:

            if writer_is_dirty:

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

                writer.commit(merge=True)

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

            else:

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

                writer.cancel()

    def build_indexes(self):

        if os.path.exists(self.index_location):

            log.debug('removing previous index')

            rmtree(self.index_location)

        if not os.path.exists(self.index_location):

            os.mkdir(self.index_location)

        chgset_idx = create_in(self.index_location, CHGSETS_SCHEMA,

                               indexname=CHGSET_IDX_NAME)

        chgset_idx_writer = chgset_idx.writer()

        file_idx = create_in(self.index_location, SCHEMA, indexname=IDX_NAME)

            parsed_url.port  # trigger netloc parsing which might raise ValueError

        except ValueError:

            raise urllib.error.URLError("Error parsing URL: %r" % url)

        # check first if it's not an local url

        if os.path.isabs(url) and os.path.isdir(url):

            return

        if parsed_url.scheme == 'git':

            # Mitigate problems elsewhere with incorrect handling of encoded paths.

            # Don't trust urllib.parse.unquote but be prepared for more flexible implementations elsewhere.

            # Space is the only allowed whitespace character - directly or % encoded. No other % or \ is allowed.

            for c in parsed_url.path.replace('%20', ' '):

                if c in '%\\':

                    raise urllib.error.URLError("Invalid escape character in path: '%s'" % c)

                if c.isspace() and c != ' ':

                    raise urllib.error.URLError("Invalid whitespace character in path: %r" % c)

            return

        if parsed_url.scheme not in ['http', 'https']:

            raise urllib.error.URLError("Unsupported protocol in URL %r" % url)

        url_obj = mercurial.util.url(safe_bytes(url))

        test_uri, handlers = get_urllib_request_handlers(url_obj)

        if not test_uri.endswith(b'info/refs'):

            test_uri = test_uri.rstrip(b'/') + b'/info/refs'

        url_obj.passwd = b'*****'

        cleaned_uri = str(url_obj)

        o = urllib.request.build_opener(*handlers)

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

        req = urllib.request.Request(

            "%s?%s" % (

                safe_str(test_uri),

                urllib.parse.urlencode({"service": 'git-upload-pack'})

            ))

        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 urllib.error.URLError("[%s] org_exc: %s" % (cleaned_uri, e))

        # now detect if it's proper git repo

        gitdata = resp.read()

        if b'service=git-upload-pack' not in gitdata:

            raise urllib.error.URLError(

                "url [%s] does not look like an git" % cleaned_uri)

    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 = self.run_git_command(cmd)

        except RepositoryError:

            # Can be raised for empty repositories

            return []

        return so.splitlines()

    def _get_all_revisions2(self):

        # alternate implementation using dulwich

        includes = [ascii_str(sha) for key, (sha, type_) in self._parsed_refs.items()

                    if type_ != b'T']

        return [c.commit.id for c in self._repo.get_walker(include=includes)]

    def _get_revision(self, revision):

        """

        Given any revision identifier, returns a 40 char string with revision hash.

        """

        if self._empty:

            raise EmptyRepositoryError("There are no changesets yet")

        if revision in (None, '', 'tip', 'HEAD', 'head', -1):

            revision = -1

        if isinstance(revision, int):

            try:

                return self.revisions[revision]

            except IndexError:

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

                raise ChangesetDoesNotExistError(msg)

        if isinstance(revision, str):

            if revision.isdigit() and (len(revision) < 12 or len(revision) == revision.count('0')):

                try:

                    return self.revisions[int(revision)]

                except IndexError:

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

                    raise ChangesetDoesNotExistError(msg)

            # get by branch/tag name

            _ref_revision = self._parsed_refs.get(safe_bytes(revision))

            if _ref_revision:  # and _ref_revision[1] in [b'H', b'RH', b'T']:

                return ascii_str(_ref_revision[0])

            if revision in self.revisions:

                return revision

            # maybe it's a tag ? we don't have them in self.revisions

            if revision in self.tags.values():

                return revision

            if SHA_PATTERN.match(revision):

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

                raise ChangesetDoesNotExistError(msg)

        raise ChangesetDoesNotExistError("Given revision %r not recognized" % revision)

    def get_ref_revision(self, ref_type, ref_name):

        """

        Returns ``GitChangeset`` 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.

        """

        if url != 'default' and '://' not in url:

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

        return url

    @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):

        return safe_str(self._repo.get_description() or b'unknown')

    @property

    def branches(self):

        if not self.revisions:

            return {}

        _branches = [(safe_str(key), ascii_str(sha))

                     for key, (sha, type_) in self._parsed_refs.items() if type_ == b'H']

        return OrderedDict(sorted(_branches, key=(lambda ctx: ctx[0]), reverse=False))

    @LazyProperty

    def closed_branches(self):

        return {}

    @LazyProperty

    def tags(self):

        return self._get_tags()

    def _get_tags(self):

        if not self.revisions:

            return {}

        _tags = [(safe_str(key), ascii_str(sha))

                 for key, (sha, type_) in self._parsed_refs.items() if type_ == b'T']

        return OrderedDict(sorted(_tags, key=(lambda ctx: ctx[0]), 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[b"refs/tags/%s" % safe_bytes(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)

        # self._repo.refs is a DiskRefsContainer, and .path gives the full absolute path of '.git'

        tagpath = os.path.join(safe_str(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):

        """Return refs as a dict, like:

        { b'v0.2.0': [b'599ba911aa24d2981225f3966eb659dfae9e9f30', b'T'] }

        """

        _repo = self._repo

        refs = _repo.get_refs()

        keys = [(b'refs/heads/', b'H'),

                (b'refs/remotes/origin/', b'RH'),

                (b'refs/tags/', b'T')]

        _refs = {}

        for ref, sha in refs.items():

            for k, type_ in keys:

                if ref.startswith(k):

                    _key = ref[len(k):]

                    if type_ == b'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 [b'refs/heads/', b'refs/remotes/origin/']:

                if key.startswith(ref_key):

                    n = key[len(ref_key):]

                    if n not in [b'HEAD']:

                        heads[n] = val

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

    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, changeset.GitChangeset):

            return revision

        return changeset.GitChangeset(repository=self, revision=self._get_revision(revision))

    def get_changesets(self, start=None, end=None, start_date=None,

           end_date=None, branch_name=None, reverse=False, max_revisions=None):

        """

        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 max_revisions:

            cmd += ['--max-count=%s' % max_revisions]

        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).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

        """

        return self._get_bookmarks()

    def _get_bookmarks(self):

        if self._empty:

            return {}

        return OrderedDict(sorted(

            ((safe_str(n), ascii_str(mercurial.node.hex(h))) for n, h in self._repo._bookmarks.items()),

            reverse=True,

            key=lambda x: x[0],  # sort by name

        ))

    def _get_all_revisions(self):

        return [ascii_str(self._repo[x].hex()) for x in self._repo.filtered(b'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``. 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 = mercurial.match.exact([safe_bytes(path)])

        else:

            file_filter = None