commits: re-implemented fetching a single commit for git case.

- faster using always cache if available
- fixed problems after changes of __bool__ of repo object which always rebuilt cache and this is not the case now
- make few error messages better

rhodecode/apps/repository/views/repo_commits.py

@@ -1,502 +1,501 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # 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 Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 
 import logging
 import collections
 
 from pyramid.httpexceptions import HTTPNotFound, HTTPBadRequest, HTTPFound
 from pyramid.view import view_config
 from pyramid.renderers import render
 from pyramid.response import Response
 
 from rhodecode.apps._base import RepoAppView
 
 from rhodecode.lib import diffs, codeblocks
 from rhodecode.lib.auth import (
     LoginRequired, HasRepoPermissionAnyDecorator, NotAnonymous, CSRFRequired)
 
 from rhodecode.lib.compat import OrderedDict
 from rhodecode.lib.diffs import (
     cache_diff, load_cached_diff, diff_cache_exist, get_diff_context,
     get_diff_whitespace_flag)
 from rhodecode.lib.exceptions import StatusChangeOnClosedPullRequestError
 import rhodecode.lib.helpers as h
 from rhodecode.lib.utils2 import safe_unicode, str2bool
 from rhodecode.lib.vcs.backends.base import EmptyCommit
 from rhodecode.lib.vcs.exceptions import (
     RepositoryError, CommitDoesNotExistError)
 from rhodecode.model.db import ChangesetComment, ChangesetStatus
 from rhodecode.model.changeset_status import ChangesetStatusModel
 from rhodecode.model.comment import CommentsModel
 from rhodecode.model.meta import Session
 from rhodecode.model.settings import VcsSettingsModel
 
 log = logging.getLogger(__name__)
 
 
 def _update_with_GET(params, request):
     for k in ['diff1', 'diff2', 'diff']:
         params[k] += request.GET.getall(k)
 
 
 
 
 
 class RepoCommitsView(RepoAppView):
     def load_default_context(self):
         c = self._get_local_tmpl_context(include_app_defaults=True)
         c.rhodecode_repo = self.rhodecode_vcs_repo
 
         return c
 
     def _is_diff_cache_enabled(self, target_repo):
         caching_enabled = self._get_general_setting(
             target_repo, 'rhodecode_diff_cache')
         log.debug('Diff caching enabled: %s', caching_enabled)
         return caching_enabled
 
     def _commit(self, commit_id_range, method):
         _ = self.request.translate
         c = self.load_default_context()
         c.fulldiff = self.request.GET.get('fulldiff')
 
         # fetch global flags of ignore ws or context lines
         diff_context = get_diff_context(self.request)
         hide_whitespace_changes = get_diff_whitespace_flag(self.request)
 
         # diff_limit will cut off the whole diff if the limit is applied
         # otherwise it will just hide the big files from the front-end
         diff_limit = c.visual.cut_off_limit_diff
         file_limit = c.visual.cut_off_limit_file
 
         # get ranges of commit ids if preset
         commit_range = commit_id_range.split('...')[:2]
 
         try:
             pre_load = ['affected_files', 'author', 'branch', 'date',
                         'message', 'parents']
 
             if len(commit_range) == 2:
                 commits = self.rhodecode_vcs_repo.get_commits(
                     start_id=commit_range[0], end_id=commit_range[1],
                     pre_load=pre_load, translate_tags=False)
                 commits = list(commits)
             else:
                 commits = [self.rhodecode_vcs_repo.get_commit(
                     commit_id=commit_id_range, pre_load=pre_load)]
 
             c.commit_ranges = commits
             if not c.commit_ranges:
-                raise RepositoryError(
-                    'The commit range returned an empty result')
-        except CommitDoesNotExistError:
-            msg = _('No such commit exists for this repository')
+                raise RepositoryError('The commit range returned an empty result')
+        except CommitDoesNotExistError as e:
+            msg = _('No such commit exists. Org exception: `{}`').format(e)
             h.flash(msg, category='error')
             raise HTTPNotFound()
         except Exception:
             log.exception("General failure")
             raise HTTPNotFound()
 
         c.changes = OrderedDict()
         c.lines_added = 0
         c.lines_deleted = 0
 
         # auto collapse if we have more than limit
         collapse_limit = diffs.DiffProcessor._collapse_commits_over
         c.collapse_all_commits = len(c.commit_ranges) > collapse_limit
 
         c.commit_statuses = ChangesetStatus.STATUSES
         c.inline_comments = []
         c.files = []
 
         c.statuses = []
         c.comments = []
         c.unresolved_comments = []
         if len(c.commit_ranges) == 1:
             commit = c.commit_ranges[0]
             c.comments = CommentsModel().get_comments(
                 self.db_repo.repo_id,
                 revision=commit.raw_id)
             c.statuses.append(ChangesetStatusModel().get_status(
                 self.db_repo.repo_id, commit.raw_id))
             # comments from PR
             statuses = ChangesetStatusModel().get_statuses(
                 self.db_repo.repo_id, commit.raw_id,
                 with_revisions=True)
             prs = set(st.pull_request for st in statuses
                       if st.pull_request is not None)
             # from associated statuses, check the pull requests, and
             # show comments from them
             for pr in prs:
                 c.comments.extend(pr.comments)
 
             c.unresolved_comments = CommentsModel()\
                 .get_commit_unresolved_todos(commit.raw_id)
 
         diff = None
         # Iterate over ranges (default commit view is always one commit)
         for commit in c.commit_ranges:
             c.changes[commit.raw_id] = []
 
             commit2 = commit
             commit1 = commit.first_parent
 
             if method == 'show':
                 inline_comments = CommentsModel().get_inline_comments(
                     self.db_repo.repo_id, revision=commit.raw_id)
                 c.inline_cnt = CommentsModel().get_inline_comments_count(
                     inline_comments)
                 c.inline_comments = inline_comments
 
                 cache_path = self.rhodecode_vcs_repo.get_create_shadow_cache_pr_path(
                     self.db_repo)
                 cache_file_path = diff_cache_exist(
                     cache_path, 'diff', commit.raw_id,
                     hide_whitespace_changes, diff_context, c.fulldiff)
 
                 caching_enabled = self._is_diff_cache_enabled(self.db_repo)
                 force_recache = str2bool(self.request.GET.get('force_recache'))
 
                 cached_diff = None
                 if caching_enabled:
                     cached_diff = load_cached_diff(cache_file_path)
 
                 has_proper_diff_cache = cached_diff and cached_diff.get('diff')
                 if not force_recache and has_proper_diff_cache:
                     diffset = cached_diff['diff']
                 else:
                     vcs_diff = self.rhodecode_vcs_repo.get_diff(
                         commit1, commit2,
                         ignore_whitespace=hide_whitespace_changes,
                         context=diff_context)
 
                     diff_processor = diffs.DiffProcessor(
                         vcs_diff, format='newdiff', diff_limit=diff_limit,
                         file_limit=file_limit, show_full_diff=c.fulldiff)
 
                     _parsed = diff_processor.prepare()
 
                     diffset = codeblocks.DiffSet(
                         repo_name=self.db_repo_name,
                         source_node_getter=codeblocks.diffset_node_getter(commit1),
                         target_node_getter=codeblocks.diffset_node_getter(commit2))
 
                     diffset = self.path_filter.render_patchset_filtered(
                         diffset, _parsed, commit1.raw_id, commit2.raw_id)
 
                     # save cached diff
                     if caching_enabled:
                         cache_diff(cache_file_path, diffset, None)
 
                 c.limited_diff = diffset.limited_diff
                 c.changes[commit.raw_id] = diffset
             else:
                 # TODO(marcink): no cache usage here...
                 _diff = self.rhodecode_vcs_repo.get_diff(
                     commit1, commit2,
                     ignore_whitespace=hide_whitespace_changes, context=diff_context)
                 diff_processor = diffs.DiffProcessor(
                     _diff, format='newdiff', diff_limit=diff_limit,
                     file_limit=file_limit, show_full_diff=c.fulldiff)
                 # downloads/raw we only need RAW diff nothing else
                 diff = self.path_filter.get_raw_patch(diff_processor)
                 c.changes[commit.raw_id] = [None, None, None, None, diff, None, None]
 
         # sort comments by how they were generated
         c.comments = sorted(c.comments, key=lambda x: x.comment_id)
 
         if len(c.commit_ranges) == 1:
             c.commit = c.commit_ranges[0]
             c.parent_tmpl = ''.join(
                 '# Parent  %s\n' % x.raw_id for x in c.commit.parents)
 
         if method == 'download':
             response = Response(diff)
             response.content_type = 'text/plain'
             response.content_disposition = (
                 'attachment; filename=%s.diff' % commit_id_range[:12])
             return response
         elif method == 'patch':
             c.diff = safe_unicode(diff)
             patch = render(
                 'rhodecode:templates/changeset/patch_changeset.mako',
                 self._get_template_context(c), self.request)
             response = Response(patch)
             response.content_type = 'text/plain'
             return response
         elif method == 'raw':
             response = Response(diff)
             response.content_type = 'text/plain'
             return response
         elif method == 'show':
             if len(c.commit_ranges) == 1:
                 html = render(
                     'rhodecode:templates/changeset/changeset.mako',
                     self._get_template_context(c), self.request)
                 return Response(html)
             else:
                 c.ancestor = None
                 c.target_repo = self.db_repo
                 html = render(
                     'rhodecode:templates/changeset/changeset_range.mako',
                     self._get_template_context(c), self.request)
                 return Response(html)
 
         raise HTTPBadRequest()
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @view_config(
         route_name='repo_commit', request_method='GET',
         renderer=None)
     def repo_commit_show(self):
         commit_id = self.request.matchdict['commit_id']
         return self._commit(commit_id, method='show')
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @view_config(
         route_name='repo_commit_raw', request_method='GET',
         renderer=None)
     @view_config(
         route_name='repo_commit_raw_deprecated', request_method='GET',
         renderer=None)
     def repo_commit_raw(self):
         commit_id = self.request.matchdict['commit_id']
         return self._commit(commit_id, method='raw')
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @view_config(
         route_name='repo_commit_patch', request_method='GET',
         renderer=None)
     def repo_commit_patch(self):
         commit_id = self.request.matchdict['commit_id']
         return self._commit(commit_id, method='patch')
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @view_config(
         route_name='repo_commit_download', request_method='GET',
         renderer=None)
     def repo_commit_download(self):
         commit_id = self.request.matchdict['commit_id']
         return self._commit(commit_id, method='download')
 
     @LoginRequired()
     @NotAnonymous()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @CSRFRequired()
     @view_config(
         route_name='repo_commit_comment_create', request_method='POST',
         renderer='json_ext')
     def repo_commit_comment_create(self):
         _ = self.request.translate
         commit_id = self.request.matchdict['commit_id']
 
         c = self.load_default_context()
         status = self.request.POST.get('changeset_status', None)
         text = self.request.POST.get('text')
         comment_type = self.request.POST.get('comment_type')
         resolves_comment_id = self.request.POST.get('resolves_comment_id', None)
 
         if status:
             text = text or (_('Status change %(transition_icon)s %(status)s')
                             % {'transition_icon': '>',
                                'status': ChangesetStatus.get_status_lbl(status)})
 
         multi_commit_ids = []
         for _commit_id in self.request.POST.get('commit_ids', '').split(','):
             if _commit_id not in ['', None, EmptyCommit.raw_id]:
                 if _commit_id not in multi_commit_ids:
                     multi_commit_ids.append(_commit_id)
 
         commit_ids = multi_commit_ids or [commit_id]
 
         comment = None
         for current_id in filter(None, commit_ids):
             comment = CommentsModel().create(
                 text=text,
                 repo=self.db_repo.repo_id,
                 user=self._rhodecode_db_user.user_id,
                 commit_id=current_id,
                 f_path=self.request.POST.get('f_path'),
                 line_no=self.request.POST.get('line'),
                 status_change=(ChangesetStatus.get_status_lbl(status)
                                if status else None),
                 status_change_type=status,
                 comment_type=comment_type,
                 resolves_comment_id=resolves_comment_id,
                 auth_user=self._rhodecode_user
             )
 
             # get status if set !
             if status:
                 # if latest status was from pull request and it's closed
                 # disallow changing status !
                 # dont_allow_on_closed_pull_request = True !
 
                 try:
                     ChangesetStatusModel().set_status(
                         self.db_repo.repo_id,
                         status,
                         self._rhodecode_db_user.user_id,
                         comment,
                         revision=current_id,
                         dont_allow_on_closed_pull_request=True
                     )
                 except StatusChangeOnClosedPullRequestError:
                     msg = _('Changing the status of a commit associated with '
                             'a closed pull request is not allowed')
                     log.exception(msg)
                     h.flash(msg, category='warning')
                     raise HTTPFound(h.route_path(
                         'repo_commit', repo_name=self.db_repo_name,
                         commit_id=current_id))
 
         # finalize, commit and redirect
         Session().commit()
 
         data = {
             'target_id': h.safeid(h.safe_unicode(
                 self.request.POST.get('f_path'))),
         }
         if comment:
             c.co = comment
             rendered_comment = render(
                 'rhodecode:templates/changeset/changeset_comment_block.mako',
                 self._get_template_context(c), self.request)
 
             data.update(comment.get_dict())
             data.update({'rendered_text': rendered_comment})
 
         return data
 
     @LoginRequired()
     @NotAnonymous()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @CSRFRequired()
     @view_config(
         route_name='repo_commit_comment_preview', request_method='POST',
         renderer='string', xhr=True)
     def repo_commit_comment_preview(self):
         # Technically a CSRF token is not needed as no state changes with this
         # call. However, as this is a POST is better to have it, so automated
         # tools don't flag it as potential CSRF.
         # Post is required because the payload could be bigger than the maximum
         # allowed by GET.
 
         text = self.request.POST.get('text')
         renderer = self.request.POST.get('renderer') or 'rst'
         if text:
             return h.render(text, renderer=renderer, mentions=True)
         return ''
 
     @LoginRequired()
     @NotAnonymous()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @CSRFRequired()
     @view_config(
         route_name='repo_commit_comment_delete', request_method='POST',
         renderer='json_ext')
     def repo_commit_comment_delete(self):
         commit_id = self.request.matchdict['commit_id']
         comment_id = self.request.matchdict['comment_id']
 
         comment = ChangesetComment.get_or_404(comment_id)
         if not comment:
             log.debug('Comment with id:%s not found, skipping', comment_id)
             # comment already deleted in another call probably
             return True
 
         is_repo_admin = h.HasRepoPermissionAny('repository.admin')(self.db_repo_name)
         super_admin = h.HasPermissionAny('hg.admin')()
         comment_owner = (comment.author.user_id == self._rhodecode_db_user.user_id)
         is_repo_comment = comment.repo.repo_name == self.db_repo_name
         comment_repo_admin = is_repo_admin and is_repo_comment
 
         if super_admin or comment_owner or comment_repo_admin:
             CommentsModel().delete(comment=comment, auth_user=self._rhodecode_user)
             Session().commit()
             return True
         else:
             log.warning('No permissions for user %s to delete comment_id: %s',
                         self._rhodecode_db_user, comment_id)
             raise HTTPNotFound()
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @view_config(
         route_name='repo_commit_data', request_method='GET',
         renderer='json_ext', xhr=True)
     def repo_commit_data(self):
         commit_id = self.request.matchdict['commit_id']
         self.load_default_context()
 
         try:
             return self.rhodecode_vcs_repo.get_commit(commit_id=commit_id)
         except CommitDoesNotExistError as e:
             return EmptyCommit(message=str(e))
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @view_config(
         route_name='repo_commit_children', request_method='GET',
         renderer='json_ext', xhr=True)
     def repo_commit_children(self):
         commit_id = self.request.matchdict['commit_id']
         self.load_default_context()
 
         try:
             commit = self.rhodecode_vcs_repo.get_commit(commit_id=commit_id)
             children = commit.children
         except CommitDoesNotExistError:
             children = []
 
         result = {"results": children}
         return result
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @view_config(
         route_name='repo_commit_parents', request_method='GET',
         renderer='json_ext')
     def repo_commit_parents(self):
         commit_id = self.request.matchdict['commit_id']
         self.load_default_context()
 
         try:
             commit = self.rhodecode_vcs_repo.get_commit(commit_id=commit_id)
             parents = commit.parents
         except CommitDoesNotExistError:
             parents = []
         result = {"results": parents}
         return result

rhodecode/lib/vcs/backends/git/inmemory.py

@@ -1,105 +1,106 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # 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 Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 GIT inmemory module
 """
 
 from rhodecode.lib.datelib import date_to_timestamp_plus_offset
 from rhodecode.lib.utils import safe_str
 from rhodecode.lib.vcs.backends import base
 
 
 class GitInMemoryCommit(base.BaseInMemoryCommit):
 
     def commit(self, message, author, parents=None, branch=None, date=None,
                **kwargs):
         """
         Performs in-memory commit (doesn't check workdir in any way) and
         returns newly created `GitCommit`. Updates repository's
         `commit_ids`.
 
         :param message: message of the commit
         :param author: full username, i.e. "Joe Doe <[e-mail]>"
         :param parents: single parent or sequence of parents from which commit
           would be derived
         :param date: `datetime.datetime` instance. Defaults to
           ``datetime.datetime.now()``.
         :param branch: branch name, as string. If none given, default backend's
           branch would be used.
 
         :raises `CommitError`: if any error occurs while committing
         """
         self.check_integrity(parents)
         if branch is None:
             branch = self.repository.DEFAULT_BRANCH_NAME
 
         ENCODING = "UTF-8"
 
         commit_tree = None
         if self.parents[0]:
             commit_tree = self.parents[0]._commit['tree']
 
         updated = []
         for node in self.added + self.changed:
             if not node.is_binary:
                 content = node.content.encode(ENCODING)
             else:
                 content = node.content
             updated.append({
                 'path': node.path,
                 'node_path': node.name.encode(ENCODING),
                 'content': content,
                 'mode': node.mode,
             })
 
         removed = [node.path for node in self.removed]
 
         date, tz = date_to_timestamp_plus_offset(date)
 
         # TODO: johbo: Make kwargs explicit and check if this is needed.
         author_time = kwargs.pop('author_time', date)
         author_tz = kwargs.pop('author_timezone', tz)
 
         commit_data = {
             'parents': [p._commit['id'] for p in self.parents if p],
             'author': safe_str(author),
             'committer': safe_str(author),
             'encoding': ENCODING,
             'message': safe_str(message),
             'commit_time': int(date),
             'author_time': int(author_time),
             'commit_timezone': tz,
             'author_timezone': author_tz,
         }
 
         commit_id = self.repository._remote.commit(
             commit_data, branch, commit_tree, updated, removed)
 
         # Update vcs repository object
+        if commit_id not in self.repository.commit_ids:
         self.repository.commit_ids.append(commit_id)
         self.repository._rebuild_cache(self.repository.commit_ids)
 
         # invalidate parsed refs after commit
         self.repository._refs = self.repository._get_refs()
         self.repository.branches = self.repository._get_branches()
         tip = self.repository.get_commit()
         self.reset()
         return tip

rhodecode/lib/vcs/backends/git/repository.py

@@ -1,1012 +1,1031 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # 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 Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 GIT repository module
 """
 
 import logging
 import os
 import re
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from rhodecode.lib.compat import OrderedDict
 from rhodecode.lib.datelib import (
     utcdate_fromtimestamp, makedate, date_astimestamp)
 from rhodecode.lib.utils import safe_unicode, safe_str
 from rhodecode.lib.vcs import connection, path as vcspath
 from rhodecode.lib.vcs.backends.base import (
     BaseRepository, CollectionGenerator, Config, MergeResponse,
     MergeFailureReason, Reference)
 from rhodecode.lib.vcs.backends.git.commit import GitCommit
 from rhodecode.lib.vcs.backends.git.diff import GitDiff
 from rhodecode.lib.vcs.backends.git.inmemory import GitInMemoryCommit
 from rhodecode.lib.vcs.exceptions import (
     CommitDoesNotExistError, EmptyRepositoryError,
     RepositoryError, TagAlreadyExistError, TagDoesNotExistError, VCSError)
 
 
 SHA_PATTERN = re.compile(r'^[[0-9a-fA-F]{12}|[0-9a-fA-F]{40}]$')
 
 log = logging.getLogger(__name__)
 
 
 class GitRepository(BaseRepository):
     """
     Git repository backend.
     """
     DEFAULT_BRANCH_NAME = 'master'
 
     contact = BaseRepository.DEFAULT_CONTACT
 
     def __init__(self, repo_path, config=None, create=False, src_url=None,
                  do_workspace_checkout=False, with_wire=None, bare=False):
 
         self.path = safe_str(os.path.abspath(repo_path))
         self.config = config if config else self.get_default_config()
         self.with_wire = with_wire
 
         self._init_repo(create, src_url, do_workspace_checkout, bare)
 
         # caches
         self._commit_ids = {}
 
     @LazyProperty
     def _remote(self):
         return connection.Git(self.path, self.config, with_wire=self.with_wire)
 
     @LazyProperty
     def bare(self):
         return self._remote.bare()
 
     @LazyProperty
     def head(self):
         return self._remote.head()
 
     @LazyProperty
     def commit_ids(self):
         """
         Returns list of commit ids, in ascending order.  Being lazy
         attribute allows external tools to inject commit ids from cache.
         """
         commit_ids = self._get_all_commit_ids()
         self._rebuild_cache(commit_ids)
         return commit_ids
 
     def _rebuild_cache(self, commit_ids):
         self._commit_ids = dict((commit_id, index)
                                 for index, commit_id in enumerate(commit_ids))
 
     def run_git_command(self, cmd, **opts):
         """
         Runs given ``cmd`` as git command and returns tuple
         (stdout, stderr).
 
         :param cmd: git command to be executed
         :param opts: env options to pass into Subprocess command
         """
         if not isinstance(cmd, list):
             raise ValueError('cmd must be a list, got %s instead' % type(cmd))
 
         skip_stderr_log = opts.pop('skip_stderr_log', False)
         out, err = self._remote.run_git_command(cmd, **opts)
         if err and not skip_stderr_log:
             log.debug('Stderr output of git command "%s":\n%s', cmd, err)
         return out, err
 
     @staticmethod
     def check_url(url, config):
         """
         Function will check given url and try to verify if it's a valid
         link. Sometimes it may happened that git will issue basic
         auth request that can cause whole API to hang when used from python
         or other external calls.
 
         On failures it'll raise urllib2.HTTPError, exception is also thrown
         when the return code is non 200
         """
         # check first if it's not an url
         if os.path.isdir(url) or url.startswith('file:'):
             return True
 
         if '+' in url.split('://', 1)[0]:
             url = url.split('+', 1)[1]
 
         # Request the _remote to verify the url
         return connection.Git.check_url(url, config.serialize())
 
     @staticmethod
     def is_valid_repository(path):
         if os.path.isdir(os.path.join(path, '.git')):
             return True
         # check case of bare repository
         try:
             GitRepository(path)
             return True
         except VCSError:
             pass
         return False
 
     def _init_repo(self, create, src_url=None, do_workspace_checkout=False,
                    bare=False):
         if create and os.path.exists(self.path):
             raise RepositoryError(
                 "Cannot create repository at %s, location already exist"
                 % self.path)
 
         if bare and do_workspace_checkout:
             raise RepositoryError("Cannot update a bare repository")
         try:
 
             if src_url:
                 # check URL before any actions
                 GitRepository.check_url(src_url, self.config)
 
             if create:
                 os.makedirs(self.path, mode=0o755)
 
                 if bare:
                     self._remote.init_bare()
                 else:
                     self._remote.init()
 
                 if src_url and bare:
                     # bare repository only allows a fetch and checkout is not allowed
                     self.fetch(src_url, commit_ids=None)
                 elif src_url:
                     self.pull(src_url, commit_ids=None,
                               update_after=do_workspace_checkout)
 
             else:
                 if not self._remote.assert_correct_path():
                     raise RepositoryError(
                         'Path "%s" does not contain a Git repository' %
                         (self.path,))
 
         # TODO: johbo: check if we have to translate the OSError here
         except OSError as err:
             raise RepositoryError(err)
 
     def _get_all_commit_ids(self, filters=None):
         # 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
 
         head = self._remote.head(show_exc=False)
         if not head:
             return []
 
         rev_filter = ['--branches', '--tags']
         extra_filter = []
 
         if filters:
             if filters.get('since'):
                 extra_filter.append('--since=%s' % (filters['since']))
             if filters.get('until'):
                 extra_filter.append('--until=%s' % (filters['until']))
             if filters.get('branch_name'):
                 rev_filter = ['--tags']
                 extra_filter.append(filters['branch_name'])
         rev_filter.extend(extra_filter)
 
             # if filters.get('start') or filters.get('end'):
             #     # skip is offset, max-count is limit
             #     if filters.get('start'):
             #         extra_filter += ' --skip=%s' % filters['start']
             #     if filters.get('end'):
             #         extra_filter += ' --max-count=%s' % (filters['end'] - (filters['start'] or 0))
 
         cmd = ['rev-list', '--reverse', '--date-order'] + rev_filter
         try:
             output, __ = self.run_git_command(cmd)
         except RepositoryError:
             # Can be raised for empty repositories
             return []
         return output.splitlines()
 
-    def _get_commit_id(self, commit_id_or_idx):
+    def _lookup_commit(self, commit_id_or_idx, translate_tag=True):
         def is_null(value):
             return len(value) == commit_id_or_idx.count('0')
 
-        if self.is_empty():
-            raise EmptyRepositoryError("There are no commits yet")
-
         if commit_id_or_idx in (None, '', 'tip', 'HEAD', 'head', -1):
             return self.commit_ids[-1]
 
         is_bstr = isinstance(commit_id_or_idx, (str, unicode))
         if ((is_bstr and commit_id_or_idx.isdigit() and len(commit_id_or_idx) < 12)
             or isinstance(commit_id_or_idx, int) or is_null(commit_id_or_idx)):
             try:
                 commit_id_or_idx = self.commit_ids[int(commit_id_or_idx)]
             except Exception:
-                msg = "Commit %s does not exist for %s" % (
-                    commit_id_or_idx, self)
+                msg = "Commit %s does not exist for %s" % (commit_id_or_idx, self.name)
                 raise CommitDoesNotExistError(msg)
 
         elif is_bstr:
             # check full path ref, eg. refs/heads/master
             ref_id = self._refs.get(commit_id_or_idx)
             if ref_id:
                 return ref_id
 
             # check branch name
             branch_ids = self.branches.values()
             ref_id = self._refs.get('refs/heads/%s' % commit_id_or_idx)
             if ref_id:
                 return ref_id
 
             # check tag name
             ref_id = self._refs.get('refs/tags/%s' % commit_id_or_idx)
             if ref_id:
                 return ref_id
 
             if (not SHA_PATTERN.match(commit_id_or_idx) or
                     commit_id_or_idx not in self.commit_ids):
-                msg = "Commit %s does not exist for %s" % (
-                    commit_id_or_idx, self)
+                msg = "Commit %s does not exist for %s" % (commit_id_or_idx, self.name)
                 raise CommitDoesNotExistError(msg)
 
         # Ensure we return full id
         if not SHA_PATTERN.match(str(commit_id_or_idx)):
             raise CommitDoesNotExistError(
                 "Given commit id %s not recognized" % commit_id_or_idx)
         return commit_id_or_idx
 
     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 last_change(self):
         """
         Returns last change made on this repository as
         `datetime.datetime` object.
         """
         try:
             return self.get_commit().date
         except RepositoryError:
             tzoffset = makedate()[1]
             return utcdate_fromtimestamp(self._get_fs_mtime(), tzoffset)
 
     def _get_fs_mtime(self):
         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):
         description = self._remote.get_description()
         return safe_unicode(description or self.DEFAULT_DESCRIPTION)
 
     def _get_refs_entries(self, prefix='', reverse=False, strip_prefix=True):
         if self.is_empty():
             return OrderedDict()
 
         result = []
         for ref, sha in self._refs.iteritems():
             if ref.startswith(prefix):
                 ref_name = ref
                 if strip_prefix:
                     ref_name = ref[len(prefix):]
                 result.append((safe_unicode(ref_name), sha))
 
         def get_name(entry):
             return entry[0]
 
         return OrderedDict(sorted(result, key=get_name, reverse=reverse))
 
     def _get_branches(self):
         return self._get_refs_entries(prefix='refs/heads/', strip_prefix=True)
 
     @LazyProperty
     def branches(self):
         return self._get_branches()
 
     @LazyProperty
     def branches_closed(self):
         return {}
 
     @LazyProperty
     def bookmarks(self):
         return {}
 
     @LazyProperty
     def branches_all(self):
         all_branches = {}
         all_branches.update(self.branches)
         all_branches.update(self.branches_closed)
         return all_branches
 
     @LazyProperty
     def tags(self):
         return self._get_tags()
 
     def _get_tags(self):
         return self._get_refs_entries(
             prefix='refs/tags/', strip_prefix=True, reverse=True)
 
     def tag(self, name, user, commit_id=None, message=None, date=None,
             **kwargs):
         # TODO: fix this method to apply annotated tags correct with message
         """
         Creates and returns a tag for the given ``commit_id``.
 
         :param name: name for new tag
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param commit_id: commit 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)
         commit = self.get_commit(commit_id=commit_id)
         message = message or "Added tag %s for commit %s" % (
             name, commit.raw_id)
         self._remote.set_refs('refs/tags/%s' % name, commit._commit['id'])
 
         self._refs = self._get_refs()
         self.tags = self._get_tags()
         return commit
 
     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 <[e-mail]>"
         :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 = vcspath.join(
             self._remote.get_refs_path(), 'refs', 'tags', name)
         try:
             os.remove(tagpath)
             self._refs = self._get_refs()
             self.tags = self._get_tags()
         except OSError as e:
             raise RepositoryError(e.strerror)
 
     def _get_refs(self):
         return self._remote.get_refs()
 
     @LazyProperty
     def _refs(self):
         return self._get_refs()
 
     @property
     def _ref_tree(self):
         node = tree = {}
         for ref, sha in self._refs.iteritems():
             path = ref.split('/')
             for bit in path[:-1]:
                 node = node.setdefault(bit, {})
             node[path[-1]] = sha
             node = tree
         return tree
 
     def get_remote_ref(self, ref_name):
         ref_key = 'refs/remotes/origin/{}'.format(safe_str(ref_name))
         try:
             return self._refs[ref_key]
         except Exception:
             return
 
     def get_commit(self, commit_id=None, commit_idx=None, pre_load=None, translate_tag=True):
         """
         Returns `GitCommit` object representing commit from git repository
         at the given `commit_id` or head (most recent commit) if None given.
         """
+        if self.is_empty():
+            raise EmptyRepositoryError("There are no commits yet")
+
         if commit_id is not None:
             self._validate_commit_id(commit_id)
+            try:
+                # we have cached idx, use it without contacting the remote
+                idx = self._commit_ids[commit_id]
+                return GitCommit(self, commit_id, idx, pre_load=pre_load)
+            except KeyError:
+                pass
+
         elif commit_idx is not None:
             self._validate_commit_idx(commit_idx)
+            try:
+                _commit_id = self.commit_ids[commit_idx]
+                if commit_idx < 0:
+                    commit_idx = self.commit_ids.index(_commit_id)
+                return GitCommit(self, _commit_id, commit_idx, pre_load=pre_load)
+            except IndexError:
             commit_id = commit_idx
-        commit_id = self._get_commit_id(commit_id)
-        try:
+        else:
+            commit_id = "tip"
+
+        commit_id = self._lookup_commit(commit_id)
+        remote_idx = None
             if translate_tag:
                 # Need to call remote to translate id for tagging scenario
-                commit_id = self._remote.get_object(commit_id)["commit_id"]
+            remote_data = self._remote.get_object(commit_id)
+            commit_id = remote_data["commit_id"]
+            remote_idx = remote_data["idx"]
+
+        try:
             idx = self._commit_ids[commit_id]
         except KeyError:
-            raise RepositoryError("Cannot get object with id %s" % commit_id)
+            idx = remote_idx or 0
 
         return GitCommit(self, commit_id, idx, pre_load=pre_load)
 
     def get_commits(
             self, start_id=None, end_id=None, start_date=None, end_date=None,
             branch_name=None, show_hidden=False, pre_load=None, translate_tags=True):
         """
         Returns generator of `GitCommit` objects from start to end (both
         are inclusive), in ascending date order.
 
         :param start_id: None, str(commit_id)
         :param end_id: None, str(commit_id)
         :param start_date: if specified, commits with commit date less than
           ``start_date`` would be filtered out from returned set
         :param end_date: if specified, commits with commit date greater than
           ``end_date`` would be filtered out from returned set
         :param branch_name: if specified, commits not reachable from given
           branch would be filtered out from returned set
         :param show_hidden: Show hidden commits such as obsolete or hidden from
             Mercurial evolve
         :raise BranchDoesNotExistError: If given `branch_name` does not
             exist.
         :raise CommitDoesNotExistError: If commits for given `start` or
           `end` could not be found.
 
         """
         if self.is_empty():
             raise EmptyRepositoryError("There are no commits yet")
+
         self._validate_branch_name(branch_name)
 
         if start_id is not None:
             self._validate_commit_id(start_id)
         if end_id is not None:
             self._validate_commit_id(end_id)
 
-        start_raw_id = self._get_commit_id(start_id)
+        start_raw_id = self._lookup_commit(start_id)
         start_pos = self._commit_ids[start_raw_id] if start_id else None
-        end_raw_id = self._get_commit_id(end_id)
+        end_raw_id = self._lookup_commit(end_id)
         end_pos = max(0, self._commit_ids[end_raw_id]) if end_id else None
 
         if None not in [start_id, end_id] and start_pos > end_pos:
             raise RepositoryError(
                 "Start commit '%s' cannot be after end commit '%s'" %
                 (start_id, end_id))
 
         if end_pos is not None:
             end_pos += 1
 
         filter_ = []
         if branch_name:
             filter_.append({'branch_name': branch_name})
         if start_date and not end_date:
             filter_.append({'since': start_date})
         if end_date and not start_date:
             filter_.append({'until': end_date})
         if start_date and end_date:
             filter_.append({'since': start_date})
             filter_.append({'until': end_date})
 
         # if start_pos or end_pos:
         #     filter_.append({'start': start_pos})
         #     filter_.append({'end': end_pos})
 
         if filter_:
             revfilters = {
                 'branch_name': branch_name,
                 'since': start_date.strftime('%m/%d/%y %H:%M:%S') if start_date else None,
                 'until': end_date.strftime('%m/%d/%y %H:%M:%S') if end_date else None,
                 'start': start_pos,
                 'end': end_pos,
             }
             commit_ids = self._get_all_commit_ids(filters=revfilters)
 
             # pure python stuff, it's slow due to walker walking whole repo
             # def get_revs(walker):
             #     for walker_entry in walker:
             #         yield walker_entry.commit.id
             # revfilters = {}
             # commit_ids = list(reversed(list(get_revs(self._repo.get_walker(**revfilters)))))
         else:
             commit_ids = self.commit_ids
 
         if start_pos or end_pos:
             commit_ids = commit_ids[start_pos: end_pos]
 
         return CollectionGenerator(self, commit_ids, pre_load=pre_load,
                                    translate_tag=translate_tags)
 
     def get_diff(
             self, commit1, commit2, path='', ignore_whitespace=False,
             context=3, path1=None):
         """
         Returns (git like) *diff*, as plain text. Shows changes introduced by
         ``commit2`` since ``commit1``.
 
         :param commit1: Entry point from which diff is shown. Can be
           ``self.EMPTY_COMMIT`` - in this case, patch showing all
           the changes since empty state of the repository until ``commit2``
         :param commit2: Until which commits 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``.
         """
         self._validate_diff_commits(commit1, commit2)
         if path1 is not None and path1 != path:
             raise ValueError("Diff of two different paths not supported.")
 
         flags = [
             '-U%s' % context, '--full-index', '--binary', '-p',
             '-M', '--abbrev=40']
         if ignore_whitespace:
             flags.append('-w')
 
         if commit1 == self.EMPTY_COMMIT:
             cmd = ['show'] + flags + [commit2.raw_id]
         else:
             cmd = ['diff'] + flags + [commit1.raw_id, commit2.raw_id]
 
         if path:
             cmd.extend(['--', path])
 
         stdout, __ = self.run_git_command(cmd)
         # If we used 'show' command, strip first few lines (until actual diff
         # starts)
         if commit1 == self.EMPTY_COMMIT:
             lines = stdout.splitlines()
             x = 0
             for line in lines:
                 if line.startswith('diff'):
                     break
                 x += 1
             # Append new line just like 'diff' command do
             stdout = '\n'.join(lines[x:]) + '\n'
         return GitDiff(stdout)
 
     def strip(self, commit_id, branch_name):
         commit = self.get_commit(commit_id=commit_id)
         if commit.merge:
             raise Exception('Cannot reset to merge commit')
 
         # parent is going to be the new head now
         commit = commit.parents[0]
         self._remote.set_refs('refs/heads/%s' % branch_name, commit.raw_id)
 
         self.commit_ids = self._get_all_commit_ids()
         self._rebuild_cache(self.commit_ids)
 
     def get_common_ancestor(self, commit_id1, commit_id2, repo2):
         if commit_id1 == commit_id2:
             return commit_id1
 
         if self != repo2:
             commits = self._remote.get_missing_revs(
                 commit_id1, commit_id2, repo2.path)
             if commits:
                 commit = repo2.get_commit(commits[-1])
                 if commit.parents:
                     ancestor_id = commit.parents[0].raw_id
                 else:
                     ancestor_id = None
             else:
                 # no commits from other repo, ancestor_id is the commit_id2
                 ancestor_id = commit_id2
         else:
             output, __ = self.run_git_command(
                 ['merge-base', commit_id1, commit_id2])
             ancestor_id = re.findall(r'[0-9a-fA-F]{40}', output)[0]
 
         return ancestor_id
 
     def compare(self, commit_id1, commit_id2, repo2, merge, pre_load=None):
         repo1 = self
         ancestor_id = None
 
         if commit_id1 == commit_id2:
             commits = []
         elif repo1 != repo2:
             missing_ids = self._remote.get_missing_revs(commit_id1, commit_id2,
                                                         repo2.path)
             commits = [
                 repo2.get_commit(commit_id=commit_id, pre_load=pre_load)
                 for commit_id in reversed(missing_ids)]
         else:
             output, __ = repo1.run_git_command(
                 ['log', '--reverse', '--pretty=format: %H', '-s',
                  '%s..%s' % (commit_id1, commit_id2)])
             commits = [
                 repo1.get_commit(commit_id=commit_id, pre_load=pre_load)
                 for commit_id in re.findall(r'[0-9a-fA-F]{40}', output)]
 
         return commits
 
     @LazyProperty
     def in_memory_commit(self):
         """
         Returns ``GitInMemoryCommit`` object for this repository.
         """
         return GitInMemoryCommit(self)
 
     def pull(self, url, commit_ids=None, update_after=False):
         """
         Pull changes from external location. Pull is different in GIT
         that fetch since it's doing a checkout
 
         :param commit_ids: Optional. Can be set to a list of commit ids
            which shall be pulled from the other repository.
         """
         refs = None
         if commit_ids is not None:
             remote_refs = self._remote.get_remote_refs(url)
             refs = [ref for ref in remote_refs if remote_refs[ref] in commit_ids]
         self._remote.pull(url, refs=refs, update_after=update_after)
         self._remote.invalidate_vcs_cache()
 
     def fetch(self, url, commit_ids=None):
         """
         Fetch all git objects from external location.
         """
         self._remote.sync_fetch(url, refs=commit_ids)
         self._remote.invalidate_vcs_cache()
 
     def push(self, url):
         refs = None
         self._remote.sync_push(url, refs=refs)
 
     def set_refs(self, ref_name, commit_id):
         self._remote.set_refs(ref_name, commit_id)
 
     def remove_ref(self, ref_name):
         self._remote.remove_ref(ref_name)
 
     def _update_server_info(self):
         """
         runs gits update-server-info command in this repo instance
         """
         self._remote.update_server_info()
 
     def _current_branch(self):
         """
         Return the name of the current branch.
 
         It only works for non bare repositories (i.e. repositories with a
         working copy)
         """
         if self.bare:
             raise RepositoryError('Bare git repos do not have active branches')
 
         if self.is_empty():
             return None
 
         stdout, _ = self.run_git_command(['rev-parse', '--abbrev-ref', 'HEAD'])
         return stdout.strip()
 
     def _checkout(self, branch_name, create=False, force=False):
         """
         Checkout a branch in the working directory.
 
         It tries to create the branch if create is True, failing if the branch
         already exists.
 
         It only works for non bare repositories (i.e. repositories with a
         working copy)
         """
         if self.bare:
             raise RepositoryError('Cannot checkout branches in a bare git repo')
 
         cmd = ['checkout']
         if force:
             cmd.append('-f')
         if create:
             cmd.append('-b')
         cmd.append(branch_name)
         self.run_git_command(cmd, fail_on_stderr=False)
 
     def _identify(self):
         """
         Return the current state of the working directory.
         """
         if self.bare:
             raise RepositoryError('Bare git repos do not have active branches')
 
         if self.is_empty():
             return None
 
         stdout, _ = self.run_git_command(['rev-parse', 'HEAD'])
         return stdout.strip()
 
     def _local_clone(self, clone_path, branch_name, source_branch=None):
         """
         Create a local clone of the current repo.
         """
         # N.B.(skreft): the --branch option is required as otherwise the shallow
         # clone will only fetch the active branch.
         cmd = ['clone', '--branch', branch_name,
                self.path, os.path.abspath(clone_path)]
 
         self.run_git_command(cmd, fail_on_stderr=False)
 
         # if we get the different source branch, make sure we also fetch it for
         # merge conditions
         if source_branch and source_branch != branch_name:
             # check if the ref exists.
             shadow_repo = GitRepository(os.path.abspath(clone_path))
             if shadow_repo.get_remote_ref(source_branch):
                 cmd = ['fetch', self.path, source_branch]
                 self.run_git_command(cmd, fail_on_stderr=False)
 
     def _local_fetch(self, repository_path, branch_name, use_origin=False):
         """
         Fetch a branch from a local repository.
         """
         repository_path = os.path.abspath(repository_path)
         if repository_path == self.path:
             raise ValueError('Cannot fetch from the same repository')
 
         if use_origin:
             branch_name = '+{branch}:refs/heads/{branch}'.format(
                 branch=branch_name)
 
         cmd = ['fetch', '--no-tags', '--update-head-ok',
                repository_path, branch_name]
         self.run_git_command(cmd, fail_on_stderr=False)
 
     def _local_reset(self, branch_name):
         branch_name = '{}'.format(branch_name)
         cmd = ['reset', '--hard', branch_name, '--']
         self.run_git_command(cmd, fail_on_stderr=False)
 
     def _last_fetch_heads(self):
         """
         Return the last fetched heads that need merging.
 
         The algorithm is defined at
         https://github.com/git/git/blob/v2.1.3/git-pull.sh#L283
         """
         if not self.bare:
             fetch_heads_path = os.path.join(self.path, '.git', 'FETCH_HEAD')
         else:
             fetch_heads_path = os.path.join(self.path, 'FETCH_HEAD')
 
         heads = []
         with open(fetch_heads_path) as f:
             for line in f:
                 if '    not-for-merge   ' in line:
                     continue
                 line = re.sub('\t.*', '', line, flags=re.DOTALL)
                 heads.append(line)
 
         return heads
 
     def _get_shadow_instance(self, shadow_repository_path, enable_hooks=False):
         return GitRepository(shadow_repository_path)
 
     def _local_pull(self, repository_path, branch_name, ff_only=True):
         """
         Pull a branch from a local repository.
         """
         if self.bare:
             raise RepositoryError('Cannot pull into a bare git repository')
         # N.B.(skreft): The --ff-only option is to make sure this is a
         # fast-forward (i.e., we are only pulling new changes and there are no
         # conflicts with our current branch)
         # Additionally, that option needs to go before --no-tags, otherwise git
         # pull complains about it being an unknown flag.
         cmd = ['pull']
         if ff_only:
             cmd.append('--ff-only')
         cmd.extend(['--no-tags', repository_path, branch_name])
         self.run_git_command(cmd, fail_on_stderr=False)
 
     def _local_merge(self, merge_message, user_name, user_email, heads):
         """
         Merge the given head into the checked out branch.
 
         It will force a merge commit.
 
         Currently it raises an error if the repo is empty, as it is not possible
         to create a merge commit in an empty repo.
 
         :param merge_message: The message to use for the merge commit.
         :param heads: the heads to merge.
         """
         if self.bare:
             raise RepositoryError('Cannot merge into a bare git repository')
 
         if not heads:
             return
 
         if self.is_empty():
             # TODO(skreft): do somehting more robust in this case.
             raise RepositoryError(
                 'Do not know how to merge into empty repositories yet')
 
         # N.B.(skreft): the --no-ff option is used to enforce the creation of a
         # commit message. We also specify the user who is doing the merge.
         cmd = ['-c', 'user.name="%s"' % safe_str(user_name),
                '-c', 'user.email=%s' % safe_str(user_email),
                'merge', '--no-ff', '-m', safe_str(merge_message)]
         cmd.extend(heads)
         try:
             output = self.run_git_command(cmd, fail_on_stderr=False)
         except RepositoryError:
             # Cleanup any merge leftovers
             self.run_git_command(['merge', '--abort'], fail_on_stderr=False)
             raise
 
     def _local_push(
             self, source_branch, repository_path, target_branch,
             enable_hooks=False, rc_scm_data=None):
         """
         Push the source_branch to the given repository and target_branch.
 
         Currently it if the target_branch is not master and the target repo is
         empty, the push will work, but then GitRepository won't be able to find
         the pushed branch or the commits. As the HEAD will be corrupted (i.e.,
         pointing to master, which does not exist).
 
         It does not run the hooks in the target repo.
         """
         # TODO(skreft): deal with the case in which the target repo is empty,
         # and the target_branch is not master.
         target_repo = GitRepository(repository_path)
         if (not target_repo.bare and
                 target_repo._current_branch() == target_branch):
             # Git prevents pushing to the checked out branch, so simulate it by
             # pulling into the target repository.
             target_repo._local_pull(self.path, source_branch)
         else:
             cmd = ['push', os.path.abspath(repository_path),
                    '%s:%s' % (source_branch, target_branch)]
             gitenv = {}
             if rc_scm_data:
                 gitenv.update({'RC_SCM_DATA': rc_scm_data})
 
             if not enable_hooks:
                 gitenv['RC_SKIP_HOOKS'] = '1'
             self.run_git_command(cmd, fail_on_stderr=False, extra_env=gitenv)
 
     def _get_new_pr_branch(self, source_branch, target_branch):
         prefix = 'pr_%s-%s_' % (source_branch, target_branch)
         pr_branches = []
         for branch in self.branches:
             if branch.startswith(prefix):
                 pr_branches.append(int(branch[len(prefix):]))
 
         if not pr_branches:
             branch_id = 0
         else:
             branch_id = max(pr_branches) + 1
 
         return '%s%d' % (prefix, branch_id)
 
     def _maybe_prepare_merge_workspace(
             self, repo_id, workspace_id, target_ref, source_ref):
         shadow_repository_path = self._get_shadow_repository_path(
             repo_id, workspace_id)
         if not os.path.exists(shadow_repository_path):
             self._local_clone(
                 shadow_repository_path, target_ref.name, source_ref.name)
             log.debug(
                 'Prepared shadow repository in %s', shadow_repository_path)
 
         return shadow_repository_path
 
     def _merge_repo(self, repo_id, workspace_id, target_ref,
                     source_repo, source_ref, merge_message,
                     merger_name, merger_email, dry_run=False,
                     use_rebase=False, close_branch=False):
 
         log.debug('Executing merge_repo with %s strategy, dry_run mode:%s',
                   'rebase' if use_rebase else 'merge', dry_run)
         if target_ref.commit_id != self.branches[target_ref.name]:
             log.warning('Target ref %s commit mismatch %s vs %s', target_ref,
                         target_ref.commit_id, self.branches[target_ref.name])
             return MergeResponse(
                 False, False, None, MergeFailureReason.TARGET_IS_NOT_HEAD,
                 metadata={'target_ref': target_ref})
 
         shadow_repository_path = self._maybe_prepare_merge_workspace(
             repo_id, workspace_id, target_ref, source_ref)
         shadow_repo = self._get_shadow_instance(shadow_repository_path)
 
         # checkout source, if it's different. Otherwise we could not
         # fetch proper commits for merge testing
         if source_ref.name != target_ref.name:
             if shadow_repo.get_remote_ref(source_ref.name):
                 shadow_repo._checkout(source_ref.name, force=True)
 
         # checkout target, and fetch changes
         shadow_repo._checkout(target_ref.name, force=True)
 
         # fetch/reset pull the target, in case it is changed
         # this handles even force changes
         shadow_repo._local_fetch(self.path, target_ref.name, use_origin=True)
         shadow_repo._local_reset(target_ref.name)
 
         # Need to reload repo to invalidate the cache, or otherwise we cannot
         # retrieve the last target commit.
         shadow_repo = self._get_shadow_instance(shadow_repository_path)
         if target_ref.commit_id != shadow_repo.branches[target_ref.name]:
             log.warning('Shadow Target ref %s commit mismatch %s vs %s',
                         target_ref, target_ref.commit_id,
                         shadow_repo.branches[target_ref.name])
             return MergeResponse(
                 False, False, None, MergeFailureReason.TARGET_IS_NOT_HEAD,
                 metadata={'target_ref': target_ref})
 
         # calculate new branch
         pr_branch = shadow_repo._get_new_pr_branch(
             source_ref.name, target_ref.name)
         log.debug('using pull-request merge branch: `%s`', pr_branch)
         # checkout to temp branch, and fetch changes
         shadow_repo._checkout(pr_branch, create=True)
         try:
             shadow_repo._local_fetch(source_repo.path, source_ref.name)
         except RepositoryError:
             log.exception('Failure when doing local fetch on '
                           'shadow repo: %s', shadow_repo)
             return MergeResponse(
                 False, False, None, MergeFailureReason.MISSING_SOURCE_REF,
                 metadata={'source_ref': source_ref})
 
         merge_ref = None
         merge_failure_reason = MergeFailureReason.NONE
         metadata = {}
         try:
             shadow_repo._local_merge(merge_message, merger_name, merger_email,
                                      [source_ref.commit_id])
             merge_possible = True
 
             # Need to reload repo to invalidate the cache, or otherwise we
             # cannot retrieve the merge commit.
             shadow_repo = GitRepository(shadow_repository_path)
             merge_commit_id = shadow_repo.branches[pr_branch]
 
             # Set a reference pointing to the merge commit. This reference may
             # be used to easily identify the last successful merge commit in
             # the shadow repository.
             shadow_repo.set_refs('refs/heads/pr-merge', merge_commit_id)
             merge_ref = Reference('branch', 'pr-merge', merge_commit_id)
         except RepositoryError:
             log.exception('Failure when doing local merge on git shadow repo')
             merge_possible = False
             merge_failure_reason = MergeFailureReason.MERGE_FAILED
 
         if merge_possible and not dry_run:
             try:
                 shadow_repo._local_push(
                     pr_branch, self.path, target_ref.name, enable_hooks=True,
                     rc_scm_data=self.config.get('rhodecode', 'RC_SCM_DATA'))
                 merge_succeeded = True
             except RepositoryError:
                 log.exception(
                     'Failure when doing local push from the shadow '
                     'repository to the target repository at %s.', self.path)
                 merge_succeeded = False
                 merge_failure_reason = MergeFailureReason.PUSH_FAILED
                 metadata['target'] = 'git shadow repo'
                 metadata['merge_commit'] = pr_branch
         else:
             merge_succeeded = False
 
         return MergeResponse(
             merge_possible, merge_succeeded, merge_ref, merge_failure_reason,
             metadata=metadata)

rhodecode/lib/vcs/backends/hg/inmemory.py

@@ -1,97 +1,98 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # 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 Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 HG inmemory module
 """
 
 from rhodecode.lib.datelib import date_to_timestamp_plus_offset
 from rhodecode.lib.utils import safe_str
 from rhodecode.lib.vcs.backends.base import BaseInMemoryCommit
 from rhodecode.lib.vcs.exceptions import RepositoryError
 
 
 class MercurialInMemoryCommit(BaseInMemoryCommit):
 
     def commit(self, message, author, parents=None, branch=None, date=None,
                **kwargs):
         """
         Performs in-memory commit (doesn't check workdir in any way) and
         returns newly created `MercurialCommit`. Updates repository's
         `commit_ids`.
 
         :param message: message of the commit
         :param author: full username, i.e. "Joe Doe <[e-mail]>"
         :param parents: single parent or sequence of parents from which commit
           would be derived
         :param date: `datetime.datetime` instance. Defaults to
           ``datetime.datetime.now()``.
         :param branch: Optional. Branch name as unicode. Will use the backend's
           default if not given.
 
         :raises `RepositoryError`: if any error occurs while committing
         """
         self.check_integrity(parents)
 
         if not isinstance(message, unicode) or not isinstance(author, unicode):
             # TODO: johbo: Should be a TypeError
             raise RepositoryError('Given message and author needs to be '
                                   'an <unicode> instance got %r & %r instead'
                                   % (type(message), type(author)))
 
         if branch is None:
             branch = self.repository.DEFAULT_BRANCH_NAME
         kwargs['branch'] = safe_str(branch)
 
         message = safe_str(message)
         author = safe_str(author)
 
         parent_ids = [p.raw_id if p else None for p in self.parents]
 
         ENCODING = "UTF-8"
 
         updated = []
         for node in self.added + self.changed:
             if node.is_binary:
                 content = node.content
             else:
                 content = node.content.encode(ENCODING)
             updated.append({
                 'path': node.path,
                 'content': content,
                 'mode': node.mode,
             })
 
         removed = [node.path for node in self.removed]
 
         date, tz = date_to_timestamp_plus_offset(date)
 
-        new_id = self.repository._remote.commitctx(
+        commit_id = self.repository._remote.commitctx(
             message=message, parents=parent_ids,
             commit_time=date, commit_timezone=tz, user=author,
             files=self.get_paths(), extra=kwargs, removed=removed,
             updated=updated)
+        if commit_id not in self.repository.commit_ids:
+            self.repository.commit_ids.append(commit_id)
+            self.repository._rebuild_cache(self.repository.commit_ids)
 
-        self.repository.commit_ids.append(new_id)
-        self.repository._rebuild_cache(self.repository.commit_ids)
         self.repository.branches = self.repository._get_branches()
         tip = self.repository.get_commit()
         self.reset()
         return tip

rhodecode/lib/vcs/backends/hg/repository.py

@@ -1,941 +1,942 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # 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 Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 HG repository module
 """
 import os
 import logging
 import binascii
 import urllib
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from rhodecode.lib.compat import OrderedDict
 from rhodecode.lib.datelib import (
     date_to_timestamp_plus_offset, utcdate_fromtimestamp, makedate)
 from rhodecode.lib.utils import safe_unicode, safe_str
 from rhodecode.lib.vcs import connection, exceptions
 from rhodecode.lib.vcs.backends.base import (
     BaseRepository, CollectionGenerator, Config, MergeResponse,
     MergeFailureReason, Reference, BasePathPermissionChecker)
 from rhodecode.lib.vcs.backends.hg.commit import MercurialCommit
 from rhodecode.lib.vcs.backends.hg.diff import MercurialDiff
 from rhodecode.lib.vcs.backends.hg.inmemory import MercurialInMemoryCommit
 from rhodecode.lib.vcs.exceptions import (
     EmptyRepositoryError, RepositoryError, TagAlreadyExistError,
     TagDoesNotExistError, CommitDoesNotExistError, SubrepoMergeError)
 from rhodecode.lib.vcs.compat import configparser
 
 hexlify = binascii.hexlify
 nullid = "\0" * 20
 
 log = logging.getLogger(__name__)
 
 
 class MercurialRepository(BaseRepository):
     """
     Mercurial repository backend
     """
     DEFAULT_BRANCH_NAME = 'default'
 
     def __init__(self, repo_path, config=None, create=False, src_url=None,
                  do_workspace_checkout=False, with_wire=None, bare=False):
         """
         Raises RepositoryError if repository could not be find at the given
         ``repo_path``.
 
         :param repo_path: local path of the repository
         :param config: config object containing the repo configuration
         :param create=False: if set to True, would try to create repository if
            it does not exist rather than raising exception
         :param src_url=None: would try to clone repository from given location
         :param do_workspace_checkout=False: sets update of working copy after
            making a clone
         :param bare: not used, compatible with other VCS
         """
 
         self.path = safe_str(os.path.abspath(repo_path))
         # mercurial since 4.4.X requires certain configuration to be present
         # because sometimes we init the repos with config we need to meet
         # special requirements
         self.config = config if config else self.get_default_config(
             default=[('extensions', 'largefiles', '1')])
         self.with_wire = with_wire
 
         self._init_repo(create, src_url, do_workspace_checkout)
 
         # caches
         self._commit_ids = {}
 
     @LazyProperty
     def _remote(self):
         return connection.Hg(self.path, self.config, with_wire=self.with_wire)
 
     @LazyProperty
     def commit_ids(self):
         """
         Returns list of commit ids, in ascending order.  Being lazy
         attribute allows external tools to inject shas from cache.
         """
         commit_ids = self._get_all_commit_ids()
         self._rebuild_cache(commit_ids)
         return commit_ids
 
     def _rebuild_cache(self, commit_ids):
         self._commit_ids = dict((commit_id, index)
                                 for index, commit_id in enumerate(commit_ids))
 
     @LazyProperty
     def branches(self):
         return self._get_branches()
 
     @LazyProperty
     def branches_closed(self):
         return self._get_branches(active=False, closed=True)
 
     @LazyProperty
     def branches_all(self):
         all_branches = {}
         all_branches.update(self.branches)
         all_branches.update(self.branches_closed)
         return all_branches
 
     def _get_branches(self, active=True, closed=False):
         """
         Gets branches for this repository
         Returns only not closed active branches by default
 
         :param active: return also active branches
         :param closed: return also closed branches
 
         """
         if self.is_empty():
             return {}
 
         def get_name(ctx):
             return ctx[0]
 
         _branches = [(safe_unicode(n), hexlify(h),) for n, h in
                      self._remote.branches(active, closed).items()]
 
         return OrderedDict(sorted(_branches, key=get_name, reverse=False))
 
     @LazyProperty
     def tags(self):
         """
         Gets tags for this repository
         """
         return self._get_tags()
 
     def _get_tags(self):
         if self.is_empty():
             return {}
 
         def get_name(ctx):
             return ctx[0]
 
         _tags = [(safe_unicode(n), hexlify(h),) for n, h in
                  self._remote.tags().items()]
 
         return OrderedDict(sorted(_tags, key=get_name, reverse=True))
 
     def tag(self, name, user, commit_id=None, message=None, date=None,
             **kwargs):
         """
         Creates and returns a tag for the given ``commit_id``.
 
         :param name: name for new tag
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param commit_id: commit 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)
         commit = self.get_commit(commit_id=commit_id)
         local = kwargs.setdefault('local', False)
 
         if message is None:
             message = "Added tag %s for commit %s" % (name, commit.short_id)
 
         date, tz = date_to_timestamp_plus_offset(date)
 
         self._remote.tag(
             name, commit.raw_id, message, local, user, date, tz)
         self._remote.invalidate_vcs_cache()
 
         # Reinitialize tags
         self.tags = self._get_tags()
         tag_id = self.tags[name]
 
         return self.get_commit(commit_id=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 <[e-mail]>"
         :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
         local = False
 
         date, tz = date_to_timestamp_plus_offset(date)
 
         self._remote.tag(name, nullid, message, local, user, date, tz)
         self._remote.invalidate_vcs_cache()
         self.tags = self._get_tags()
 
     @LazyProperty
     def bookmarks(self):
         """
         Gets bookmarks for this repository
         """
         return self._get_bookmarks()
 
     def _get_bookmarks(self):
         if self.is_empty():
             return {}
 
         def get_name(ctx):
             return ctx[0]
 
         _bookmarks = [
             (safe_unicode(n), hexlify(h)) for n, h in
             self._remote.bookmarks().items()]
 
         return OrderedDict(sorted(_bookmarks, key=get_name))
 
     def _get_all_commit_ids(self):
         return self._remote.get_all_commit_ids('visible')
 
     def get_diff(
             self, commit1, commit2, path='', ignore_whitespace=False,
             context=3, path1=None):
         """
         Returns (git like) *diff*, as plain text. Shows changes introduced by
         `commit2` since `commit1`.
 
         :param commit1: Entry point from which diff is shown. Can be
           ``self.EMPTY_COMMIT`` - in this case, patch showing all
           the changes since empty state of the repository until `commit2`
         :param commit2: Until which commit 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``.
         """
         self._validate_diff_commits(commit1, commit2)
         if path1 is not None and path1 != path:
             raise ValueError("Diff of two different paths not supported.")
 
         if path:
             file_filter = [self.path, path]
         else:
             file_filter = None
 
         diff = self._remote.diff(
             commit1.raw_id, commit2.raw_id, file_filter=file_filter,
             opt_git=True, opt_ignorews=ignore_whitespace,
             context=context)
         return MercurialDiff(diff)
 
     def strip(self, commit_id, branch=None):
         self._remote.strip(commit_id, update=False, backup="none")
 
         self._remote.invalidate_vcs_cache()
         self.commit_ids = self._get_all_commit_ids()
         self._rebuild_cache(self.commit_ids)
 
     def verify(self):
         verify = self._remote.verify()
 
         self._remote.invalidate_vcs_cache()
         return verify
 
     def get_common_ancestor(self, commit_id1, commit_id2, repo2):
         if commit_id1 == commit_id2:
             return commit_id1
 
         ancestors = self._remote.revs_from_revspec(
             "ancestor(id(%s), id(%s))", commit_id1, commit_id2,
             other_path=repo2.path)
         return repo2[ancestors[0]].raw_id if ancestors else None
 
     def compare(self, commit_id1, commit_id2, repo2, merge, pre_load=None):
         if commit_id1 == commit_id2:
             commits = []
         else:
             if merge:
                 indexes = self._remote.revs_from_revspec(
                     "ancestors(id(%s)) - ancestors(id(%s)) - id(%s)",
                     commit_id2, commit_id1, commit_id1, other_path=repo2.path)
             else:
                 indexes = self._remote.revs_from_revspec(
                     "id(%s)..id(%s) - id(%s)", commit_id1, commit_id2,
                     commit_id1, other_path=repo2.path)
 
             commits = [repo2.get_commit(commit_idx=idx, pre_load=pre_load)
                        for idx in indexes]
 
         return commits
 
     @staticmethod
     def check_url(url, config):
         """
         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
 
         # Request the _remote to verify the url
         return connection.Hg.check_url(url, config.serialize())
 
     @staticmethod
     def is_valid_repository(path):
         return os.path.isdir(os.path.join(path, '.hg'))
 
     def _init_repo(self, create, src_url=None, do_workspace_checkout=False):
         """
         Function will check for mercurial repository in given path. 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.
 
         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 `do_workspace_checkout` flag.
         """
         if create and os.path.exists(self.path):
             raise RepositoryError(
                 "Cannot create repository at %s, location already exist"
                 % self.path)
 
         if src_url:
             url = str(self._get_url(src_url))
             MercurialRepository.check_url(url, self.config)
 
             self._remote.clone(url, self.path, do_workspace_checkout)
 
             # Don't try to create if we've already cloned repo
             create = False
 
         if create:
             os.makedirs(self.path, mode=0o755)
 
         self._remote.localrepository(create)
 
     @LazyProperty
     def in_memory_commit(self):
         return MercurialInMemoryCommit(self)
 
     @LazyProperty
     def description(self):
         description = self._remote.get_config_value(
             'web', 'description', untrusted=True)
         return safe_unicode(description or self.DEFAULT_DESCRIPTION)
 
     @LazyProperty
     def contact(self):
         contact = (
             self._remote.get_config_value("web", "contact") or
             self._remote.get_config_value("ui", "username"))
         return safe_unicode(contact or self.DEFAULT_CONTACT)
 
     @LazyProperty
     def last_change(self):
         """
         Returns last change made on this repository as
         `datetime.datetime` object.
         """
         try:
             return self.get_commit().date
         except RepositoryError:
             tzoffset = makedate()[1]
             return utcdate_fromtimestamp(self._get_fs_mtime(), tzoffset)
 
     def _get_fs_mtime(self):
         # 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_url(self, url):
         """
         Returns normalized url. If schema is not given, would fall
         to filesystem
         (``file:///``) schema.
         """
         url = url.encode('utf8')
         if url != 'default' and '://' not in 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_commit(self, commit_id=None, commit_idx=None, pre_load=None, translate_tag=None):
         """
         Returns ``MercurialCommit`` object representing repository's
         commit at the given `commit_id` or `commit_idx`.
         """
         if self.is_empty():
             raise EmptyRepositoryError("There are no commits yet")
 
         if commit_id is not None:
             self._validate_commit_id(commit_id)
             try:
+                # we have cached idx, use it without contacting the remote
                 idx = self._commit_ids[commit_id]
                 return MercurialCommit(self, commit_id, idx, pre_load=pre_load)
             except KeyError:
                 pass
+
         elif commit_idx is not None:
             self._validate_commit_idx(commit_idx)
             try:
-                id_ = self.commit_ids[commit_idx]
+                _commit_id = self.commit_ids[commit_idx]
                 if commit_idx < 0:
-                    commit_idx += len(self.commit_ids)
-                return MercurialCommit(
-                    self, id_, commit_idx, pre_load=pre_load)
+                    commit_idx = self.commit_ids.index(_commit_id)
+
+                return MercurialCommit(self, _commit_id, commit_idx, pre_load=pre_load)
             except IndexError:
                 commit_id = commit_idx
         else:
             commit_id = "tip"
 
         if isinstance(commit_id, unicode):
             commit_id = safe_str(commit_id)
 
         try:
             raw_id, idx = self._remote.lookup(commit_id, both=True)
         except CommitDoesNotExistError:
-            msg = "Commit %s does not exist for %s" % (
-                commit_id, self)
+            msg = "Commit %s does not exist for %s" % (commit_id, self.name)
             raise CommitDoesNotExistError(msg)
 
         return MercurialCommit(self, raw_id, idx, pre_load=pre_load)
 
     def get_commits(
             self, start_id=None, end_id=None, start_date=None, end_date=None,
             branch_name=None, show_hidden=False, pre_load=None, translate_tags=None):
         """
         Returns generator of ``MercurialCommit`` objects from start to end
         (both are inclusive)
 
         :param start_id: None, str(commit_id)
         :param end_id: None, str(commit_id)
         :param start_date: if specified, commits with commit date less than
           ``start_date`` would be filtered out from returned set
         :param end_date: if specified, commits with commit date greater than
           ``end_date`` would be filtered out from returned set
         :param branch_name: if specified, commits not reachable from given
           branch would be filtered out from returned set
         :param show_hidden: Show hidden commits such as obsolete or hidden from
             Mercurial evolve
         :raise BranchDoesNotExistError: If given ``branch_name`` does not
             exist.
         :raise CommitDoesNotExistError: If commit for given ``start`` or
           ``end`` could not be found.
         """
         # actually we should check now if it's not an empty repo
         branch_ancestors = False
         if self.is_empty():
             raise EmptyRepositoryError("There are no commits yet")
         self._validate_branch_name(branch_name)
 
         if start_id is not None:
             self._validate_commit_id(start_id)
             c_start = self.get_commit(commit_id=start_id)
             start_pos = self._commit_ids[c_start.raw_id]
         else:
             start_pos = None
 
         if end_id is not None:
             self._validate_commit_id(end_id)
             c_end = self.get_commit(commit_id=end_id)
             end_pos = max(0, self._commit_ids[c_end.raw_id])
         else:
             end_pos = None
 
         if None not in [start_id, end_id] and start_pos > end_pos:
             raise RepositoryError(
                 "Start commit '%s' cannot be after end commit '%s'" %
                 (start_id, end_id))
 
         if end_pos is not None:
             end_pos += 1
 
         commit_filter = []
 
         if branch_name and not branch_ancestors:
             commit_filter.append('branch("%s")' % (branch_name,))
         elif branch_name and branch_ancestors:
             commit_filter.append('ancestors(branch("%s"))' % (branch_name,))
 
         if start_date and not end_date:
             commit_filter.append('date(">%s")' % (start_date,))
         if end_date and not start_date:
             commit_filter.append('date("<%s")' % (end_date,))
         if start_date and end_date:
             commit_filter.append(
                 'date(">%s") and date("<%s")' % (start_date, end_date))
 
         if not show_hidden:
             commit_filter.append('not obsolete()')
             commit_filter.append('not hidden()')
 
         # TODO: johbo: Figure out a simpler way for this solution
         collection_generator = CollectionGenerator
         if commit_filter:
             commit_filter = ' and '.join(map(safe_str, commit_filter))
             revisions = self._remote.rev_range([commit_filter])
             collection_generator = MercurialIndexBasedCollectionGenerator
         else:
             revisions = self.commit_ids
 
         if start_pos or end_pos:
             revisions = revisions[start_pos:end_pos]
 
         return collection_generator(self, revisions, pre_load=pre_load)
 
     def pull(self, url, commit_ids=None):
         """
         Pull changes from external location.
 
         :param commit_ids: Optional. Can be set to a list of commit ids
            which shall be pulled from the other repository.
         """
         url = self._get_url(url)
         self._remote.pull(url, commit_ids=commit_ids)
         self._remote.invalidate_vcs_cache()
 
     def fetch(self, url, commit_ids=None):
         """
         Backward compatibility with GIT fetch==pull
         """
         return self.pull(url, commit_ids=commit_ids)
 
     def push(self, url):
         url = self._get_url(url)
         self._remote.sync_push(url)
 
     def _local_clone(self, clone_path):
         """
         Create a local clone of the current repo.
         """
         self._remote.clone(self.path, clone_path, update_after_clone=True,
                            hooks=False)
 
     def _update(self, revision, clean=False):
         """
         Update the working copy to the specified revision.
         """
         log.debug('Doing checkout to commit: `%s` for %s', revision, self)
         self._remote.update(revision, clean=clean)
 
     def _identify(self):
         """
         Return the current state of the working directory.
         """
         return self._remote.identify().strip().rstrip('+')
 
     def _heads(self, branch=None):
         """
         Return the commit ids of the repository heads.
         """
         return self._remote.heads(branch=branch).strip().split(' ')
 
     def _ancestor(self, revision1, revision2):
         """
         Return the common ancestor of the two revisions.
         """
         return self._remote.ancestor(revision1, revision2)
 
     def _local_push(
             self, revision, repository_path, push_branches=False,
             enable_hooks=False):
         """
         Push the given revision to the specified repository.
 
         :param push_branches: allow to create branches in the target repo.
         """
         self._remote.push(
             [revision], repository_path, hooks=enable_hooks,
             push_branches=push_branches)
 
     def _local_merge(self, target_ref, merge_message, user_name, user_email,
                      source_ref, use_rebase=False, dry_run=False):
         """
         Merge the given source_revision into the checked out revision.
 
         Returns the commit id of the merge and a boolean indicating if the
         commit needs to be pushed.
         """
         self._update(target_ref.commit_id, clean=True)
 
         ancestor = self._ancestor(target_ref.commit_id, source_ref.commit_id)
         is_the_same_branch = self._is_the_same_branch(target_ref, source_ref)
 
         if ancestor == source_ref.commit_id:
             # Nothing to do, the changes were already integrated
             return target_ref.commit_id, False
 
         elif ancestor == target_ref.commit_id and is_the_same_branch:
             # In this case we should force a commit message
             return source_ref.commit_id, True
 
         if use_rebase:
             try:
                 bookmark_name = 'rcbook%s%s' % (source_ref.commit_id,
                                                 target_ref.commit_id)
                 self.bookmark(bookmark_name, revision=source_ref.commit_id)
                 self._remote.rebase(
                     source=source_ref.commit_id, dest=target_ref.commit_id)
                 self._remote.invalidate_vcs_cache()
                 self._update(bookmark_name, clean=True)
                 return self._identify(), True
             except RepositoryError:
                 # The rebase-abort may raise another exception which 'hides'
                 # the original one, therefore we log it here.
                 log.exception('Error while rebasing shadow repo during merge.')
 
                 # Cleanup any rebase leftovers
                 self._remote.invalidate_vcs_cache()
                 self._remote.rebase(abort=True)
                 self._remote.invalidate_vcs_cache()
                 self._remote.update(clean=True)
                 raise
         else:
             try:
                 self._remote.merge(source_ref.commit_id)
                 self._remote.invalidate_vcs_cache()
                 self._remote.commit(
                     message=safe_str(merge_message),
                     username=safe_str('%s <%s>' % (user_name, user_email)))
                 self._remote.invalidate_vcs_cache()
                 return self._identify(), True
             except RepositoryError:
                 # Cleanup any merge leftovers
                 self._remote.update(clean=True)
                 raise
 
     def _local_close(self, target_ref, user_name, user_email,
                      source_ref, close_message=''):
         """
         Close the branch of the given source_revision
 
         Returns the commit id of the close and a boolean indicating if the
         commit needs to be pushed.
         """
         self._update(source_ref.commit_id)
         message = close_message or "Closing branch: `{}`".format(source_ref.name)
         try:
             self._remote.commit(
                 message=safe_str(message),
                 username=safe_str('%s <%s>' % (user_name, user_email)),
                 close_branch=True)
             self._remote.invalidate_vcs_cache()
             return self._identify(), True
         except RepositoryError:
             # Cleanup any commit leftovers
             self._remote.update(clean=True)
             raise
 
     def _is_the_same_branch(self, target_ref, source_ref):
         return (
             self._get_branch_name(target_ref) ==
             self._get_branch_name(source_ref))
 
     def _get_branch_name(self, ref):
         if ref.type == 'branch':
             return ref.name
         return self._remote.ctx_branch(ref.commit_id)
 
     def _maybe_prepare_merge_workspace(
             self, repo_id, workspace_id, unused_target_ref, unused_source_ref):
         shadow_repository_path = self._get_shadow_repository_path(
             repo_id, workspace_id)
         if not os.path.exists(shadow_repository_path):
             self._local_clone(shadow_repository_path)
             log.debug(
                 'Prepared shadow repository in %s', shadow_repository_path)
 
         return shadow_repository_path
 
     def _merge_repo(self, repo_id, workspace_id, target_ref,
                     source_repo, source_ref, merge_message,
                     merger_name, merger_email, dry_run=False,
                     use_rebase=False, close_branch=False):
 
         log.debug('Executing merge_repo with %s strategy, dry_run mode:%s',
                   'rebase' if use_rebase else 'merge', dry_run)
         if target_ref.commit_id not in self._heads():
             return MergeResponse(
                 False, False, None, MergeFailureReason.TARGET_IS_NOT_HEAD,
                 metadata={'target_ref': target_ref})
 
         try:
             if target_ref.type == 'branch' and len(self._heads(target_ref.name)) != 1:
                 heads = '\n,'.join(self._heads(target_ref.name))
                 metadata = {
                     'target_ref': target_ref,
                     'source_ref': source_ref,
                     'heads': heads
                 }
                 return MergeResponse(
                     False, False, None,
                     MergeFailureReason.HG_TARGET_HAS_MULTIPLE_HEADS,
                     metadata=metadata)
         except CommitDoesNotExistError:
             log.exception('Failure when looking up branch heads on hg target')
             return MergeResponse(
                 False, False, None, MergeFailureReason.MISSING_TARGET_REF,
                 metadata={'target_ref': target_ref})
 
         shadow_repository_path = self._maybe_prepare_merge_workspace(
             repo_id, workspace_id, target_ref, source_ref)
         shadow_repo = self._get_shadow_instance(shadow_repository_path)
 
         log.debug('Pulling in target reference %s', target_ref)
         self._validate_pull_reference(target_ref)
         shadow_repo._local_pull(self.path, target_ref)
 
         try:
             log.debug('Pulling in source reference %s', source_ref)
             source_repo._validate_pull_reference(source_ref)
             shadow_repo._local_pull(source_repo.path, source_ref)
         except CommitDoesNotExistError:
             log.exception('Failure when doing local pull on hg shadow repo')
             return MergeResponse(
                 False, False, None, MergeFailureReason.MISSING_SOURCE_REF,
                 metadata={'source_ref': source_ref})
 
         merge_ref = None
         merge_commit_id = None
         close_commit_id = None
         merge_failure_reason = MergeFailureReason.NONE
         metadata = {}
 
         # enforce that close branch should be used only in case we source from
         # an actual Branch
         close_branch = close_branch and source_ref.type == 'branch'
 
         # don't allow to close branch if source and target are the same
         close_branch = close_branch and source_ref.name != target_ref.name
 
         needs_push_on_close = False
         if close_branch and not use_rebase and not dry_run:
             try:
                 close_commit_id, needs_push_on_close = shadow_repo._local_close(
                     target_ref, merger_name, merger_email, source_ref)
                 merge_possible = True
             except RepositoryError:
                 log.exception('Failure when doing close branch on '
                               'shadow repo: %s', shadow_repo)
                 merge_possible = False
                 merge_failure_reason = MergeFailureReason.MERGE_FAILED
         else:
             merge_possible = True
 
         needs_push = False
         if merge_possible:
             try:
                 merge_commit_id, needs_push = shadow_repo._local_merge(
                     target_ref, merge_message, merger_name, merger_email,
                     source_ref, use_rebase=use_rebase, dry_run=dry_run)
                 merge_possible = True
 
                 # read the state of the close action, if it
                 # maybe required a push
                 needs_push = needs_push or needs_push_on_close
 
                 # Set a bookmark pointing to the merge commit. This bookmark
                 # may be used to easily identify the last successful merge
                 # commit in the shadow repository.
                 shadow_repo.bookmark('pr-merge', revision=merge_commit_id)
                 merge_ref = Reference('book', 'pr-merge', merge_commit_id)
             except SubrepoMergeError:
                 log.exception(
                     'Subrepo merge error during local merge on hg shadow repo.')
                 merge_possible = False
                 merge_failure_reason = MergeFailureReason.SUBREPO_MERGE_FAILED
                 needs_push = False
             except RepositoryError:
                 log.exception('Failure when doing local merge on hg shadow repo')
                 merge_possible = False
                 merge_failure_reason = MergeFailureReason.MERGE_FAILED
                 needs_push = False
 
         if merge_possible and not dry_run:
             if needs_push:
                 # In case the target is a bookmark, update it, so after pushing
                 # the bookmarks is also updated in the target.
                 if target_ref.type == 'book':
                     shadow_repo.bookmark(
                         target_ref.name, revision=merge_commit_id)
                 try:
                     shadow_repo_with_hooks = self._get_shadow_instance(
                         shadow_repository_path,
                         enable_hooks=True)
                     # This is the actual merge action, we push from shadow
                     # into origin.
                     # Note: the push_branches option will push any new branch
                     # defined in the source repository to the target. This may
                     # be dangerous as branches are permanent in Mercurial.
                     # This feature was requested in issue #441.
                     shadow_repo_with_hooks._local_push(
                         merge_commit_id, self.path, push_branches=True,
                         enable_hooks=True)
 
                     # maybe we also need to push the close_commit_id
                     if close_commit_id:
                         shadow_repo_with_hooks._local_push(
                             close_commit_id, self.path, push_branches=True,
                             enable_hooks=True)
                     merge_succeeded = True
                 except RepositoryError:
                     log.exception(
                         'Failure when doing local push from the shadow '
                         'repository to the target repository at %s.', self.path)
                     merge_succeeded = False
                     merge_failure_reason = MergeFailureReason.PUSH_FAILED
                     metadata['target'] = 'hg shadow repo'
                     metadata['merge_commit'] = merge_commit_id
             else:
                 merge_succeeded = True
         else:
             merge_succeeded = False
 
         return MergeResponse(
             merge_possible, merge_succeeded, merge_ref, merge_failure_reason,
             metadata=metadata)
 
     def _get_shadow_instance(self, shadow_repository_path, enable_hooks=False):
         config = self.config.copy()
         if not enable_hooks:
             config.clear_section('hooks')
         return MercurialRepository(shadow_repository_path, config)
 
     def _validate_pull_reference(self, reference):
         if not (reference.name in self.bookmarks or
                 reference.name in self.branches or
                 self.get_commit(reference.commit_id)):
             raise CommitDoesNotExistError(
                 'Unknown branch, bookmark or commit id')
 
     def _local_pull(self, repository_path, reference):
         """
         Fetch a branch, bookmark or commit from a local repository.
         """
         repository_path = os.path.abspath(repository_path)
         if repository_path == self.path:
             raise ValueError('Cannot pull from the same repository')
 
         reference_type_to_option_name = {
             'book': 'bookmark',
             'branch': 'branch',
         }
         option_name = reference_type_to_option_name.get(
             reference.type, 'revision')
 
         if option_name == 'revision':
             ref = reference.commit_id
         else:
             ref = reference.name
 
         options = {option_name: [ref]}
         self._remote.pull_cmd(repository_path, hooks=False, **options)
         self._remote.invalidate_vcs_cache()
 
     def bookmark(self, bookmark, revision=None):
         if isinstance(bookmark, unicode):
             bookmark = safe_str(bookmark)
         self._remote.bookmark(bookmark, revision=revision)
         self._remote.invalidate_vcs_cache()
 
     def get_path_permissions(self, username):
         hgacl_file = os.path.join(self.path, '.hg/hgacl')
 
         def read_patterns(suffix):
             svalue = None
             for section, option in [
                 ('narrowacl', username + suffix),
                 ('narrowacl', 'default' + suffix),
                 ('narrowhgacl', username + suffix),
                 ('narrowhgacl', 'default' + suffix)
             ]:
                 try:
                     svalue = hgacl.get(section, option)
                     break  # stop at the first value we find
                 except configparser.NoOptionError:
                     pass
             if not svalue:
                 return None
             result = ['/']
             for pattern in svalue.split():
                 result.append(pattern)
                 if '*' not in pattern and '?' not in pattern:
                     result.append(pattern + '/*')
             return result
 
         if os.path.exists(hgacl_file):
             try:
                 hgacl = configparser.RawConfigParser()
                 hgacl.read(hgacl_file)
 
                 includes = read_patterns('.includes')
                 excludes = read_patterns('.excludes')
                 return BasePathPermissionChecker.create_from_patterns(
                     includes, excludes)
             except BaseException as e:
                 msg = 'Cannot read ACL settings from {} on {}: {}'.format(
                     hgacl_file, self.name, e)
                 raise exceptions.RepositoryRequirementError(msg)
         else:
             return None
 
 
 class MercurialIndexBasedCollectionGenerator(CollectionGenerator):
 
     def _commit_factory(self, commit_id):
         return self.repo.get_commit(
             commit_idx=commit_id, pre_load=self.pre_load)

rhodecode/model/pull_request.py

@@ -1,1745 +1,1742 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2012-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # 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 Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 
 """
 pull request model for RhodeCode
 """
 
 
 import json
 import logging
 import datetime
 import urllib
 import collections
 
 from pyramid import compat
 from pyramid.threadlocal import get_current_request
 
 from rhodecode import events
 from rhodecode.translation import lazy_ugettext
 from rhodecode.lib import helpers as h, hooks_utils, diffs
 from rhodecode.lib import audit_logger
 from rhodecode.lib.compat import OrderedDict
 from rhodecode.lib.hooks_daemon import prepare_callback_daemon
 from rhodecode.lib.markup_renderer import (
     DEFAULT_COMMENTS_RENDERER, RstTemplateRenderer)
 from rhodecode.lib.utils2 import safe_unicode, safe_str, md5_safe
 from rhodecode.lib.vcs.backends.base import (
     Reference, MergeResponse, MergeFailureReason, UpdateFailureReason)
 from rhodecode.lib.vcs.conf import settings as vcs_settings
 from rhodecode.lib.vcs.exceptions import (
     CommitDoesNotExistError, EmptyRepositoryError)
 from rhodecode.model import BaseModel
 from rhodecode.model.changeset_status import ChangesetStatusModel
 from rhodecode.model.comment import CommentsModel
 from rhodecode.model.db import (
     or_, PullRequest, PullRequestReviewers, ChangesetStatus,
     PullRequestVersion, ChangesetComment, Repository, RepoReviewRule)
 from rhodecode.model.meta import Session
 from rhodecode.model.notification import NotificationModel, \
     EmailNotificationModel
 from rhodecode.model.scm import ScmModel
 from rhodecode.model.settings import VcsSettingsModel
 
 
 log = logging.getLogger(__name__)
 
 
 # Data structure to hold the response data when updating commits during a pull
 # request update.
 UpdateResponse = collections.namedtuple('UpdateResponse', [
     'executed', 'reason', 'new', 'old', 'changes',
     'source_changed', 'target_changed'])
 
 
 class PullRequestModel(BaseModel):
 
     cls = PullRequest
 
     DIFF_CONTEXT = diffs.DEFAULT_CONTEXT
 
     UPDATE_STATUS_MESSAGES = {
         UpdateFailureReason.NONE: lazy_ugettext(
             'Pull request update successful.'),
         UpdateFailureReason.UNKNOWN: lazy_ugettext(
             'Pull request update failed because of an unknown error.'),
         UpdateFailureReason.NO_CHANGE: lazy_ugettext(
             'No update needed because the source and target have not changed.'),
         UpdateFailureReason.WRONG_REF_TYPE: lazy_ugettext(
             'Pull request cannot be updated because the reference type is '
             'not supported for an update. Only Branch, Tag or Bookmark is allowed.'),
         UpdateFailureReason.MISSING_TARGET_REF: lazy_ugettext(
             'This pull request cannot be updated because the target '
             'reference is missing.'),
         UpdateFailureReason.MISSING_SOURCE_REF: lazy_ugettext(
             'This pull request cannot be updated because the source '
             'reference is missing.'),
     }
     REF_TYPES = ['bookmark', 'book', 'tag', 'branch']
     UPDATABLE_REF_TYPES = ['bookmark', 'book', 'branch']
 
     def __get_pull_request(self, pull_request):
         return self._get_instance((
             PullRequest, PullRequestVersion), pull_request)
 
     def _check_perms(self, perms, pull_request, user, api=False):
         if not api:
             return h.HasRepoPermissionAny(*perms)(
                 user=user, repo_name=pull_request.target_repo.repo_name)
         else:
             return h.HasRepoPermissionAnyApi(*perms)(
                 user=user, repo_name=pull_request.target_repo.repo_name)
 
     def check_user_read(self, pull_request, user, api=False):
         _perms = ('repository.admin', 'repository.write', 'repository.read',)
         return self._check_perms(_perms, pull_request, user, api)
 
     def check_user_merge(self, pull_request, user, api=False):
         _perms = ('repository.admin', 'repository.write', 'hg.admin',)
         return self._check_perms(_perms, pull_request, user, api)
 
     def check_user_update(self, pull_request, user, api=False):
         owner = user.user_id == pull_request.user_id
         return self.check_user_merge(pull_request, user, api) or owner
 
     def check_user_delete(self, pull_request, user):
         owner = user.user_id == pull_request.user_id
         _perms = ('repository.admin',)
         return self._check_perms(_perms, pull_request, user) or owner
 
     def check_user_change_status(self, pull_request, user, api=False):
         reviewer = user.user_id in [x.user_id for x in
                                     pull_request.reviewers]
         return self.check_user_update(pull_request, user, api) or reviewer
 
     def check_user_comment(self, pull_request, user):
         owner = user.user_id == pull_request.user_id
         return self.check_user_read(pull_request, user) or owner
 
     def get(self, pull_request):
         return self.__get_pull_request(pull_request)
 
     def _prepare_get_all_query(self, repo_name, source=False, statuses=None,
                                opened_by=None, order_by=None,
                                order_dir='desc', only_created=True):
         repo = None
         if repo_name:
             repo = self._get_repo(repo_name)
 
         q = PullRequest.query()
 
         # source or target
         if repo and source:
             q = q.filter(PullRequest.source_repo == repo)
         elif repo:
             q = q.filter(PullRequest.target_repo == repo)
 
         # closed,opened
         if statuses:
             q = q.filter(PullRequest.status.in_(statuses))
 
         # opened by filter
         if opened_by:
             q = q.filter(PullRequest.user_id.in_(opened_by))
 
         # only get those that are in "created" state
         if only_created:
             q = q.filter(PullRequest.pull_request_state == PullRequest.STATE_CREATED)
 
         if order_by:
             order_map = {
                 'name_raw': PullRequest.pull_request_id,
                 'id': PullRequest.pull_request_id,
                 'title': PullRequest.title,
                 'updated_on_raw': PullRequest.updated_on,
                 'target_repo': PullRequest.target_repo_id
             }
             if order_dir == 'asc':
                 q = q.order_by(order_map[order_by].asc())
             else:
                 q = q.order_by(order_map[order_by].desc())
 
         return q
 
     def count_all(self, repo_name, source=False, statuses=None,
                   opened_by=None):
         """
         Count the number of pull requests for a specific repository.
 
         :param repo_name: target or source repo
         :param source: boolean flag to specify if repo_name refers to source
         :param statuses: list of pull request statuses
         :param opened_by: author user of the pull request
         :returns: int number of pull requests
         """
         q = self._prepare_get_all_query(
             repo_name, source=source, statuses=statuses, opened_by=opened_by)
 
         return q.count()
 
     def get_all(self, repo_name, source=False, statuses=None, opened_by=None,
                 offset=0, length=None, order_by=None, order_dir='desc'):
         """
         Get all pull requests for a specific repository.
 
         :param repo_name: target or source repo
         :param source: boolean flag to specify if repo_name refers to source
         :param statuses: list of pull request statuses
         :param opened_by: author user of the pull request
         :param offset: pagination offset
         :param length: length of returned list
         :param order_by: order of the returned list
         :param order_dir: 'asc' or 'desc' ordering direction
         :returns: list of pull requests
         """
         q = self._prepare_get_all_query(
             repo_name, source=source, statuses=statuses, opened_by=opened_by,
             order_by=order_by, order_dir=order_dir)
 
         if length:
             pull_requests = q.limit(length).offset(offset).all()
         else:
             pull_requests = q.all()
 
         return pull_requests
 
     def count_awaiting_review(self, repo_name, source=False, statuses=None,
                               opened_by=None):
         """
         Count the number of pull requests for a specific repository that are
         awaiting review.
 
         :param repo_name: target or source repo
         :param source: boolean flag to specify if repo_name refers to source
         :param statuses: list of pull request statuses
         :param opened_by: author user of the pull request
         :returns: int number of pull requests
         """
         pull_requests = self.get_awaiting_review(
             repo_name, source=source, statuses=statuses, opened_by=opened_by)
 
         return len(pull_requests)
 
     def get_awaiting_review(self, repo_name, source=False, statuses=None,
                             opened_by=None, offset=0, length=None,
                             order_by=None, order_dir='desc'):
         """
         Get all pull requests for a specific repository that are awaiting
         review.
 
         :param repo_name: target or source repo
         :param source: boolean flag to specify if repo_name refers to source
         :param statuses: list of pull request statuses
         :param opened_by: author user of the pull request
         :param offset: pagination offset
         :param length: length of returned list
         :param order_by: order of the returned list
         :param order_dir: 'asc' or 'desc' ordering direction
         :returns: list of pull requests
         """
         pull_requests = self.get_all(
             repo_name, source=source, statuses=statuses, opened_by=opened_by,
             order_by=order_by, order_dir=order_dir)
 
         _filtered_pull_requests = []
         for pr in pull_requests:
             status = pr.calculated_review_status()
             if status in [ChangesetStatus.STATUS_NOT_REVIEWED,
                           ChangesetStatus.STATUS_UNDER_REVIEW]:
                 _filtered_pull_requests.append(pr)
         if length:
             return _filtered_pull_requests[offset:offset+length]
         else:
             return _filtered_pull_requests
 
     def count_awaiting_my_review(self, repo_name, source=False, statuses=None,
                                  opened_by=None, user_id=None):
         """
         Count the number of pull requests for a specific repository that are
         awaiting review from a specific user.
 
         :param repo_name: target or source repo
         :param source: boolean flag to specify if repo_name refers to source
         :param statuses: list of pull request statuses
         :param opened_by: author user of the pull request
         :param user_id: reviewer user of the pull request
         :returns: int number of pull requests
         """
         pull_requests = self.get_awaiting_my_review(
             repo_name, source=source, statuses=statuses, opened_by=opened_by,
             user_id=user_id)
 
         return len(pull_requests)
 
     def get_awaiting_my_review(self, repo_name, source=False, statuses=None,
                                opened_by=None, user_id=None, offset=0,
                                length=None, order_by=None, order_dir='desc'):
         """
         Get all pull requests for a specific repository that are awaiting
         review from a specific user.
 
         :param repo_name: target or source repo
         :param source: boolean flag to specify if repo_name refers to source
         :param statuses: list of pull request statuses
         :param opened_by: author user of the pull request
         :param user_id: reviewer user of the pull request
         :param offset: pagination offset
         :param length: length of returned list
         :param order_by: order of the returned list
         :param order_dir: 'asc' or 'desc' ordering direction
         :returns: list of pull requests
         """
         pull_requests = self.get_all(
             repo_name, source=source, statuses=statuses, opened_by=opened_by,
             order_by=order_by, order_dir=order_dir)
 
         _my = PullRequestModel().get_not_reviewed(user_id)
         my_participation = []
         for pr in pull_requests:
             if pr in _my:
                 my_participation.append(pr)
         _filtered_pull_requests = my_participation
         if length:
             return _filtered_pull_requests[offset:offset+length]
         else:
             return _filtered_pull_requests
 
     def get_not_reviewed(self, user_id):
         return [
             x.pull_request for x in PullRequestReviewers.query().filter(
                 PullRequestReviewers.user_id == user_id).all()
         ]
 
     def _prepare_participating_query(self, user_id=None, statuses=None,
                                      order_by=None, order_dir='desc'):
         q = PullRequest.query()
         if user_id:
             reviewers_subquery = Session().query(
                 PullRequestReviewers.pull_request_id).filter(
                 PullRequestReviewers.user_id == user_id).subquery()
             user_filter = or_(
                 PullRequest.user_id == user_id,
                 PullRequest.pull_request_id.in_(reviewers_subquery)
             )
             q = PullRequest.query().filter(user_filter)
 
         # closed,opened
         if statuses:
             q = q.filter(PullRequest.status.in_(statuses))
 
         if order_by:
             order_map = {
                 'name_raw': PullRequest.pull_request_id,
                 'title': PullRequest.title,
                 'updated_on_raw': PullRequest.updated_on,
                 'target_repo': PullRequest.target_repo_id
             }
             if order_dir == 'asc':
                 q = q.order_by(order_map[order_by].asc())
             else:
                 q = q.order_by(order_map[order_by].desc())
 
         return q
 
     def count_im_participating_in(self, user_id=None, statuses=None):
         q = self._prepare_participating_query(user_id, statuses=statuses)
         return q.count()
 
     def get_im_participating_in(
             self, user_id=None, statuses=None, offset=0,
             length=None, order_by=None, order_dir='desc'):
         """
         Get all Pull requests that i'm participating in, or i have opened
         """
 
         q = self._prepare_participating_query(
             user_id, statuses=statuses, order_by=order_by,
             order_dir=order_dir)
 
         if length:
             pull_requests = q.limit(length).offset(offset).all()
         else:
             pull_requests = q.all()
 
         return pull_requests
 
     def get_versions(self, pull_request):
         """
         returns version of pull request sorted by ID descending
         """
         return PullRequestVersion.query()\
             .filter(PullRequestVersion.pull_request == pull_request)\
             .order_by(PullRequestVersion.pull_request_version_id.asc())\
             .all()
 
     def get_pr_version(self, pull_request_id, version=None):
         at_version = None
 
         if version and version == 'latest':
             pull_request_ver = PullRequest.get(pull_request_id)
             pull_request_obj = pull_request_ver
             _org_pull_request_obj = pull_request_obj
             at_version = 'latest'
         elif version:
             pull_request_ver = PullRequestVersion.get_or_404(version)
             pull_request_obj = pull_request_ver
             _org_pull_request_obj = pull_request_ver.pull_request
             at_version = pull_request_ver.pull_request_version_id
         else:
             _org_pull_request_obj = pull_request_obj = PullRequest.get_or_404(
                 pull_request_id)
 
         pull_request_display_obj = PullRequest.get_pr_display_object(
             pull_request_obj, _org_pull_request_obj)
 
         return _org_pull_request_obj, pull_request_obj, \
                pull_request_display_obj, at_version
 
     def create(self, created_by, source_repo, source_ref, target_repo,
                target_ref, revisions, reviewers, title, description=None,
                description_renderer=None,
                reviewer_data=None, translator=None, auth_user=None):
         translator = translator or get_current_request().translate
 
         created_by_user = self._get_user(created_by)
         auth_user = auth_user or created_by_user.AuthUser()
         source_repo = self._get_repo(source_repo)
         target_repo = self._get_repo(target_repo)
 
         pull_request = PullRequest()
         pull_request.source_repo = source_repo
         pull_request.source_ref = source_ref
         pull_request.target_repo = target_repo
         pull_request.target_ref = target_ref
         pull_request.revisions = revisions
         pull_request.title = title
         pull_request.description = description
         pull_request.description_renderer = description_renderer
         pull_request.author = created_by_user
         pull_request.reviewer_data = reviewer_data
         pull_request.pull_request_state = pull_request.STATE_CREATING
         Session().add(pull_request)
         Session().flush()
 
         reviewer_ids = set()
         # members / reviewers
         for reviewer_object in reviewers:
             user_id, reasons, mandatory, rules = reviewer_object
             user = self._get_user(user_id)
 
             # skip duplicates
             if user.user_id in reviewer_ids:
                 continue
 
             reviewer_ids.add(user.user_id)
 
             reviewer = PullRequestReviewers()
             reviewer.user = user
             reviewer.pull_request = pull_request
             reviewer.reasons = reasons
             reviewer.mandatory = mandatory
 
             # NOTE(marcink): pick only first rule for now
             rule_id = list(rules)[0] if rules else None
             rule = RepoReviewRule.get(rule_id) if rule_id else None
             if rule:
                 review_group = rule.user_group_vote_rule(user_id)
                 # we check if this particular reviewer is member of a voting group
                 if review_group:
                     # NOTE(marcink):
                     # can be that user is member of more but we pick the first same,
                     # same as default reviewers algo
                     review_group = review_group[0]
 
                     rule_data = {
                         'rule_name':
                             rule.review_rule_name,
                         'rule_user_group_entry_id':
                             review_group.repo_review_rule_users_group_id,
                         'rule_user_group_name':
                             review_group.users_group.users_group_name,
                         'rule_user_group_members':
                             [x.user.username for x in review_group.users_group.members],
                         'rule_user_group_members_id':
                             [x.user.user_id for x in review_group.users_group.members],
                     }
                     # e.g {'vote_rule': -1, 'mandatory': True}
                     rule_data.update(review_group.rule_data())
 
                     reviewer.rule_data = rule_data
 
             Session().add(reviewer)
             Session().flush()
 
         # Set approval status to "Under Review" for all commits which are
         # part of this pull request.
         ChangesetStatusModel().set_status(
             repo=target_repo,
             status=ChangesetStatus.STATUS_UNDER_REVIEW,
             user=created_by_user,
             pull_request=pull_request
         )
         # we commit early at this point. This has to do with a fact
         # that before queries do some row-locking. And because of that
         # we need to commit and finish transaction before below validate call
         # that for large repos could be long resulting in long row locks
         Session().commit()
 
         # prepare workspace, and run initial merge simulation. Set state during that
         # operation
         pull_request = PullRequest.get(pull_request.pull_request_id)
 
         # set as merging, for simulation, and if finished to created so we mark
         # simulation is working fine
         with pull_request.set_state(PullRequest.STATE_MERGING,
                                     final_state=PullRequest.STATE_CREATED):
             MergeCheck.validate(
                 pull_request, auth_user=auth_user, translator=translator)
 
         self.notify_reviewers(pull_request, reviewer_ids)
         self.trigger_pull_request_hook(
             pull_request, created_by_user, 'create')
 
         creation_data = pull_request.get_api_data(with_merge_state=False)
         self._log_audit_action(
             'repo.pull_request.create', {'data': creation_data},
             auth_user, pull_request)
 
         return pull_request
 
     def trigger_pull_request_hook(self, pull_request, user, action, data=None):
         pull_request = self.__get_pull_request(pull_request)
         target_scm = pull_request.target_repo.scm_instance()
         if action == 'create':
             trigger_hook = hooks_utils.trigger_log_create_pull_request_hook
         elif action == 'merge':
             trigger_hook = hooks_utils.trigger_log_merge_pull_request_hook
         elif action == 'close':
             trigger_hook = hooks_utils.trigger_log_close_pull_request_hook
         elif action == 'review_status_change':
             trigger_hook = hooks_utils.trigger_log_review_pull_request_hook
         elif action == 'update':
             trigger_hook = hooks_utils.trigger_log_update_pull_request_hook
         elif action == 'comment':
             # dummy hook ! for comment. We want this function to handle all cases
             def trigger_hook(*args, **kwargs):
                 pass
             comment = data['comment']
             events.trigger(events.PullRequestCommentEvent(pull_request, comment))
         else:
             return
 
         trigger_hook(
             username=user.username,
             repo_name=pull_request.target_repo.repo_name,
             repo_alias=target_scm.alias,
             pull_request=pull_request,
             data=data)
 
     def _get_commit_ids(self, pull_request):
         """
         Return the commit ids of the merged pull request.
 
         This method is not dealing correctly yet with the lack of autoupdates
         nor with the implicit target updates.
         For example: if a commit in the source repo is already in the target it
         will be reported anyways.
         """
         merge_rev = pull_request.merge_rev
         if merge_rev is None:
             raise ValueError('This pull request was not merged yet')
 
         commit_ids = list(pull_request.revisions)
         if merge_rev not in commit_ids:
             commit_ids.append(merge_rev)
 
         return commit_ids
 
     def merge_repo(self, pull_request, user, extras):
         log.debug("Merging pull request %s", pull_request.pull_request_id)
         extras['user_agent'] = 'internal-merge'
         merge_state = self._merge_pull_request(pull_request, user, extras)
         if merge_state.executed:
             log.debug("Merge was successful, updating the pull request comments.")
             self._comment_and_close_pr(pull_request, user, merge_state)
 
             self._log_audit_action(
                 'repo.pull_request.merge',
                 {'merge_state': merge_state.__dict__},
                 user, pull_request)
 
         else:
             log.warn("Merge failed, not updating the pull request.")
         return merge_state
 
     def _merge_pull_request(self, pull_request, user, extras, merge_msg=None):
         target_vcs = pull_request.target_repo.scm_instance()
         source_vcs = pull_request.source_repo.scm_instance()
 
         message = safe_unicode(merge_msg or vcs_settings.MERGE_MESSAGE_TMPL).format(
             pr_id=pull_request.pull_request_id,
             pr_title=pull_request.title,
             source_repo=source_vcs.name,
             source_ref_name=pull_request.source_ref_parts.name,
             target_repo=target_vcs.name,
             target_ref_name=pull_request.target_ref_parts.name,
         )
 
         workspace_id = self._workspace_id(pull_request)
         repo_id = pull_request.target_repo.repo_id
         use_rebase = self._use_rebase_for_merging(pull_request)
         close_branch = self._close_branch_before_merging(pull_request)
 
         target_ref = self._refresh_reference(
             pull_request.target_ref_parts, target_vcs)
 
         callback_daemon, extras = prepare_callback_daemon(
             extras, protocol=vcs_settings.HOOKS_PROTOCOL,
             host=vcs_settings.HOOKS_HOST,
             use_direct_calls=vcs_settings.HOOKS_DIRECT_CALLS)
 
         with callback_daemon:
             # TODO: johbo: Implement a clean way to run a config_override
             # for a single call.
             target_vcs.config.set(
                 'rhodecode', 'RC_SCM_DATA', json.dumps(extras))
 
             user_name = user.short_contact
             merge_state = target_vcs.merge(
                 repo_id, workspace_id, target_ref, source_vcs,
                 pull_request.source_ref_parts,
                 user_name=user_name, user_email=user.email,
                 message=message, use_rebase=use_rebase,
                 close_branch=close_branch)
         return merge_state
 
     def _comment_and_close_pr(self, pull_request, user, merge_state, close_msg=None):
         pull_request.merge_rev = merge_state.merge_ref.commit_id
         pull_request.updated_on = datetime.datetime.now()
         close_msg = close_msg or 'Pull request merged and closed'
 
         CommentsModel().create(
             text=safe_unicode(close_msg),
             repo=pull_request.target_repo.repo_id,
             user=user.user_id,
             pull_request=pull_request.pull_request_id,
             f_path=None,
             line_no=None,
             closing_pr=True
         )
 
         Session().add(pull_request)
         Session().flush()
         # TODO: paris: replace invalidation with less radical solution
         ScmModel().mark_for_invalidation(
             pull_request.target_repo.repo_name)
         self.trigger_pull_request_hook(pull_request, user, 'merge')
 
     def has_valid_update_type(self, pull_request):
         source_ref_type = pull_request.source_ref_parts.type
         return source_ref_type in self.REF_TYPES
 
     def update_commits(self, pull_request):
         """
         Get the updated list of commits for the pull request
         and return the new pull request version and the list
         of commits processed by this update action
         """
         pull_request = self.__get_pull_request(pull_request)
         source_ref_type = pull_request.source_ref_parts.type
         source_ref_name = pull_request.source_ref_parts.name
         source_ref_id = pull_request.source_ref_parts.commit_id
 
         target_ref_type = pull_request.target_ref_parts.type
         target_ref_name = pull_request.target_ref_parts.name
         target_ref_id = pull_request.target_ref_parts.commit_id
 
         if not self.has_valid_update_type(pull_request):
             log.debug("Skipping update of pull request %s due to ref type: %s",
                       pull_request, source_ref_type)
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.WRONG_REF_TYPE,
                 old=pull_request, new=None, changes=None,
                 source_changed=False, target_changed=False)
 
         # source repo
         source_repo = pull_request.source_repo.scm_instance()
-        source_repo.count()  # cache rebuild
 
         try:
             source_commit = source_repo.get_commit(commit_id=source_ref_name)
         except CommitDoesNotExistError:
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.MISSING_SOURCE_REF,
                 old=pull_request, new=None, changes=None,
                 source_changed=False, target_changed=False)
 
         source_changed = source_ref_id != source_commit.raw_id
 
         # target repo
         target_repo = pull_request.target_repo.scm_instance()
-        target_repo.count()  # cache rebuild
 
         try:
             target_commit = target_repo.get_commit(commit_id=target_ref_name)
         except CommitDoesNotExistError:
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.MISSING_TARGET_REF,
                 old=pull_request, new=None, changes=None,
                 source_changed=False, target_changed=False)
         target_changed = target_ref_id != target_commit.raw_id
 
         if not (source_changed or target_changed):
             log.debug("Nothing changed in pull request %s", pull_request)
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.NO_CHANGE,
                 old=pull_request, new=None, changes=None,
                 source_changed=target_changed, target_changed=source_changed)
 
         change_in_found = 'target repo' if target_changed else 'source repo'
         log.debug('Updating pull request because of change in %s detected',
                   change_in_found)
 
         # Finally there is a need for an update, in case of source change
         # we create a new version, else just an update
         if source_changed:
             pull_request_version = self._create_version_from_snapshot(pull_request)
             self._link_comments_to_version(pull_request_version)
         else:
             try:
                 ver = pull_request.versions[-1]
             except IndexError:
                 ver = None
 
             pull_request.pull_request_version_id = \
                 ver.pull_request_version_id if ver else None
             pull_request_version = pull_request
 
         try:
             if target_ref_type in self.REF_TYPES:
                 target_commit = target_repo.get_commit(target_ref_name)
             else:
                 target_commit = target_repo.get_commit(target_ref_id)
         except CommitDoesNotExistError:
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.MISSING_TARGET_REF,
                 old=pull_request, new=None, changes=None,
                 source_changed=source_changed, target_changed=target_changed)
 
         # re-compute commit ids
         old_commit_ids = pull_request.revisions
         pre_load = ["author", "branch", "date", "message"]
         commit_ranges = target_repo.compare(
             target_commit.raw_id, source_commit.raw_id, source_repo, merge=True,
             pre_load=pre_load)
 
         ancestor = source_repo.get_common_ancestor(
             source_commit.raw_id, target_commit.raw_id, target_repo)
 
         pull_request.source_ref = '%s:%s:%s' % (
             source_ref_type, source_ref_name, source_commit.raw_id)
         pull_request.target_ref = '%s:%s:%s' % (
             target_ref_type, target_ref_name, ancestor)
 
         pull_request.revisions = [
             commit.raw_id for commit in reversed(commit_ranges)]
         pull_request.updated_on = datetime.datetime.now()
         Session().add(pull_request)
         new_commit_ids = pull_request.revisions
 
         old_diff_data, new_diff_data = self._generate_update_diffs(
             pull_request, pull_request_version)
 
         # calculate commit and file changes
         changes = self._calculate_commit_id_changes(
             old_commit_ids, new_commit_ids)
         file_changes = self._calculate_file_changes(
             old_diff_data, new_diff_data)
 
         # set comments as outdated if DIFFS changed
         CommentsModel().outdate_comments(
             pull_request, old_diff_data=old_diff_data,
             new_diff_data=new_diff_data)
 
         commit_changes = (changes.added or changes.removed)
         file_node_changes = (
             file_changes.added or file_changes.modified or file_changes.removed)
         pr_has_changes = commit_changes or file_node_changes
 
         # Add an automatic comment to the pull request, in case
         # anything has changed
         if pr_has_changes:
             update_comment = CommentsModel().create(
                 text=self._render_update_message(changes, file_changes),
                 repo=pull_request.target_repo,
                 user=pull_request.author,
                 pull_request=pull_request,
                 send_email=False, renderer=DEFAULT_COMMENTS_RENDERER)
 
             # Update status to "Under Review" for added commits
             for commit_id in changes.added:
                 ChangesetStatusModel().set_status(
                     repo=pull_request.source_repo,
                     status=ChangesetStatus.STATUS_UNDER_REVIEW,
                     comment=update_comment,
                     user=pull_request.author,
                     pull_request=pull_request,
                     revision=commit_id)
 
         log.debug(
             'Updated pull request %s, added_ids: %s, common_ids: %s, '
             'removed_ids: %s', pull_request.pull_request_id,
             changes.added, changes.common, changes.removed)
         log.debug(
             'Updated pull request with the following file changes: %s',
             file_changes)
 
         log.info(
             "Updated pull request %s from commit %s to commit %s, "
             "stored new version %s of this pull request.",
             pull_request.pull_request_id, source_ref_id,
             pull_request.source_ref_parts.commit_id,
             pull_request_version.pull_request_version_id)
         Session().commit()
         self.trigger_pull_request_hook(pull_request, pull_request.author, 'update')
 
         return UpdateResponse(
             executed=True, reason=UpdateFailureReason.NONE,
             old=pull_request, new=pull_request_version, changes=changes,
             source_changed=source_changed, target_changed=target_changed)
 
     def _create_version_from_snapshot(self, pull_request):
         version = PullRequestVersion()
         version.title = pull_request.title
         version.description = pull_request.description
         version.status = pull_request.status
         version.pull_request_state = pull_request.pull_request_state
         version.created_on = datetime.datetime.now()
         version.updated_on = pull_request.updated_on
         version.user_id = pull_request.user_id
         version.source_repo = pull_request.source_repo
         version.source_ref = pull_request.source_ref
         version.target_repo = pull_request.target_repo
         version.target_ref = pull_request.target_ref
 
         version._last_merge_source_rev = pull_request._last_merge_source_rev
         version._last_merge_target_rev = pull_request._last_merge_target_rev
         version.last_merge_status = pull_request.last_merge_status
         version.shadow_merge_ref = pull_request.shadow_merge_ref
         version.merge_rev = pull_request.merge_rev
         version.reviewer_data = pull_request.reviewer_data
 
         version.revisions = pull_request.revisions
         version.pull_request = pull_request
         Session().add(version)
         Session().flush()
 
         return version
 
     def _generate_update_diffs(self, pull_request, pull_request_version):
 
         diff_context = (
             self.DIFF_CONTEXT +
             CommentsModel.needed_extra_diff_context())
         hide_whitespace_changes = False
         source_repo = pull_request_version.source_repo
         source_ref_id = pull_request_version.source_ref_parts.commit_id
         target_ref_id = pull_request_version.target_ref_parts.commit_id
         old_diff = self._get_diff_from_pr_or_version(
             source_repo, source_ref_id, target_ref_id,
             hide_whitespace_changes=hide_whitespace_changes, diff_context=diff_context)
 
         source_repo = pull_request.source_repo
         source_ref_id = pull_request.source_ref_parts.commit_id
         target_ref_id = pull_request.target_ref_parts.commit_id
 
         new_diff = self._get_diff_from_pr_or_version(
             source_repo, source_ref_id, target_ref_id,
             hide_whitespace_changes=hide_whitespace_changes, diff_context=diff_context)
 
         old_diff_data = diffs.DiffProcessor(old_diff)
         old_diff_data.prepare()
         new_diff_data = diffs.DiffProcessor(new_diff)
         new_diff_data.prepare()
 
         return old_diff_data, new_diff_data
 
     def _link_comments_to_version(self, pull_request_version):
         """
         Link all unlinked comments of this pull request to the given version.
 
         :param pull_request_version: The `PullRequestVersion` to which
             the comments shall be linked.
 
         """
         pull_request = pull_request_version.pull_request
         comments = ChangesetComment.query()\
             .filter(
                 # TODO: johbo: Should we query for the repo at all here?
                 # Pending decision on how comments of PRs are to be related
                 # to either the source repo, the target repo or no repo at all.
                 ChangesetComment.repo_id == pull_request.target_repo.repo_id,
                 ChangesetComment.pull_request == pull_request,
                 ChangesetComment.pull_request_version == None)\
             .order_by(ChangesetComment.comment_id.asc())
 
         # TODO: johbo: Find out why this breaks if it is done in a bulk
         # operation.
         for comment in comments:
             comment.pull_request_version_id = (
                 pull_request_version.pull_request_version_id)
             Session().add(comment)
 
     def _calculate_commit_id_changes(self, old_ids, new_ids):
         added = [x for x in new_ids if x not in old_ids]
         common = [x for x in new_ids if x in old_ids]
         removed = [x for x in old_ids if x not in new_ids]
         total = new_ids
         return ChangeTuple(added, common, removed, total)
 
     def _calculate_file_changes(self, old_diff_data, new_diff_data):
 
         old_files = OrderedDict()
         for diff_data in old_diff_data.parsed_diff:
             old_files[diff_data['filename']] = md5_safe(diff_data['raw_diff'])
 
         added_files = []
         modified_files = []
         removed_files = []
         for diff_data in new_diff_data.parsed_diff:
             new_filename = diff_data['filename']
             new_hash = md5_safe(diff_data['raw_diff'])
 
             old_hash = old_files.get(new_filename)
             if not old_hash:
                 # file is not present in old diff, means it's added
                 added_files.append(new_filename)
             else:
                 if new_hash != old_hash:
                     modified_files.append(new_filename)
                 # now remove a file from old, since we have seen it already
                 del old_files[new_filename]
 
         # removed files is when there are present in old, but not in NEW,
         # since we remove old files that are present in new diff, left-overs
         # if any should be the removed files
         removed_files.extend(old_files.keys())
 
         return FileChangeTuple(added_files, modified_files, removed_files)
 
     def _render_update_message(self, changes, file_changes):
         """
         render the message using DEFAULT_COMMENTS_RENDERER (RST renderer),
         so it's always looking the same disregarding on which default
         renderer system is using.
 
         :param changes: changes named tuple
         :param file_changes: file changes named tuple
 
         """
         new_status = ChangesetStatus.get_status_lbl(
             ChangesetStatus.STATUS_UNDER_REVIEW)
 
         changed_files = (
             file_changes.added + file_changes.modified + file_changes.removed)
 
         params = {
             'under_review_label': new_status,
             'added_commits': changes.added,
             'removed_commits': changes.removed,
             'changed_files': changed_files,
             'added_files': file_changes.added,
             'modified_files': file_changes.modified,
             'removed_files': file_changes.removed,
         }
         renderer = RstTemplateRenderer()
         return renderer.render('pull_request_update.mako', **params)
 
     def edit(self, pull_request, title, description, description_renderer, user):
         pull_request = self.__get_pull_request(pull_request)
         old_data = pull_request.get_api_data(with_merge_state=False)
         if pull_request.is_closed():
             raise ValueError('This pull request is closed')
         if title:
             pull_request.title = title
         pull_request.description = description
         pull_request.updated_on = datetime.datetime.now()
         pull_request.description_renderer = description_renderer
         Session().add(pull_request)
         self._log_audit_action(
             'repo.pull_request.edit', {'old_data': old_data},
             user, pull_request)
 
     def update_reviewers(self, pull_request, reviewer_data, user):
         """
         Update the reviewers in the pull request
 
         :param pull_request: the pr to update
         :param reviewer_data: list of tuples
             [(user, ['reason1', 'reason2'], mandatory_flag, [rules])]
         """
         pull_request = self.__get_pull_request(pull_request)
         if pull_request.is_closed():
             raise ValueError('This pull request is closed')
 
         reviewers = {}
         for user_id, reasons, mandatory, rules in reviewer_data:
             if isinstance(user_id, (int, compat.string_types)):
                 user_id = self._get_user(user_id).user_id
             reviewers[user_id] = {
                 'reasons': reasons, 'mandatory': mandatory}
 
         reviewers_ids = set(reviewers.keys())
         current_reviewers = PullRequestReviewers.query()\
             .filter(PullRequestReviewers.pull_request ==
                     pull_request).all()
         current_reviewers_ids = set([x.user.user_id for x in current_reviewers])
 
         ids_to_add = reviewers_ids.difference(current_reviewers_ids)
         ids_to_remove = current_reviewers_ids.difference(reviewers_ids)
 
         log.debug("Adding %s reviewers", ids_to_add)
         log.debug("Removing %s reviewers", ids_to_remove)
         changed = False
         added_audit_reviewers = []
         removed_audit_reviewers = []
 
         for uid in ids_to_add:
             changed = True
             _usr = self._get_user(uid)
             reviewer = PullRequestReviewers()
             reviewer.user = _usr
             reviewer.pull_request = pull_request
             reviewer.reasons = reviewers[uid]['reasons']
             # NOTE(marcink): mandatory shouldn't be changed now
             # reviewer.mandatory = reviewers[uid]['reasons']
             Session().add(reviewer)
             added_audit_reviewers.append(reviewer.get_dict())
 
         for uid in ids_to_remove:
             changed = True
             # NOTE(marcink): we fetch "ALL" reviewers using .all(). This is an edge case
             # that prevents and fixes cases that we added the same reviewer twice.
             # this CAN happen due to the lack of DB checks
             reviewers = PullRequestReviewers.query()\
                 .filter(PullRequestReviewers.user_id == uid,
                         PullRequestReviewers.pull_request == pull_request)\
                 .all()
 
             for obj in reviewers:
                 added_audit_reviewers.append(obj.get_dict())
                 Session().delete(obj)
 
         if changed:
             Session().expire_all()
             pull_request.updated_on = datetime.datetime.now()
             Session().add(pull_request)
 
         # finally store audit logs
         for user_data in added_audit_reviewers:
             self._log_audit_action(
                 'repo.pull_request.reviewer.add', {'data': user_data},
                 user, pull_request)
         for user_data in removed_audit_reviewers:
             self._log_audit_action(
                 'repo.pull_request.reviewer.delete', {'old_data': user_data},
                 user, pull_request)
 
         self.notify_reviewers(pull_request, ids_to_add)
         return ids_to_add, ids_to_remove
 
     def get_url(self, pull_request, request=None, permalink=False):
         if not request:
             request = get_current_request()
 
         if permalink:
             return request.route_url(
                 'pull_requests_global',
                 pull_request_id=pull_request.pull_request_id,)
         else:
             return request.route_url('pullrequest_show',
                 repo_name=safe_str(pull_request.target_repo.repo_name),
                 pull_request_id=pull_request.pull_request_id,)
 
     def get_shadow_clone_url(self, pull_request, request=None):
         """
         Returns qualified url pointing to the shadow repository. If this pull
         request is closed there is no shadow repository and ``None`` will be
         returned.
         """
         if pull_request.is_closed():
             return None
         else:
             pr_url = urllib.unquote(self.get_url(pull_request, request=request))
             return safe_unicode('{pr_url}/repository'.format(pr_url=pr_url))
 
     def notify_reviewers(self, pull_request, reviewers_ids):
         # notification to reviewers
         if not reviewers_ids:
             return
 
         pull_request_obj = pull_request
         # get the current participants of this pull request
         recipients = reviewers_ids
         notification_type = EmailNotificationModel.TYPE_PULL_REQUEST
 
         pr_source_repo = pull_request_obj.source_repo
         pr_target_repo = pull_request_obj.target_repo
 
         pr_url = h.route_url('pullrequest_show',
             repo_name=pr_target_repo.repo_name,
             pull_request_id=pull_request_obj.pull_request_id,)
 
         # set some variables for email notification
         pr_target_repo_url = h.route_url(
             'repo_summary', repo_name=pr_target_repo.repo_name)
 
         pr_source_repo_url = h.route_url(
             'repo_summary', repo_name=pr_source_repo.repo_name)
 
         # pull request specifics
         pull_request_commits = [
             (x.raw_id, x.message)
             for x in map(pr_source_repo.get_commit, pull_request.revisions)]
 
         kwargs = {
             'user': pull_request.author,
             'pull_request': pull_request_obj,
             'pull_request_commits': pull_request_commits,
 
             'pull_request_target_repo': pr_target_repo,
             'pull_request_target_repo_url': pr_target_repo_url,
 
             'pull_request_source_repo': pr_source_repo,
             'pull_request_source_repo_url': pr_source_repo_url,
 
             'pull_request_url': pr_url,
         }
 
         # pre-generate the subject for notification itself
         (subject,
          _h, _e,  # we don't care about those
          body_plaintext) = EmailNotificationModel().render_email(
             notification_type, **kwargs)
 
         # create notification objects, and emails
         NotificationModel().create(
             created_by=pull_request.author,
             notification_subject=subject,
             notification_body=body_plaintext,
             notification_type=notification_type,
             recipients=recipients,
             email_kwargs=kwargs,
         )
 
     def delete(self, pull_request, user):
         pull_request = self.__get_pull_request(pull_request)
         old_data = pull_request.get_api_data(with_merge_state=False)
         self._cleanup_merge_workspace(pull_request)
         self._log_audit_action(
             'repo.pull_request.delete', {'old_data': old_data},
             user, pull_request)
         Session().delete(pull_request)
 
     def close_pull_request(self, pull_request, user):
         pull_request = self.__get_pull_request(pull_request)
         self._cleanup_merge_workspace(pull_request)
         pull_request.status = PullRequest.STATUS_CLOSED
         pull_request.updated_on = datetime.datetime.now()
         Session().add(pull_request)
         self.trigger_pull_request_hook(
             pull_request, pull_request.author, 'close')
 
         pr_data = pull_request.get_api_data(with_merge_state=False)
         self._log_audit_action(
             'repo.pull_request.close', {'data': pr_data}, user, pull_request)
 
     def close_pull_request_with_comment(
             self, pull_request, user, repo, message=None, auth_user=None):
 
         pull_request_review_status = pull_request.calculated_review_status()
 
         if pull_request_review_status == ChangesetStatus.STATUS_APPROVED:
             # approved only if we have voting consent
             status = ChangesetStatus.STATUS_APPROVED
         else:
             status = ChangesetStatus.STATUS_REJECTED
         status_lbl = ChangesetStatus.get_status_lbl(status)
 
         default_message = (
             'Closing with status change {transition_icon} {status}.'
         ).format(transition_icon='>', status=status_lbl)
         text = message or default_message
 
         # create a comment, and link it to new status
         comment = CommentsModel().create(
             text=text,
             repo=repo.repo_id,
             user=user.user_id,
             pull_request=pull_request.pull_request_id,
             status_change=status_lbl,
             status_change_type=status,
             closing_pr=True,
             auth_user=auth_user,
         )
 
         # calculate old status before we change it
         old_calculated_status = pull_request.calculated_review_status()
         ChangesetStatusModel().set_status(
             repo.repo_id,
             status,
             user.user_id,
             comment=comment,
             pull_request=pull_request.pull_request_id
         )
 
         Session().flush()
         events.trigger(events.PullRequestCommentEvent(pull_request, comment))
         # we now calculate the status of pull request again, and based on that
         # calculation trigger status change. This might happen in cases
         # that non-reviewer admin closes a pr, which means his vote doesn't
         # change the status, while if he's a reviewer this might change it.
         calculated_status = pull_request.calculated_review_status()
         if old_calculated_status != calculated_status:
             self.trigger_pull_request_hook(
                 pull_request, user, 'review_status_change',
                 data={'status': calculated_status})
 
         # finally close the PR
         PullRequestModel().close_pull_request(
             pull_request.pull_request_id, user)
 
         return comment, status
 
     def merge_status(self, pull_request, translator=None,
                      force_shadow_repo_refresh=False):
         _ = translator or get_current_request().translate
 
         if not self._is_merge_enabled(pull_request):
             return False, _('Server-side pull request merging is disabled.')
         if pull_request.is_closed():
             return False, _('This pull request is closed.')
         merge_possible, msg = self._check_repo_requirements(
             target=pull_request.target_repo, source=pull_request.source_repo,
             translator=_)
         if not merge_possible:
             return merge_possible, msg
 
         try:
             resp = self._try_merge(
                 pull_request,
                 force_shadow_repo_refresh=force_shadow_repo_refresh)
             log.debug("Merge response: %s", resp)
             status = resp.possible, resp.merge_status_message
         except NotImplementedError:
             status = False, _('Pull request merging is not supported.')
 
         return status
 
     def _check_repo_requirements(self, target, source, translator):
         """
         Check if `target` and `source` have compatible requirements.
 
         Currently this is just checking for largefiles.
         """
         _ = translator
         target_has_largefiles = self._has_largefiles(target)
         source_has_largefiles = self._has_largefiles(source)
         merge_possible = True
         message = u''
 
         if target_has_largefiles != source_has_largefiles:
             merge_possible = False
             if source_has_largefiles:
                 message = _(
                     'Target repository large files support is disabled.')
             else:
                 message = _(
                     'Source repository large files support is disabled.')
 
         return merge_possible, message
 
     def _has_largefiles(self, repo):
         largefiles_ui = VcsSettingsModel(repo=repo).get_ui_settings(
             'extensions', 'largefiles')
         return largefiles_ui and largefiles_ui[0].active
 
     def _try_merge(self, pull_request, force_shadow_repo_refresh=False):
         """
         Try to merge the pull request and return the merge status.
         """
         log.debug(
             "Trying out if the pull request %s can be merged. Force_refresh=%s",
             pull_request.pull_request_id, force_shadow_repo_refresh)
         target_vcs = pull_request.target_repo.scm_instance()
         # Refresh the target reference.
         try:
             target_ref = self._refresh_reference(
                 pull_request.target_ref_parts, target_vcs)
         except CommitDoesNotExistError:
             merge_state = MergeResponse(
                 False, False, None, MergeFailureReason.MISSING_TARGET_REF,
                 metadata={'target_ref': pull_request.target_ref_parts})
             return merge_state
 
         target_locked = pull_request.target_repo.locked
         if target_locked and target_locked[0]:
             locked_by = 'user:{}'.format(target_locked[0])
             log.debug("The target repository is locked by %s.", locked_by)
             merge_state = MergeResponse(
                 False, False, None, MergeFailureReason.TARGET_IS_LOCKED,
                 metadata={'locked_by': locked_by})
         elif force_shadow_repo_refresh or self._needs_merge_state_refresh(
                 pull_request, target_ref):
             log.debug("Refreshing the merge status of the repository.")
             merge_state = self._refresh_merge_state(
                 pull_request, target_vcs, target_ref)
         else:
             possible = pull_request.last_merge_status == MergeFailureReason.NONE
             metadata = {
                 'target_ref': pull_request.target_ref_parts,
                 'source_ref': pull_request.source_ref_parts,
             }
             if not possible and target_ref.type == 'branch':
                 # NOTE(marcink): case for mercurial multiple heads on branch
                 heads = target_vcs._heads(target_ref.name)
                 if len(heads) != 1:
                     heads = '\n,'.join(target_vcs._heads(target_ref.name))
                     metadata.update({
                         'heads': heads
                     })
             merge_state = MergeResponse(
                 possible, False, None, pull_request.last_merge_status, metadata=metadata)
 
         return merge_state
 
     def _refresh_reference(self, reference, vcs_repository):
         if reference.type in self.UPDATABLE_REF_TYPES:
             name_or_id = reference.name
         else:
             name_or_id = reference.commit_id
 
-        vcs_repository.count() # cache rebuild
         refreshed_commit = vcs_repository.get_commit(name_or_id)
         refreshed_reference = Reference(
             reference.type, reference.name, refreshed_commit.raw_id)
         return refreshed_reference
 
     def _needs_merge_state_refresh(self, pull_request, target_reference):
         return not(
             pull_request.revisions and
             pull_request.revisions[0] == pull_request._last_merge_source_rev and
             target_reference.commit_id == pull_request._last_merge_target_rev)
 
     def _refresh_merge_state(self, pull_request, target_vcs, target_reference):
         workspace_id = self._workspace_id(pull_request)
         source_vcs = pull_request.source_repo.scm_instance()
         repo_id = pull_request.target_repo.repo_id
         use_rebase = self._use_rebase_for_merging(pull_request)
         close_branch = self._close_branch_before_merging(pull_request)
         merge_state = target_vcs.merge(
             repo_id, workspace_id,
             target_reference, source_vcs, pull_request.source_ref_parts,
             dry_run=True, use_rebase=use_rebase,
             close_branch=close_branch)
 
         # Do not store the response if there was an unknown error.
         if merge_state.failure_reason != MergeFailureReason.UNKNOWN:
             pull_request._last_merge_source_rev = \
                 pull_request.source_ref_parts.commit_id
             pull_request._last_merge_target_rev = target_reference.commit_id
             pull_request.last_merge_status = merge_state.failure_reason
             pull_request.shadow_merge_ref = merge_state.merge_ref
             Session().add(pull_request)
             Session().commit()
 
         return merge_state
 
     def _workspace_id(self, pull_request):
         workspace_id = 'pr-%s' % pull_request.pull_request_id
         return workspace_id
 
     def generate_repo_data(self, repo, commit_id=None, branch=None,
                            bookmark=None, translator=None):
         from rhodecode.model.repo import RepoModel
 
         all_refs, selected_ref = \
             self._get_repo_pullrequest_sources(
                 repo.scm_instance(), commit_id=commit_id,
                 branch=branch, bookmark=bookmark, translator=translator)
 
         refs_select2 = []
         for element in all_refs:
             children = [{'id': x[0], 'text': x[1]} for x in element[0]]
             refs_select2.append({'text': element[1], 'children': children})
 
         return {
             'user': {
                 'user_id': repo.user.user_id,
                 'username': repo.user.username,
                 'firstname': repo.user.first_name,
                 'lastname': repo.user.last_name,
                 'gravatar_link': h.gravatar_url(repo.user.email, 14),
             },
             'name': repo.repo_name,
             'link': RepoModel().get_url(repo),
             'description': h.chop_at_smart(repo.description_safe, '\n'),
             'refs': {
                 'all_refs': all_refs,
                 'selected_ref': selected_ref,
                 'select2_refs': refs_select2
             }
         }
 
     def generate_pullrequest_title(self, source, source_ref, target):
         return u'{source}#{at_ref} to {target}'.format(
             source=source,
             at_ref=source_ref,
             target=target,
         )
 
     def _cleanup_merge_workspace(self, pull_request):
         # Merging related cleanup
         repo_id = pull_request.target_repo.repo_id
         target_scm = pull_request.target_repo.scm_instance()
         workspace_id = self._workspace_id(pull_request)
 
         try:
             target_scm.cleanup_merge_workspace(repo_id, workspace_id)
         except NotImplementedError:
             pass
 
     def _get_repo_pullrequest_sources(
             self, repo, commit_id=None, branch=None, bookmark=None,
             translator=None):
         """
         Return a structure with repo's interesting commits, suitable for
         the selectors in pullrequest controller
 
         :param commit_id: a commit that must be in the list somehow
             and selected by default
         :param branch: a branch that must be in the list and selected
             by default - even if closed
         :param bookmark: a bookmark that must be in the list and selected
         """
         _ = translator or get_current_request().translate
 
         commit_id = safe_str(commit_id) if commit_id else None
         branch = safe_unicode(branch) if branch else None
         bookmark = safe_unicode(bookmark) if bookmark else None
 
         selected = None
 
         # order matters: first source that has commit_id in it will be selected
         sources = []
         sources.append(('book', repo.bookmarks.items(), _('Bookmarks'), bookmark))
         sources.append(('branch', repo.branches.items(), _('Branches'), branch))
 
         if commit_id:
             ref_commit = (h.short_id(commit_id), commit_id)
             sources.append(('rev', [ref_commit], _('Commit IDs'), commit_id))
 
         sources.append(
             ('branch', repo.branches_closed.items(), _('Closed Branches'), branch),
         )
 
         groups = []
 
         for group_key, ref_list, group_name, match in sources:
             group_refs = []
             for ref_name, ref_id in ref_list:
                 ref_key = u'{}:{}:{}'.format(group_key, ref_name, ref_id)
                 group_refs.append((ref_key, ref_name))
 
                 if not selected:
                     if set([commit_id, match]) & set([ref_id, ref_name]):
                         selected = ref_key
 
             if group_refs:
                 groups.append((group_refs, group_name))
 
         if not selected:
             ref = commit_id or branch or bookmark
             if ref:
                 raise CommitDoesNotExistError(
                     u'No commit refs could be found matching: {}'.format(ref))
             elif repo.DEFAULT_BRANCH_NAME in repo.branches:
                 selected = u'branch:{}:{}'.format(
                     safe_unicode(repo.DEFAULT_BRANCH_NAME),
                     safe_unicode(repo.branches[repo.DEFAULT_BRANCH_NAME])
                 )
             elif repo.commit_ids:
                 # make the user select in this case
                 selected = None
             else:
                 raise EmptyRepositoryError()
         return groups, selected
 
     def get_diff(self, source_repo, source_ref_id, target_ref_id,
                  hide_whitespace_changes, diff_context):
 
         return self._get_diff_from_pr_or_version(
             source_repo, source_ref_id, target_ref_id,
             hide_whitespace_changes=hide_whitespace_changes, diff_context=diff_context)
 
     def _get_diff_from_pr_or_version(
             self, source_repo, source_ref_id, target_ref_id,
             hide_whitespace_changes, diff_context):
 
         target_commit = source_repo.get_commit(
             commit_id=safe_str(target_ref_id))
         source_commit = source_repo.get_commit(
             commit_id=safe_str(source_ref_id))
         if isinstance(source_repo, Repository):
             vcs_repo = source_repo.scm_instance()
         else:
             vcs_repo = source_repo
 
         # TODO: johbo: In the context of an update, we cannot reach
         # the old commit anymore with our normal mechanisms. It needs
         # some sort of special support in the vcs layer to avoid this
         # workaround.
         if (source_commit.raw_id == vcs_repo.EMPTY_COMMIT_ID and
                 vcs_repo.alias == 'git'):
             source_commit.raw_id = safe_str(source_ref_id)
 
         log.debug('calculating diff between '
                   'source_ref:%s and target_ref:%s for repo `%s`',
                   target_ref_id, source_ref_id,
                   safe_unicode(vcs_repo.path))
 
         vcs_diff = vcs_repo.get_diff(
             commit1=target_commit, commit2=source_commit,
             ignore_whitespace=hide_whitespace_changes, context=diff_context)
         return vcs_diff
 
     def _is_merge_enabled(self, pull_request):
         return self._get_general_setting(
             pull_request, 'rhodecode_pr_merge_enabled')
 
     def _use_rebase_for_merging(self, pull_request):
         repo_type = pull_request.target_repo.repo_type
         if repo_type == 'hg':
             return self._get_general_setting(
                 pull_request, 'rhodecode_hg_use_rebase_for_merging')
         elif repo_type == 'git':
             return self._get_general_setting(
                 pull_request, 'rhodecode_git_use_rebase_for_merging')
 
         return False
 
     def _close_branch_before_merging(self, pull_request):
         repo_type = pull_request.target_repo.repo_type
         if repo_type == 'hg':
             return self._get_general_setting(
                 pull_request, 'rhodecode_hg_close_branch_before_merging')
         elif repo_type == 'git':
             return self._get_general_setting(
                 pull_request, 'rhodecode_git_close_branch_before_merging')
 
         return False
 
     def _get_general_setting(self, pull_request, settings_key, default=False):
         settings_model = VcsSettingsModel(repo=pull_request.target_repo)
         settings = settings_model.get_general_settings()
         return settings.get(settings_key, default)
 
     def _log_audit_action(self, action, action_data, user, pull_request):
         audit_logger.store(
             action=action,
             action_data=action_data,
             user=user,
             repo=pull_request.target_repo)
 
     def get_reviewer_functions(self):
         """
         Fetches functions for validation and fetching default reviewers.
         If available we use the EE package, else we fallback to CE
         package functions
         """
         try:
             from rc_reviewers.utils import get_default_reviewers_data
             from rc_reviewers.utils import validate_default_reviewers
         except ImportError:
             from rhodecode.apps.repository.utils import get_default_reviewers_data
             from rhodecode.apps.repository.utils import validate_default_reviewers
 
         return get_default_reviewers_data, validate_default_reviewers
 
 
 class MergeCheck(object):
     """
     Perform Merge Checks and returns a check object which stores information
     about merge errors, and merge conditions
     """
     TODO_CHECK = 'todo'
     PERM_CHECK = 'perm'
     REVIEW_CHECK = 'review'
     MERGE_CHECK = 'merge'
 
     def __init__(self):
         self.review_status = None
         self.merge_possible = None
         self.merge_msg = ''
         self.failed = None
         self.errors = []
         self.error_details = OrderedDict()
 
     def push_error(self, error_type, message, error_key, details):
         self.failed = True
         self.errors.append([error_type, message])
         self.error_details[error_key] = dict(
             details=details,
             error_type=error_type,
             message=message
         )
 
     @classmethod
     def validate(cls, pull_request, auth_user, translator, fail_early=False,
                  force_shadow_repo_refresh=False):
         _ = translator
         merge_check = cls()
 
         # permissions to merge
         user_allowed_to_merge = PullRequestModel().check_user_merge(
             pull_request, auth_user)
         if not user_allowed_to_merge:
             log.debug("MergeCheck: cannot merge, approval is pending.")
 
             msg = _('User `{}` not allowed to perform merge.').format(auth_user.username)
             merge_check.push_error('error', msg, cls.PERM_CHECK, auth_user.username)
             if fail_early:
                 return merge_check
 
         # permission to merge into the target branch
         target_commit_id = pull_request.target_ref_parts.commit_id
         if pull_request.target_ref_parts.type == 'branch':
             branch_name = pull_request.target_ref_parts.name
         else:
             # for mercurial we can always figure out the branch from the commit
             # in case of bookmark
             target_commit = pull_request.target_repo.get_commit(target_commit_id)
             branch_name = target_commit.branch
 
         rule, branch_perm = auth_user.get_rule_and_branch_permission(
             pull_request.target_repo.repo_name, branch_name)
         if branch_perm and branch_perm == 'branch.none':
             msg = _('Target branch `{}` changes rejected by rule {}.').format(
                 branch_name, rule)
             merge_check.push_error('error', msg, cls.PERM_CHECK, auth_user.username)
             if fail_early:
                 return merge_check
 
         # review status, must be always present
         review_status = pull_request.calculated_review_status()
         merge_check.review_status = review_status
 
         status_approved = review_status == ChangesetStatus.STATUS_APPROVED
         if not status_approved:
             log.debug("MergeCheck: cannot merge, approval is pending.")
 
             msg = _('Pull request reviewer approval is pending.')
 
             merge_check.push_error('warning', msg, cls.REVIEW_CHECK, review_status)
 
             if fail_early:
                 return merge_check
 
         # left over TODOs
         todos = CommentsModel().get_pull_request_unresolved_todos(pull_request)
         if todos:
             log.debug("MergeCheck: cannot merge, {} "
                       "unresolved TODOs left.".format(len(todos)))
 
             if len(todos) == 1:
                 msg = _('Cannot merge, {} TODO still not resolved.').format(
                     len(todos))
             else:
                 msg = _('Cannot merge, {} TODOs still not resolved.').format(
                     len(todos))
 
             merge_check.push_error('warning', msg, cls.TODO_CHECK, todos)
 
             if fail_early:
                 return merge_check
 
         # merge possible, here is the filesystem simulation + shadow repo
         merge_status, msg = PullRequestModel().merge_status(
             pull_request, translator=translator,
             force_shadow_repo_refresh=force_shadow_repo_refresh)
         merge_check.merge_possible = merge_status
         merge_check.merge_msg = msg
         if not merge_status:
             log.debug("MergeCheck: cannot merge, pull request merge not possible.")
             merge_check.push_error('warning', msg, cls.MERGE_CHECK, None)
 
             if fail_early:
                 return merge_check
 
         log.debug('MergeCheck: is failed: %s', merge_check.failed)
         return merge_check
 
     @classmethod
     def get_merge_conditions(cls, pull_request, translator):
         _ = translator
         merge_details = {}
 
         model = PullRequestModel()
         use_rebase = model._use_rebase_for_merging(pull_request)
 
         if use_rebase:
             merge_details['merge_strategy'] = dict(
                 details={},
                 message=_('Merge strategy: rebase')
             )
         else:
             merge_details['merge_strategy'] = dict(
                 details={},
                 message=_('Merge strategy: explicit merge commit')
             )
 
         close_branch = model._close_branch_before_merging(pull_request)
         if close_branch:
             repo_type = pull_request.target_repo.repo_type
             close_msg = ''
             if repo_type == 'hg':
                 close_msg = _('Source branch will be closed after merge.')
             elif repo_type == 'git':
                 close_msg = _('Source branch will be deleted after merge.')
 
             merge_details['close_branch'] = dict(
                 details={},
                 message=close_msg
             )
 
         return merge_details
 
 
 ChangeTuple = collections.namedtuple(
     'ChangeTuple', ['added', 'common', 'removed', 'total'])
 
 FileChangeTuple = collections.namedtuple(
     'FileChangeTuple', ['added', 'modified', 'removed'])

rhodecode/tests/models/test_scm.py

@@ -1,196 +1,195 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # 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 Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import os
 import stat
 import sys
 
 import pytest
 from mock import Mock, patch, DEFAULT
 
 import rhodecode
 from rhodecode.model import db, scm
 from rhodecode.tests import no_newline_id_generator
 
 
 def test_scm_instance_config(backend):
     repo = backend.create_repo()
     with patch.multiple('rhodecode.model.db.Repository',
                         _get_instance=DEFAULT,
                         _get_instance_cached=DEFAULT) as mocks:
         repo.scm_instance()
         mocks['_get_instance'].assert_called_with(
             config=None, cache=False)
 
         config = {'some': 'value'}
         repo.scm_instance(config=config)
         mocks['_get_instance'].assert_called_with(
             config=config, cache=False)
 
         with patch.dict(rhodecode.CONFIG, {'vcs_full_cache': 'true'}):
             repo.scm_instance(config=config)
             mocks['_get_instance_cached'].assert_called()
 
 
 def test__get_instance_config(backend):
     repo = backend.create_repo()
     vcs_class = Mock()
     with patch.multiple('rhodecode.lib.vcs.backends',
                         get_scm=DEFAULT,
                         get_backend=DEFAULT) as mocks:
         mocks['get_scm'].return_value = backend.alias
         mocks['get_backend'].return_value = vcs_class
         with patch('rhodecode.model.db.Repository._config') as config_mock:
             repo._get_instance()
             vcs_class.assert_called_with(
                 repo_path=repo.repo_full_path, config=config_mock,
                 create=False, with_wire={'cache': True})
 
         new_config = {'override': 'old_config'}
         repo._get_instance(config=new_config)
         vcs_class.assert_called_with(
             repo_path=repo.repo_full_path, config=new_config, create=False,
             with_wire={'cache': True})
 
 
 def test_mark_for_invalidation_config(backend):
     repo = backend.create_repo()
     with patch('rhodecode.model.db.Repository.update_commit_cache') as _mock:
         scm.ScmModel().mark_for_invalidation(repo.repo_name)
         _, kwargs = _mock.call_args
         assert kwargs['config'].__dict__ == repo._config.__dict__
 
 
 def test_mark_for_invalidation_with_delete_updates_last_commit(backend):
     commits = [{'message': 'A'}, {'message': 'B'}]
     repo = backend.create_repo(commits=commits)
     scm.ScmModel().mark_for_invalidation(repo.repo_name, delete=True)
     assert repo.changeset_cache['revision'] == 1
 
 
 def test_mark_for_invalidation_with_delete_updates_last_commit_empty(backend):
     repo = backend.create_repo()
     scm.ScmModel().mark_for_invalidation(repo.repo_name, delete=True)
     assert repo.changeset_cache['revision'] == -1
 
 
 def test_strip_with_multiple_heads(backend_hg):
     commits = [
         {'message': 'A'},
         {'message': 'a'},
         {'message': 'b'},
         {'message': 'B', 'parents': ['A']},
         {'message': 'a1'},
     ]
     repo = backend_hg.create_repo(commits=commits)
     commit_ids = backend_hg.commit_ids
 
     model = scm.ScmModel()
     model.strip(repo, commit_ids['b'], branch=None)
 
     vcs_repo = repo.scm_instance()
     rest_commit_ids = [c.raw_id for c in vcs_repo.get_commits()]
     assert len(rest_commit_ids) == 4
     assert commit_ids['b'] not in rest_commit_ids
 
 
 def test_strip_with_single_heads(backend_hg):
     commits = [
         {'message': 'A'},
         {'message': 'a'},
         {'message': 'b'},
     ]
     repo = backend_hg.create_repo(commits=commits)
     commit_ids = backend_hg.commit_ids
 
     model = scm.ScmModel()
     model.strip(repo, commit_ids['b'], branch=None)
 
     vcs_repo = repo.scm_instance()
     rest_commit_ids = [c.raw_id for c in vcs_repo.get_commits()]
     assert len(rest_commit_ids) == 2
     assert commit_ids['b'] not in rest_commit_ids
 
 
-def test_get_nodes_returns_unicode_flat(backend_random):
-    repo = backend_random.repo
-    directories, files = scm.ScmModel().get_nodes(
-        repo.repo_name, repo.get_commit(commit_idx=0).raw_id,
-        flat=True)
+def test_get_nodes_returns_unicode_flat(backend):
+    repo = backend.repo
+    commit_id = repo.get_commit(commit_idx=0).raw_id
+    directories, files = scm.ScmModel().get_nodes(repo.repo_name, commit_id, flat=True)
     assert_contains_only_unicode(directories)
     assert_contains_only_unicode(files)
 
 
-def test_get_nodes_returns_unicode_non_flat(backend_random):
-    repo = backend_random.repo
-    directories, files = scm.ScmModel().get_nodes(
-        repo.repo_name, repo.get_commit(commit_idx=0).raw_id,
-        flat=False)
+def test_get_nodes_returns_unicode_non_flat(backend):
+    repo = backend.repo
+    commit_id = repo.get_commit(commit_idx=0).raw_id
+
+    directories, files = scm.ScmModel().get_nodes(repo.repo_name, commit_id, flat=False)
     # johbo: Checking only the names for now, since that is the critical
     # part.
     assert_contains_only_unicode([d['name'] for d in directories])
     assert_contains_only_unicode([f['name'] for f in files])
 
 
 def test_get_nodes_max_file_bytes(backend_random):
     repo = backend_random.repo
     max_file_bytes = 10
     directories, files = scm.ScmModel().get_nodes(
         repo.repo_name, repo.get_commit(commit_idx=0).raw_id, content=True,
         extended_info=True, flat=False)
     assert any(file['content'] and len(file['content']) > max_file_bytes
                for file in files)
 
     directories, files = scm.ScmModel().get_nodes(
         repo.repo_name, repo.get_commit(commit_idx=0).raw_id, content=True,
         extended_info=True, flat=False, max_file_bytes=max_file_bytes)
     assert all(
         file['content'] is None if file['size'] > max_file_bytes else True
         for file in files)
 
 
 def assert_contains_only_unicode(structure):
     assert structure
     for value in structure:
         assert isinstance(value, unicode)
 
 
 @pytest.mark.backends("hg", "git")
 def test_get_non_unicode_reference(backend):
     model = scm.ScmModel()
     non_unicode_list = ["Adını".decode("cp1254")]
 
     def scm_instance():
         return Mock(
             branches=non_unicode_list, bookmarks=non_unicode_list,
             tags=non_unicode_list, alias=backend.alias)
 
     repo = Mock(__class__=db.Repository, scm_instance=scm_instance)
     choices, __ = model.get_repo_landing_revs(translator=lambda s: s, repo=repo)
     if backend.alias == 'hg':
         valid_choices = [
             'rev:tip', u'branch:Ad\xc4\xb1n\xc4\xb1',
             u'book:Ad\xc4\xb1n\xc4\xb1', u'tag:Ad\xc4\xb1n\xc4\xb1']
     else:
         valid_choices = [
             'rev:tip', u'branch:Ad\xc4\xb1n\xc4\xb1',
             u'tag:Ad\xc4\xb1n\xc4\xb1']
 
     assert choices == valid_choices

rhodecode/tests/vcs/conftest.py

@@ -1,257 +1,256 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # 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 Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import time
 import shutil
 import datetime
 
 import pytest
 
 from rhodecode.lib.vcs.backends import get_backend
 from rhodecode.lib.vcs.backends.base import Config
 from rhodecode.lib.vcs.nodes import FileNode
 from rhodecode.tests import get_new_dir
 from rhodecode.tests.utils import check_skip_backends, check_xfail_backends
 
 
 @pytest.fixture()
 def vcs_repository_support(
         request, backend_alias, baseapp, _vcs_repo_container):
     """
     Provide a test repository for the test run.
 
     Depending on the value of `recreate_repo_per_test` a new repo for each
     test will be created.
 
     The parameter `--backends` can be used to limit this fixture to specific
     backend implementations.
     """
     cls = request.cls
 
     check_skip_backends(request.node, backend_alias)
     check_xfail_backends(request.node, backend_alias)
 
     if _should_create_repo_per_test(cls):
         _vcs_repo_container = _create_vcs_repo_container(request)
 
     repo = _vcs_repo_container.get_repo(cls, backend_alias=backend_alias)
 
     # TODO: johbo: Supporting old test class api, think about removing this
     cls.repo = repo
     cls.repo_path = repo.path
     cls.default_branch = repo.DEFAULT_BRANCH_NAME
     cls.Backend = cls.backend_class = repo.__class__
     cls.imc = repo.in_memory_commit
 
     return backend_alias, repo
 
 
 @pytest.fixture(scope='class')
 def _vcs_repo_container(request):
     """
     Internal fixture intended to help support class based scoping on demand.
     """
     return _create_vcs_repo_container(request)
 
 
 def _create_vcs_repo_container(request):
     repo_container = VcsRepoContainer()
     if not request.config.getoption('--keep-tmp-path'):
         request.addfinalizer(repo_container.cleanup)
     return repo_container
 
 
 class VcsRepoContainer(object):
 
     def __init__(self):
         self._cleanup_paths = []
         self._repos = {}
 
     def get_repo(self, test_class, backend_alias):
         if backend_alias not in self._repos:
             repo = _create_empty_repository(test_class, backend_alias)
 
             self._cleanup_paths.append(repo.path)
             self._repos[backend_alias] = repo
         return self._repos[backend_alias]
 
     def cleanup(self):
         for repo_path in reversed(self._cleanup_paths):
             shutil.rmtree(repo_path)
 
 
 def _should_create_repo_per_test(cls):
     return getattr(cls, 'recreate_repo_per_test', False)
 
 
 def _create_empty_repository(cls, backend_alias=None):
     Backend = get_backend(backend_alias or cls.backend_alias)
     repo_path = get_new_dir(str(time.time()))
     repo = Backend(repo_path, create=True)
     if hasattr(cls, '_get_commits'):
         commits = cls._get_commits()
         cls.tip = _add_commits_to_repo(repo, commits)
 
     return repo
 
 
 @pytest.fixture
 def config():
     """
     Instance of a repository config.
 
     The instance contains only one value:
 
     - Section: "section-a"
     - Key:     "a-1"
     - Value:   "value-a-1"
 
     The intended usage is for cases where a config instance is needed but no
     specific content is required.
     """
     config = Config()
     config.set('section-a', 'a-1', 'value-a-1')
     return config
 
 
 def _add_commits_to_repo(repo, commits):
     imc = repo.in_memory_commit
     tip = None
 
     for commit in commits:
         for node in commit.get('added', []):
             imc.add(FileNode(node.path, content=node.content))
         for node in commit.get('changed', []):
             imc.change(FileNode(node.path, content=node.content))
         for node in commit.get('removed', []):
             imc.remove(FileNode(node.path))
 
         tip = imc.commit(
             message=unicode(commit['message']),
             author=unicode(commit['author']),
             date=commit['date'],
             branch=commit.get('branch'))
-
     return tip
 
 
 @pytest.fixture
 def vcs_repo(request, backend_alias):
     Backend = get_backend(backend_alias)
     repo_path = get_new_dir(str(time.time()))
     repo = Backend(repo_path, create=True)
 
     @request.addfinalizer
     def cleanup():
         shutil.rmtree(repo_path)
 
     return repo
 
 
 @pytest.fixture
 def generate_repo_with_commits(vcs_repo):
     """
     Creates a fabric to generate N comits with some file nodes on a randomly
     generated repository
     """
 
     def commit_generator(num):
         start_date = datetime.datetime(2010, 1, 1, 20)
         for x in xrange(num):
             yield {
                 'message': 'Commit %d' % x,
                 'author': 'Joe Doe <[e-mail]>',
                 'date': start_date + datetime.timedelta(hours=12 * x),
                 'added': [
                     FileNode('file_%d.txt' % x, content='Foobar %d' % x),
                 ],
                 'modified': [
                     FileNode('file_%d.txt' % x,
                              content='Foobar %d modified' % (x-1)),
                 ]
             }
 
     def commit_maker(num=5):
         _add_commits_to_repo(vcs_repo, commit_generator(num))
         return vcs_repo
 
     return commit_maker
 
 
 @pytest.fixture
 def hg_repo(request, vcs_repo):
     repo = vcs_repo
 
     commits = repo._get_commits()
     _add_commits_to_repo(repo, commits)
 
     return repo
 
 
 @pytest.fixture
 def hg_commit(hg_repo):
     return hg_repo.get_commit()
 
 
 class BackendTestMixin(object):
     """
     This is a backend independent test case class which should be created
     with ``type`` method.
 
     It is required to set following attributes at subclass:
 
     - ``backend_alias``: alias of used backend (see ``vcs.BACKENDS``)
     - ``repo_path``: path to the repository which would be created for set of
       tests
     - ``recreate_repo_per_test``: If set to ``False``, repo would NOT be
       created
       before every single test. Defaults to ``True``.
     """
     recreate_repo_per_test = True
 
     @classmethod
     def _get_commits(cls):
         commits = [
             {
                 'message': u'Initial commit',
                 'author': u'Joe Doe <[e-mail]>',
                 'date': datetime.datetime(2010, 1, 1, 20),
                 'added': [
                     FileNode('foobar', content='Foobar'),
                     FileNode('foobar2', content='Foobar II'),
                     FileNode('foo/bar/baz', content='baz here!'),
                 ],
             },
             {
                 'message': u'Changes...',
                 'author': u'Jane Doe <[e-mail]>',
                 'date': datetime.datetime(2010, 1, 1, 21),
                 'added': [
                     FileNode('some/new.txt', content='news...'),
                 ],
                 'changed': [
                     FileNode('foobar', 'Foobar I'),
                 ],
                 'removed': [],
             },
         ]
         return commits
 
 

rhodecode/tests/vcs/test_commits.py

@@ -1,593 +1,592 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # 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 Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import datetime
 import time
 
 import pytest
 
 from rhodecode.lib.vcs.backends.base import (
     CollectionGenerator, FILEMODE_DEFAULT, EmptyCommit)
 from rhodecode.lib.vcs.exceptions import (
     BranchDoesNotExistError, CommitDoesNotExistError,
     RepositoryError, EmptyRepositoryError)
 from rhodecode.lib.vcs.nodes import (
     FileNode, AddedFileNodesGenerator,
     ChangedFileNodesGenerator, RemovedFileNodesGenerator)
 from rhodecode.tests import get_new_dir
 from rhodecode.tests.vcs.conftest import BackendTestMixin
 
 
 class TestBaseChangeset:
 
     def test_is_deprecated(self):
         from rhodecode.lib.vcs.backends.base import BaseChangeset
         pytest.deprecated_call(BaseChangeset)
 
 
 class TestEmptyCommit(object):
 
     def test_branch_without_alias_returns_none(self):
         commit = EmptyCommit()
         assert commit.branch is None
 
 
 @pytest.mark.usefixtures("vcs_repository_support")
 class TestCommitsInNonEmptyRepo(BackendTestMixin):
     recreate_repo_per_test = True
 
     @classmethod
     def _get_commits(cls):
         start_date = datetime.datetime(2010, 1, 1, 20)
         for x in xrange(5):
             yield {
                 'message': 'Commit %d' % x,
                 'author': 'Joe Doe <[e-mail]>',
                 'date': start_date + datetime.timedelta(hours=12 * x),
                 'added': [
                     FileNode('file_%d.txt' % x, content='Foobar %d' % x),
                 ],
             }
 
     def test_walk_returns_empty_list_in_case_of_file(self):
         result = list(self.tip.walk('file_0.txt'))
         assert result == []
 
     @pytest.mark.backends("git", "hg")
     def test_new_branch(self):
         self.imc.add(FileNode('docs/index.txt',
                      content='Documentation\n'))
         foobar_tip = self.imc.commit(
             message=u'New branch: foobar',
             author=u'joe',
             branch='foobar',
         )
         assert 'foobar' in self.repo.branches
         assert foobar_tip.branch == 'foobar'
         # 'foobar' should be the only branch that contains the new commit
         branch = self.repo.branches.values()
         assert branch[0] != branch[1]
 
     @pytest.mark.backends("git", "hg")
     def test_new_head_in_default_branch(self):
         tip = self.repo.get_commit()
         self.imc.add(FileNode('docs/index.txt',
                      content='Documentation\n'))
         foobar_tip = self.imc.commit(
             message=u'New branch: foobar',
             author=u'joe',
             branch='foobar',
             parents=[tip],
         )
         self.imc.change(FileNode('docs/index.txt',
                         content='Documentation\nand more...\n'))
         newtip = self.imc.commit(
             message=u'At default branch',
             author=u'joe',
             branch=foobar_tip.branch,
             parents=[foobar_tip],
         )
 
         newest_tip = self.imc.commit(
             message=u'Merged with %s' % foobar_tip.raw_id,
             author=u'joe',
             branch=self.backend_class.DEFAULT_BRANCH_NAME,
             parents=[newtip, foobar_tip],
         )
 
         assert newest_tip.branch == self.backend_class.DEFAULT_BRANCH_NAME
 
     @pytest.mark.backends("git", "hg")
     def test_get_commits_respects_branch_name(self):
         """
         * e1930d0 (HEAD, master) Back in default branch
         | * e1930d0 (docs) New Branch: docs2
         | * dcc14fa New branch: docs
         |/
         * e63c41a Initial commit
         ...
         * 624d3db Commit 0
 
         :return:
         """
         DEFAULT_BRANCH = self.repo.DEFAULT_BRANCH_NAME
         TEST_BRANCH = 'docs'
         org_tip = self.repo.get_commit()
 
         self.imc.add(FileNode('readme.txt', content='Document\n'))
         initial = self.imc.commit(
             message=u'Initial commit',
             author=u'joe',
             parents=[org_tip],
             branch=DEFAULT_BRANCH,)
 
         self.imc.add(FileNode('newdoc.txt', content='foobar\n'))
         docs_branch_commit1 = self.imc.commit(
             message=u'New branch: docs',
             author=u'joe',
             parents=[initial],
             branch=TEST_BRANCH,)
 
         self.imc.add(FileNode('newdoc2.txt', content='foobar2\n'))
         docs_branch_commit2 = self.imc.commit(
             message=u'New branch: docs2',
             author=u'joe',
             parents=[docs_branch_commit1],
             branch=TEST_BRANCH,)
 
         self.imc.add(FileNode('newfile', content='hello world\n'))
         self.imc.commit(
             message=u'Back in default branch',
             author=u'joe',
             parents=[initial],
             branch=DEFAULT_BRANCH,)
 
-        default_branch_commits = self.repo.get_commits(
-            branch_name=DEFAULT_BRANCH)
+        default_branch_commits = self.repo.get_commits(branch_name=DEFAULT_BRANCH)
         assert docs_branch_commit1 not in list(default_branch_commits)
         assert docs_branch_commit2 not in list(default_branch_commits)
 
         docs_branch_commits = self.repo.get_commits(
             start_id=self.repo.commit_ids[0], end_id=self.repo.commit_ids[-1],
             branch_name=TEST_BRANCH)
         assert docs_branch_commit1 in list(docs_branch_commits)
         assert docs_branch_commit2 in list(docs_branch_commits)
 
     @pytest.mark.backends("svn")
     def test_get_commits_respects_branch_name_svn(self, vcsbackend_svn):
         repo = vcsbackend_svn['svn-simple-layout']
         commits = repo.get_commits(branch_name='trunk')
         commit_indexes = [c.idx for c in commits]
         assert commit_indexes == [1, 2, 3, 7, 12, 15]
 
     def test_get_commit_by_branch(self):
         for branch, commit_id in self.repo.branches.iteritems():
             assert commit_id == self.repo.get_commit(branch).raw_id
 
     def test_get_commit_by_tag(self):
         for tag, commit_id in self.repo.tags.iteritems():
             assert commit_id == self.repo.get_commit(tag).raw_id
 
     def test_get_commit_parents(self):
         repo = self.repo
         for test_idx in [1, 2, 3]:
             commit = repo.get_commit(commit_idx=test_idx - 1)
             assert [commit] == repo.get_commit(commit_idx=test_idx).parents
 
     def test_get_commit_children(self):
         repo = self.repo
         for test_idx in [1, 2, 3]:
             commit = repo.get_commit(commit_idx=test_idx + 1)
             assert [commit] == repo.get_commit(commit_idx=test_idx).children
 
 
 @pytest.mark.usefixtures("vcs_repository_support")
 class TestCommits(BackendTestMixin):
     recreate_repo_per_test = False
 
     @classmethod
     def _get_commits(cls):
         start_date = datetime.datetime(2010, 1, 1, 20)
         for x in xrange(5):
             yield {
                 'message': u'Commit %d' % x,
                 'author': u'Joe Doe <[e-mail]>',
                 'date': start_date + datetime.timedelta(hours=12 * x),
                 'added': [
                     FileNode('file_%d.txt' % x, content='Foobar %d' % x),
                 ],
             }
 
     def test_simple(self):
         tip = self.repo.get_commit()
         assert tip.date, datetime.datetime(2010, 1, 3 == 20)
 
     def test_simple_serialized_commit(self):
         tip = self.repo.get_commit()
         # json.dumps(tip) uses .__json__() method
         data = tip.__json__()
         assert 'branch' in data
         assert data['revision']
 
     def test_retrieve_tip(self):
         tip = self.repo.get_commit('tip')
         assert tip == self.repo.get_commit()
 
     def test_invalid(self):
         with pytest.raises(CommitDoesNotExistError):
             self.repo.get_commit(commit_idx=123456789)
 
     def test_idx(self):
         commit = self.repo[0]
         assert commit.idx == 0
 
     def test_negative_idx(self):
         commit = self.repo.get_commit(commit_idx=-1)
         assert commit.idx >= 0
 
     def test_revision_is_deprecated(self):
         def get_revision(commit):
             return commit.revision
 
         commit = self.repo[0]
         pytest.deprecated_call(get_revision, commit)
 
     def test_size(self):
         tip = self.repo.get_commit()
         size = 5 * len('Foobar N')  # Size of 5 files
         assert tip.size == size
 
     def test_size_at_commit(self):
         tip = self.repo.get_commit()
         size = 5 * len('Foobar N')  # Size of 5 files
         assert self.repo.size_at_commit(tip.raw_id) == size
 
     def test_size_at_first_commit(self):
         commit = self.repo[0]
         size = len('Foobar N')  # Size of 1 file
         assert self.repo.size_at_commit(commit.raw_id) == size
 
     def test_author(self):
         tip = self.repo.get_commit()
         assert_text_equal(tip.author, u'Joe Doe <[e-mail]>')
 
     def test_author_name(self):
         tip = self.repo.get_commit()
         assert_text_equal(tip.author_name, u'Joe Doe')
 
     def test_author_email(self):
         tip = self.repo.get_commit()
         assert_text_equal(tip.author_email, u'[e-mail]')
 
     def test_message(self):
         tip = self.repo.get_commit()
         assert_text_equal(tip.message, u'Commit 4')
 
     def test_diff(self):
         tip = self.repo.get_commit()
         diff = tip.diff()
         assert "+Foobar 4" in diff.raw
 
     def test_prev(self):
         tip = self.repo.get_commit()
         prev_commit = tip.prev()
         assert prev_commit.message == 'Commit 3'
 
     def test_prev_raises_on_first_commit(self):
         commit = self.repo.get_commit(commit_idx=0)
         with pytest.raises(CommitDoesNotExistError):
             commit.prev()
 
     def test_prev_works_on_second_commit_issue_183(self):
         commit = self.repo.get_commit(commit_idx=1)
         prev_commit = commit.prev()
         assert prev_commit.idx == 0
 
     def test_next(self):
         commit = self.repo.get_commit(commit_idx=2)
         next_commit = commit.next()
         assert next_commit.message == 'Commit 3'
 
     def test_next_raises_on_tip(self):
         commit = self.repo.get_commit()
         with pytest.raises(CommitDoesNotExistError):
             commit.next()
 
     def test_get_path_commit(self):
         commit = self.repo.get_commit()
         commit.get_path_commit('file_4.txt')
         assert commit.message == 'Commit 4'
 
     def test_get_filenodes_generator(self):
         tip = self.repo.get_commit()
         filepaths = [node.path for node in tip.get_filenodes_generator()]
         assert filepaths == ['file_%d.txt' % x for x in xrange(5)]
 
     def test_get_file_annotate(self):
         file_added_commit = self.repo.get_commit(commit_idx=3)
         annotations = list(file_added_commit.get_file_annotate('file_3.txt'))
 
         line_no, commit_id, commit_loader, line = annotations[0]
 
         assert line_no == 1
         assert commit_id == file_added_commit.raw_id
         assert commit_loader() == file_added_commit
         assert 'Foobar 3' in line
 
     def test_get_file_annotate_does_not_exist(self):
         file_added_commit = self.repo.get_commit(commit_idx=2)
         # TODO: Should use a specific exception class here?
         with pytest.raises(Exception):
             list(file_added_commit.get_file_annotate('file_3.txt'))
 
     def test_get_file_annotate_tip(self):
         tip = self.repo.get_commit()
         commit = self.repo.get_commit(commit_idx=3)
         expected_values = list(commit.get_file_annotate('file_3.txt'))
         annotations = list(tip.get_file_annotate('file_3.txt'))
 
         # Note: Skip index 2 because the loader function is not the same
         for idx in (0, 1, 3):
             assert annotations[0][idx] == expected_values[0][idx]
 
     def test_get_commits_is_ordered_by_date(self):
         commits = self.repo.get_commits()
         assert isinstance(commits, CollectionGenerator)
         assert len(commits) == 0 or len(commits) != 0
         commits = list(commits)
         ordered_by_date = sorted(commits, key=lambda commit: commit.date)
         assert commits == ordered_by_date
 
     def test_get_commits_respects_start(self):
         second_id = self.repo.commit_ids[1]
         commits = self.repo.get_commits(start_id=second_id)
         assert isinstance(commits, CollectionGenerator)
         commits = list(commits)
         assert len(commits) == 4
 
     def test_get_commits_includes_start_commit(self):
         second_id = self.repo.commit_ids[1]
         commits = self.repo.get_commits(start_id=second_id)
         assert isinstance(commits, CollectionGenerator)
         commits = list(commits)
         assert commits[0].raw_id == second_id
 
     def test_get_commits_respects_end(self):
         second_id = self.repo.commit_ids[1]
         commits = self.repo.get_commits(end_id=second_id)
         assert isinstance(commits, CollectionGenerator)
         commits = list(commits)
         assert commits[-1].raw_id == second_id
         assert len(commits) == 2
 
     def test_get_commits_respects_both_start_and_end(self):
         second_id = self.repo.commit_ids[1]
         third_id = self.repo.commit_ids[2]
         commits = self.repo.get_commits(start_id=second_id, end_id=third_id)
         assert isinstance(commits, CollectionGenerator)
         commits = list(commits)
         assert len(commits) == 2
 
     def test_get_commits_on_empty_repo_raises_EmptyRepository_error(self):
         repo_path = get_new_dir(str(time.time()))
         repo = self.Backend(repo_path, create=True)
 
         with pytest.raises(EmptyRepositoryError):
             list(repo.get_commits(start_id='foobar'))
 
     def test_get_commits_respects_hidden(self):
         commits = self.repo.get_commits(show_hidden=True)
         assert isinstance(commits, CollectionGenerator)
         assert len(commits) == 5
 
     def test_get_commits_includes_end_commit(self):
         second_id = self.repo.commit_ids[1]
         commits = self.repo.get_commits(end_id=second_id)
         assert isinstance(commits, CollectionGenerator)
         assert len(commits) == 2
         commits = list(commits)
         assert commits[-1].raw_id == second_id
 
     def test_get_commits_respects_start_date(self):
         start_date = datetime.datetime(2010, 1, 2)
         commits = self.repo.get_commits(start_date=start_date)
         assert isinstance(commits, CollectionGenerator)
         # Should be 4 commits after 2010-01-02 00:00:00
         assert len(commits) == 4
         for c in commits:
             assert c.date >= start_date
 
     def test_get_commits_respects_start_date_with_branch(self):
         start_date = datetime.datetime(2010, 1, 2)
         commits = self.repo.get_commits(
             start_date=start_date, branch_name=self.repo.DEFAULT_BRANCH_NAME)
         assert isinstance(commits, CollectionGenerator)
         # Should be 4 commits after 2010-01-02 00:00:00
         assert len(commits) == 4
         for c in commits:
             assert c.date >= start_date
 
     def test_get_commits_respects_start_date_and_end_date(self):
         start_date = datetime.datetime(2010, 1, 2)
         end_date = datetime.datetime(2010, 1, 3)
         commits = self.repo.get_commits(start_date=start_date,
                                         end_date=end_date)
         assert isinstance(commits, CollectionGenerator)
         assert len(commits) == 2
         for c in commits:
             assert c.date >= start_date
             assert c.date <= end_date
 
     def test_get_commits_respects_end_date(self):
         end_date = datetime.datetime(2010, 1, 2)
         commits = self.repo.get_commits(end_date=end_date)
         assert isinstance(commits, CollectionGenerator)
         assert len(commits) == 1
         for c in commits:
             assert c.date <= end_date
 
     def test_get_commits_respects_reverse(self):
         commits = self.repo.get_commits()  # no longer reverse support
         assert isinstance(commits, CollectionGenerator)
         assert len(commits) == 5
         commit_ids = reversed([c.raw_id for c in commits])
         assert list(commit_ids) == list(reversed(self.repo.commit_ids))
 
     def test_get_commits_slice_generator(self):
         commits = self.repo.get_commits(
             branch_name=self.repo.DEFAULT_BRANCH_NAME)
         assert isinstance(commits, CollectionGenerator)
         commit_slice = list(commits[1:3])
         assert len(commit_slice) == 2
 
     def test_get_commits_raise_commitdoesnotexist_for_wrong_start(self):
         with pytest.raises(CommitDoesNotExistError):
             list(self.repo.get_commits(start_id='foobar'))
 
     def test_get_commits_raise_commitdoesnotexist_for_wrong_end(self):
         with pytest.raises(CommitDoesNotExistError):
             list(self.repo.get_commits(end_id='foobar'))
 
     def test_get_commits_raise_branchdoesnotexist_for_wrong_branch_name(self):
         with pytest.raises(BranchDoesNotExistError):
             list(self.repo.get_commits(branch_name='foobar'))
 
     def test_get_commits_raise_repositoryerror_for_wrong_start_end(self):
         start_id = self.repo.commit_ids[-1]
         end_id = self.repo.commit_ids[0]
         with pytest.raises(RepositoryError):
             list(self.repo.get_commits(start_id=start_id, end_id=end_id))
 
     def test_get_commits_raises_for_numerical_ids(self):
         with pytest.raises(TypeError):
             self.repo.get_commits(start_id=1, end_id=2)
 
     def test_commit_equality(self):
         commit1 = self.repo.get_commit(self.repo.commit_ids[0])
         commit2 = self.repo.get_commit(self.repo.commit_ids[1])
 
         assert commit1 == commit1
         assert commit2 == commit2
         assert commit1 != commit2
         assert commit2 != commit1
         assert commit1 != None
         assert None != commit1
         assert 1 != commit1
         assert 'string' != commit1
 
 
 @pytest.mark.parametrize("filename, expected", [
     ("README.rst", False),
     ("README", True),
 ])
 def test_commit_is_link(vcsbackend, filename, expected):
     commit = vcsbackend.repo.get_commit()
     link_status = commit.is_link(filename)
     assert link_status is expected
 
 
 @pytest.mark.usefixtures("vcs_repository_support")
 class TestCommitsChanges(BackendTestMixin):
     recreate_repo_per_test = False
 
     @classmethod
     def _get_commits(cls):
         return [
             {
                 'message': u'Initial',
                 'author': u'Joe Doe <[e-mail]>',
                 'date': datetime.datetime(2010, 1, 1, 20),
                 'added': [
                     FileNode('foo/bar', content='foo'),
                     FileNode('foo/bał', content='foo'),
                     FileNode('foobar', content='foo'),
                     FileNode('qwe', content='foo'),
                 ],
             },
             {
                 'message': u'Massive changes',
                 'author': u'Joe Doe <[e-mail]>',
                 'date': datetime.datetime(2010, 1, 1, 22),
                 'added': [FileNode('fallout', content='War never changes')],
                 'changed': [
                     FileNode('foo/bar', content='baz'),
                     FileNode('foobar', content='baz'),
                 ],
                 'removed': [FileNode('qwe')],
             },
         ]
 
     def test_initial_commit(self, local_dt_to_utc):
         commit = self.repo.get_commit(commit_idx=0)
         assert set(commit.added) == set([
             commit.get_node('foo/bar'),
             commit.get_node('foo/bał'),
             commit.get_node('foobar'),
             commit.get_node('qwe'),
         ])
         assert set(commit.changed) == set()
         assert set(commit.removed) == set()
         assert set(commit.affected_files) == set(
             ['foo/bar', 'foo/bał', 'foobar', 'qwe'])
         assert commit.date == local_dt_to_utc(
             datetime.datetime(2010, 1, 1, 20, 0))
 
     def test_head_added(self):
         commit = self.repo.get_commit()
         assert isinstance(commit.added, AddedFileNodesGenerator)
         assert set(commit.added) == set([commit.get_node('fallout')])
         assert isinstance(commit.changed, ChangedFileNodesGenerator)
         assert set(commit.changed) == set([
             commit.get_node('foo/bar'),
             commit.get_node('foobar'),
         ])
         assert isinstance(commit.removed, RemovedFileNodesGenerator)
         assert len(commit.removed) == 1
         assert list(commit.removed)[0].path == 'qwe'
 
     def test_get_filemode(self):
         commit = self.repo.get_commit()
         assert FILEMODE_DEFAULT == commit.get_file_mode('foo/bar')
 
     def test_get_filemode_non_ascii(self):
         commit = self.repo.get_commit()
         assert FILEMODE_DEFAULT == commit.get_file_mode('foo/bał')
         assert FILEMODE_DEFAULT == commit.get_file_mode(u'foo/bał')
 
     def test_get_path_history(self):
         commit = self.repo.get_commit()
         history = commit.get_path_history('foo/bar')
         assert len(history) == 2
 
     def test_get_path_history_with_limit(self):
         commit = self.repo.get_commit()
         history = commit.get_path_history('foo/bar', limit=1)
         assert len(history) == 1
 
     def test_get_path_history_first_commit(self):
         commit = self.repo[0]
         history = commit.get_path_history('foo/bar')
         assert len(history) == 1
 
 
 def assert_text_equal(expected, given):
     assert expected == given
     assert isinstance(expected, unicode)
     assert isinstance(given, unicode)

rhodecode/tests/vcs/test_git.py

@@ -1,1288 +1,1288 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # 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 Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import datetime
 import mock
 import os
 import sys
 import shutil
 
 import pytest
 
 from rhodecode.lib.utils import make_db_config
 from rhodecode.lib.vcs.backends.base import Reference
 from rhodecode.lib.vcs.backends.git import (
     GitRepository, GitCommit, discover_git_version)
 from rhodecode.lib.vcs.exceptions import (
     RepositoryError, VCSError, NodeDoesNotExistError)
 from rhodecode.lib.vcs.nodes import (
     NodeKind, FileNode, DirNode, NodeState, SubModuleNode)
 from rhodecode.tests import TEST_GIT_REPO, TEST_GIT_REPO_CLONE, get_new_dir
 from rhodecode.tests.vcs.conftest import BackendTestMixin
 
 
 pytestmark = pytest.mark.backends("git")
 
 
 def repo_path_generator():
     """
     Return a different path to be used for cloning repos.
     """
     i = 0
     while True:
         i += 1
         yield '%s-%d' % (TEST_GIT_REPO_CLONE, i)
 
 
 REPO_PATH_GENERATOR = repo_path_generator()
 
 
 class TestGitRepository:
 
     # pylint: disable=protected-access
 
     def __check_for_existing_repo(self):
         if os.path.exists(TEST_GIT_REPO_CLONE):
             self.fail('Cannot test git clone repo as location %s already '
                       'exists. You should manually remove it first.'
                       % TEST_GIT_REPO_CLONE)
 
     @pytest.fixture(autouse=True)
     def prepare(self, request, baseapp):
         self.repo = GitRepository(TEST_GIT_REPO, bare=True)
 
     def get_clone_repo(self):
         """
         Return a non bare clone of the base repo.
         """
         clone_path = next(REPO_PATH_GENERATOR)
         repo_clone = GitRepository(
             clone_path, create=True, src_url=self.repo.path, bare=False)
 
         return repo_clone
 
     def get_empty_repo(self, bare=False):
         """
         Return a non bare empty repo.
         """
         return GitRepository(next(REPO_PATH_GENERATOR), create=True, bare=bare)
 
     def test_wrong_repo_path(self):
         wrong_repo_path = '/tmp/errorrepo_git'
         with pytest.raises(RepositoryError):
             GitRepository(wrong_repo_path)
 
     def test_repo_clone(self):
         self.__check_for_existing_repo()
         repo = GitRepository(TEST_GIT_REPO)
         repo_clone = GitRepository(
             TEST_GIT_REPO_CLONE,
             src_url=TEST_GIT_REPO, create=True, do_workspace_checkout=True)
         assert len(repo.commit_ids) == len(repo_clone.commit_ids)
         # Checking hashes of commits should be enough
         for commit in repo.get_commits():
             raw_id = commit.raw_id
             assert raw_id == repo_clone.get_commit(raw_id).raw_id
 
     def test_repo_clone_without_create(self):
         with pytest.raises(RepositoryError):
             GitRepository(
                 TEST_GIT_REPO_CLONE + '_wo_create', src_url=TEST_GIT_REPO)
 
     def test_repo_clone_with_update(self):
         repo = GitRepository(TEST_GIT_REPO)
         clone_path = TEST_GIT_REPO_CLONE + '_with_update'
         repo_clone = GitRepository(
             clone_path,
             create=True, src_url=TEST_GIT_REPO, do_workspace_checkout=True)
         assert len(repo.commit_ids) == len(repo_clone.commit_ids)
 
         # check if current workdir was updated
         fpath = os.path.join(clone_path, 'MANIFEST.in')
         assert os.path.isfile(fpath)
 
     def test_repo_clone_without_update(self):
         repo = GitRepository(TEST_GIT_REPO)
         clone_path = TEST_GIT_REPO_CLONE + '_without_update'
         repo_clone = GitRepository(
             clone_path,
             create=True, src_url=TEST_GIT_REPO, do_workspace_checkout=False)
         assert len(repo.commit_ids) == len(repo_clone.commit_ids)
         # check if current workdir was *NOT* updated
         fpath = os.path.join(clone_path, 'MANIFEST.in')
         # Make sure it's not bare repo
         assert not repo_clone.bare
         assert not os.path.isfile(fpath)
 
     def test_repo_clone_into_bare_repo(self):
         repo = GitRepository(TEST_GIT_REPO)
         clone_path = TEST_GIT_REPO_CLONE + '_bare.git'
         repo_clone = GitRepository(
             clone_path, create=True, src_url=repo.path, bare=True)
         assert repo_clone.bare
 
     def test_create_repo_is_not_bare_by_default(self):
         repo = GitRepository(get_new_dir('not-bare-by-default'), create=True)
         assert not repo.bare
 
     def test_create_bare_repo(self):
         repo = GitRepository(get_new_dir('bare-repo'), create=True, bare=True)
         assert repo.bare
 
     def test_update_server_info(self):
         self.repo._update_server_info()
 
     def test_fetch(self, vcsbackend_git):
         # Note: This is a git specific part of the API, it's only implemented
         # by the git backend.
         source_repo = vcsbackend_git.repo
         target_repo = vcsbackend_git.create_repo(bare=True)
         target_repo.fetch(source_repo.path)
         # Note: Get a fresh instance, avoids caching trouble
         target_repo = vcsbackend_git.backend(target_repo.path)
         assert len(source_repo.commit_ids) == len(target_repo.commit_ids)
 
     def test_commit_ids(self):
         # there are 112 commits (by now)
         # so we can assume they would be available from now on
         subset = {'c1214f7e79e02fc37156ff215cd71275450cffc3',
                   '38b5fe81f109cb111f549bfe9bb6b267e10bc557',
                   'fa6600f6848800641328adbf7811fd2372c02ab2',
                   '102607b09cdd60e2793929c4f90478be29f85a17',
                   '49d3fd156b6f7db46313fac355dca1a0b94a0017',
                   '2d1028c054665b962fa3d307adfc923ddd528038',
                   'd7e0d30fbcae12c90680eb095a4f5f02505ce501',
                   'ff7ca51e58c505fec0dd2491de52c622bb7a806b',
                   'dd80b0f6cf5052f17cc738c2951c4f2070200d7f',
                   '8430a588b43b5d6da365400117c89400326e7992',
                   'd955cd312c17b02143c04fa1099a352b04368118',
                   'f67b87e5c629c2ee0ba58f85197e423ff28d735b',
                   'add63e382e4aabc9e1afdc4bdc24506c269b7618',
                   'f298fe1189f1b69779a4423f40b48edf92a703fc',
                   'bd9b619eb41994cac43d67cf4ccc8399c1125808',
                   '6e125e7c890379446e98980d8ed60fba87d0f6d1',
                   'd4a54db9f745dfeba6933bf5b1e79e15d0af20bd',
                   '0b05e4ed56c802098dfc813cbe779b2f49e92500',
                   '191caa5b2c81ed17c0794bf7bb9958f4dcb0b87e',
                   '45223f8f114c64bf4d6f853e3c35a369a6305520',
                   'ca1eb7957a54bce53b12d1a51b13452f95bc7c7e',
                   'f5ea29fc42ef67a2a5a7aecff10e1566699acd68',
                   '27d48942240f5b91dfda77accd2caac94708cc7d',
                   '622f0eb0bafd619d2560c26f80f09e3b0b0d78af',
                   'e686b958768ee96af8029fe19c6050b1a8dd3b2b'}
         assert subset.issubset(set(self.repo.commit_ids))
 
     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)]:
             commit_ids = list(self.repo[sfrom:sto])
             assert len(commit_ids) == size
             assert commit_ids[0] == self.repo.get_commit(commit_idx=sfrom)
             assert commit_ids[-1] == self.repo.get_commit(commit_idx=sto - 1)
 
     def test_branches(self):
         # TODO: Need more tests here
         # Removed (those are 'remotes' branches for cloned repo)
         # assert 'master' in self.repo.branches
         # assert 'gittree' in self.repo.branches
         # assert 'web-branch' in self.repo.branches
         for __, commit_id in self.repo.branches.items():
             assert isinstance(self.repo.get_commit(commit_id), GitCommit)
 
     def test_tags(self):
         # TODO: Need more tests here
         assert 'v0.1.1' in self.repo.tags
         assert 'v0.1.2' in self.repo.tags
         for __, commit_id in self.repo.tags.items():
             assert isinstance(self.repo.get_commit(commit_id), GitCommit)
 
     def _test_single_commit_cache(self, commit_id):
         commit = self.repo.get_commit(commit_id)
         assert commit_id in self.repo.commits
         assert commit is self.repo.commits[commit_id]
 
     def test_initial_commit(self):
         commit_id = self.repo.commit_ids[0]
         init_commit = self.repo.get_commit(commit_id)
         init_author = init_commit.author
 
         assert init_commit.message == 'initial import\n'
         assert init_author == 'Marcin Kuzminski <[e-mail]>'
         assert init_author == init_commit.committer
         for path in ('vcs/__init__.py',
                      'vcs/backends/BaseRepository.py',
                      'vcs/backends/__init__.py'):
             assert isinstance(init_commit.get_node(path), FileNode)
         for path in ('', 'vcs', 'vcs/backends'):
             assert isinstance(init_commit.get_node(path), DirNode)
 
         with pytest.raises(NodeDoesNotExistError):
             init_commit.get_node(path='foobar')
 
         node = init_commit.get_node('vcs/')
         assert hasattr(node, 'kind')
         assert node.kind == NodeKind.DIR
 
         node = init_commit.get_node('vcs')
         assert hasattr(node, 'kind')
         assert node.kind == NodeKind.DIR
 
         node = init_commit.get_node('vcs/__init__.py')
         assert hasattr(node, 'kind')
         assert node.kind == NodeKind.FILE
 
     def test_not_existing_commit(self):
         with pytest.raises(RepositoryError):
             self.repo.get_commit('f' * 40)
 
     def test_commit10(self):
 
         commit10 = self.repo.get_commit(self.repo.commit_ids[9])
         README = """===
 VCS
 ===
 
 Various Version Control System management abstraction layer for Python.
 
 Introduction
 ------------
 
 TODO: To be written...
 
 """
         node = commit10.get_node('README.rst')
         assert node.kind == NodeKind.FILE
         assert node.content == README
 
     def test_head(self):
         assert self.repo.head == self.repo.get_commit().raw_id
 
     def test_checkout_with_create(self):
         repo_clone = self.get_clone_repo()
 
         new_branch = 'new_branch'
         assert repo_clone._current_branch() == 'master'
         assert set(repo_clone.branches) == {'master'}
         repo_clone._checkout(new_branch, create=True)
 
         # Branches is a lazy property so we need to recrete the Repo object.
         repo_clone = GitRepository(repo_clone.path)
         assert set(repo_clone.branches) == {'master', new_branch}
         assert repo_clone._current_branch() == new_branch
 
     def test_checkout(self):
         repo_clone = self.get_clone_repo()
 
         repo_clone._checkout('new_branch', create=True)
         repo_clone._checkout('master')
 
         assert repo_clone._current_branch() == 'master'
 
     def test_checkout_same_branch(self):
         repo_clone = self.get_clone_repo()
 
         repo_clone._checkout('master')
         assert repo_clone._current_branch() == 'master'
 
     def test_checkout_branch_already_exists(self):
         repo_clone = self.get_clone_repo()
 
         with pytest.raises(RepositoryError):
             repo_clone._checkout('master', create=True)
 
     def test_checkout_bare_repo(self):
         with pytest.raises(RepositoryError):
             self.repo._checkout('master')
 
     def test_current_branch_bare_repo(self):
         with pytest.raises(RepositoryError):
             self.repo._current_branch()
 
     def test_current_branch_empty_repo(self):
         repo = self.get_empty_repo()
         assert repo._current_branch() is None
 
     def test_local_clone(self):
         clone_path = next(REPO_PATH_GENERATOR)
         self.repo._local_clone(clone_path, 'master')
         repo_clone = GitRepository(clone_path)
 
         assert self.repo.commit_ids == repo_clone.commit_ids
 
     def test_local_clone_with_specific_branch(self):
         source_repo = self.get_clone_repo()
 
         # Create a new branch in source repo
         new_branch_commit = source_repo.commit_ids[-3]
         source_repo._checkout(new_branch_commit)
         source_repo._checkout('new_branch', create=True)
 
         clone_path = next(REPO_PATH_GENERATOR)
         source_repo._local_clone(clone_path, 'new_branch')
         repo_clone = GitRepository(clone_path)
 
         assert source_repo.commit_ids[:-3 + 1] == repo_clone.commit_ids
 
         clone_path = next(REPO_PATH_GENERATOR)
         source_repo._local_clone(clone_path, 'master')
         repo_clone = GitRepository(clone_path)
 
         assert source_repo.commit_ids == repo_clone.commit_ids
 
     def test_local_clone_fails_if_target_exists(self):
         with pytest.raises(RepositoryError):
             self.repo._local_clone(self.repo.path, 'master')
 
     def test_local_fetch(self):
         target_repo = self.get_empty_repo()
         source_repo = self.get_clone_repo()
 
         # Create a new branch in source repo
         master_commit = source_repo.commit_ids[-1]
         new_branch_commit = source_repo.commit_ids[-3]
         source_repo._checkout(new_branch_commit)
         source_repo._checkout('new_branch', create=True)
 
         target_repo._local_fetch(source_repo.path, 'new_branch')
         assert target_repo._last_fetch_heads() == [new_branch_commit]
 
         target_repo._local_fetch(source_repo.path, 'master')
         assert target_repo._last_fetch_heads() == [master_commit]
 
     def test_local_fetch_from_bare_repo(self):
         target_repo = self.get_empty_repo()
         target_repo._local_fetch(self.repo.path, 'master')
 
         master_commit = self.repo.commit_ids[-1]
         assert target_repo._last_fetch_heads() == [master_commit]
 
     def test_local_fetch_from_same_repo(self):
         with pytest.raises(ValueError):
             self.repo._local_fetch(self.repo.path, 'master')
 
     def test_local_fetch_branch_does_not_exist(self):
         target_repo = self.get_empty_repo()
 
         with pytest.raises(RepositoryError):
             target_repo._local_fetch(self.repo.path, 'new_branch')
 
     def test_local_pull(self):
         target_repo = self.get_empty_repo()
         source_repo = self.get_clone_repo()
 
         # Create a new branch in source repo
         master_commit = source_repo.commit_ids[-1]
         new_branch_commit = source_repo.commit_ids[-3]
         source_repo._checkout(new_branch_commit)
         source_repo._checkout('new_branch', create=True)
 
         target_repo._local_pull(source_repo.path, 'new_branch')
         target_repo = GitRepository(target_repo.path)
         assert target_repo.head == new_branch_commit
 
         target_repo._local_pull(source_repo.path, 'master')
         target_repo = GitRepository(target_repo.path)
         assert target_repo.head == master_commit
 
     def test_local_pull_in_bare_repo(self):
         with pytest.raises(RepositoryError):
             self.repo._local_pull(self.repo.path, 'master')
 
     def test_local_merge(self):
         target_repo = self.get_empty_repo()
         source_repo = self.get_clone_repo()
 
         # Create a new branch in source repo
         master_commit = source_repo.commit_ids[-1]
         new_branch_commit = source_repo.commit_ids[-3]
         source_repo._checkout(new_branch_commit)
         source_repo._checkout('new_branch', create=True)
 
         # This is required as one cannot do a -ff-only merge in an empty repo.
         target_repo._local_pull(source_repo.path, 'new_branch')
 
         target_repo._local_fetch(source_repo.path, 'master')
         merge_message = 'Merge message\n\nDescription:...'
         user_name = 'Albert Einstein'
         user_email = '[e-mail]'
         target_repo._local_merge(merge_message, user_name, user_email,
                                  target_repo._last_fetch_heads())
 
         target_repo = GitRepository(target_repo.path)
         assert target_repo.commit_ids[-2] == master_commit
         last_commit = target_repo.get_commit(target_repo.head)
         assert last_commit.message.strip() == merge_message
         assert last_commit.author == '%s <%s>' % (user_name, user_email)
 
         assert not os.path.exists(
             os.path.join(target_repo.path, '.git', 'MERGE_HEAD'))
 
     def test_local_merge_raises_exception_on_conflict(self, vcsbackend_git):
         target_repo = vcsbackend_git.create_repo(number_of_commits=1)
         vcsbackend_git.ensure_file('README', 'I will conflict with you!!!')
 
         target_repo._local_fetch(self.repo.path, 'master')
         with pytest.raises(RepositoryError):
             target_repo._local_merge(
                 'merge_message', 'user name', '[e-mail]',
                 target_repo._last_fetch_heads())
 
         # Check we are not left in an intermediate merge state
         assert not os.path.exists(
             os.path.join(target_repo.path, '.git', 'MERGE_HEAD'))
 
     def test_local_merge_into_empty_repo(self):
         target_repo = self.get_empty_repo()
 
         # This is required as one cannot do a -ff-only merge in an empty repo.
         target_repo._local_fetch(self.repo.path, 'master')
         with pytest.raises(RepositoryError):
             target_repo._local_merge(
                 'merge_message', 'user name', '[e-mail]',
                 target_repo._last_fetch_heads())
 
     def test_local_merge_in_bare_repo(self):
         with pytest.raises(RepositoryError):
             self.repo._local_merge(
                 'merge_message', 'user name', '[e-mail]', None)
 
     def test_local_push_non_bare(self):
         target_repo = self.get_empty_repo()
 
         pushed_branch = 'pushed_branch'
         self.repo._local_push('master', target_repo.path, pushed_branch)
         # Fix the HEAD of the target repo, or otherwise GitRepository won't
         # report any branches.
         with open(os.path.join(target_repo.path, '.git', 'HEAD'), 'w') as f:
             f.write('ref: refs/heads/%s' % pushed_branch)
 
         target_repo = GitRepository(target_repo.path)
 
         assert (target_repo.branches[pushed_branch] ==
                 self.repo.branches['master'])
 
     def test_local_push_bare(self):
         target_repo = self.get_empty_repo(bare=True)
 
         pushed_branch = 'pushed_branch'
         self.repo._local_push('master', target_repo.path, pushed_branch)
         # Fix the HEAD of the target repo, or otherwise GitRepository won't
         # report any branches.
         with open(os.path.join(target_repo.path, 'HEAD'), 'w') as f:
             f.write('ref: refs/heads/%s' % pushed_branch)
 
         target_repo = GitRepository(target_repo.path)
 
         assert (target_repo.branches[pushed_branch] ==
                 self.repo.branches['master'])
 
     def test_local_push_non_bare_target_branch_is_checked_out(self):
         target_repo = self.get_clone_repo()
 
         pushed_branch = 'pushed_branch'
         # Create a new branch in source repo
         new_branch_commit = target_repo.commit_ids[-3]
         target_repo._checkout(new_branch_commit)
         target_repo._checkout(pushed_branch, create=True)
 
         self.repo._local_push('master', target_repo.path, pushed_branch)
 
         target_repo = GitRepository(target_repo.path)
 
         assert (target_repo.branches[pushed_branch] ==
                 self.repo.branches['master'])
 
     def test_local_push_raises_exception_on_conflict(self, vcsbackend_git):
         target_repo = vcsbackend_git.create_repo(number_of_commits=1)
         with pytest.raises(RepositoryError):
             self.repo._local_push('master', target_repo.path, 'master')
 
     def test_hooks_can_be_enabled_via_env_variable_for_local_push(self):
         target_repo = self.get_empty_repo(bare=True)
 
         with mock.patch.object(self.repo, 'run_git_command') as run_mock:
             self.repo._local_push(
                 'master', target_repo.path, 'master', enable_hooks=True)
         env = run_mock.call_args[1]['extra_env']
         assert 'RC_SKIP_HOOKS' not in env
 
     def _add_failing_hook(self, repo_path, hook_name, bare=False):
         path_components = (
             ['hooks', hook_name] if bare else ['.git', 'hooks', hook_name])
         hook_path = os.path.join(repo_path, *path_components)
         with open(hook_path, 'w') as f:
             script_lines = [
                 '#!%s' % sys.executable,
                 'import os',
                 'import sys',
                 'if os.environ.get("RC_SKIP_HOOKS"):',
                 '    sys.exit(0)',
                 'sys.exit(1)',
             ]
             f.write('\n'.join(script_lines))
         os.chmod(hook_path, 0o755)
 
     def test_local_push_does_not_execute_hook(self):
         target_repo = self.get_empty_repo()
 
         pushed_branch = 'pushed_branch'
         self._add_failing_hook(target_repo.path, 'pre-receive')
         self.repo._local_push('master', target_repo.path, pushed_branch)
         # Fix the HEAD of the target repo, or otherwise GitRepository won't
         # report any branches.
         with open(os.path.join(target_repo.path, '.git', 'HEAD'), 'w') as f:
             f.write('ref: refs/heads/%s' % pushed_branch)
 
         target_repo = GitRepository(target_repo.path)
 
         assert (target_repo.branches[pushed_branch] ==
                 self.repo.branches['master'])
 
     def test_local_push_executes_hook(self):
         target_repo = self.get_empty_repo(bare=True)
         self._add_failing_hook(target_repo.path, 'pre-receive', bare=True)
         with pytest.raises(RepositoryError):
             self.repo._local_push(
                 'master', target_repo.path, 'master', enable_hooks=True)
 
     def test_maybe_prepare_merge_workspace(self):
         workspace = self.repo._maybe_prepare_merge_workspace(
             2, 'pr2', Reference('branch', 'master', 'unused'),
             Reference('branch', 'master', 'unused'))
 
         assert os.path.isdir(workspace)
         workspace_repo = GitRepository(workspace)
         assert workspace_repo.branches == self.repo.branches
 
         # Calling it a second time should also succeed
         workspace = self.repo._maybe_prepare_merge_workspace(
             2, 'pr2', Reference('branch', 'master', 'unused'),
             Reference('branch', 'master', 'unused'))
         assert os.path.isdir(workspace)
 
     def test_maybe_prepare_merge_workspace_different_refs(self):
         workspace = self.repo._maybe_prepare_merge_workspace(
             2, 'pr2', Reference('branch', 'master', 'unused'),
             Reference('branch', 'develop', 'unused'))
 
         assert os.path.isdir(workspace)
         workspace_repo = GitRepository(workspace)
         assert workspace_repo.branches == self.repo.branches
 
         # Calling it a second time should also succeed
         workspace = self.repo._maybe_prepare_merge_workspace(
             2, 'pr2', Reference('branch', 'master', 'unused'),
             Reference('branch', 'develop', 'unused'))
         assert os.path.isdir(workspace)
 
     def test_cleanup_merge_workspace(self):
         workspace = self.repo._maybe_prepare_merge_workspace(
             2, 'pr3', Reference('branch', 'master', 'unused'),
             Reference('branch', 'master', 'unused'))
         self.repo.cleanup_merge_workspace(2, 'pr3')
 
         assert not os.path.exists(workspace)
 
     def test_cleanup_merge_workspace_invalid_workspace_id(self):
         # No assert: because in case of an inexistent workspace this function
         # should still succeed.
         self.repo.cleanup_merge_workspace(1, 'pr4')
 
     def test_set_refs(self):
         test_ref = 'refs/test-refs/abcde'
         test_commit_id = 'ecb86e1f424f2608262b130db174a7dfd25a6623'
 
         self.repo.set_refs(test_ref, test_commit_id)
         stdout, _ = self.repo.run_git_command(['show-ref'])
         assert test_ref in stdout
         assert test_commit_id in stdout
 
     def test_remove_ref(self):
         test_ref = 'refs/test-refs/abcde'
         test_commit_id = 'ecb86e1f424f2608262b130db174a7dfd25a6623'
         self.repo.set_refs(test_ref, test_commit_id)
         stdout, _ = self.repo.run_git_command(['show-ref'])
         assert test_ref in stdout
         assert test_commit_id in stdout
 
         self.repo.remove_ref(test_ref)
         stdout, _ = self.repo.run_git_command(['show-ref'])
         assert test_ref not in stdout
         assert test_commit_id not in stdout
 
 
 class TestGitCommit(object):
 
     @pytest.fixture(autouse=True)
     def prepare(self):
         self.repo = GitRepository(TEST_GIT_REPO)
 
     def test_default_commit(self):
         tip = self.repo.get_commit()
         assert tip == self.repo.get_commit(None)
         assert tip == self.repo.get_commit('tip')
 
     def test_root_node(self):
         tip = self.repo.get_commit()
         assert tip.root is tip.get_node('')
 
     def test_lazy_fetch(self):
         """
         Test if commit's nodes expands and are cached as we walk through
         the commit. This test is somewhat hard to write as order of tests
         is a key here. Written by running command after command in a shell.
         """
         commit_id = '2a13f185e4525f9d4b59882791a2d397b90d5ddc'
         assert commit_id in self.repo.commit_ids
         commit = self.repo.get_commit(commit_id)
         assert len(commit.nodes) == 0
         root = commit.root
         assert len(commit.nodes) == 1
         assert len(root.nodes) == 8
         # accessing root.nodes updates commit.nodes
         assert len(commit.nodes) == 9
 
         docs = root.get_node('docs')
         # we haven't yet accessed anything new as docs dir was already cached
         assert len(commit.nodes) == 9
         assert len(docs.nodes) == 8
         # accessing docs.nodes updates commit.nodes
         assert len(commit.nodes) == 17
 
         assert docs is commit.get_node('docs')
         assert docs is root.nodes[0]
         assert docs is root.dirs[0]
         assert docs is commit.get_node('docs')
 
     def test_nodes_with_commit(self):
         commit_id = '2a13f185e4525f9d4b59882791a2d397b90d5ddc'
         commit = self.repo.get_commit(commit_id)
         root = commit.root
         docs = root.get_node('docs')
         assert docs is commit.get_node('docs')
         api = docs.get_node('api')
         assert api is commit.get_node('docs/api')
         index = api.get_node('index.rst')
         assert index is commit.get_node('docs/api/index.rst')
         assert index is commit.get_node('docs')\
             .get_node('api')\
             .get_node('index.rst')
 
     def test_branch_and_tags(self):
         """
         rev0 = self.repo.commit_ids[0]
         commit0 = self.repo.get_commit(rev0)
         assert commit0.branch == 'master'
         assert commit0.tags == []
 
         rev10 = self.repo.commit_ids[10]
         commit10 = self.repo.get_commit(rev10)
         assert commit10.branch == 'master'
         assert commit10.tags == []
 
         rev44 = self.repo.commit_ids[44]
         commit44 = self.repo.get_commit(rev44)
         assert commit44.branch == 'web-branch'
 
         tip = self.repo.get_commit('tip')
         assert 'tip' in tip.tags
         """
         # Those tests would fail - branches are now going
         # to be changed at main API in order to support git backend
         pass
 
     def test_file_size(self):
         to_check = (
             ('c1214f7e79e02fc37156ff215cd71275450cffc3',
                 'vcs/backends/BaseRepository.py', 502),
             ('d7e0d30fbcae12c90680eb095a4f5f02505ce501',
                 'vcs/backends/hg.py', 854),
             ('6e125e7c890379446e98980d8ed60fba87d0f6d1',
                 'setup.py', 1068),
 
             ('d955cd312c17b02143c04fa1099a352b04368118',
                 'vcs/backends/base.py', 2921),
             ('ca1eb7957a54bce53b12d1a51b13452f95bc7c7e',
                 'vcs/backends/base.py', 3936),
             ('f50f42baeed5af6518ef4b0cb2f1423f3851a941',
                 'vcs/backends/base.py', 6189),
         )
         for commit_id, path, size in to_check:
             node = self.repo.get_commit(commit_id).get_node(path)
             assert node.is_file()
             assert node.size == size
 
     def test_file_history_from_commits(self):
         node = self.repo[10].get_node('setup.py')
         commit_ids = [commit.raw_id for commit in node.history]
         assert ['ff7ca51e58c505fec0dd2491de52c622bb7a806b'] == commit_ids
 
         node = self.repo[20].get_node('setup.py')
         node_ids = [commit.raw_id for commit in node.history]
         assert ['191caa5b2c81ed17c0794bf7bb9958f4dcb0b87e',
                 'ff7ca51e58c505fec0dd2491de52c622bb7a806b'] == node_ids
 
         # special case we check history from commit that has this particular
         # file changed this means we check if it's included as well
         node = self.repo.get_commit('191caa5b2c81ed17c0794bf7bb9958f4dcb0b87e') \
             .get_node('setup.py')
         node_ids = [commit.raw_id for commit in node.history]
         assert ['191caa5b2c81ed17c0794bf7bb9958f4dcb0b87e',
                 'ff7ca51e58c505fec0dd2491de52c622bb7a806b'] == node_ids
 
     def test_file_history(self):
         # we can only check if those commits are present in the history
         # as we cannot update this test every time file is changed
         files = {
             'setup.py': [
                 '54386793436c938cff89326944d4c2702340037d',
                 '51d254f0ecf5df2ce50c0b115741f4cf13985dab',
                 '998ed409c795fec2012b1c0ca054d99888b22090',
                 '5e0eb4c47f56564395f76333f319d26c79e2fb09',
                 '0115510b70c7229dbc5dc49036b32e7d91d23acd',
                 '7cb3fd1b6d8c20ba89e2264f1c8baebc8a52d36e',
                 '2a13f185e4525f9d4b59882791a2d397b90d5ddc',
                 '191caa5b2c81ed17c0794bf7bb9958f4dcb0b87e',
                 'ff7ca51e58c505fec0dd2491de52c622bb7a806b',
             ],
             'vcs/nodes.py': [
                 '33fa3223355104431402a888fa77a4e9956feb3e',
                 'fa014c12c26d10ba682fadb78f2a11c24c8118e1',
                 'e686b958768ee96af8029fe19c6050b1a8dd3b2b',
                 'ab5721ca0a081f26bf43d9051e615af2cc99952f',
                 'c877b68d18e792a66b7f4c529ea02c8f80801542',
                 '4313566d2e417cb382948f8d9d7c765330356054',
                 '6c2303a793671e807d1cfc70134c9ca0767d98c2',
                 '54386793436c938cff89326944d4c2702340037d',
                 '54000345d2e78b03a99d561399e8e548de3f3203',
                 '1c6b3677b37ea064cb4b51714d8f7498f93f4b2b',
                 '2d03ca750a44440fb5ea8b751176d1f36f8e8f46',
                 '2a08b128c206db48c2f0b8f70df060e6db0ae4f8',
                 '30c26513ff1eb8e5ce0e1c6b477ee5dc50e2f34b',
                 'ac71e9503c2ca95542839af0ce7b64011b72ea7c',
                 '12669288fd13adba2a9b7dd5b870cc23ffab92d2',
                 '5a0c84f3e6fe3473e4c8427199d5a6fc71a9b382',
                 '12f2f5e2b38e6ff3fbdb5d722efed9aa72ecb0d5',
                 '5eab1222a7cd4bfcbabc218ca6d04276d4e27378',
                 'f50f42baeed5af6518ef4b0cb2f1423f3851a941',
                 'd7e390a45f6aa96f04f5e7f583ad4f867431aa25',
                 'f15c21f97864b4f071cddfbf2750ec2e23859414',
                 'e906ef056cf539a4e4e5fc8003eaf7cf14dd8ade',
                 'ea2b108b48aa8f8c9c4a941f66c1a03315ca1c3b',
                 '84dec09632a4458f79f50ddbbd155506c460b4f9',
                 '0115510b70c7229dbc5dc49036b32e7d91d23acd',
                 '2a13f185e4525f9d4b59882791a2d397b90d5ddc',
                 '3bf1c5868e570e39569d094f922d33ced2fa3b2b',
                 'b8d04012574729d2c29886e53b1a43ef16dd00a1',
                 '6970b057cffe4aab0a792aa634c89f4bebf01441',
                 'dd80b0f6cf5052f17cc738c2951c4f2070200d7f',
                 'ff7ca51e58c505fec0dd2491de52c622bb7a806b',
             ],
             'vcs/backends/git.py': [
                 '4cf116ad5a457530381135e2f4c453e68a1b0105',
                 '9a751d84d8e9408e736329767387f41b36935153',
                 'cb681fb539c3faaedbcdf5ca71ca413425c18f01',
                 '428f81bb652bcba8d631bce926e8834ff49bdcc6',
                 '180ab15aebf26f98f714d8c68715e0f05fa6e1c7',
                 '2b8e07312a2e89e92b90426ab97f349f4bce2a3a',
                 '50e08c506174d8645a4bb517dd122ac946a0f3bf',
                 '54000345d2e78b03a99d561399e8e548de3f3203',
             ],
         }
         for path, commit_ids in files.items():
             node = self.repo.get_commit(commit_ids[0]).get_node(path)
             node_ids = [commit.raw_id for commit in node.history]
             assert set(commit_ids).issubset(set(node_ids)), (
                 "We assumed that %s is subset of commit_ids for which file %s "
                 "has been changed, and history of that node returned: %s"
                 % (commit_ids, path, node_ids))
 
     def test_file_annotate(self):
         files = {
             'vcs/backends/__init__.py': {
                 'c1214f7e79e02fc37156ff215cd71275450cffc3': {
                     'lines_no': 1,
                     'commits': [
                         'c1214f7e79e02fc37156ff215cd71275450cffc3',
                     ],
                 },
                 '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647': {
                     'lines_no': 21,
                     'commits': [
                         '49d3fd156b6f7db46313fac355dca1a0b94a0017',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                     ],
                 },
                 'e29b67bd158580fc90fc5e9111240b90e6e86064': {
                     'lines_no': 32,
                     'commits': [
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '5eab1222a7cd4bfcbabc218ca6d04276d4e27378',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '54000345d2e78b03a99d561399e8e548de3f3203',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '78c3f0c23b7ee935ec276acb8b8212444c33c396',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '2a13f185e4525f9d4b59882791a2d397b90d5ddc',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '78c3f0c23b7ee935ec276acb8b8212444c33c396',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                     ],
                 },
             },
         }
 
         for fname, commit_dict in files.items():
             for commit_id, __ in commit_dict.items():
                 commit = self.repo.get_commit(commit_id)
 
                 l1_1 = [x[1] for x in commit.get_file_annotate(fname)]
                 l1_2 = [x[2]().raw_id for x in commit.get_file_annotate(fname)]
                 assert l1_1 == l1_2
                 l1 = l1_1
                 l2 = files[fname][commit_id]['commits']
                 assert l1 == l2, (
                     "The lists of commit_ids for %s@commit_id %s"
                     "from annotation list should match each other, "
                     "got \n%s \nvs \n%s " % (fname, commit_id, l1, l2))
 
     def test_files_state(self):
         """
         Tests state of FileNodes.
         """
         node = self.repo\
             .get_commit('e6ea6d16e2f26250124a1f4b4fe37a912f9d86a0')\
             .get_node('vcs/utils/diffs.py')
         assert node.state, NodeState.ADDED
         assert node.added
         assert not node.changed
         assert not node.not_changed
         assert not node.removed
 
         node = self.repo\
             .get_commit('33fa3223355104431402a888fa77a4e9956feb3e')\
             .get_node('.hgignore')
         assert node.state, NodeState.CHANGED
         assert not node.added
         assert node.changed
         assert not node.not_changed
         assert not node.removed
 
         node = self.repo\
             .get_commit('e29b67bd158580fc90fc5e9111240b90e6e86064')\
             .get_node('setup.py')
         assert node.state, NodeState.NOT_CHANGED
         assert not node.added
         assert not node.changed
         assert node.not_changed
         assert not node.removed
 
         # If node has REMOVED state then trying to fetch it would raise
         # CommitError exception
         commit = self.repo.get_commit(
             'fa6600f6848800641328adbf7811fd2372c02ab2')
         path = 'vcs/backends/BaseRepository.py'
         with pytest.raises(NodeDoesNotExistError):
             commit.get_node(path)
         # but it would be one of ``removed`` (commit's attribute)
         assert path in [rf.path for rf in commit.removed]
 
         commit = self.repo.get_commit(
             '54386793436c938cff89326944d4c2702340037d')
         changed = [
             'setup.py', 'tests/test_nodes.py', 'vcs/backends/hg.py',
             'vcs/nodes.py']
         assert set(changed) == set([f.path for f in commit.changed])
 
     def test_unicode_branch_refs(self):
         unicode_branches = {
             'refs/heads/unicode': '6c0ce52b229aa978889e91b38777f800e85f330b',
             u'refs/heads/uniçö∂e': 'ürl',
         }
         with mock.patch(
             ("rhodecode.lib.vcs.backends.git.repository"
                 ".GitRepository._refs"),
                 unicode_branches):
             branches = self.repo.branches
 
         assert 'unicode' in branches
         assert u'uniçö∂e' in branches
 
     def test_unicode_tag_refs(self):
         unicode_tags = {
             'refs/tags/unicode': '6c0ce52b229aa978889e91b38777f800e85f330b',
             u'refs/tags/uniçö∂e': '6c0ce52b229aa978889e91b38777f800e85f330b',
         }
         with mock.patch(
             ("rhodecode.lib.vcs.backends.git.repository"
                 ".GitRepository._refs"),
                 unicode_tags):
             tags = self.repo.tags
 
         assert 'unicode' in tags
         assert u'uniçö∂e' in tags
 
     def test_commit_message_is_unicode(self):
         for commit in self.repo:
             assert type(commit.message) == unicode
 
     def test_commit_author_is_unicode(self):
         for commit in self.repo:
             assert type(commit.author) == unicode
 
     def test_repo_files_content_is_unicode(self):
         commit = self.repo.get_commit()
         for node in commit.get_node('/'):
             if node.is_file():
                 assert 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_commit()
         with pytest.raises(VCSError):
             tip.get_node(path)
 
     @pytest.mark.parametrize("author_email, commit_id", [
         ('[e-mail]', 'c1214f7e79e02fc37156ff215cd71275450cffc3'),
         ('[e-mail]',
          'ff7ca51e58c505fec0dd2491de52c622bb7a806b'),
         ('none@none', '8430a588b43b5d6da365400117c89400326e7992'),
     ])
     def test_author_email(self, author_email, commit_id):
         commit = self.repo.get_commit(commit_id)
         assert author_email == commit.author_email
 
     @pytest.mark.parametrize("author, commit_id", [
         ('Marcin Kuzminski', 'c1214f7e79e02fc37156ff215cd71275450cffc3'),
         ('Lukasz Balcerzak', 'ff7ca51e58c505fec0dd2491de52c622bb7a806b'),
         ('marcink', '8430a588b43b5d6da365400117c89400326e7992'),
     ])
     def test_author_username(self, author, commit_id):
         commit = self.repo.get_commit(commit_id)
         assert author == commit.author_name
 
 
 class TestLargeFileRepo(object):
 
     def test_large_file(self, backend_git):
         conf = make_db_config()
         repo = backend_git.create_test_repo('largefiles', conf)
 
         tip = repo.scm_instance().get_commit()
 
         # extract stored LF node into the origin cache
         lfs_store = os.path.join(repo.repo_path, repo.repo_name, 'lfs_store')
 
         oid = '7b331c02e313c7599d5a90212e17e6d3cb729bd2e1c9b873c302a63c95a2f9bf'
         oid_path = os.path.join(lfs_store, oid)
         oid_destination = os.path.join(
             conf.get('vcs_git_lfs', 'store_location'), oid)
         shutil.copy(oid_path, oid_destination)
 
         node = tip.get_node('1MB.zip')
 
         lf_node = node.get_largefile_node()
 
         assert lf_node.is_largefile() is True
         assert lf_node.size == 1024000
         assert lf_node.name == '1MB.zip'
 
 
 @pytest.mark.usefixtures("vcs_repository_support")
 class TestGitSpecificWithRepo(BackendTestMixin):
 
     @classmethod
     def _get_commits(cls):
         return [
             {
                 'message': 'Initial',
                 'author': 'Joe Doe <[e-mail]>',
                 'date': datetime.datetime(2010, 1, 1, 20),
                 'added': [
                     FileNode('foobar/static/js/admin/base.js', content='base'),
                     FileNode(
                         'foobar/static/admin', content='admin',
                         mode=0o120000),  # this is a link
                     FileNode('foo', content='foo'),
                 ],
             },
             {
                 'message': 'Second',
                 'author': 'Joe Doe <[e-mail]>',
                 'date': datetime.datetime(2010, 1, 1, 22),
                 'added': [
                     FileNode('foo2', content='foo2'),
                 ],
             },
         ]
 
     def test_paths_slow_traversing(self):
         commit = self.repo.get_commit()
         assert commit.get_node('foobar').get_node('static').get_node('js')\
             .get_node('admin').get_node('base.js').content == 'base'
 
     def test_paths_fast_traversing(self):
         commit = self.repo.get_commit()
         assert (
             commit.get_node('foobar/static/js/admin/base.js').content ==
             'base')
 
     def test_get_diff_runs_git_command_with_hashes(self):
         self.repo.run_git_command = mock.Mock(return_value=['', ''])
         self.repo.get_diff(self.repo[0], self.repo[1])
         self.repo.run_git_command.assert_called_once_with(
             ['diff', '-U3', '--full-index', '--binary', '-p', '-M',
-             '--abbrev=40', self.repo._get_commit_id(0),
-             self.repo._get_commit_id(1)])
+             '--abbrev=40', self.repo._lookup_commit(0),
+             self.repo._lookup_commit(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_COMMIT, self.repo[1])
         self.repo.run_git_command.assert_called_once_with(
             ['show', '-U3', '--full-index', '--binary', '-p', '-M',
-             '--abbrev=40', self.repo._get_commit_id(1)])
+             '--abbrev=40', self.repo._lookup_commit(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(self.repo[0], self.repo[1], 'foo')
         self.repo.run_git_command.assert_called_once_with(
             ['diff', '-U3', '--full-index', '--binary', '-p', '-M',
-             '--abbrev=40', self.repo._get_commit_id(0),
-             self.repo._get_commit_id(1), '--', 'foo'])
+             '--abbrev=40', self.repo._lookup_commit(0),
+             self.repo._lookup_commit(1), '--', 'foo'])
 
 
 @pytest.mark.usefixtures("vcs_repository_support")
 class TestGitRegression(BackendTestMixin):
 
     @classmethod
     def _get_commits(cls):
         return [
             {
                 'message': 'Initial',
                 'author': 'Joe Doe <[e-mail]>',
                 '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 <[e-mail]>',
                 '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'),
                 ],
             },
         ]
 
     @pytest.mark.parametrize("path, expected_paths", [
         ('bot', [
             'bot/build',
             'bot/templates',
             'bot/__init__.py']),
         ('bot/build', [
             'bot/build/migrations',
             'bot/build/static',
             'bot/build/templates']),
         ('bot/build/static', [
             'bot/build/static/templates']),
         ('bot/build/static/templates', [
             'bot/build/static/templates/f.html',
             'bot/build/static/templates/f1.html']),
         ('bot/build/templates', [
             'bot/build/templates/err.html',
             'bot/build/templates/err2.html']),
         ('bot/templates/', [
             'bot/templates/404.html',
             'bot/templates/500.html']),
     ])
     def test_similar_paths(self, path, expected_paths):
         commit = self.repo.get_commit()
         paths = [n.path for n in commit.get_nodes(path)]
         assert paths == expected_paths
 
 
 class TestDiscoverGitVersion(object):
 
     def test_returns_git_version(self, baseapp):
         version = discover_git_version()
         assert version
 
     def test_returns_empty_string_without_vcsserver(self):
         mock_connection = mock.Mock()
         mock_connection.discover_git_version = mock.Mock(
             side_effect=Exception)
         with mock.patch('rhodecode.lib.vcs.connection.Git', mock_connection):
             version = discover_git_version()
         assert version == ''
 
 
 class TestGetSubmoduleUrl(object):
     def test_submodules_file_found(self):
         commit = GitCommit(repository=mock.Mock(), raw_id='abcdef12', idx=1)
         node = mock.Mock()
         with mock.patch.object(
                 commit, 'get_node', return_value=node) as get_node_mock:
             node.content = (
                 '[submodule "subrepo1"]\n'
                 '\tpath = subrepo1\n'
                 '\turl = https://code.rhodecode.com/dulwich\n'
             )
             result = commit._get_submodule_url('subrepo1')
         get_node_mock.assert_called_once_with('.gitmodules')
         assert result == 'https://code.rhodecode.com/dulwich'
 
     def test_complex_submodule_path(self):
         commit = GitCommit(repository=mock.Mock(), raw_id='abcdef12', idx=1)
         node = mock.Mock()
         with mock.patch.object(
                 commit, 'get_node', return_value=node) as get_node_mock:
             node.content = (
                 '[submodule "complex/subrepo/path"]\n'
                 '\tpath = complex/subrepo/path\n'
                 '\turl = https://code.rhodecode.com/dulwich\n'
             )
             result = commit._get_submodule_url('complex/subrepo/path')
         get_node_mock.assert_called_once_with('.gitmodules')
         assert result == 'https://code.rhodecode.com/dulwich'
 
     def test_submodules_file_not_found(self):
         commit = GitCommit(repository=mock.Mock(), raw_id='abcdef12', idx=1)
         with mock.patch.object(
                 commit, 'get_node', side_effect=NodeDoesNotExistError):
             result = commit._get_submodule_url('complex/subrepo/path')
         assert result is None
 
     def test_path_not_found(self):
         commit = GitCommit(repository=mock.Mock(), raw_id='abcdef12', idx=1)
         node = mock.Mock()
         with mock.patch.object(
                 commit, 'get_node', return_value=node) as get_node_mock:
             node.content = (
                 '[submodule "subrepo1"]\n'
                 '\tpath = subrepo1\n'
                 '\turl = https://code.rhodecode.com/dulwich\n'
             )
             result = commit._get_submodule_url('subrepo2')
         get_node_mock.assert_called_once_with('.gitmodules')
         assert result is None
 
     def test_returns_cached_values(self):
         commit = GitCommit(repository=mock.Mock(), raw_id='abcdef12', idx=1)
         node = mock.Mock()
         with mock.patch.object(
                 commit, 'get_node', return_value=node) as get_node_mock:
             node.content = (
                 '[submodule "subrepo1"]\n'
                 '\tpath = subrepo1\n'
                 '\turl = https://code.rhodecode.com/dulwich\n'
             )
             for _ in range(3):
                 commit._get_submodule_url('subrepo1')
         get_node_mock.assert_called_once_with('.gitmodules')
 
     def test_get_node_returns_a_link(self):
         repository = mock.Mock()
         repository.alias = 'git'
         commit = GitCommit(repository=repository, raw_id='abcdef12', idx=1)
         submodule_url = 'https://code.rhodecode.com/dulwich'
         get_id_patch = mock.patch.object(
             commit, '_get_id_for_path', return_value=(1, 'link'))
         get_submodule_patch = mock.patch.object(
             commit, '_get_submodule_url', return_value=submodule_url)
 
         with get_id_patch, get_submodule_patch as submodule_mock:
             node = commit.get_node('/abcde')
 
         submodule_mock.assert_called_once_with('/abcde')
         assert type(node) == SubModuleNode
         assert node.url == submodule_url
 
     def test_get_nodes_returns_links(self):
         repository = mock.MagicMock()
         repository.alias = 'git'
         repository._remote.tree_items.return_value = [
             ('subrepo', 'stat', 1, 'link')
         ]
         commit = GitCommit(repository=repository, raw_id='abcdef12', idx=1)
         submodule_url = 'https://code.rhodecode.com/dulwich'
         get_id_patch = mock.patch.object(
             commit, '_get_id_for_path', return_value=(1, 'tree'))
         get_submodule_patch = mock.patch.object(
             commit, '_get_submodule_url', return_value=submodule_url)
 
         with get_id_patch, get_submodule_patch as submodule_mock:
             nodes = commit.get_nodes('/abcde')
 
         submodule_mock.assert_called_once_with('/abcde/subrepo')
         assert len(nodes) == 1
         assert type(nodes[0]) == SubModuleNode
         assert nodes[0].url == submodule_url