repos: cleannup and fix landing-ref code.

with latest git update this exposed a bug when we allowed to set references like tip for git repos
now the defaults are loaded from a single place controlled by default branch for git.

Repos without master are now working just fine with new git and new logic.

rhodecode/api/tests/test_update_repo.py

@@ -1,203 +1,210 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2020 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 mock
 import pytest
 
 from rhodecode.model.repo import RepoModel
+from rhodecode.model.scm import ScmModel
 from rhodecode.tests import TEST_USER_ADMIN_LOGIN, TEST_USER_REGULAR_LOGIN
 from rhodecode.api.tests.utils import (
     build_data, api_call, assert_error, assert_ok, crash, jsonify)
 from rhodecode.tests.fixture import Fixture
 from rhodecode.tests.plugin import plain_http_host_only_stub
 
 fixture = Fixture()
 
 UPDATE_REPO_NAME = 'api_update_me'
 
 
 class SAME_AS_UPDATES(object):
     """ Constant used for tests below """
 
 
 @pytest.mark.usefixtures("testuser_api", "app")
 class TestApiUpdateRepo(object):
 
     @pytest.mark.parametrize("updates, expected", [
         ({'owner': TEST_USER_REGULAR_LOGIN},
          SAME_AS_UPDATES),
 
         ({'description': 'new description'},
          SAME_AS_UPDATES),
 
         ({'clone_uri': 'http://foo.com/repo'},
          SAME_AS_UPDATES),
 
         ({'clone_uri': None},
          {'clone_uri': ''}),
 
         ({'clone_uri': ''},
          {'clone_uri': ''}),
 
         ({'clone_uri': 'http://example.com/repo_pull'},
          {'clone_uri': 'http://example.com/repo_pull'}),
 
         ({'push_uri': ''},
          {'push_uri': ''}),
 
         ({'push_uri': 'http://example.com/repo_push'},
          {'push_uri': 'http://example.com/repo_push'}),
 
-        ({'landing_rev': 'rev:tip'},
-         {'landing_rev': ['rev', 'tip']}),
+        ({'landing_rev': None},  # auto-updated based on type of repo
+         {'landing_rev': [None, None]}),
 
         ({'enable_statistics': True},
          SAME_AS_UPDATES),
 
         ({'enable_locking': True},
          SAME_AS_UPDATES),
 
         ({'enable_downloads': True},
          SAME_AS_UPDATES),
 
         ({'repo_name': 'new_repo_name'},
          {
             'repo_name': 'new_repo_name',
             'url': 'http://{}/new_repo_name'.format(plain_http_host_only_stub())
          }),
 
         ({'repo_name': 'test_group_for_update/{}'.format(UPDATE_REPO_NAME),
           '_group': 'test_group_for_update'},
          {
             'repo_name': 'test_group_for_update/{}'.format(UPDATE_REPO_NAME),
             'url': 'http://{}/test_group_for_update/{}'.format(
                 plain_http_host_only_stub(), UPDATE_REPO_NAME)
          }),
     ])
     def test_api_update_repo(self, updates, expected, backend):
         repo_name = UPDATE_REPO_NAME
         repo = fixture.create_repo(repo_name, repo_type=backend.alias)
         if updates.get('_group'):
             fixture.create_repo_group(updates['_group'])
 
+        if 'landing_rev' in updates:
+            default_landing_ref, _lbl = ScmModel.backend_landing_ref(backend.alias)
+            _type, _name = default_landing_ref.split(':')
+            updates['landing_rev'] = default_landing_ref
+            expected['landing_rev'] = [_type, _name]
+
         expected_api_data = repo.get_api_data(include_secrets=True)
         if expected is SAME_AS_UPDATES:
             expected_api_data.update(updates)
         else:
             expected_api_data.update(expected)
 
         id_, params = build_data(
             self.apikey, 'update_repo', repoid=repo_name, **updates)
 
         with mock.patch('rhodecode.model.validation_schema.validators.url_validator'):
             response = api_call(self.app, params)
 
         if updates.get('repo_name'):
             repo_name = updates['repo_name']
 
         try:
             expected = {
                 'msg': 'updated repo ID:%s %s' % (repo.repo_id, repo_name),
                 'repository': jsonify(expected_api_data)
             }
             assert_ok(id_, expected, given=response.body)
         finally:
             fixture.destroy_repo(repo_name)
             if updates.get('_group'):
                 fixture.destroy_repo_group(updates['_group'])
 
     def test_api_update_repo_fork_of_field(self, backend):
         master_repo = backend.create_repo()
         repo = backend.create_repo()
         updates = {
             'fork_of': master_repo.repo_name,
             'fork_of_id': master_repo.repo_id
         }
         expected_api_data = repo.get_api_data(include_secrets=True)
         expected_api_data.update(updates)
 
         id_, params = build_data(
             self.apikey, 'update_repo', repoid=repo.repo_name, **updates)
         response = api_call(self.app, params)
         expected = {
             'msg': 'updated repo ID:%s %s' % (repo.repo_id, repo.repo_name),
             'repository': jsonify(expected_api_data)
         }
         assert_ok(id_, expected, given=response.body)
         result = response.json['result']['repository']
         assert result['fork_of'] == master_repo.repo_name
         assert result['fork_of_id'] == master_repo.repo_id
 
     def test_api_update_repo_fork_of_not_found(self, backend):
         master_repo_name = 'fake-parent-repo'
         repo = backend.create_repo()
         updates = {
             'fork_of': master_repo_name
         }
         id_, params = build_data(
             self.apikey, 'update_repo', repoid=repo.repo_name, **updates)
         response = api_call(self.app, params)
         expected = {
             'repo_fork_of': 'Fork with id `{}` does not exists'.format(
                 master_repo_name)}
         assert_error(id_, expected, given=response.body)
 
     def test_api_update_repo_with_repo_group_not_existing(self):
         repo_name = 'admin_owned'
         fake_repo_group = 'test_group_for_update'
         fixture.create_repo(repo_name)
         updates = {'repo_name': '{}/{}'.format(fake_repo_group, repo_name)}
         id_, params = build_data(
             self.apikey, 'update_repo', repoid=repo_name, **updates)
         response = api_call(self.app, params)
         try:
             expected = {
                 'repo_group': 'Repository group `{}` does not exist'.format(fake_repo_group)
             }
             assert_error(id_, expected, given=response.body)
         finally:
             fixture.destroy_repo(repo_name)
 
     def test_api_update_repo_regular_user_not_allowed(self):
         repo_name = 'admin_owned'
         fixture.create_repo(repo_name)
         updates = {'active': False}
         id_, params = build_data(
             self.apikey_regular, 'update_repo', repoid=repo_name, **updates)
         response = api_call(self.app, params)
         try:
             expected = 'repository `%s` does not exist' % (repo_name,)
             assert_error(id_, expected, given=response.body)
         finally:
             fixture.destroy_repo(repo_name)
 
     @mock.patch.object(RepoModel, 'update', crash)
     def test_api_update_repo_exception_occurred(self, backend):
         repo_name = UPDATE_REPO_NAME
         fixture.create_repo(repo_name, repo_type=backend.alias)
         id_, params = build_data(
             self.apikey, 'update_repo', repoid=repo_name,
             owner=TEST_USER_ADMIN_LOGIN,)
         response = api_call(self.app, params)
         try:
             expected = 'failed to update repo `%s`' % (repo_name,)
             assert_error(id_, expected, given=response.body)
         finally:
             fixture.destroy_repo(repo_name)

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

@@ -1,1051 +1,1052 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2020 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.utils2 import CachedProperty
 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, UnresolvedFilesInRepo)
 
 
 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 = os.environ.get('GIT_DEFAULT_BRANCH_NAME') or 'master'
+    DEFAULT_REF = 'branch:{}'.format(DEFAULT_BRANCH_NAME)
 
     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 or {"cache": False}  # default should not use cache
 
         self._init_repo(create, src_url, do_workspace_checkout, bare)
 
         # caches
         self._commit_ids = {}
 
     @LazyProperty
     def _remote(self):
         repo_id = self.path
         return connection.Git(self.path, repo_id, self.config, with_wire=self.with_wire)
 
     @LazyProperty
     def bare(self):
         return self._remote.bare()
 
     @LazyProperty
     def head(self):
         return self._remote.head()
 
     @CachedProperty
     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):
         return self._remote.get_all_commit_ids()
 
     def _get_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 = []
                 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 _lookup_commit(self, commit_id_or_idx, translate_tag=True, maybe_unreachable=False, reference_obj=None):
 
         def is_null(value):
             return len(value) == commit_id_or_idx.count('0')
 
         if commit_id_or_idx in (None, '', 'tip', 'HEAD', 'head', -1):
             return self.commit_ids[-1]
 
         commit_missing_err = "Commit {} does not exist for `{}`".format(
             *map(safe_str, [commit_id_or_idx, self.name]))
 
         is_bstr = isinstance(commit_id_or_idx, (str, unicode))
         is_branch = reference_obj and reference_obj.branch
 
         lookup_ok = False
         if is_bstr:
             # Need to call remote to translate id for tagging scenarios,
             # or branch that are numeric
             try:
                 remote_data = self._remote.get_object(commit_id_or_idx,
                                                       maybe_unreachable=maybe_unreachable)
                 commit_id_or_idx = remote_data["commit_id"]
                 lookup_ok = True
             except (CommitDoesNotExistError,):
                 lookup_ok = False
 
         if lookup_ok is False:
             is_numeric_idx = \
                 (is_bstr and commit_id_or_idx.isdigit() and len(commit_id_or_idx) < 12) \
                 or isinstance(commit_id_or_idx, int)
             if not is_branch and (is_numeric_idx or is_null(commit_id_or_idx)):
                 try:
                     commit_id_or_idx = self.commit_ids[int(commit_id_or_idx)]
                     lookup_ok = True
                 except Exception:
                     raise CommitDoesNotExistError(commit_missing_err)
 
         # we failed regular lookup, and by integer number lookup
         if lookup_ok is False:
             raise CommitDoesNotExistError(commit_missing_err)
 
         # 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)
 
     @CachedProperty
     def branches(self):
         return self._get_branches()
 
     @CachedProperty
     def branches_closed(self):
         return {}
 
     @CachedProperty
     def bookmarks(self):
         return {}
 
     @CachedProperty
     def branches_all(self):
         all_branches = {}
         all_branches.update(self.branches)
         all_branches.update(self.branches_closed)
         return all_branches
 
     @CachedProperty
     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.raw_id)
 
         self._invalidate_prop_cache('tags')
         self._invalidate_prop_cache('_refs')
 
         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)
 
         self._remote.tag_remove(name)
         self._invalidate_prop_cache('tags')
         self._invalidate_prop_cache('_refs')
 
     def _get_refs(self):
         return self._remote.get_refs()
 
     @CachedProperty
     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, maybe_unreachable=False, reference_obj=None):
         """
         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
         else:
             commit_id = "tip"
 
         if translate_tag:
             commit_id = self._lookup_commit(
                 commit_id, maybe_unreachable=maybe_unreachable,
                 reference_obj=reference_obj)
 
         try:
             idx = self._commit_ids[commit_id]
         except KeyError:
             idx = -1
 
         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._lookup_commit(start_id)
         start_pos = self._commit_ids[start_raw_id] if start_id else None
         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_commit_ids(filters=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.")
 
         if path:
             file_filter = path
         else:
             file_filter = None
 
         diff = self._remote.diff(
             commit1.raw_id, commit2.raw_id, file_filter=file_filter,
             opt_ignorews=ignore_whitespace,
             context=context)
         return GitDiff(diff)
 
     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)
 
         # clear cached properties
         self._invalidate_prop_cache('commit_ids')
         self._invalidate_prop_cache('_refs')
         self._invalidate_prop_cache('branches')
 
         return len(self.commit_ids)
 
     def get_common_ancestor(self, commit_id1, commit_id2, repo2):
         log.debug('Calculating common ancestor between %sc1:%s and %sc2:%s',
                   self, commit_id1, repo2, commit_id2)
 
         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 = self.COMMIT_ID_PAT.findall(output)[0]
 
         log.debug('Found common ancestor with sha: %s', ancestor_id)
 
         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 self.COMMIT_ID_PAT.findall(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)
         self._invalidate_prop_cache('_refs')
 
     def remove_ref(self, ref_name):
         self._remote.remove_ref(ref_name)
         self._invalidate_prop_cache('_refs')
 
     def run_gc(self, prune=True):
         cmd = ['gc', '--aggressive']
         if prune:
             cmd += ['--prune=now']
         _stdout, stderr = self.run_git_command(cmd, fail_on_stderr=False)
         return stderr
 
     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 _create_branch(self, branch_name, commit_id):
         """
         creates a branch in a GIT repo
         """
         self._remote.create_branch(branch_name, commit_id)
 
     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, cache=False):
         return GitRepository(shadow_repository_path, with_wire={"cache": cache})
 
     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 something more robust in this case.
             raise RepositoryError('Do not know how to merge into empty repositories yet')
         unresolved = None
 
         # 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)]
 
         merge_cmd = cmd + heads
 
         try:
             self.run_git_command(merge_cmd, fail_on_stderr=False)
         except RepositoryError:
             files = self.run_git_command(['diff', '--name-only', '--diff-filter', 'U'],
                                          fail_on_stderr=False)[0].splitlines()
             # NOTE(marcink): we add U notation for consistent with HG backend output
             unresolved = ['U {}'.format(f) for f in files]
 
             # Cleanup any merge leftovers
             self._remote.invalidate_vcs_cache()
             self.run_git_command(['merge', '--abort'], fail_on_stderr=False)
 
             if unresolved:
                 raise UnresolvedFilesInRepo(unresolved)
             else:
                 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(
             self.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 %s shadow repository in %s',
                       self.alias, 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 invalidate the cache, or otherwise we
             # cannot retrieve the merge commit.
             shadow_repo = shadow_repo.get_shadow_instance(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 as e:
             log.exception('Failure when doing local merge on git shadow repo')
             if isinstance(e, UnresolvedFilesInRepo):
                 metadata['unresolved_files'] = '\n* conflict: ' + ('\n * conflict: '.join(e.args[0]))
 
             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/model/repo.py

@@ -1,1195 +1,1197 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2020 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 re
 import shutil
 import time
 import logging
 import traceback
 import datetime
 
 from pyramid.threadlocal import get_current_request
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from rhodecode import events
 from rhodecode.lib.auth import HasUserGroupPermissionAny
 from rhodecode.lib.caching_query import FromCache
 from rhodecode.lib.exceptions import AttachedForksError, AttachedPullRequestsError
 from rhodecode.lib import hooks_base
 from rhodecode.lib.user_log_filter import user_log_filter
 from rhodecode.lib.utils import make_db_config
 from rhodecode.lib.utils2 import (
     safe_str, safe_unicode, remove_prefix, obfuscate_url_pw,
     get_current_rhodecode_user, safe_int, action_logger_generic)
 from rhodecode.lib.vcs.backends import get_backend
 from rhodecode.model import BaseModel
 from rhodecode.model.db import (
     _hash_key, func, case, joinedload, or_, in_filter_generator,
     Session, Repository, UserRepoToPerm, UserGroupRepoToPerm,
     UserRepoGroupToPerm, UserGroupRepoGroupToPerm, User, Permission,
     Statistics, UserGroup, RepoGroup, RepositoryField, UserLog)
 from rhodecode.model.permission import PermissionModel
 from rhodecode.model.settings import VcsSettingsModel
 
 log = logging.getLogger(__name__)
 
 
 class RepoModel(BaseModel):
 
     cls = Repository
 
     def _get_user_group(self, users_group):
         return self._get_instance(UserGroup, users_group,
                                   callback=UserGroup.get_by_group_name)
 
     def _get_repo_group(self, repo_group):
         return self._get_instance(RepoGroup, repo_group,
                                   callback=RepoGroup.get_by_group_name)
 
     def _create_default_perms(self, repository, private):
         # create default permission
         default = 'repository.read'
         def_user = User.get_default_user()
         for p in def_user.user_perms:
             if p.permission.permission_name.startswith('repository.'):
                 default = p.permission.permission_name
                 break
 
         default_perm = 'repository.none' if private else default
 
         repo_to_perm = UserRepoToPerm()
         repo_to_perm.permission = Permission.get_by_key(default_perm)
 
         repo_to_perm.repository = repository
         repo_to_perm.user_id = def_user.user_id
 
         return repo_to_perm
 
     @LazyProperty
     def repos_path(self):
         """
         Gets the repositories root path from database
         """
         settings_model = VcsSettingsModel(sa=self.sa)
         return settings_model.get_repos_location()
 
     def get(self, repo_id):
         repo = self.sa.query(Repository) \
             .filter(Repository.repo_id == repo_id)
 
         return repo.scalar()
 
     def get_repo(self, repository):
         return self._get_repo(repository)
 
     def get_by_repo_name(self, repo_name, cache=False):
         repo = self.sa.query(Repository) \
             .filter(Repository.repo_name == repo_name)
 
         if cache:
             name_key = _hash_key(repo_name)
             repo = repo.options(
                 FromCache("sql_cache_short", "get_repo_%s" % name_key))
         return repo.scalar()
 
     def _extract_id_from_repo_name(self, repo_name):
         if repo_name.startswith('/'):
             repo_name = repo_name.lstrip('/')
         by_id_match = re.match(r'^_(\d{1,})', repo_name)
         if by_id_match:
             return by_id_match.groups()[0]
 
     def get_repo_by_id(self, repo_name):
         """
         Extracts repo_name by id from special urls.
         Example url is _11/repo_name
 
         :param repo_name:
         :return: repo object if matched else None
         """
         _repo_id = None
         try:
             _repo_id = self._extract_id_from_repo_name(repo_name)
             if _repo_id:
                 return self.get(_repo_id)
         except Exception:
             log.exception('Failed to extract repo_name from URL')
             if _repo_id:
                 Session().rollback()
 
         return None
 
     def get_repos_for_root(self, root, traverse=False):
         if traverse:
             like_expression = u'{}%'.format(safe_unicode(root))
             repos = Repository.query().filter(
                 Repository.repo_name.like(like_expression)).all()
         else:
             if root and not isinstance(root, RepoGroup):
                 raise ValueError(
                     'Root must be an instance '
                     'of RepoGroup, got:{} instead'.format(type(root)))
             repos = Repository.query().filter(Repository.group == root).all()
         return repos
 
     def get_url(self, repo, request=None, permalink=False):
         if not request:
             request = get_current_request()
 
         if not request:
             return
 
         if permalink:
             return request.route_url(
                 'repo_summary', repo_name='_{}'.format(safe_str(repo.repo_id)))
         else:
             return request.route_url(
                 'repo_summary', repo_name=safe_str(repo.repo_name))
 
     def get_commit_url(self, repo, commit_id, request=None, permalink=False):
         if not request:
             request = get_current_request()
 
         if not request:
             return
 
         if permalink:
             return request.route_url(
                 'repo_commit', repo_name=safe_str(repo.repo_id),
                 commit_id=commit_id)
 
         else:
             return request.route_url(
                 'repo_commit', repo_name=safe_str(repo.repo_name),
                 commit_id=commit_id)
 
     def get_repo_log(self, repo, filter_term):
         repo_log = UserLog.query()\
             .filter(or_(UserLog.repository_id == repo.repo_id,
                         UserLog.repository_name == repo.repo_name))\
             .options(joinedload(UserLog.user))\
             .options(joinedload(UserLog.repository))\
             .order_by(UserLog.action_date.desc())
 
         repo_log = user_log_filter(repo_log, filter_term)
         return repo_log
 
     @classmethod
     def update_commit_cache(cls, repositories=None):
         if not repositories:
             repositories = Repository.getAll()
         for repo in repositories:
             repo.update_commit_cache()
 
     def get_repos_as_dict(self, repo_list=None, admin=False,
                           super_user_actions=False, short_name=None):
 
         _render = get_current_request().get_partial_renderer(
             'rhodecode:templates/data_table/_dt_elements.mako')
         c = _render.get_call_context()
         h = _render.get_helpers()
 
         def quick_menu(repo_name):
             return _render('quick_menu', repo_name)
 
         def repo_lnk(name, rtype, rstate, private, archived, fork_of):
             if short_name is not None:
                 short_name_var = short_name
             else:
                 short_name_var = not admin
             return _render('repo_name', name, rtype, rstate, private, archived, fork_of,
                            short_name=short_name_var, admin=False)
 
         def last_change(last_change):
             if admin and isinstance(last_change, datetime.datetime) and not last_change.tzinfo:
                 ts = time.time()
                 utc_offset = (datetime.datetime.fromtimestamp(ts)
                               - datetime.datetime.utcfromtimestamp(ts)).total_seconds()
                 last_change = last_change + datetime.timedelta(seconds=utc_offset)
 
             return _render("last_change", last_change)
 
         def rss_lnk(repo_name):
             return _render("rss", repo_name)
 
         def atom_lnk(repo_name):
             return _render("atom", repo_name)
 
         def last_rev(repo_name, cs_cache):
             return _render('revision', repo_name, cs_cache.get('revision'),
                            cs_cache.get('raw_id'), cs_cache.get('author'),
                            cs_cache.get('message'), cs_cache.get('date'))
 
         def desc(desc):
             return _render('repo_desc', desc, c.visual.stylify_metatags)
 
         def state(repo_state):
             return _render("repo_state", repo_state)
 
         def repo_actions(repo_name):
             return _render('repo_actions', repo_name, super_user_actions)
 
         def user_profile(username):
             return _render('user_profile', username)
 
         repos_data = []
         for repo in repo_list:
             # NOTE(marcink): because we use only raw column we need to load it like that
             changeset_cache = Repository._load_changeset_cache(
                 repo.repo_id, repo._changeset_cache)
 
             row = {
                 "menu": quick_menu(repo.repo_name),
 
                 "name": repo_lnk(repo.repo_name, repo.repo_type, repo.repo_state,
                                  repo.private, repo.archived, repo.fork),
 
                 "desc": desc(h.escape(repo.description)),
 
                 "last_change": last_change(repo.updated_on),
 
                 "last_changeset": last_rev(repo.repo_name, changeset_cache),
                 "last_changeset_raw": changeset_cache.get('revision'),
 
                 "owner": user_profile(repo.User.username),
 
                 "state": state(repo.repo_state),
                 "rss": rss_lnk(repo.repo_name),
                 "atom": atom_lnk(repo.repo_name),
             }
             if admin:
                 row.update({
                     "action": repo_actions(repo.repo_name),
                 })
             repos_data.append(row)
 
         return repos_data
 
     def get_repos_data_table(
             self, draw, start, limit,
             search_q, order_by, order_dir,
             auth_user, repo_group_id):
         from rhodecode.model.scm import RepoList
 
         _perms = ['repository.read', 'repository.write', 'repository.admin']
 
         repos = Repository.query() \
             .filter(Repository.group_id == repo_group_id) \
             .all()
         auth_repo_list = RepoList(
             repos, perm_set=_perms,
             extra_kwargs=dict(user=auth_user))
 
         allowed_ids = [-1]
         for repo in auth_repo_list:
             allowed_ids.append(repo.repo_id)
 
         repos_data_total_count = Repository.query() \
             .filter(Repository.group_id == repo_group_id) \
             .filter(or_(
                 # generate multiple IN to fix limitation problems
                 *in_filter_generator(Repository.repo_id, allowed_ids))
             ) \
             .count()
 
         base_q = Session.query(
             Repository.repo_id,
             Repository.repo_name,
             Repository.description,
             Repository.repo_type,
             Repository.repo_state,
             Repository.private,
             Repository.archived,
             Repository.fork,
             Repository.updated_on,
             Repository._changeset_cache,
             User,
             ) \
             .filter(Repository.group_id == repo_group_id) \
             .filter(or_(
                 # generate multiple IN to fix limitation problems
                 *in_filter_generator(Repository.repo_id, allowed_ids))
             ) \
             .join(User, User.user_id == Repository.user_id) \
             .group_by(Repository, User)
 
         repos_data_total_filtered_count = base_q.count()
 
         sort_defined = False
         if order_by == 'repo_name':
             sort_col = func.lower(Repository.repo_name)
             sort_defined = True
         elif order_by == 'user_username':
             sort_col = User.username
         else:
             sort_col = getattr(Repository, order_by, None)
 
         if sort_defined or sort_col:
             if order_dir == 'asc':
                 sort_col = sort_col.asc()
             else:
                 sort_col = sort_col.desc()
 
         base_q = base_q.order_by(sort_col)
         base_q = base_q.offset(start).limit(limit)
 
         repos_list = base_q.all()
 
         repos_data = RepoModel().get_repos_as_dict(
             repo_list=repos_list, admin=False)
 
         data = ({
             'draw': draw,
             'data': repos_data,
             'recordsTotal': repos_data_total_count,
             'recordsFiltered': repos_data_total_filtered_count,
         })
         return data
 
     def _get_defaults(self, repo_name):
         """
         Gets information about repository, and returns a dict for
         usage in forms
 
         :param repo_name:
         """
 
         repo_info = Repository.get_by_repo_name(repo_name)
 
         if repo_info is None:
             return None
 
         defaults = repo_info.get_dict()
         defaults['repo_name'] = repo_info.just_name
 
         groups = repo_info.groups_with_parents
         parent_group = groups[-1] if groups else None
 
         # we use -1 as this is how in HTML, we mark an empty group
         defaults['repo_group'] = getattr(parent_group, 'group_id', -1)
 
         keys_to_process = (
             {'k': 'repo_type', 'strip': False},
             {'k': 'repo_enable_downloads', 'strip': True},
             {'k': 'repo_description', 'strip': True},
             {'k': 'repo_enable_locking', 'strip': True},
             {'k': 'repo_landing_rev', 'strip': True},
             {'k': 'clone_uri', 'strip': False},
             {'k': 'push_uri', 'strip': False},
             {'k': 'repo_private', 'strip': True},
             {'k': 'repo_enable_statistics', 'strip': True}
         )
 
         for item in keys_to_process:
             attr = item['k']
             if item['strip']:
                 attr = remove_prefix(item['k'], 'repo_')
 
             val = defaults[attr]
             if item['k'] == 'repo_landing_rev':
                 val = ':'.join(defaults[attr])
             defaults[item['k']] = val
             if item['k'] == 'clone_uri':
                 defaults['clone_uri_hidden'] = repo_info.clone_uri_hidden
             if item['k'] == 'push_uri':
                 defaults['push_uri_hidden'] = repo_info.push_uri_hidden
 
         # fill owner
         if repo_info.user:
             defaults.update({'user': repo_info.user.username})
         else:
             replacement_user = User.get_first_super_admin().username
             defaults.update({'user': replacement_user})
 
         return defaults
 
     def update(self, repo, **kwargs):
         try:
             cur_repo = self._get_repo(repo)
             source_repo_name = cur_repo.repo_name
 
             affected_user_ids = []
             if 'user' in kwargs:
                 old_owner_id = cur_repo.user.user_id
                 new_owner = User.get_by_username(kwargs['user'])
                 cur_repo.user = new_owner
 
                 if old_owner_id != new_owner.user_id:
                     affected_user_ids = [new_owner.user_id, old_owner_id]
 
             if 'repo_group' in kwargs:
                 cur_repo.group = RepoGroup.get(kwargs['repo_group'])
             log.debug('Updating repo %s with params:%s', cur_repo, kwargs)
 
             update_keys = [
                 (1, 'repo_description'),
                 (1, 'repo_landing_rev'),
                 (1, 'repo_private'),
                 (1, 'repo_enable_downloads'),
                 (1, 'repo_enable_locking'),
                 (1, 'repo_enable_statistics'),
                 (0, 'clone_uri'),
                 (0, 'push_uri'),
                 (0, 'fork_id')
             ]
             for strip, k in update_keys:
                 if k in kwargs:
                     val = kwargs[k]
                     if strip:
                         k = remove_prefix(k, 'repo_')
 
                     setattr(cur_repo, k, val)
 
             new_name = cur_repo.get_new_name(kwargs['repo_name'])
             cur_repo.repo_name = new_name
 
             # if private flag is set, reset default permission to NONE
             if kwargs.get('repo_private'):
                 EMPTY_PERM = 'repository.none'
                 RepoModel().grant_user_permission(
                     repo=cur_repo, user=User.DEFAULT_USER, perm=EMPTY_PERM
                 )
             if kwargs.get('repo_landing_rev'):
                 landing_rev_val = kwargs['repo_landing_rev']
                 RepoModel().set_landing_rev(cur_repo, landing_rev_val)
 
             # handle extra fields
             for field in filter(lambda k: k.startswith(RepositoryField.PREFIX), kwargs):
                 k = RepositoryField.un_prefix_key(field)
                 ex_field = RepositoryField.get_by_key_name(
                     key=k, repo=cur_repo)
                 if ex_field:
                     ex_field.field_value = kwargs[field]
                     self.sa.add(ex_field)
 
             self.sa.add(cur_repo)
 
             if source_repo_name != new_name:
                 # rename repository
                 self._rename_filesystem_repo(
                     old=source_repo_name, new=new_name)
 
             if affected_user_ids:
                 PermissionModel().trigger_permission_flush(affected_user_ids)
 
             return cur_repo
         except Exception:
             log.error(traceback.format_exc())
             raise
 
     def _create_repo(self, repo_name, repo_type, description, owner,
                      private=False, clone_uri=None, repo_group=None,
-                     landing_rev='rev:tip', fork_of=None,
+                     landing_rev=None, fork_of=None,
                      copy_fork_permissions=False, enable_statistics=False,
                      enable_locking=False, enable_downloads=False,
                      copy_group_permissions=False,
                      state=Repository.STATE_PENDING):
         """
         Create repository inside database with PENDING state, this should be
         only executed by create() repo. With exception of importing existing
         repos
         """
         from rhodecode.model.scm import ScmModel
 
         owner = self._get_user(owner)
         fork_of = self._get_repo(fork_of)
         repo_group = self._get_repo_group(safe_int(repo_group))
+        default_landing_ref, _lbl = ScmModel.backend_landing_ref(repo_type)
+        landing_rev = landing_rev or default_landing_ref
 
         try:
             repo_name = safe_unicode(repo_name)
             description = safe_unicode(description)
             # repo name is just a name of repository
             # while repo_name_full is a full qualified name that is combined
             # with name and path of group
             repo_name_full = repo_name
             repo_name = repo_name.split(Repository.NAME_SEP)[-1]
 
             new_repo = Repository()
             new_repo.repo_state = state
             new_repo.enable_statistics = False
             new_repo.repo_name = repo_name_full
             new_repo.repo_type = repo_type
             new_repo.user = owner
             new_repo.group = repo_group
             new_repo.description = description or repo_name
             new_repo.private = private
             new_repo.archived = False
             new_repo.clone_uri = clone_uri
             new_repo.landing_rev = landing_rev
 
             new_repo.enable_statistics = enable_statistics
             new_repo.enable_locking = enable_locking
             new_repo.enable_downloads = enable_downloads
 
             if repo_group:
                 new_repo.enable_locking = repo_group.enable_locking
 
             if fork_of:
                 parent_repo = fork_of
                 new_repo.fork = parent_repo
 
             events.trigger(events.RepoPreCreateEvent(new_repo))
 
             self.sa.add(new_repo)
 
             EMPTY_PERM = 'repository.none'
             if fork_of and copy_fork_permissions:
                 repo = fork_of
                 user_perms = UserRepoToPerm.query() \
                     .filter(UserRepoToPerm.repository == repo).all()
                 group_perms = UserGroupRepoToPerm.query() \
                     .filter(UserGroupRepoToPerm.repository == repo).all()
 
                 for perm in user_perms:
                     UserRepoToPerm.create(
                         perm.user, new_repo, perm.permission)
 
                 for perm in group_perms:
                     UserGroupRepoToPerm.create(
                         perm.users_group, new_repo, perm.permission)
                 # in case we copy permissions and also set this repo to private
                 # override the default user permission to make it a private repo
                 if private:
                     RepoModel(self.sa).grant_user_permission(
                         repo=new_repo, user=User.DEFAULT_USER, perm=EMPTY_PERM)
 
             elif repo_group and copy_group_permissions:
                 user_perms = UserRepoGroupToPerm.query() \
                     .filter(UserRepoGroupToPerm.group == repo_group).all()
 
                 group_perms = UserGroupRepoGroupToPerm.query() \
                     .filter(UserGroupRepoGroupToPerm.group == repo_group).all()
 
                 for perm in user_perms:
                     perm_name = perm.permission.permission_name.replace(
                         'group.', 'repository.')
                     perm_obj = Permission.get_by_key(perm_name)
                     UserRepoToPerm.create(perm.user, new_repo, perm_obj)
 
                 for perm in group_perms:
                     perm_name = perm.permission.permission_name.replace(
                         'group.', 'repository.')
                     perm_obj = Permission.get_by_key(perm_name)
                     UserGroupRepoToPerm.create(perm.users_group, new_repo, perm_obj)
 
                 if private:
                     RepoModel(self.sa).grant_user_permission(
                         repo=new_repo, user=User.DEFAULT_USER, perm=EMPTY_PERM)
 
             else:
                 perm_obj = self._create_default_perms(new_repo, private)
                 self.sa.add(perm_obj)
 
             # now automatically start following this repository as owner
             ScmModel(self.sa).toggle_following_repo(new_repo.repo_id, owner.user_id)
 
             # we need to flush here, in order to check if database won't
             # throw any exceptions, create filesystem dirs at the very end
             self.sa.flush()
             events.trigger(events.RepoCreateEvent(new_repo))
             return new_repo
 
         except Exception:
             log.error(traceback.format_exc())
             raise
 
     def create(self, form_data, cur_user):
         """
         Create repository using celery tasks
 
         :param form_data:
         :param cur_user:
         """
         from rhodecode.lib.celerylib import tasks, run_task
         return run_task(tasks.create_repo, form_data, cur_user)
 
     def update_permissions(self, repo, perm_additions=None, perm_updates=None,
                            perm_deletions=None, check_perms=True,
                            cur_user=None):
         if not perm_additions:
             perm_additions = []
         if not perm_updates:
             perm_updates = []
         if not perm_deletions:
             perm_deletions = []
 
         req_perms = ('usergroup.read', 'usergroup.write', 'usergroup.admin')
 
         changes = {
             'added': [],
             'updated': [],
             'deleted': [],
             'default_user_changed': None
         }
 
         repo = self._get_repo(repo)
 
         # update permissions
         for member_id, perm, member_type in perm_updates:
             member_id = int(member_id)
             if member_type == 'user':
                 member_name = User.get(member_id).username
                 if member_name == User.DEFAULT_USER:
                     # NOTE(dan): detect if we changed permissions for default user
                     perm_obj = self.sa.query(UserRepoToPerm) \
                         .filter(UserRepoToPerm.user_id == member_id) \
                         .filter(UserRepoToPerm.repository == repo) \
                         .scalar()
                     if perm_obj and perm_obj.permission.permission_name != perm:
                         changes['default_user_changed'] = True
 
                 # this updates also current one if found
                 self.grant_user_permission(
                     repo=repo, user=member_id, perm=perm)
             elif member_type == 'user_group':
                 # check if we have permissions to alter this usergroup
                 member_name = UserGroup.get(member_id).users_group_name
                 if not check_perms or HasUserGroupPermissionAny(
                         *req_perms)(member_name, user=cur_user):
                     self.grant_user_group_permission(
                         repo=repo, group_name=member_id, perm=perm)
             else:
                 raise ValueError("member_type must be 'user' or 'user_group' "
                                  "got {} instead".format(member_type))
             changes['updated'].append({'type': member_type, 'id': member_id,
                                        'name': member_name, 'new_perm': perm})
 
         # set new permissions
         for member_id, perm, member_type in perm_additions:
             member_id = int(member_id)
             if member_type == 'user':
                 member_name = User.get(member_id).username
                 self.grant_user_permission(
                     repo=repo, user=member_id, perm=perm)
             elif member_type == 'user_group':
                 # check if we have permissions to alter this usergroup
                 member_name = UserGroup.get(member_id).users_group_name
                 if not check_perms or HasUserGroupPermissionAny(
                         *req_perms)(member_name, user=cur_user):
                     self.grant_user_group_permission(
                         repo=repo, group_name=member_id, perm=perm)
             else:
                 raise ValueError("member_type must be 'user' or 'user_group' "
                                  "got {} instead".format(member_type))
 
             changes['added'].append({'type': member_type, 'id': member_id,
                                      'name': member_name, 'new_perm': perm})
         # delete permissions
         for member_id, perm, member_type in perm_deletions:
             member_id = int(member_id)
             if member_type == 'user':
                 member_name = User.get(member_id).username
                 self.revoke_user_permission(repo=repo, user=member_id)
             elif member_type == 'user_group':
                 # check if we have permissions to alter this usergroup
                 member_name = UserGroup.get(member_id).users_group_name
                 if not check_perms or HasUserGroupPermissionAny(
                         *req_perms)(member_name, user=cur_user):
                     self.revoke_user_group_permission(
                         repo=repo, group_name=member_id)
             else:
                 raise ValueError("member_type must be 'user' or 'user_group' "
                                  "got {} instead".format(member_type))
 
             changes['deleted'].append({'type': member_type, 'id': member_id,
                                        'name': member_name, 'new_perm': perm})
         return changes
 
     def create_fork(self, form_data, cur_user):
         """
         Simple wrapper into executing celery task for fork creation
 
         :param form_data:
         :param cur_user:
         """
         from rhodecode.lib.celerylib import tasks, run_task
         return run_task(tasks.create_repo_fork, form_data, cur_user)
 
     def archive(self, repo):
         """
         Archive given repository. Set archive flag.
 
         :param repo:
         """
         repo = self._get_repo(repo)
         if repo:
 
             try:
                 repo.archived = True
                 self.sa.add(repo)
                 self.sa.commit()
             except Exception:
                 log.error(traceback.format_exc())
                 raise
 
     def delete(self, repo, forks=None, pull_requests=None, fs_remove=True, cur_user=None):
         """
         Delete given repository, forks parameter defines what do do with
         attached forks. Throws AttachedForksError if deleted repo has attached
         forks
 
         :param repo:
         :param forks: str 'delete' or 'detach'
         :param pull_requests: str 'delete' or None
         :param fs_remove: remove(archive) repo from filesystem
         """
         if not cur_user:
             cur_user = getattr(get_current_rhodecode_user(), 'username', None)
         repo = self._get_repo(repo)
         if repo:
             if forks == 'detach':
                 for r in repo.forks:
                     r.fork = None
                     self.sa.add(r)
             elif forks == 'delete':
                 for r in repo.forks:
                     self.delete(r, forks='delete')
             elif [f for f in repo.forks]:
                 raise AttachedForksError()
 
             # check for pull requests
             pr_sources = repo.pull_requests_source
             pr_targets = repo.pull_requests_target
             if pull_requests != 'delete' and (pr_sources or pr_targets):
                 raise AttachedPullRequestsError()
 
             old_repo_dict = repo.get_dict()
             events.trigger(events.RepoPreDeleteEvent(repo))
             try:
                 self.sa.delete(repo)
                 if fs_remove:
                     self._delete_filesystem_repo(repo)
                 else:
                     log.debug('skipping removal from filesystem')
                 old_repo_dict.update({
                     'deleted_by': cur_user,
                     'deleted_on': time.time(),
                 })
                 hooks_base.delete_repository(**old_repo_dict)
                 events.trigger(events.RepoDeleteEvent(repo))
             except Exception:
                 log.error(traceback.format_exc())
                 raise
 
     def grant_user_permission(self, repo, user, perm):
         """
         Grant permission for user on given repository, or update existing one
         if found
 
         :param repo: Instance of Repository, repository_id, or repository name
         :param user: Instance of User, user_id or username
         :param perm: Instance of Permission, or permission_name
         """
         user = self._get_user(user)
         repo = self._get_repo(repo)
         permission = self._get_perm(perm)
 
         # check if we have that permission already
         obj = self.sa.query(UserRepoToPerm) \
             .filter(UserRepoToPerm.user == user) \
             .filter(UserRepoToPerm.repository == repo) \
             .scalar()
         if obj is None:
             # create new !
             obj = UserRepoToPerm()
         obj.repository = repo
         obj.user = user
         obj.permission = permission
         self.sa.add(obj)
         log.debug('Granted perm %s to %s on %s', perm, user, repo)
         action_logger_generic(
             'granted permission: {} to user: {} on repo: {}'.format(
                 perm, user, repo), namespace='security.repo')
         return obj
 
     def revoke_user_permission(self, repo, user):
         """
         Revoke permission for user on given repository
 
         :param repo: Instance of Repository, repository_id, or repository name
         :param user: Instance of User, user_id or username
         """
 
         user = self._get_user(user)
         repo = self._get_repo(repo)
 
         obj = self.sa.query(UserRepoToPerm) \
             .filter(UserRepoToPerm.repository == repo) \
             .filter(UserRepoToPerm.user == user) \
             .scalar()
         if obj:
             self.sa.delete(obj)
             log.debug('Revoked perm on %s on %s', repo, user)
             action_logger_generic(
                 'revoked permission from user: {} on repo: {}'.format(
                     user, repo), namespace='security.repo')
 
     def grant_user_group_permission(self, repo, group_name, perm):
         """
         Grant permission for user group on given repository, or update
         existing one if found
 
         :param repo: Instance of Repository, repository_id, or repository name
         :param group_name: Instance of UserGroup, users_group_id,
             or user group name
         :param perm: Instance of Permission, or permission_name
         """
         repo = self._get_repo(repo)
         group_name = self._get_user_group(group_name)
         permission = self._get_perm(perm)
 
         # check if we have that permission already
         obj = self.sa.query(UserGroupRepoToPerm) \
             .filter(UserGroupRepoToPerm.users_group == group_name) \
             .filter(UserGroupRepoToPerm.repository == repo) \
             .scalar()
 
         if obj is None:
             # create new
             obj = UserGroupRepoToPerm()
 
         obj.repository = repo
         obj.users_group = group_name
         obj.permission = permission
         self.sa.add(obj)
         log.debug('Granted perm %s to %s on %s', perm, group_name, repo)
         action_logger_generic(
             'granted permission: {} to usergroup: {} on repo: {}'.format(
                 perm, group_name, repo), namespace='security.repo')
 
         return obj
 
     def revoke_user_group_permission(self, repo, group_name):
         """
         Revoke permission for user group on given repository
 
         :param repo: Instance of Repository, repository_id, or repository name
         :param group_name: Instance of UserGroup, users_group_id,
             or user group name
         """
         repo = self._get_repo(repo)
         group_name = self._get_user_group(group_name)
 
         obj = self.sa.query(UserGroupRepoToPerm) \
             .filter(UserGroupRepoToPerm.repository == repo) \
             .filter(UserGroupRepoToPerm.users_group == group_name) \
             .scalar()
         if obj:
             self.sa.delete(obj)
             log.debug('Revoked perm to %s on %s', repo, group_name)
             action_logger_generic(
                 'revoked permission from usergroup: {} on repo: {}'.format(
                     group_name, repo), namespace='security.repo')
 
     def delete_stats(self, repo_name):
         """
         removes stats for given repo
 
         :param repo_name:
         """
         repo = self._get_repo(repo_name)
         try:
             obj = self.sa.query(Statistics) \
                 .filter(Statistics.repository == repo).scalar()
             if obj:
                 self.sa.delete(obj)
         except Exception:
             log.error(traceback.format_exc())
             raise
 
     def add_repo_field(self, repo_name, field_key, field_label, field_value='',
                        field_type='str', field_desc=''):
 
         repo = self._get_repo(repo_name)
 
         new_field = RepositoryField()
         new_field.repository = repo
         new_field.field_key = field_key
         new_field.field_type = field_type  # python type
         new_field.field_value = field_value
         new_field.field_desc = field_desc
         new_field.field_label = field_label
         self.sa.add(new_field)
         return new_field
 
     def delete_repo_field(self, repo_name, field_key):
         repo = self._get_repo(repo_name)
         field = RepositoryField.get_by_key_name(field_key, repo)
         if field:
             self.sa.delete(field)
 
     def set_landing_rev(self, repo, landing_rev_name):
         if landing_rev_name.startswith('branch:'):
             landing_rev_name = landing_rev_name.split('branch:')[-1]
         scm_instance = repo.scm_instance()
         if scm_instance:
             return scm_instance._remote.set_head_ref(landing_rev_name)
 
     def _create_filesystem_repo(self, repo_name, repo_type, repo_group,
                                 clone_uri=None, repo_store_location=None,
                                 use_global_config=False, install_hooks=True):
         """
         makes repository on filesystem. It's group aware means it'll create
         a repository within a group, and alter the paths accordingly of
         group location
 
         :param repo_name:
         :param alias:
         :param parent:
         :param clone_uri:
         :param repo_store_location:
         """
         from rhodecode.lib.utils import is_valid_repo, is_valid_repo_group
         from rhodecode.model.scm import ScmModel
 
         if Repository.NAME_SEP in repo_name:
             raise ValueError(
                 'repo_name must not contain groups got `%s`' % repo_name)
 
         if isinstance(repo_group, RepoGroup):
             new_parent_path = os.sep.join(repo_group.full_path_splitted)
         else:
             new_parent_path = repo_group or ''
 
         if repo_store_location:
             _paths = [repo_store_location]
         else:
             _paths = [self.repos_path, new_parent_path, repo_name]
             # we need to make it str for mercurial
         repo_path = os.path.join(*map(lambda x: safe_str(x), _paths))
 
         # check if this path is not a repository
         if is_valid_repo(repo_path, self.repos_path):
             raise Exception('This path %s is a valid repository' % repo_path)
 
         # check if this path is a group
         if is_valid_repo_group(repo_path, self.repos_path):
             raise Exception('This path %s is a valid group' % repo_path)
 
         log.info('creating repo %s in %s from url: `%s`',
                  repo_name, safe_unicode(repo_path),
                  obfuscate_url_pw(clone_uri))
 
         backend = get_backend(repo_type)
 
         config_repo = None if use_global_config else repo_name
         if config_repo and new_parent_path:
             config_repo = Repository.NAME_SEP.join(
                 (new_parent_path, config_repo))
         config = make_db_config(clear_session=False, repo=config_repo)
         config.set('extensions', 'largefiles', '')
 
         # patch and reset hooks section of UI config to not run any
         # hooks on creating remote repo
         config.clear_section('hooks')
 
         # TODO: johbo: Unify this, hardcoded "bare=True" does not look nice
         if repo_type == 'git':
             repo = backend(
                 repo_path, config=config, create=True, src_url=clone_uri, bare=True,
                 with_wire={"cache": False})
         else:
             repo = backend(
                 repo_path, config=config, create=True, src_url=clone_uri,
                 with_wire={"cache": False})
 
         if install_hooks:
             repo.install_hooks()
 
         log.debug('Created repo %s with %s backend',
                   safe_unicode(repo_name), safe_unicode(repo_type))
         return repo
 
     def _rename_filesystem_repo(self, old, new):
         """
         renames repository on filesystem
 
         :param old: old name
         :param new: new name
         """
         log.info('renaming repo from %s to %s', old, new)
 
         old_path = os.path.join(self.repos_path, old)
         new_path = os.path.join(self.repos_path, new)
         if os.path.isdir(new_path):
             raise Exception(
                 'Was trying to rename to already existing dir %s' % new_path
             )
         shutil.move(old_path, new_path)
 
     def _delete_filesystem_repo(self, repo):
         """
         removes repo from filesystem, the removal is acctually made by
         added rm__ prefix into dir, and rename internat .hg/.git dirs so this
         repository is no longer valid for rhodecode, can be undeleted later on
         by reverting the renames on this repository
 
         :param repo: repo object
         """
         rm_path = os.path.join(self.repos_path, repo.repo_name)
         repo_group = repo.group
         log.info("Removing repository %s", rm_path)
         # disable hg/git internal that it doesn't get detected as repo
         alias = repo.repo_type
 
         config = make_db_config(clear_session=False)
         config.set('extensions', 'largefiles', '')
         bare = getattr(repo.scm_instance(config=config), 'bare', False)
 
         # skip this for bare git repos
         if not bare:
             # disable VCS repo
             vcs_path = os.path.join(rm_path, '.%s' % alias)
             if os.path.exists(vcs_path):
                 shutil.move(vcs_path, os.path.join(rm_path, 'rm__.%s' % alias))
 
         _now = datetime.datetime.now()
         _ms = str(_now.microsecond).rjust(6, '0')
         _d = 'rm__%s__%s' % (_now.strftime('%Y%m%d_%H%M%S_' + _ms),
                              repo.just_name)
         if repo_group:
             # if repository is in group, prefix the removal path with the group
             args = repo_group.full_path_splitted + [_d]
             _d = os.path.join(*args)
 
         if os.path.isdir(rm_path):
             shutil.move(rm_path, os.path.join(self.repos_path, _d))
 
         # finally cleanup diff-cache if it exists
         cached_diffs_dir = repo.cached_diffs_dir
         if os.path.isdir(cached_diffs_dir):
             shutil.rmtree(cached_diffs_dir)
 
 
 class ReadmeFinder:
     """
     Utility which knows how to find a readme for a specific commit.
 
     The main idea is that this is a configurable algorithm. When creating an
     instance you can define parameters, currently only the `default_renderer`.
     Based on this configuration the method :meth:`search` behaves slightly
     different.
     """
 
     readme_re = re.compile(r'^readme(\.[^\.]+)?$', re.IGNORECASE)
     path_re = re.compile(r'^docs?', re.IGNORECASE)
 
     default_priorities = {
         None: 0,
         '.text': 2,
         '.txt': 3,
         '.rst': 1,
         '.rest': 2,
         '.md': 1,
         '.mkdn': 2,
         '.mdown': 3,
         '.markdown': 4,
     }
 
     path_priority = {
         'doc': 0,
         'docs': 1,
     }
 
     FALLBACK_PRIORITY = 99
 
     RENDERER_TO_EXTENSION = {
         'rst': ['.rst', '.rest'],
         'markdown': ['.md', 'mkdn', '.mdown', '.markdown'],
     }
 
     def __init__(self, default_renderer=None):
         self._default_renderer = default_renderer
         self._renderer_extensions = self.RENDERER_TO_EXTENSION.get(
             default_renderer, [])
 
     def search(self, commit, path=u'/'):
         """
         Find a readme in the given `commit`.
         """
         nodes = commit.get_nodes(path)
         matches = self._match_readmes(nodes)
         matches = self._sort_according_to_priority(matches)
         if matches:
             return matches[0].node
 
         paths = self._match_paths(nodes)
         paths = self._sort_paths_according_to_priority(paths)
         for path in paths:
             match = self.search(commit, path=path)
             if match:
                 return match
 
         return None
 
     def _match_readmes(self, nodes):
         for node in nodes:
             if not node.is_file():
                 continue
             path = node.path.rsplit('/', 1)[-1]
             match = self.readme_re.match(path)
             if match:
                 extension = match.group(1)
                 yield ReadmeMatch(node, match, self._priority(extension))
 
     def _match_paths(self, nodes):
         for node in nodes:
             if not node.is_dir():
                 continue
             match = self.path_re.match(node.path)
             if match:
                 yield node.path
 
     def _priority(self, extension):
         renderer_priority = (
             0 if extension in self._renderer_extensions else 1)
         extension_priority = self.default_priorities.get(
             extension, self.FALLBACK_PRIORITY)
         return (renderer_priority, extension_priority)
 
     def _sort_according_to_priority(self, matches):
 
         def priority_and_path(match):
             return (match.priority, match.path)
 
         return sorted(matches, key=priority_and_path)
 
     def _sort_paths_according_to_priority(self, paths):
 
         def priority_and_path(path):
             return (self.path_priority.get(path, self.FALLBACK_PRIORITY), path)
 
         return sorted(paths, key=priority_and_path)
 
 
 class ReadmeMatch:
 
     def __init__(self, node, match, priority):
         self.node = node
         self._match = match
         self.priority = priority
 
     @property
     def path(self):
         return self.node.path
 
     def __repr__(self):
         return '<ReadmeMatch {} priority={}'.format(self.path, self.priority)

rhodecode/model/scm.py

@@ -1,1025 +1,1028 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2020 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/
 
 """
 Scm model for RhodeCode
 """
 
 import os.path
 import traceback
 import logging
 import cStringIO
 
 from sqlalchemy import func
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 import rhodecode
 from rhodecode.lib.vcs import get_backend
 from rhodecode.lib.vcs.exceptions import RepositoryError, NodeNotChangedError
 from rhodecode.lib.vcs.nodes import FileNode
 from rhodecode.lib.vcs.backends.base import EmptyCommit
 from rhodecode.lib import helpers as h, rc_cache
 from rhodecode.lib.auth import (
     HasRepoPermissionAny, HasRepoGroupPermissionAny,
     HasUserGroupPermissionAny)
 from rhodecode.lib.exceptions import NonRelativePathError, IMCCommitError
 from rhodecode.lib import hooks_utils
 from rhodecode.lib.utils import (
     get_filesystem_repos, make_db_config)
 from rhodecode.lib.utils2 import (safe_str, safe_unicode)
 from rhodecode.lib.system_info import get_system_info
 from rhodecode.model import BaseModel
 from rhodecode.model.db import (
     or_, false,
     Repository, CacheKey, UserFollowing, UserLog, User, RepoGroup,
     PullRequest, FileStore)
 from rhodecode.model.settings import VcsSettingsModel
 from rhodecode.model.validation_schema.validators import url_validator, InvalidCloneUrl
 
 log = logging.getLogger(__name__)
 
 
 class UserTemp(object):
     def __init__(self, user_id):
         self.user_id = user_id
 
     def __repr__(self):
         return "<%s('id:%s')>" % (self.__class__.__name__, self.user_id)
 
 
 class RepoTemp(object):
     def __init__(self, repo_id):
         self.repo_id = repo_id
 
     def __repr__(self):
         return "<%s('id:%s')>" % (self.__class__.__name__, self.repo_id)
 
 
 class SimpleCachedRepoList(object):
     """
     Lighter version of of iteration of repos without the scm initialisation,
     and with cache usage
     """
     def __init__(self, db_repo_list, repos_path, order_by=None, perm_set=None):
         self.db_repo_list = db_repo_list
         self.repos_path = repos_path
         self.order_by = order_by
         self.reversed = (order_by or '').startswith('-')
         if not perm_set:
             perm_set = ['repository.read', 'repository.write',
                         'repository.admin']
         self.perm_set = perm_set
 
     def __len__(self):
         return len(self.db_repo_list)
 
     def __repr__(self):
         return '<%s (%s)>' % (self.__class__.__name__, self.__len__())
 
     def __iter__(self):
         for dbr in self.db_repo_list:
             # check permission at this level
             has_perm = HasRepoPermissionAny(*self.perm_set)(
                 dbr.repo_name, 'SimpleCachedRepoList check')
             if not has_perm:
                 continue
 
             tmp_d = {
                 'name': dbr.repo_name,
                 'dbrepo': dbr.get_dict(),
                 'dbrepo_fork': dbr.fork.get_dict() if dbr.fork else {}
             }
             yield tmp_d
 
 
 class _PermCheckIterator(object):
 
     def __init__(
             self, obj_list, obj_attr, perm_set, perm_checker,
             extra_kwargs=None):
         """
         Creates iterator from given list of objects, additionally
         checking permission for them from perm_set var
 
         :param obj_list: list of db objects
         :param obj_attr: attribute of object to pass into perm_checker
         :param perm_set: list of permissions to check
         :param perm_checker: callable to check permissions against
         """
         self.obj_list = obj_list
         self.obj_attr = obj_attr
         self.perm_set = perm_set
         self.perm_checker = perm_checker(*self.perm_set)
         self.extra_kwargs = extra_kwargs or {}
 
     def __len__(self):
         return len(self.obj_list)
 
     def __repr__(self):
         return '<%s (%s)>' % (self.__class__.__name__, self.__len__())
 
     def __iter__(self):
         for db_obj in self.obj_list:
             # check permission at this level
             # NOTE(marcink): the __dict__.get() is ~4x faster then getattr()
             name = db_obj.__dict__.get(self.obj_attr, None)
             if not self.perm_checker(name, self.__class__.__name__, **self.extra_kwargs):
                 continue
 
             yield db_obj
 
 
 class RepoList(_PermCheckIterator):
 
     def __init__(self, db_repo_list, perm_set=None, extra_kwargs=None):
         if not perm_set:
             perm_set = ['repository.read', 'repository.write', 'repository.admin']
 
         super(RepoList, self).__init__(
             obj_list=db_repo_list,
             obj_attr='_repo_name', perm_set=perm_set,
             perm_checker=HasRepoPermissionAny,
             extra_kwargs=extra_kwargs)
 
 
 class RepoGroupList(_PermCheckIterator):
 
     def __init__(self, db_repo_group_list, perm_set=None, extra_kwargs=None):
         if not perm_set:
             perm_set = ['group.read', 'group.write', 'group.admin']
 
         super(RepoGroupList, self).__init__(
             obj_list=db_repo_group_list,
             obj_attr='_group_name', perm_set=perm_set,
             perm_checker=HasRepoGroupPermissionAny,
             extra_kwargs=extra_kwargs)
 
 
 class UserGroupList(_PermCheckIterator):
 
     def __init__(self, db_user_group_list, perm_set=None, extra_kwargs=None):
         if not perm_set:
             perm_set = ['usergroup.read', 'usergroup.write', 'usergroup.admin']
 
         super(UserGroupList, self).__init__(
             obj_list=db_user_group_list,
             obj_attr='users_group_name', perm_set=perm_set,
             perm_checker=HasUserGroupPermissionAny,
             extra_kwargs=extra_kwargs)
 
 
 class ScmModel(BaseModel):
     """
     Generic Scm Model
     """
 
     @LazyProperty
     def repos_path(self):
         """
         Gets the repositories root path from database
         """
 
         settings_model = VcsSettingsModel(sa=self.sa)
         return settings_model.get_repos_location()
 
     def repo_scan(self, repos_path=None):
         """
         Listing of repositories in given path. This path should not be a
         repository itself. Return a dictionary of repository objects
 
         :param repos_path: path to directory containing repositories
         """
 
         if repos_path is None:
             repos_path = self.repos_path
 
         log.info('scanning for repositories in %s', repos_path)
 
         config = make_db_config()
         config.set('extensions', 'largefiles', '')
         repos = {}
 
         for name, path in get_filesystem_repos(repos_path, recursive=True):
             # name need to be decomposed and put back together using the /
             # since this is internal storage separator for rhodecode
             name = Repository.normalize_repo_name(name)
 
             try:
                 if name in repos:
                     raise RepositoryError('Duplicate repository name %s '
                                           'found in %s' % (name, path))
                 elif path[0] in rhodecode.BACKENDS:
                     backend = get_backend(path[0])
                     repos[name] = backend(path[1], config=config,
                                           with_wire={"cache": False})
             except OSError:
                 continue
             except RepositoryError:
                 log.exception('Failed to create a repo')
                 continue
 
         log.debug('found %s paths with repositories', len(repos))
         return repos
 
     def get_repos(self, all_repos=None, sort_key=None):
         """
         Get all repositories from db and for each repo create it's
         backend instance and fill that backed with information from database
 
         :param all_repos: list of repository names as strings
             give specific repositories list, good for filtering
 
         :param sort_key: initial sorting of repositories
         """
         if all_repos is None:
             all_repos = self.sa.query(Repository)\
                 .filter(Repository.group_id == None)\
                 .order_by(func.lower(Repository.repo_name)).all()
         repo_iter = SimpleCachedRepoList(
             all_repos, repos_path=self.repos_path, order_by=sort_key)
         return repo_iter
 
     def get_repo_groups(self, all_groups=None):
         if all_groups is None:
             all_groups = RepoGroup.query()\
                 .filter(RepoGroup.group_parent_id == None).all()
         return [x for x in RepoGroupList(all_groups)]
 
     def mark_for_invalidation(self, repo_name, delete=False):
         """
         Mark caches of this repo invalid in the database. `delete` flag
         removes the cache entries
 
         :param repo_name: the repo_name for which caches should be marked
             invalid, or deleted
         :param delete: delete the entry keys instead of setting bool
             flag on them, and also purge caches used by the dogpile
         """
         repo = Repository.get_by_repo_name(repo_name)
 
         if repo:
             invalidation_namespace = CacheKey.REPO_INVALIDATION_NAMESPACE.format(
                 repo_id=repo.repo_id)
             CacheKey.set_invalidate(invalidation_namespace, delete=delete)
 
             repo_id = repo.repo_id
             config = repo._config
             config.set('extensions', 'largefiles', '')
             repo.update_commit_cache(config=config, cs_cache=None)
             if delete:
                 cache_namespace_uid = 'cache_repo.{}'.format(repo_id)
                 rc_cache.clear_cache_namespace(
                     'cache_repo', cache_namespace_uid, invalidate=True)
 
     def toggle_following_repo(self, follow_repo_id, user_id):
 
         f = self.sa.query(UserFollowing)\
             .filter(UserFollowing.follows_repo_id == follow_repo_id)\
             .filter(UserFollowing.user_id == user_id).scalar()
 
         if f is not None:
             try:
                 self.sa.delete(f)
                 return
             except Exception:
                 log.error(traceback.format_exc())
                 raise
 
         try:
             f = UserFollowing()
             f.user_id = user_id
             f.follows_repo_id = follow_repo_id
             self.sa.add(f)
         except Exception:
             log.error(traceback.format_exc())
             raise
 
     def toggle_following_user(self, follow_user_id, user_id):
         f = self.sa.query(UserFollowing)\
             .filter(UserFollowing.follows_user_id == follow_user_id)\
             .filter(UserFollowing.user_id == user_id).scalar()
 
         if f is not None:
             try:
                 self.sa.delete(f)
                 return
             except Exception:
                 log.error(traceback.format_exc())
                 raise
 
         try:
             f = UserFollowing()
             f.user_id = user_id
             f.follows_user_id = follow_user_id
             self.sa.add(f)
         except Exception:
             log.error(traceback.format_exc())
             raise
 
     def is_following_repo(self, repo_name, user_id, cache=False):
         r = self.sa.query(Repository)\
             .filter(Repository.repo_name == repo_name).scalar()
 
         f = self.sa.query(UserFollowing)\
             .filter(UserFollowing.follows_repository == r)\
             .filter(UserFollowing.user_id == user_id).scalar()
 
         return f is not None
 
     def is_following_user(self, username, user_id, cache=False):
         u = User.get_by_username(username)
 
         f = self.sa.query(UserFollowing)\
             .filter(UserFollowing.follows_user == u)\
             .filter(UserFollowing.user_id == user_id).scalar()
 
         return f is not None
 
     def get_followers(self, repo):
         repo = self._get_repo(repo)
 
         return self.sa.query(UserFollowing)\
             .filter(UserFollowing.follows_repository == repo).count()
 
     def get_forks(self, repo):
         repo = self._get_repo(repo)
         return self.sa.query(Repository)\
             .filter(Repository.fork == repo).count()
 
     def get_pull_requests(self, repo):
         repo = self._get_repo(repo)
         return self.sa.query(PullRequest)\
             .filter(PullRequest.target_repo == repo)\
             .filter(PullRequest.status != PullRequest.STATUS_CLOSED).count()
 
     def get_artifacts(self, repo):
         repo = self._get_repo(repo)
         return self.sa.query(FileStore)\
             .filter(FileStore.repo == repo)\
             .filter(or_(FileStore.hidden == None, FileStore.hidden == false())).count()
 
     def mark_as_fork(self, repo, fork, user):
         repo = self._get_repo(repo)
         fork = self._get_repo(fork)
         if fork and repo.repo_id == fork.repo_id:
             raise Exception("Cannot set repository as fork of itself")
 
         if fork and repo.repo_type != fork.repo_type:
             raise RepositoryError(
                 "Cannot set repository as fork of repository with other type")
 
         repo.fork = fork
         self.sa.add(repo)
         return repo
 
     def pull_changes(self, repo, username, remote_uri=None, validate_uri=True):
         dbrepo = self._get_repo(repo)
         remote_uri = remote_uri or dbrepo.clone_uri
         if not remote_uri:
             raise Exception("This repository doesn't have a clone uri")
 
         repo = dbrepo.scm_instance(cache=False)
         repo.config.clear_section('hooks')
 
         try:
             # NOTE(marcink): add extra validation so we skip invalid urls
             # this is due this tasks can be executed via scheduler without
             # proper validation of remote_uri
             if validate_uri:
                 config = make_db_config(clear_session=False)
                 url_validator(remote_uri, dbrepo.repo_type, config)
         except InvalidCloneUrl:
             raise
 
         repo_name = dbrepo.repo_name
         try:
             # TODO: we need to make sure those operations call proper hooks !
             repo.fetch(remote_uri)
 
             self.mark_for_invalidation(repo_name)
         except Exception:
             log.error(traceback.format_exc())
             raise
 
     def push_changes(self, repo, username, remote_uri=None, validate_uri=True):
         dbrepo = self._get_repo(repo)
         remote_uri = remote_uri or dbrepo.push_uri
         if not remote_uri:
             raise Exception("This repository doesn't have a clone uri")
 
         repo = dbrepo.scm_instance(cache=False)
         repo.config.clear_section('hooks')
 
         try:
             # NOTE(marcink): add extra validation so we skip invalid urls
             # this is due this tasks can be executed via scheduler without
             # proper validation of remote_uri
             if validate_uri:
                 config = make_db_config(clear_session=False)
                 url_validator(remote_uri, dbrepo.repo_type, config)
         except InvalidCloneUrl:
             raise
 
         try:
             repo.push(remote_uri)
         except Exception:
             log.error(traceback.format_exc())
             raise
 
     def commit_change(self, repo, repo_name, commit, user, author, message,
                       content, f_path):
         """
         Commits changes
 
         :param repo: SCM instance
 
         """
         user = self._get_user(user)
 
         # decoding here will force that we have proper encoded values
         # in any other case this will throw exceptions and deny commit
         content = safe_str(content)
         path = safe_str(f_path)
         # message and author needs to be unicode
         # proper backend should then translate that into required type
         message = safe_unicode(message)
         author = safe_unicode(author)
         imc = repo.in_memory_commit
         imc.change(FileNode(path, content, mode=commit.get_file_mode(f_path)))
         try:
             # TODO: handle pre-push action !
             tip = imc.commit(
                 message=message, author=author, parents=[commit],
                 branch=commit.branch)
         except Exception as e:
             log.error(traceback.format_exc())
             raise IMCCommitError(str(e))
         finally:
             # always clear caches, if commit fails we want fresh object also
             self.mark_for_invalidation(repo_name)
 
         # We trigger the post-push action
         hooks_utils.trigger_post_push_hook(
             username=user.username, action='push_local', hook_type='post_push',
             repo_name=repo_name, repo_type=repo.alias, commit_ids=[tip.raw_id])
         return tip
 
     def _sanitize_path(self, f_path):
         if f_path.startswith('/') or f_path.startswith('./') or '../' in f_path:
             raise NonRelativePathError('%s is not an relative path' % f_path)
         if f_path:
             f_path = os.path.normpath(f_path)
         return f_path
 
     def get_dirnode_metadata(self, request, commit, dir_node):
         if not dir_node.is_dir():
             return []
 
         data = []
         for node in dir_node:
             if not node.is_file():
                 # we skip file-nodes
                 continue
 
             last_commit = node.last_commit
             last_commit_date = last_commit.date
             data.append({
                 'name': node.name,
                 'size': h.format_byte_size_binary(node.size),
                 'modified_at': h.format_date(last_commit_date),
                 'modified_ts': last_commit_date.isoformat(),
                 'revision': last_commit.revision,
                 'short_id': last_commit.short_id,
                 'message': h.escape(last_commit.message),
                 'author': h.escape(last_commit.author),
                 'user_profile': h.gravatar_with_user(
                     request, last_commit.author),
             })
 
         return data
 
     def get_nodes(self, repo_name, commit_id, root_path='/', flat=True,
                   extended_info=False, content=False, max_file_bytes=None):
         """
         recursive walk in root dir and return a set of all path in that dir
         based on repository walk function
 
         :param repo_name: name of repository
         :param commit_id: commit id for which to list nodes
         :param root_path: root path to list
         :param flat: return as a list, if False returns a dict with description
         :param extended_info: show additional info such as md5, binary, size etc
         :param content: add nodes content to the return data
         :param max_file_bytes: will not return file contents over this limit
 
         """
         _files = list()
         _dirs = list()
         try:
             _repo = self._get_repo(repo_name)
             commit = _repo.scm_instance().get_commit(commit_id=commit_id)
             root_path = root_path.lstrip('/')
             for __, dirs, files in commit.walk(root_path):
 
                 for f in files:
                     _content = None
                     _data = f_name = f.unicode_path
 
                     if not flat:
                         _data = {
                             "name": h.escape(f_name),
                             "type": "file",
                             }
                         if extended_info:
                             _data.update({
                                 "md5": f.md5,
                                 "binary": f.is_binary,
                                 "size": f.size,
                                 "extension": f.extension,
                                 "mimetype": f.mimetype,
                                 "lines": f.lines()[0]
                             })
 
                         if content:
                             over_size_limit = (max_file_bytes is not None
                                                and f.size > max_file_bytes)
                             full_content = None
                             if not f.is_binary and not over_size_limit:
                                 full_content = safe_str(f.content)
 
                             _data.update({
                                 "content": full_content,
                             })
                     _files.append(_data)
 
                 for d in dirs:
                     _data = d_name = d.unicode_path
                     if not flat:
                         _data = {
                             "name": h.escape(d_name),
                             "type": "dir",
                             }
                     if extended_info:
                         _data.update({
                             "md5": None,
                             "binary": None,
                             "size": None,
                             "extension": None,
                         })
                     if content:
                         _data.update({
                             "content": None
                         })
                     _dirs.append(_data)
         except RepositoryError:
             log.exception("Exception in get_nodes")
             raise
 
         return _dirs, _files
 
     def get_quick_filter_nodes(self, repo_name, commit_id, root_path='/'):
         """
         Generate files for quick filter in files view
         """
 
         _files = list()
         _dirs = list()
         try:
             _repo = self._get_repo(repo_name)
             commit = _repo.scm_instance().get_commit(commit_id=commit_id)
             root_path = root_path.lstrip('/')
             for __, dirs, files in commit.walk(root_path):
 
                 for f in files:
 
                     _data = {
                         "name": h.escape(f.unicode_path),
                         "type": "file",
                         }
 
                     _files.append(_data)
 
                 for d in dirs:
 
                     _data = {
                         "name": h.escape(d.unicode_path),
                         "type": "dir",
                         }
 
                     _dirs.append(_data)
         except RepositoryError:
             log.exception("Exception in get_quick_filter_nodes")
             raise
 
         return _dirs, _files
 
     def get_node(self, repo_name, commit_id, file_path,
                  extended_info=False, content=False, max_file_bytes=None, cache=True):
         """
         retrieve single node from commit
         """
         try:
 
             _repo = self._get_repo(repo_name)
             commit = _repo.scm_instance().get_commit(commit_id=commit_id)
 
             file_node = commit.get_node(file_path)
             if file_node.is_dir():
                 raise RepositoryError('The given path is a directory')
 
             _content = None
             f_name = file_node.unicode_path
 
             file_data = {
                 "name": h.escape(f_name),
                 "type": "file",
             }
 
             if extended_info:
                 file_data.update({
                     "extension": file_node.extension,
                     "mimetype": file_node.mimetype,
                 })
 
                 if cache:
                     md5 = file_node.md5
                     is_binary = file_node.is_binary
                     size = file_node.size
                 else:
                     is_binary, md5, size, _content = file_node.metadata_uncached()
 
                 file_data.update({
                     "md5": md5,
                     "binary": is_binary,
                     "size": size,
                 })
 
             if content and cache:
                 # get content + cache
                 size = file_node.size
                 over_size_limit = (max_file_bytes is not None and size > max_file_bytes)
                 full_content = None
                 all_lines = 0
                 if not file_node.is_binary and not over_size_limit:
                     full_content = safe_unicode(file_node.content)
                     all_lines, empty_lines = file_node.count_lines(full_content)
 
                 file_data.update({
                     "content": full_content,
                     "lines": all_lines
                 })
             elif content:
                 # get content *without* cache
                 if _content is None:
                     is_binary, md5, size, _content = file_node.metadata_uncached()
 
                 over_size_limit = (max_file_bytes is not None and size > max_file_bytes)
                 full_content = None
                 all_lines = 0
                 if not is_binary and not over_size_limit:
                     full_content = safe_unicode(_content)
                     all_lines, empty_lines = file_node.count_lines(full_content)
 
                 file_data.update({
                     "content": full_content,
                     "lines": all_lines
                 })
 
         except RepositoryError:
             log.exception("Exception in get_node")
             raise
 
         return file_data
 
     def get_fts_data(self, repo_name, commit_id, root_path='/'):
         """
         Fetch node tree for usage in full text search
         """
 
         tree_info = list()
 
         try:
             _repo = self._get_repo(repo_name)
             commit = _repo.scm_instance().get_commit(commit_id=commit_id)
             root_path = root_path.lstrip('/')
             for __, dirs, files in commit.walk(root_path):
 
                 for f in files:
                     is_binary, md5, size, _content = f.metadata_uncached()
                     _data = {
                         "name": f.unicode_path,
                         "md5": md5,
                         "extension": f.extension,
                         "binary": is_binary,
                         "size": size
                     }
 
                     tree_info.append(_data)
 
         except RepositoryError:
             log.exception("Exception in get_nodes")
             raise
 
         return tree_info
 
     def create_nodes(self, user, repo, message, nodes, parent_commit=None,
                      author=None, trigger_push_hook=True):
         """
         Commits given multiple nodes into repo
 
         :param user: RhodeCode User object or user_id, the commiter
         :param repo: RhodeCode Repository object
         :param message: commit message
         :param nodes: mapping {filename:{'content':content},...}
         :param parent_commit: parent commit, can be empty than it's
            initial commit
         :param author: author of commit, cna be different that commiter
            only for git
         :param trigger_push_hook: trigger push hooks
 
         :returns: new commited commit
         """
 
         user = self._get_user(user)
         scm_instance = repo.scm_instance(cache=False)
 
         processed_nodes = []
         for f_path in nodes:
             f_path = self._sanitize_path(f_path)
             content = nodes[f_path]['content']
             f_path = safe_str(f_path)
             # decoding here will force that we have proper encoded values
             # in any other case this will throw exceptions and deny commit
             if isinstance(content, (basestring,)):
                 content = safe_str(content)
             elif isinstance(content, (file, cStringIO.OutputType,)):
                 content = content.read()
             else:
                 raise Exception('Content is of unrecognized type %s' % (
                     type(content)
                 ))
             processed_nodes.append((f_path, content))
 
         message = safe_unicode(message)
         commiter = user.full_contact
         author = safe_unicode(author) if author else commiter
 
         imc = scm_instance.in_memory_commit
 
         if not parent_commit:
             parent_commit = EmptyCommit(alias=scm_instance.alias)
 
         if isinstance(parent_commit, EmptyCommit):
             # EmptyCommit means we we're editing empty repository
             parents = None
         else:
             parents = [parent_commit]
         # add multiple nodes
         for path, content in processed_nodes:
             imc.add(FileNode(path, content=content))
         # TODO: handle pre push scenario
         tip = imc.commit(message=message,
                          author=author,
                          parents=parents,
                          branch=parent_commit.branch)
 
         self.mark_for_invalidation(repo.repo_name)
         if trigger_push_hook:
             hooks_utils.trigger_post_push_hook(
                 username=user.username, action='push_local',
                 repo_name=repo.repo_name, repo_type=scm_instance.alias,
                 hook_type='post_push',
                 commit_ids=[tip.raw_id])
         return tip
 
     def update_nodes(self, user, repo, message, nodes, parent_commit=None,
                      author=None, trigger_push_hook=True):
         user = self._get_user(user)
         scm_instance = repo.scm_instance(cache=False)
 
         message = safe_unicode(message)
         commiter = user.full_contact
         author = safe_unicode(author) if author else commiter
 
         imc = scm_instance.in_memory_commit
 
         if not parent_commit:
             parent_commit = EmptyCommit(alias=scm_instance.alias)
 
         if isinstance(parent_commit, EmptyCommit):
             # EmptyCommit means we we're editing empty repository
             parents = None
         else:
             parents = [parent_commit]
 
         # add multiple nodes
         for _filename, data in nodes.items():
             # new filename, can be renamed from the old one, also sanitaze
             # the path for any hack around relative paths like ../../ etc.
             filename = self._sanitize_path(data['filename'])
             old_filename = self._sanitize_path(_filename)
             content = data['content']
             file_mode = data.get('mode')
             filenode = FileNode(old_filename, content=content, mode=file_mode)
             op = data['op']
             if op == 'add':
                 imc.add(filenode)
             elif op == 'del':
                 imc.remove(filenode)
             elif op == 'mod':
                 if filename != old_filename:
                     # TODO: handle renames more efficient, needs vcs lib changes
                     imc.remove(filenode)
                     imc.add(FileNode(filename, content=content, mode=file_mode))
                 else:
                     imc.change(filenode)
 
         try:
             # TODO: handle pre push scenario commit changes
             tip = imc.commit(message=message,
                              author=author,
                              parents=parents,
                              branch=parent_commit.branch)
         except NodeNotChangedError:
             raise
         except Exception as e:
             log.exception("Unexpected exception during call to imc.commit")
             raise IMCCommitError(str(e))
         finally:
             # always clear caches, if commit fails we want fresh object also
             self.mark_for_invalidation(repo.repo_name)
 
         if trigger_push_hook:
             hooks_utils.trigger_post_push_hook(
                 username=user.username, action='push_local', hook_type='post_push',
                 repo_name=repo.repo_name, repo_type=scm_instance.alias,
                 commit_ids=[tip.raw_id])
 
         return tip
 
     def delete_nodes(self, user, repo, message, nodes, parent_commit=None,
                      author=None, trigger_push_hook=True):
         """
         Deletes given multiple nodes into `repo`
 
         :param user: RhodeCode User object or user_id, the committer
         :param repo: RhodeCode Repository object
         :param message: commit message
         :param nodes: mapping {filename:{'content':content},...}
         :param parent_commit: parent commit, can be empty than it's initial
            commit
         :param author: author of commit, cna be different that commiter only
            for git
         :param trigger_push_hook: trigger push hooks
 
         :returns: new commit after deletion
         """
 
         user = self._get_user(user)
         scm_instance = repo.scm_instance(cache=False)
 
         processed_nodes = []
         for f_path in nodes:
             f_path = self._sanitize_path(f_path)
             # content can be empty but for compatabilty it allows same dicts
             # structure as add_nodes
             content = nodes[f_path].get('content')
             processed_nodes.append((f_path, content))
 
         message = safe_unicode(message)
         commiter = user.full_contact
         author = safe_unicode(author) if author else commiter
 
         imc = scm_instance.in_memory_commit
 
         if not parent_commit:
             parent_commit = EmptyCommit(alias=scm_instance.alias)
 
         if isinstance(parent_commit, EmptyCommit):
             # EmptyCommit means we we're editing empty repository
             parents = None
         else:
             parents = [parent_commit]
         # add multiple nodes
         for path, content in processed_nodes:
             imc.remove(FileNode(path, content=content))
 
         # TODO: handle pre push scenario
         tip = imc.commit(message=message,
                          author=author,
                          parents=parents,
                          branch=parent_commit.branch)
 
         self.mark_for_invalidation(repo.repo_name)
         if trigger_push_hook:
             hooks_utils.trigger_post_push_hook(
                 username=user.username, action='push_local', hook_type='post_push',
                 repo_name=repo.repo_name, repo_type=scm_instance.alias,
                 commit_ids=[tip.raw_id])
         return tip
 
     def strip(self, repo, commit_id, branch):
         scm_instance = repo.scm_instance(cache=False)
         scm_instance.config.clear_section('hooks')
         scm_instance.strip(commit_id, branch)
         self.mark_for_invalidation(repo.repo_name)
 
     def get_unread_journal(self):
         return self.sa.query(UserLog).count()
 
     @classmethod
     def backend_landing_ref(cls, repo_type):
         """
         Return a default landing ref based on a repository type.
         """
 
         landing_ref = {
             'hg': ('branch:default', 'default'),
             'git': ('branch:master', 'master'),
             'svn': ('rev:tip', 'latest tip'),
             'default': ('rev:tip', 'latest tip'),
         }
 
         return landing_ref.get(repo_type) or landing_ref['default']
 
     def get_repo_landing_revs(self, translator, repo=None):
         """
         Generates select option with tags branches and bookmarks (for hg only)
         grouped by type
 
         :param repo:
         """
+        from rhodecode.lib.vcs.backends.git import GitRepository
+
         _ = translator
         repo = self._get_repo(repo)
 
         if repo:
             repo_type = repo.repo_type
         else:
             repo_type = 'default'
 
         default_landing_ref, landing_ref_lbl = self.backend_landing_ref(repo_type)
 
         default_ref_options = [
             [default_landing_ref, landing_ref_lbl]
         ]
         default_choices = [
             default_landing_ref
         ]
 
         if not repo:
+            # presented at NEW repo creation
             return default_choices, default_ref_options
 
         repo = repo.scm_instance()
 
-        ref_options = [('rev:tip', 'latest tip')]
-        choices = ['rev:tip']
+        ref_options = [(default_landing_ref, landing_ref_lbl)]
+        choices = [default_landing_ref]
 
         # branches
         branch_group = [(u'branch:%s' % safe_unicode(b), safe_unicode(b)) for b in repo.branches]
         if not branch_group:
             # new repo, or without maybe a branch?
             branch_group = default_ref_options
 
         branches_group = (branch_group, _("Branches"))
         ref_options.append(branches_group)
         choices.extend([x[0] for x in branches_group[0]])
 
         # bookmarks for HG
         if repo.alias == 'hg':
             bookmarks_group = (
                 [(u'book:%s' % safe_unicode(b), safe_unicode(b))
                     for b in repo.bookmarks],
                 _("Bookmarks"))
             ref_options.append(bookmarks_group)
             choices.extend([x[0] for x in bookmarks_group[0]])
 
         # tags
         tags_group = (
             [(u'tag:%s' % safe_unicode(t), safe_unicode(t))
                 for t in repo.tags],
             _("Tags"))
         ref_options.append(tags_group)
         choices.extend([x[0] for x in tags_group[0]])
 
         return choices, ref_options
 
     def get_server_info(self, environ=None):
         server_info = get_system_info(environ)
         return server_info

rhodecode/model/validation_schema/schemas/repo_schema.py

@@ -1,454 +1,450 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2016-2020 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 colander
 import deform.widget
 
 from rhodecode.translation import _
 from rhodecode.model.validation_schema.utils import convert_to_optgroup, username_converter
 from rhodecode.model.validation_schema import validators, preparers, types
 
 DEFAULT_LANDING_REF = 'rev:tip'
-DEFAULT_BACKEND_LANDING_REF = {
-    'hg': 'branch:default',
-    'git': 'branch:master',
-    'svn': 'rev:tip',
-}
 
 
 def get_group_and_repo(repo_name):
     from rhodecode.model.repo_group import RepoGroupModel
     return RepoGroupModel()._get_group_name_and_parent(
         repo_name, get_object=True)
 
 
 def get_repo_group(repo_group_id):
     from rhodecode.model.repo_group import RepoGroup
     return RepoGroup.get(repo_group_id), RepoGroup.CHOICES_SEPARATOR
 
 
 @colander.deferred
 def deferred_repo_type_validator(node, kw):
     options = kw.get('repo_type_options', [])
     return colander.OneOf([x for x in options])
 
 
 @colander.deferred
 def deferred_repo_owner_validator(node, kw):
 
     def repo_owner_validator(node, value):
         from rhodecode.model.db import User
         value = username_converter(value)
         existing = User.get_by_username(value)
         if not existing:
             msg = _(u'Repo owner with id `{}` does not exists').format(value)
             raise colander.Invalid(node, msg)
 
     return repo_owner_validator
 
 
 @colander.deferred
 def deferred_landing_ref_validator(node, kw):
     options = kw.get(
         'repo_ref_options', [DEFAULT_LANDING_REF])
     return colander.OneOf([x for x in options])
 
 
 @colander.deferred
 def deferred_sync_uri_validator(node, kw):
     repo_type = kw.get('repo_type')
     validator = validators.CloneUriValidator(repo_type)
     return validator
 
 
 @colander.deferred
 def deferred_landing_ref_widget(node, kw):
+    from rhodecode.model.scm import ScmModel
+
     repo_type = kw.get('repo_type')
     default_opts = []
     if repo_type:
-        default_opts.append(
-            (DEFAULT_BACKEND_LANDING_REF[repo_type],
-             DEFAULT_BACKEND_LANDING_REF[repo_type]))
+        default_landing_ref, _lbl = ScmModel.backend_landing_ref(repo_type)
+        default_opts.append((default_landing_ref, default_landing_ref))
 
     items = kw.get('repo_ref_items', default_opts)
     items = convert_to_optgroup(items)
     return deform.widget.Select2Widget(values=items)
 
 
 @colander.deferred
 def deferred_fork_of_validator(node, kw):
     old_values = kw.get('old_values') or {}
 
     def fork_of_validator(node, value):
         from rhodecode.model.db import Repository, RepoGroup
         existing = Repository.get_by_repo_name(value)
         if not existing:
             msg = _(u'Fork with id `{}` does not exists').format(value)
             raise colander.Invalid(node, msg)
         elif old_values['repo_name'] == existing.repo_name:
             msg = _(u'Cannot set fork of '
                     u'parameter of this repository to itself').format(value)
             raise colander.Invalid(node, msg)
 
     return fork_of_validator
 
 
 @colander.deferred
 def deferred_can_write_to_group_validator(node, kw):
     request_user = kw.get('user')
     old_values = kw.get('old_values') or {}
 
     def can_write_to_group_validator(node, value):
         """
         Checks if given repo path is writable by user. This includes checks if
         user is allowed to create repositories under root path or under
         repo group paths
         """
 
         from rhodecode.lib.auth import (
             HasPermissionAny, HasRepoGroupPermissionAny)
         from rhodecode.model.repo_group import RepoGroupModel
 
         messages = {
             'invalid_repo_group':
                 _(u"Repository group `{}` does not exist"),
             # permissions denied we expose as not existing, to prevent
             # resource discovery
             'permission_denied':
                 _(u"Repository group `{}` does not exist"),
             'permission_denied_root':
                 _(u"You do not have the permission to store "
                   u"repositories in the root location.")
         }
 
         value = value['repo_group_name']
 
         is_root_location = value is types.RootLocation
         # NOT initialized validators, we must call them
         can_create_repos_at_root = HasPermissionAny('hg.admin', 'hg.create.repository')
 
         # if values is root location, we simply need to check if we can write
         # to root location !
         if is_root_location:
 
             if can_create_repos_at_root(user=request_user):
                 # we can create repo group inside tool-level. No more checks
                 # are required
                 return
             else:
                 old_name = old_values.get('repo_name')
                 if old_name and old_name == old_values.get('submitted_repo_name'):
                     # since we didn't change the name, we can skip validation and
                     # allow current users without store-in-root permissions to update
                     return
 
                 # "fake" node name as repo_name, otherwise we oddly report
                 # the error as if it was coming form repo_group
                 # however repo_group is empty when using root location.
                 node.name = 'repo_name'
                 raise colander.Invalid(node, messages['permission_denied_root'])
 
         # parent group not exists ? throw an error
         repo_group = RepoGroupModel().get_by_group_name(value)
         if value and not repo_group:
             raise colander.Invalid(
                 node, messages['invalid_repo_group'].format(value))
 
         gr_name = repo_group.group_name
 
         # create repositories with write permission on group is set to true
         create_on_write = HasPermissionAny(
             'hg.create.write_on_repogroup.true')(user=request_user)
 
         group_admin = HasRepoGroupPermissionAny('group.admin')(
             gr_name, 'can write into group validator', user=request_user)
         group_write = HasRepoGroupPermissionAny('group.write')(
             gr_name, 'can write into group validator', user=request_user)
 
         forbidden = not (group_admin or (group_write and create_on_write))
 
         # TODO: handling of old values, and detecting no-change in path
         # to skip permission checks in such cases. This only needs to be
         # implemented if we use this schema in forms as well
 
         # gid = (old_data['repo_group'].get('group_id')
         #        if (old_data and 'repo_group' in old_data) else None)
         # value_changed = gid != safe_int(value)
         # new = not old_data
 
         # do check if we changed the value, there's a case that someone got
         # revoked write permissions to a repository, he still created, we
         # don't need to check permission if he didn't change the value of
         # groups in form box
         # if value_changed or new:
         #     # parent group need to be existing
         # TODO: ENDS HERE
 
         if repo_group and forbidden:
             msg = messages['permission_denied'].format(value)
             raise colander.Invalid(node, msg)
 
     return can_write_to_group_validator
 
 
 @colander.deferred
 def deferred_unique_name_validator(node, kw):
     request_user = kw.get('user')
     old_values = kw.get('old_values') or {}
 
     def unique_name_validator(node, value):
         from rhodecode.model.db import Repository, RepoGroup
         name_changed = value != old_values.get('repo_name')
 
         existing = Repository.get_by_repo_name(value)
         if name_changed and existing:
             msg = _(u'Repository with name `{}` already exists').format(value)
             raise colander.Invalid(node, msg)
 
         existing_group = RepoGroup.get_by_group_name(value)
         if name_changed and existing_group:
             msg = _(u'Repository group with name `{}` already exists').format(
                 value)
             raise colander.Invalid(node, msg)
     return unique_name_validator
 
 
 @colander.deferred
 def deferred_repo_name_validator(node, kw):
     def no_git_suffix_validator(node, value):
         if value.endswith('.git'):
             msg = _('Repository name cannot end with .git')
             raise colander.Invalid(node, msg)
     return colander.All(
         no_git_suffix_validator, validators.valid_name_validator)
 
 
 @colander.deferred
 def deferred_repo_group_validator(node, kw):
     options = kw.get(
         'repo_repo_group_options')
     return colander.OneOf([x for x in options])
 
 
 @colander.deferred
 def deferred_repo_group_widget(node, kw):
     items = kw.get('repo_repo_group_items')
     return deform.widget.Select2Widget(values=items)
 
 
 class GroupType(colander.Mapping):
     def _validate(self, node, value):
         try:
             return dict(repo_group_name=value)
         except Exception as e:
             raise colander.Invalid(
                 node, '"${val}" is not a mapping type: ${err}'.format(
                     val=value, err=e))
 
     def deserialize(self, node, cstruct):
         if cstruct is colander.null:
             return cstruct
 
         appstruct = super(GroupType, self).deserialize(node, cstruct)
         validated_name = appstruct['repo_group_name']
 
         # inject group based on once deserialized data
         (repo_name_without_group,
          parent_group_name,
          parent_group) = get_group_and_repo(validated_name)
 
         appstruct['repo_name_with_group'] = validated_name
         appstruct['repo_name_without_group'] = repo_name_without_group
         appstruct['repo_group_name'] = parent_group_name or types.RootLocation
 
         if parent_group:
             appstruct['repo_group_id'] = parent_group.group_id
 
         return appstruct
 
 
 class GroupSchema(colander.SchemaNode):
     schema_type = GroupType
     validator = deferred_can_write_to_group_validator
     missing = colander.null
 
 
 class RepoGroup(GroupSchema):
     repo_group_name = colander.SchemaNode(
         types.GroupNameType())
     repo_group_id = colander.SchemaNode(
         colander.String(), missing=None)
     repo_name_without_group = colander.SchemaNode(
         colander.String(), missing=None)
 
 
 class RepoGroupAccessSchema(colander.MappingSchema):
     repo_group = RepoGroup()
 
 
 class RepoNameUniqueSchema(colander.MappingSchema):
     unique_repo_name = colander.SchemaNode(
         colander.String(),
         validator=deferred_unique_name_validator)
 
 
 class RepoSchema(colander.MappingSchema):
 
     repo_name = colander.SchemaNode(
         types.RepoNameType(),
         validator=deferred_repo_name_validator)
 
     repo_type = colander.SchemaNode(
         colander.String(),
         validator=deferred_repo_type_validator)
 
     repo_owner = colander.SchemaNode(
         colander.String(),
         validator=deferred_repo_owner_validator,
         widget=deform.widget.TextInputWidget())
 
     repo_description = colander.SchemaNode(
         colander.String(), missing='',
         widget=deform.widget.TextAreaWidget())
 
     repo_landing_commit_ref = colander.SchemaNode(
         colander.String(),
         validator=deferred_landing_ref_validator,
         preparers=[preparers.strip_preparer],
         missing=DEFAULT_LANDING_REF,
         widget=deferred_landing_ref_widget)
 
     repo_clone_uri = colander.SchemaNode(
         colander.String(),
         validator=deferred_sync_uri_validator,
         preparers=[preparers.strip_preparer],
         missing='')
 
     repo_push_uri = colander.SchemaNode(
         colander.String(),
         validator=deferred_sync_uri_validator,
         preparers=[preparers.strip_preparer],
         missing='')
 
     repo_fork_of = colander.SchemaNode(
         colander.String(),
         validator=deferred_fork_of_validator,
         missing=None)
 
     repo_private = colander.SchemaNode(
         types.StringBooleanType(),
         missing=False, widget=deform.widget.CheckboxWidget())
     repo_copy_permissions = colander.SchemaNode(
         types.StringBooleanType(),
         missing=False, widget=deform.widget.CheckboxWidget())
     repo_enable_statistics = colander.SchemaNode(
         types.StringBooleanType(),
         missing=False, widget=deform.widget.CheckboxWidget())
     repo_enable_downloads = colander.SchemaNode(
         types.StringBooleanType(),
         missing=False, widget=deform.widget.CheckboxWidget())
     repo_enable_locking = colander.SchemaNode(
         types.StringBooleanType(),
         missing=False, widget=deform.widget.CheckboxWidget())
 
     def deserialize(self, cstruct):
         """
         Custom deserialize that allows to chain validation, and verify
         permissions, and as last step uniqueness
         """
 
         # first pass, to validate given data
         appstruct = super(RepoSchema, self).deserialize(cstruct)
         validated_name = appstruct['repo_name']
 
         # second pass to validate permissions to repo_group
         if 'old_values' in self.bindings:
             # save current repo name for name change checks
             self.bindings['old_values']['submitted_repo_name'] = validated_name
         second = RepoGroupAccessSchema().bind(**self.bindings)
         appstruct_second = second.deserialize({'repo_group': validated_name})
         # save result
         appstruct['repo_group'] = appstruct_second['repo_group']
 
         # thirds to validate uniqueness
         third = RepoNameUniqueSchema().bind(**self.bindings)
         third.deserialize({'unique_repo_name': validated_name})
 
         return appstruct
 
 
 class RepoSettingsSchema(RepoSchema):
     repo_group = colander.SchemaNode(
         colander.Integer(),
         validator=deferred_repo_group_validator,
         widget=deferred_repo_group_widget,
         missing='')
 
     repo_clone_uri_change = colander.SchemaNode(
         colander.String(),
         missing='NEW')
 
     repo_clone_uri = colander.SchemaNode(
         colander.String(),
         preparers=[preparers.strip_preparer],
         validator=deferred_sync_uri_validator,
         missing='')
 
     repo_push_uri_change = colander.SchemaNode(
         colander.String(),
         missing='NEW')
 
     repo_push_uri = colander.SchemaNode(
         colander.String(),
         preparers=[preparers.strip_preparer],
         validator=deferred_sync_uri_validator,
         missing='')
 
     def deserialize(self, cstruct):
         """
         Custom deserialize that allows to chain validation, and verify
         permissions, and as last step uniqueness
         """
 
         # first pass, to validate given data
         appstruct = super(RepoSchema, self).deserialize(cstruct)
         validated_name = appstruct['repo_name']
         # because of repoSchema adds repo-group as an ID, we inject it as
         # full name here because validators require it, it's unwrapped later
         # so it's safe to use and final name is going to be without group anyway
 
         group, separator = get_repo_group(appstruct['repo_group'])
         if group:
             validated_name = separator.join([group.group_name, validated_name])
 
         # second pass to validate permissions to repo_group
         if 'old_values' in self.bindings:
             # save current repo name for name change checks
             self.bindings['old_values']['submitted_repo_name'] = validated_name
         second = RepoGroupAccessSchema().bind(**self.bindings)
         appstruct_second = second.deserialize({'repo_group': validated_name})
         # save result
         appstruct['repo_group'] = appstruct_second['repo_group']
 
         # thirds to validate uniqueness
         third = RepoNameUniqueSchema().bind(**self.bindings)
         third.deserialize({'unique_repo_name': validated_name})
 
         return appstruct

rhodecode/tests/fixture.py

@@ -1,419 +1,426 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2020 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/
 
 """
 Helpers for fixture generation
 """
 
 import os
 import time
 import tempfile
 import shutil
 
 import configobj
 
 from rhodecode.model.settings import SettingsModel
 from rhodecode.tests import *
 from rhodecode.model.db import Repository, User, RepoGroup, UserGroup, Gist, UserEmailMap
 from rhodecode.model.meta import Session
 from rhodecode.model.repo import RepoModel
 from rhodecode.model.user import UserModel
 from rhodecode.model.repo_group import RepoGroupModel
 from rhodecode.model.user_group import UserGroupModel
 from rhodecode.model.gist import GistModel
 from rhodecode.model.auth_token import AuthTokenModel
+from rhodecode.model.scm import ScmModel
 from rhodecode.authentication.plugins.auth_rhodecode import \
     RhodeCodeAuthPlugin
 
 dn = os.path.dirname
 FIXTURES = os.path.join(dn(dn(os.path.abspath(__file__))), 'tests', 'fixtures')
 
 
 def error_function(*args, **kwargs):
     raise Exception('Total Crash !')
 
 
 class TestINI(object):
     """
     Allows to create a new test.ini file as a copy of existing one with edited
     data. Example usage::
 
         with TestINI('test.ini', [{'section':{'key':val'}]) as new_test_ini_path:
             print('paster server %s' % new_test_ini)
     """
 
     def __init__(self, ini_file_path, ini_params, new_file_prefix='DEFAULT',
                  destroy=True, dir=None):
         self.ini_file_path = ini_file_path
         self.ini_params = ini_params
         self.new_path = None
         self.new_path_prefix = new_file_prefix
         self._destroy = destroy
         self._dir = dir
 
     def __enter__(self):
         return self.create()
 
     def __exit__(self, exc_type, exc_val, exc_tb):
         self.destroy()
 
     def create(self):
         config = configobj.ConfigObj(
             self.ini_file_path, file_error=True, write_empty_values=True)
 
         for data in self.ini_params:
             section, ini_params = data.items()[0]
             for key, val in ini_params.items():
                 config[section][key] = val
         with tempfile.NamedTemporaryFile(
                 prefix=self.new_path_prefix, suffix='.ini', dir=self._dir,
                 delete=False) as new_ini_file:
             config.write(new_ini_file)
             self.new_path = new_ini_file.name
 
         return self.new_path
 
     def destroy(self):
         if self._destroy:
             os.remove(self.new_path)
 
 
 class Fixture(object):
 
     def anon_access(self, status):
         """
         Context process for disabling anonymous access. use like:
         fixture = Fixture()
         with fixture.anon_access(False):
             #tests
 
         after this block anon access will be set to `not status`
         """
 
         class context(object):
             def __enter__(self):
                 anon = User.get_default_user()
                 anon.active = status
                 Session().add(anon)
                 Session().commit()
                 time.sleep(1.5)  # must sleep for cache (1s to expire)
 
             def __exit__(self, exc_type, exc_val, exc_tb):
                 anon = User.get_default_user()
                 anon.active = not status
                 Session().add(anon)
                 Session().commit()
 
         return context()
 
     def auth_restriction(self, registry, auth_restriction):
         """
         Context process for changing the builtin rhodecode plugin auth restrictions.
         Use like:
         fixture = Fixture()
         with fixture.auth_restriction('super_admin'):
             #tests
 
         after this block auth restriction will be taken off
         """
 
         class context(object):
             def _get_plugin(self):
                 plugin_id = 'egg:rhodecode-enterprise-ce#{}'.format(RhodeCodeAuthPlugin.uid)
                 plugin = RhodeCodeAuthPlugin(plugin_id)
                 return plugin
 
             def __enter__(self):
 
                 plugin = self._get_plugin()
                 plugin.create_or_update_setting('auth_restriction', auth_restriction)
                 Session().commit()
                 SettingsModel().invalidate_settings_cache()
 
             def __exit__(self, exc_type, exc_val, exc_tb):
 
                 plugin = self._get_plugin()
                 plugin.create_or_update_setting(
                     'auth_restriction', RhodeCodeAuthPlugin.AUTH_RESTRICTION_NONE)
                 Session().commit()
                 SettingsModel().invalidate_settings_cache()
 
         return context()
 
     def scope_restriction(self, registry, scope_restriction):
         """
         Context process for changing the builtin rhodecode plugin scope restrictions.
         Use like:
         fixture = Fixture()
         with fixture.scope_restriction('scope_http'):
             #tests
 
         after this block scope restriction will be taken off
         """
 
         class context(object):
             def _get_plugin(self):
                 plugin_id = 'egg:rhodecode-enterprise-ce#{}'.format(RhodeCodeAuthPlugin.uid)
                 plugin = RhodeCodeAuthPlugin(plugin_id)
                 return plugin
 
             def __enter__(self):
                 plugin = self._get_plugin()
                 plugin.create_or_update_setting('scope_restriction', scope_restriction)
                 Session().commit()
                 SettingsModel().invalidate_settings_cache()
 
             def __exit__(self, exc_type, exc_val, exc_tb):
                 plugin = self._get_plugin()
                 plugin.create_or_update_setting(
                     'scope_restriction', RhodeCodeAuthPlugin.AUTH_RESTRICTION_SCOPE_ALL)
                 Session().commit()
                 SettingsModel().invalidate_settings_cache()
 
         return context()
 
     def _get_repo_create_params(self, **custom):
+        repo_type = custom.get('repo_type') or 'hg'
+
+        default_landing_ref, landing_ref_lbl = ScmModel.backend_landing_ref(repo_type)
+
         defs = {
             'repo_name': None,
-            'repo_type': 'hg',
+            'repo_type': repo_type,
             'clone_uri': '',
             'push_uri': '',
             'repo_group': '-1',
             'repo_description': 'DESC',
             'repo_private': False,
-            'repo_landing_rev': 'rev:tip',
+            'repo_landing_commit_ref': default_landing_ref,
             'repo_copy_permissions': False,
             'repo_state': Repository.STATE_CREATED,
         }
         defs.update(custom)
         if 'repo_name_full' not in custom:
             defs.update({'repo_name_full': defs['repo_name']})
 
         # fix the repo name if passed as repo_name_full
         if defs['repo_name']:
             defs['repo_name'] = defs['repo_name'].split('/')[-1]
 
         return defs
 
     def _get_group_create_params(self, **custom):
         defs = {
             'group_name': None,
             'group_description': 'DESC',
             'perm_updates': [],
             'perm_additions': [],
             'perm_deletions': [],
             'group_parent_id': -1,
             'enable_locking': False,
             'recursive': False,
         }
         defs.update(custom)
 
         return defs
 
     def _get_user_create_params(self, name, **custom):
         defs = {
             'username': name,
             'password': 'qweqwe',
             'email': '[e-mail]' % name,
             'firstname': 'TestUser',
             'lastname': 'Test',
             'description': 'test description',
             'active': True,
             'admin': False,
             'extern_type': 'rhodecode',
             'extern_name': None,
         }
         defs.update(custom)
 
         return defs
 
     def _get_user_group_create_params(self, name, **custom):
         defs = {
             'users_group_name': name,
             'user_group_description': 'DESC',
             'users_group_active': True,
             'user_group_data': {},
         }
         defs.update(custom)
 
         return defs
 
     def create_repo(self, name, **kwargs):
         repo_group = kwargs.get('repo_group')
         if isinstance(repo_group, RepoGroup):
             kwargs['repo_group'] = repo_group.group_id
             name = name.split(Repository.NAME_SEP)[-1]
             name = Repository.NAME_SEP.join((repo_group.group_name, name))
 
         if 'skip_if_exists' in kwargs:
             del kwargs['skip_if_exists']
             r = Repository.get_by_repo_name(name)
             if r:
                 return r
 
         form_data = self._get_repo_create_params(repo_name=name, **kwargs)
         cur_user = kwargs.get('cur_user', TEST_USER_ADMIN_LOGIN)
         RepoModel().create(form_data, cur_user)
         Session().commit()
         repo = Repository.get_by_repo_name(name)
         assert repo
         return repo
 
     def create_fork(self, repo_to_fork, fork_name, **kwargs):
         repo_to_fork = Repository.get_by_repo_name(repo_to_fork)
 
-        form_data = self._get_repo_create_params(repo_name=fork_name,
+        form_data = self._get_repo_create_params(
+            repo_name=fork_name,
                                             fork_parent_id=repo_to_fork.repo_id,
                                             repo_type=repo_to_fork.repo_type,
                                             **kwargs)
+
         #TODO: fix it !!
         form_data['description'] = form_data['repo_description']
         form_data['private'] = form_data['repo_private']
-        form_data['landing_rev'] = form_data['repo_landing_rev']
+        form_data['landing_rev'] = form_data['repo_landing_commit_ref']
 
         owner = kwargs.get('cur_user', TEST_USER_ADMIN_LOGIN)
         RepoModel().create_fork(form_data, cur_user=owner)
         Session().commit()
         r = Repository.get_by_repo_name(fork_name)
         assert r
         return r
 
     def destroy_repo(self, repo_name, **kwargs):
         RepoModel().delete(repo_name, pull_requests='delete', **kwargs)
         Session().commit()
 
     def destroy_repo_on_filesystem(self, repo_name):
         rm_path = os.path.join(RepoModel().repos_path, repo_name)
         if os.path.isdir(rm_path):
             shutil.rmtree(rm_path)
 
     def create_repo_group(self, name, **kwargs):
         if 'skip_if_exists' in kwargs:
             del kwargs['skip_if_exists']
             gr = RepoGroup.get_by_group_name(group_name=name)
             if gr:
                 return gr
         form_data = self._get_group_create_params(group_name=name, **kwargs)
         owner = kwargs.get('cur_user', TEST_USER_ADMIN_LOGIN)
         gr = RepoGroupModel().create(
             group_name=form_data['group_name'],
             group_description=form_data['group_name'],
             owner=owner)
         Session().commit()
         gr = RepoGroup.get_by_group_name(gr.group_name)
         return gr
 
     def destroy_repo_group(self, repogroupid):
         RepoGroupModel().delete(repogroupid)
         Session().commit()
 
     def create_user(self, name, **kwargs):
         if 'skip_if_exists' in kwargs:
             del kwargs['skip_if_exists']
             user = User.get_by_username(name)
             if user:
                 return user
         form_data = self._get_user_create_params(name, **kwargs)
         user = UserModel().create(form_data)
 
         # create token for user
         AuthTokenModel().create(
                 user=user, description=u'TEST_USER_TOKEN')
 
         Session().commit()
         user = User.get_by_username(user.username)
         return user
 
     def destroy_user(self, userid):
         UserModel().delete(userid)
         Session().commit()
 
     def create_additional_user_email(self, user, email):
         uem = UserEmailMap()
         uem.user = user
         uem.email = email
         Session().add(uem)
         return uem
 
     def destroy_users(self, userid_iter):
         for user_id in userid_iter:
             if User.get_by_username(user_id):
                 UserModel().delete(user_id)
                 Session().commit()
 
     def create_user_group(self, name, **kwargs):
         if 'skip_if_exists' in kwargs:
             del kwargs['skip_if_exists']
             gr = UserGroup.get_by_group_name(group_name=name)
             if gr:
                 return gr
         # map active flag to the real attribute. For API consistency of fixtures
         if 'active' in kwargs:
             kwargs['users_group_active'] = kwargs['active']
             del kwargs['active']
         form_data = self._get_user_group_create_params(name, **kwargs)
         owner = kwargs.get('cur_user', TEST_USER_ADMIN_LOGIN)
         user_group = UserGroupModel().create(
             name=form_data['users_group_name'],
             description=form_data['user_group_description'],
             owner=owner, active=form_data['users_group_active'],
             group_data=form_data['user_group_data'])
         Session().commit()
         user_group = UserGroup.get_by_group_name(user_group.users_group_name)
         return user_group
 
     def destroy_user_group(self, usergroupid):
         UserGroupModel().delete(user_group=usergroupid, force=True)
         Session().commit()
 
     def create_gist(self, **kwargs):
         form_data = {
             'description': 'new-gist',
             'owner': TEST_USER_ADMIN_LOGIN,
             'gist_type': GistModel.cls.GIST_PUBLIC,
             'lifetime': -1,
             'acl_level': Gist.ACL_LEVEL_PUBLIC,
             'gist_mapping': {'filename1.txt': {'content': 'hello world'},}
         }
         form_data.update(kwargs)
         gist = GistModel().create(
             description=form_data['description'], owner=form_data['owner'],
             gist_mapping=form_data['gist_mapping'], gist_type=form_data['gist_type'],
             lifetime=form_data['lifetime'], gist_acl_level=form_data['acl_level']
         )
         Session().commit()
         return gist
 
     def destroy_gists(self, gistid=None):
         for g in GistModel.cls.get_all():
             if gistid:
                 if gistid == g.gist_access_id:
                     GistModel().delete(g)
             else:
                 GistModel().delete(g)
         Session().commit()
 
     def load_resource(self, resource_name, strip=False):
         with open(os.path.join(FIXTURES, resource_name)) as f:
             source = f.read()
             if strip:
                 source = source.strip()
 
         return source