shadow-repos: skip init of full repo to generate shadow repo path.

rhodecode/lib/middleware/simplevcs.py

@@ -1,678 +1,675 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 SimpleVCS middleware for handling protocol request (push/clone etc.)
 It's implemented with basic auth function
 """
 
 import os
 import re
 import logging
 import importlib
 from functools import wraps
 from StringIO import StringIO
 from lxml import etree
 
 import time
 from paste.httpheaders import REMOTE_USER, AUTH_TYPE
 
 from pyramid.httpexceptions import (
     HTTPNotFound, HTTPForbidden, HTTPNotAcceptable, HTTPInternalServerError)
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 import rhodecode
 from rhodecode.authentication.base import authenticate, VCS_TYPE, loadplugin
 from rhodecode.lib import rc_cache
 from rhodecode.lib.auth import AuthUser, HasPermissionAnyMiddleware
 from rhodecode.lib.base import (
     BasicAuth, get_ip_addr, get_user_agent, vcs_operation_context)
 from rhodecode.lib.exceptions import (UserCreationError, NotAllowedToCreateUserError)
 from rhodecode.lib.hooks_daemon import prepare_callback_daemon
 from rhodecode.lib.middleware import appenlight
 from rhodecode.lib.middleware.utils import scm_app_http
 from rhodecode.lib.utils import is_valid_repo, SLUG_RE
 from rhodecode.lib.utils2 import safe_str, fix_PATH, str2bool, safe_unicode
 from rhodecode.lib.vcs.conf import settings as vcs_settings
 from rhodecode.lib.vcs.backends import base
 
 from rhodecode.model import meta
 from rhodecode.model.db import User, Repository, PullRequest
 from rhodecode.model.scm import ScmModel
 from rhodecode.model.pull_request import PullRequestModel
 from rhodecode.model.settings import SettingsModel, VcsSettingsModel
 
 log = logging.getLogger(__name__)
 
 
 def extract_svn_txn_id(acl_repo_name, data):
     """
     Helper method for extraction of svn txn_id from submitted XML data during
     POST operations
     """
     try:
         root = etree.fromstring(data)
         pat = re.compile(r'/txn/(?P<txn_id>.*)')
         for el in root:
             if el.tag == '{DAV:}source':
                 for sub_el in el:
                     if sub_el.tag == '{DAV:}href':
                         match = pat.search(sub_el.text)
                         if match:
                             svn_tx_id = match.groupdict()['txn_id']
                             txn_id = rc_cache.utils.compute_key_from_params(
                                 acl_repo_name, svn_tx_id)
                             return txn_id
     except Exception:
         log.exception('Failed to extract txn_id')
 
 
 def initialize_generator(factory):
     """
     Initializes the returned generator by draining its first element.
 
     This can be used to give a generator an initializer, which is the code
     up to the first yield statement. This decorator enforces that the first
     produced element has the value ``"__init__"`` to make its special
     purpose very explicit in the using code.
     """
 
     @wraps(factory)
     def wrapper(*args, **kwargs):
         gen = factory(*args, **kwargs)
         try:
             init = gen.next()
         except StopIteration:
             raise ValueError('Generator must yield at least one element.')
         if init != "__init__":
             raise ValueError('First yielded element must be "__init__".')
         return gen
     return wrapper
 
 
 class SimpleVCS(object):
     """Common functionality for SCM HTTP handlers."""
 
     SCM = 'unknown'
 
     acl_repo_name = None
     url_repo_name = None
     vcs_repo_name = None
     rc_extras = {}
 
     # We have to handle requests to shadow repositories different than requests
     # to normal repositories. Therefore we have to distinguish them. To do this
     # we use this regex which will match only on URLs pointing to shadow
     # repositories.
     shadow_repo_re = re.compile(
         '(?P<groups>(?:{slug_pat}/)*)'  # repo groups
         '(?P<target>{slug_pat})/'       # target repo
         'pull-request/(?P<pr_id>\d+)/'  # pull request
         'repository$'                   # shadow repo
         .format(slug_pat=SLUG_RE.pattern))
 
     def __init__(self, config, registry):
         self.registry = registry
         self.config = config
         # re-populated by specialized middleware
         self.repo_vcs_config = base.Config()
         self.rhodecode_settings = SettingsModel().get_all_settings(cache=True)
 
         registry.rhodecode_settings = self.rhodecode_settings
         # authenticate this VCS request using authfunc
         auth_ret_code_detection = \
             str2bool(self.config.get('auth_ret_code_detection', False))
         self.authenticate = BasicAuth(
             '', authenticate, registry, config.get('auth_ret_code'),
             auth_ret_code_detection)
         self.ip_addr = '0.0.0.0'
 
     @LazyProperty
     def global_vcs_config(self):
         try:
             return VcsSettingsModel().get_ui_settings_as_config_obj()
         except Exception:
             return base.Config()
 
     @property
     def base_path(self):
         settings_path = self.repo_vcs_config.get(*VcsSettingsModel.PATH_SETTING)
 
         if not settings_path:
             settings_path = self.global_vcs_config.get(*VcsSettingsModel.PATH_SETTING)
 
         if not settings_path:
             # try, maybe we passed in explicitly as config option
             settings_path = self.config.get('base_path')
 
         if not settings_path:
             raise ValueError('FATAL: base_path is empty')
         return settings_path
 
     def set_repo_names(self, environ):
         """
         This will populate the attributes acl_repo_name, url_repo_name,
         vcs_repo_name and is_shadow_repo. In case of requests to normal (non
         shadow) repositories all names are equal. In case of requests to a
         shadow repository the acl-name points to the target repo of the pull
         request and the vcs-name points to the shadow repo file system path.
         The url-name is always the URL used by the vcs client program.
 
         Example in case of a shadow repo:
             acl_repo_name = RepoGroup/MyRepo
             url_repo_name = RepoGroup/MyRepo/pull-request/3/repository
             vcs_repo_name = /repo/base/path/RepoGroup/.__shadow_MyRepo_pr-3'
         """
         # First we set the repo name from URL for all attributes. This is the
         # default if handling normal (non shadow) repo requests.
         self.url_repo_name = self._get_repository_name(environ)
         self.acl_repo_name = self.vcs_repo_name = self.url_repo_name
         self.is_shadow_repo = False
 
         # Check if this is a request to a shadow repository.
         match = self.shadow_repo_re.match(self.url_repo_name)
         if match:
             match_dict = match.groupdict()
 
             # Build acl repo name from regex match.
             acl_repo_name = safe_unicode('{groups}{target}'.format(
                 groups=match_dict['groups'] or '',
                 target=match_dict['target']))
 
             # Retrieve pull request instance by ID from regex match.
             pull_request = PullRequest.get(match_dict['pr_id'])
 
             # Only proceed if we got a pull request and if acl repo name from
             # URL equals the target repo name of the pull request.
-            if pull_request and \
-                    (acl_repo_name == pull_request.target_repo.repo_name):
-                repo_id = pull_request.target_repo.repo_id
+            if pull_request and (acl_repo_name == pull_request.target_repo.repo_name):
+
                 # Get file system path to shadow repository.
                 workspace_id = PullRequestModel()._workspace_id(pull_request)
-                target_vcs = pull_request.target_repo.scm_instance()
-                vcs_repo_name = target_vcs._get_shadow_repository_path(
-                    repo_id, workspace_id)
+                vcs_repo_name = pull_request.target_repo.get_shadow_repository_path(workspace_id)
 
                 # Store names for later usage.
                 self.vcs_repo_name = vcs_repo_name
                 self.acl_repo_name = acl_repo_name
                 self.is_shadow_repo = True
 
         log.debug('Setting all VCS repository names: %s', {
             'acl_repo_name': self.acl_repo_name,
             'url_repo_name': self.url_repo_name,
             'vcs_repo_name': self.vcs_repo_name,
         })
 
     @property
     def scm_app(self):
         custom_implementation = self.config['vcs.scm_app_implementation']
         if custom_implementation == 'http':
             log.debug('Using HTTP implementation of scm app.')
             scm_app_impl = scm_app_http
         else:
             log.debug('Using custom implementation of scm_app: "{}"'.format(
                 custom_implementation))
             scm_app_impl = importlib.import_module(custom_implementation)
         return scm_app_impl
 
     def _get_by_id(self, repo_name):
         """
         Gets a special pattern _<ID> from clone url and tries to replace it
         with a repository_name for support of _<ID> non changeable urls
         """
 
         data = repo_name.split('/')
         if len(data) >= 2:
             from rhodecode.model.repo import RepoModel
             by_id_match = RepoModel().get_repo_by_id(repo_name)
             if by_id_match:
                 data[1] = by_id_match.repo_name
 
         return safe_str('/'.join(data))
 
     def _invalidate_cache(self, repo_name):
         """
         Set's cache for this repository for invalidation on next access
 
         :param repo_name: full repo name, also a cache key
         """
         ScmModel().mark_for_invalidation(repo_name)
 
     def is_valid_and_existing_repo(self, repo_name, base_path, scm_type):
         db_repo = Repository.get_by_repo_name(repo_name)
         if not db_repo:
             log.debug('Repository `%s` not found inside the database.',
                       repo_name)
             return False
 
         if db_repo.repo_type != scm_type:
             log.warning(
                 'Repository `%s` have incorrect scm_type, expected %s got %s',
                 repo_name, db_repo.repo_type, scm_type)
             return False
 
         config = db_repo._config
         config.set('extensions', 'largefiles', '')
         return is_valid_repo(
             repo_name, base_path,
             explicit_scm=scm_type, expect_scm=scm_type, config=config)
 
     def valid_and_active_user(self, user):
         """
         Checks if that user is not empty, and if it's actually object it checks
         if he's active.
 
         :param user: user object or None
         :return: boolean
         """
         if user is None:
             return False
 
         elif user.active:
             return True
 
         return False
 
     @property
     def is_shadow_repo_dir(self):
         return os.path.isdir(self.vcs_repo_name)
 
     def _check_permission(self, action, user, auth_user, repo_name, ip_addr=None,
                           plugin_id='', plugin_cache_active=False, cache_ttl=0):
         """
         Checks permissions using action (push/pull) user and repository
         name. If plugin_cache and ttl is set it will use the plugin which
         authenticated the user to store the cached permissions result for N
         amount of seconds as in cache_ttl
 
         :param action: push or pull action
         :param user: user instance
         :param repo_name: repository name
         """
 
         log.debug('AUTH_CACHE_TTL for permissions `%s` active: %s (TTL: %s)',
                   plugin_id, plugin_cache_active, cache_ttl)
 
         user_id = user.user_id
         cache_namespace_uid = 'cache_user_auth.{}'.format(user_id)
         region = rc_cache.get_or_create_region('cache_perms', cache_namespace_uid)
 
         @region.conditional_cache_on_arguments(namespace=cache_namespace_uid,
                                                expiration_time=cache_ttl,
                                                condition=plugin_cache_active)
         def compute_perm_vcs(
                 cache_name, plugin_id, action, user_id, repo_name, ip_addr):
 
             log.debug('auth: calculating permission access now...')
             # check IP
             inherit = user.inherit_default_permissions
             ip_allowed = AuthUser.check_ip_allowed(
                 user_id, ip_addr, inherit_from_default=inherit)
             if ip_allowed:
                 log.info('Access for IP:%s allowed', ip_addr)
             else:
                 return False
 
             if action == 'push':
                 perms = ('repository.write', 'repository.admin')
                 if not HasPermissionAnyMiddleware(*perms)(auth_user, repo_name):
                     return False
 
             else:
                 # any other action need at least read permission
                 perms = (
                     'repository.read', 'repository.write', 'repository.admin')
                 if not HasPermissionAnyMiddleware(*perms)(auth_user, repo_name):
                     return False
 
             return True
 
         start = time.time()
         log.debug('Running plugin `%s` permissions check', plugin_id)
 
         # for environ based auth, password can be empty, but then the validation is
         # on the server that fills in the env data needed for authentication
         perm_result = compute_perm_vcs(
             'vcs_permissions', plugin_id, action, user.user_id, repo_name, ip_addr)
 
         auth_time = time.time() - start
         log.debug('Permissions for plugin `%s` completed in %.4fs, '
                   'expiration time of fetched cache %.1fs.',
                   plugin_id, auth_time, cache_ttl)
 
         return perm_result
 
     def _get_http_scheme(self, environ):
         try:
             return environ['wsgi.url_scheme']
         except Exception:
             log.exception('Failed to read http scheme')
             return 'http'
 
     def _check_ssl(self, environ, start_response):
         """
         Checks the SSL check flag and returns False if SSL is not present
         and required True otherwise
         """
         org_proto = environ['wsgi._org_proto']
         # check if we have SSL required  ! if not it's a bad request !
         require_ssl = str2bool(self.repo_vcs_config.get('web', 'push_ssl'))
         if require_ssl and org_proto == 'http':
             log.debug(
                 'Bad request: detected protocol is `%s` and '
                 'SSL/HTTPS is required.', org_proto)
             return False
         return True
 
     def _get_default_cache_ttl(self):
         # take AUTH_CACHE_TTL from the `rhodecode` auth plugin
         plugin = loadplugin('egg:rhodecode-enterprise-ce#rhodecode')
         plugin_settings = plugin.get_settings()
         plugin_cache_active, cache_ttl = plugin.get_ttl_cache(
             plugin_settings) or (False, 0)
         return plugin_cache_active, cache_ttl
 
     def __call__(self, environ, start_response):
         try:
             return self._handle_request(environ, start_response)
         except Exception:
             log.exception("Exception while handling request")
             appenlight.track_exception(environ)
             return HTTPInternalServerError()(environ, start_response)
         finally:
             meta.Session.remove()
 
     def _handle_request(self, environ, start_response):
         if not self._check_ssl(environ, start_response):
             reason = ('SSL required, while RhodeCode was unable '
                       'to detect this as SSL request')
             log.debug('User not allowed to proceed, %s', reason)
             return HTTPNotAcceptable(reason)(environ, start_response)
 
         if not self.url_repo_name:
             log.warning('Repository name is empty: %s', self.url_repo_name)
             # failed to get repo name, we fail now
             return HTTPNotFound()(environ, start_response)
         log.debug('Extracted repo name is %s', self.url_repo_name)
 
         ip_addr = get_ip_addr(environ)
         user_agent = get_user_agent(environ)
         username = None
 
         # skip passing error to error controller
         environ['pylons.status_code_redirect'] = True
 
         # ======================================================================
         # GET ACTION PULL or PUSH
         # ======================================================================
         action = self._get_action(environ)
 
         # ======================================================================
         # Check if this is a request to a shadow repository of a pull request.
         # In this case only pull action is allowed.
         # ======================================================================
         if self.is_shadow_repo and action != 'pull':
             reason = 'Only pull action is allowed for shadow repositories.'
             log.debug('User not allowed to proceed, %s', reason)
             return HTTPNotAcceptable(reason)(environ, start_response)
 
         # Check if the shadow repo actually exists, in case someone refers
         # to it, and it has been deleted because of successful merge.
         if self.is_shadow_repo and not self.is_shadow_repo_dir:
             log.debug(
                 'Shadow repo detected, and shadow repo dir `%s` is missing',
                 self.is_shadow_repo_dir)
             return HTTPNotFound()(environ, start_response)
 
         # ======================================================================
         # CHECK ANONYMOUS PERMISSION
         # ======================================================================
         detect_force_push = False
         check_branch_perms = False
         if action in ['pull', 'push']:
             user_obj = anonymous_user = User.get_default_user()
             auth_user = user_obj.AuthUser()
             username = anonymous_user.username
             if anonymous_user.active:
                 plugin_cache_active, cache_ttl = self._get_default_cache_ttl()
                 # ONLY check permissions if the user is activated
                 anonymous_perm = self._check_permission(
                     action, anonymous_user, auth_user, self.acl_repo_name, ip_addr,
                     plugin_id='anonymous_access',
                     plugin_cache_active=plugin_cache_active,
                     cache_ttl=cache_ttl,
                 )
             else:
                 anonymous_perm = False
 
             if not anonymous_user.active or not anonymous_perm:
                 if not anonymous_user.active:
                     log.debug('Anonymous access is disabled, running '
                               'authentication')
 
                 if not anonymous_perm:
                     log.debug('Not enough credentials to access this '
                               'repository as anonymous user')
 
                 username = None
                 # ==============================================================
                 # DEFAULT PERM FAILED OR ANONYMOUS ACCESS IS DISABLED SO WE
                 # NEED TO AUTHENTICATE AND ASK FOR AUTH USER PERMISSIONS
                 # ==============================================================
 
                 # try to auth based on environ, container auth methods
                 log.debug('Running PRE-AUTH for container based authentication')
                 pre_auth = authenticate(
                     '', '', environ, VCS_TYPE, registry=self.registry,
                     acl_repo_name=self.acl_repo_name)
                 if pre_auth and pre_auth.get('username'):
                     username = pre_auth['username']
                 log.debug('PRE-AUTH got %s as username', username)
                 if pre_auth:
                     log.debug('PRE-AUTH successful from %s',
                               pre_auth.get('auth_data', {}).get('_plugin'))
 
                 # If not authenticated by the container, running basic auth
                 # before inject the calling repo_name for special scope checks
                 self.authenticate.acl_repo_name = self.acl_repo_name
 
                 plugin_cache_active, cache_ttl = False, 0
                 plugin = None
                 if not username:
                     self.authenticate.realm = self.authenticate.get_rc_realm()
 
                     try:
                         auth_result = self.authenticate(environ)
                     except (UserCreationError, NotAllowedToCreateUserError) as e:
                         log.error(e)
                         reason = safe_str(e)
                         return HTTPNotAcceptable(reason)(environ, start_response)
 
                     if isinstance(auth_result, dict):
                         AUTH_TYPE.update(environ, 'basic')
                         REMOTE_USER.update(environ, auth_result['username'])
                         username = auth_result['username']
                         plugin = auth_result.get('auth_data', {}).get('_plugin')
                         log.info(
                             'MAIN-AUTH successful for user `%s` from %s plugin',
                             username, plugin)
 
                         plugin_cache_active, cache_ttl = auth_result.get(
                             'auth_data', {}).get('_ttl_cache') or (False, 0)
                     else:
                         return auth_result.wsgi_application(environ, start_response)
 
                 # ==============================================================
                 # CHECK PERMISSIONS FOR THIS REQUEST USING GIVEN USERNAME
                 # ==============================================================
                 user = User.get_by_username(username)
                 if not self.valid_and_active_user(user):
                     return HTTPForbidden()(environ, start_response)
                 username = user.username
                 user_id = user.user_id
 
                 # check user attributes for password change flag
                 user_obj = user
                 auth_user = user_obj.AuthUser()
                 if user_obj and user_obj.username != User.DEFAULT_USER and \
                         user_obj.user_data.get('force_password_change'):
                     reason = 'password change required'
                     log.debug('User not allowed to authenticate, %s', reason)
                     return HTTPNotAcceptable(reason)(environ, start_response)
 
                 # check permissions for this repository
                 perm = self._check_permission(
                     action, user, auth_user, self.acl_repo_name, ip_addr,
                     plugin, plugin_cache_active, cache_ttl)
                 if not perm:
                     return HTTPForbidden()(environ, start_response)
                 environ['rc_auth_user_id'] = user_id
 
             if action == 'push':
                 perms = auth_user.get_branch_permissions(self.acl_repo_name)
                 if perms:
                     check_branch_perms = True
                     detect_force_push = True
 
         # extras are injected into UI object and later available
         # in hooks executed by RhodeCode
         check_locking = _should_check_locking(environ.get('QUERY_STRING'))
 
         extras = vcs_operation_context(
             environ, repo_name=self.acl_repo_name, username=username,
             action=action, scm=self.SCM, check_locking=check_locking,
             is_shadow_repo=self.is_shadow_repo, check_branch_perms=check_branch_perms,
             detect_force_push=detect_force_push
         )
 
         # ======================================================================
         # REQUEST HANDLING
         # ======================================================================
         repo_path = os.path.join(
             safe_str(self.base_path), safe_str(self.vcs_repo_name))
         log.debug('Repository path is %s', repo_path)
 
         fix_PATH()
 
         log.info(
             '%s action on %s repo "%s" by "%s" from %s %s',
             action, self.SCM, safe_str(self.url_repo_name),
             safe_str(username), ip_addr, user_agent)
 
         return self._generate_vcs_response(
             environ, start_response, repo_path, extras, action)
 
     @initialize_generator
     def _generate_vcs_response(
             self, environ, start_response, repo_path, extras, action):
         """
         Returns a generator for the response content.
 
         This method is implemented as a generator, so that it can trigger
         the cache validation after all content sent back to the client. It
         also handles the locking exceptions which will be triggered when
         the first chunk is produced by the underlying WSGI application.
         """
         txn_id = ''
         if 'CONTENT_LENGTH' in environ and environ['REQUEST_METHOD'] == 'MERGE':
             # case for SVN, we want to re-use the callback daemon port
             # so we use the txn_id, for this we peek the body, and still save
             # it as wsgi.input
             data = environ['wsgi.input'].read()
             environ['wsgi.input'] = StringIO(data)
             txn_id = extract_svn_txn_id(self.acl_repo_name, data)
 
         callback_daemon, extras = self._prepare_callback_daemon(
             extras, environ, action, txn_id=txn_id)
         log.debug('HOOKS extras is %s', extras)
 
         http_scheme = self._get_http_scheme(environ)
 
         config = self._create_config(extras, self.acl_repo_name, scheme=http_scheme)
         app = self._create_wsgi_app(repo_path, self.url_repo_name, config)
         with callback_daemon:
             app.rc_extras = extras
 
             try:
                 response = app(environ, start_response)
             finally:
                 # This statement works together with the decorator
                 # "initialize_generator" above. The decorator ensures that
                 # we hit the first yield statement before the generator is
                 # returned back to the WSGI server. This is needed to
                 # ensure that the call to "app" above triggers the
                 # needed callback to "start_response" before the
                 # generator is actually used.
                 yield "__init__"
 
             # iter content
             for chunk in response:
                 yield chunk
 
             try:
                 # invalidate cache on push
                 if action == 'push':
                     self._invalidate_cache(self.url_repo_name)
             finally:
                 meta.Session.remove()
 
     def _get_repository_name(self, environ):
         """Get repository name out of the environmnent
 
         :param environ: WSGI environment
         """
         raise NotImplementedError()
 
     def _get_action(self, environ):
         """Map request commands into a pull or push command.
 
         :param environ: WSGI environment
         """
         raise NotImplementedError()
 
     def _create_wsgi_app(self, repo_path, repo_name, config):
         """Return the WSGI app that will finally handle the request."""
         raise NotImplementedError()
 
     def _create_config(self, extras, repo_name, scheme='http'):
         """Create a safe config representation."""
         raise NotImplementedError()
 
     def _should_use_callback_daemon(self, extras, environ, action):
         return True
 
     def _prepare_callback_daemon(self, extras, environ, action, txn_id=None):
         direct_calls = vcs_settings.HOOKS_DIRECT_CALLS
         if not self._should_use_callback_daemon(extras, environ, action):
             # disable callback daemon for actions that don't require it
             direct_calls = True
 
         return prepare_callback_daemon(
             extras, protocol=vcs_settings.HOOKS_PROTOCOL,
             host=vcs_settings.HOOKS_HOST, use_direct_calls=direct_calls, txn_id=txn_id)
 
 
 def _should_check_locking(query_string):
     # this is kind of hacky, but due to how mercurial handles client-server
     # server see all operation on commit; bookmarks, phases and
     # obsolescence marker in different transaction, we don't want to check
     # locking on those
     return query_string not in ['cmd=listkeys']

rhodecode/lib/vcs/backends/base.py

@@ -1,1896 +1,1899 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 Base module for all VCS systems
 """
 import os
 import re
 import time
 import shutil
 import datetime
 import fnmatch
 import itertools
 import logging
 import collections
 import warnings
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from pyramid import compat
 
 import rhodecode
 from rhodecode.translation import lazy_ugettext
 from rhodecode.lib.utils2 import safe_str, safe_unicode, CachedProperty
 from rhodecode.lib.vcs import connection
 from rhodecode.lib.vcs.utils import author_name, author_email
 from rhodecode.lib.vcs.conf import settings
 from rhodecode.lib.vcs.exceptions import (
     CommitError, EmptyRepositoryError, NodeAlreadyAddedError,
     NodeAlreadyChangedError, NodeAlreadyExistsError, NodeAlreadyRemovedError,
     NodeDoesNotExistError, NodeNotChangedError, VCSError,
     ImproperArchiveTypeError, BranchDoesNotExistError, CommitDoesNotExistError,
     RepositoryError)
 
 
 log = logging.getLogger(__name__)
 
 
 FILEMODE_DEFAULT = 0o100644
 FILEMODE_EXECUTABLE = 0o100755
 EMPTY_COMMIT_ID = '0' * 40
 
 Reference = collections.namedtuple('Reference', ('type', 'name', 'commit_id'))
 
 
 class MergeFailureReason(object):
     """
     Enumeration with all the reasons why the server side merge could fail.
 
     DO NOT change the number of the reasons, as they may be stored in the
     database.
 
     Changing the name of a reason is acceptable and encouraged to deprecate old
     reasons.
     """
 
     # Everything went well.
     NONE = 0
 
     # An unexpected exception was raised. Check the logs for more details.
     UNKNOWN = 1
 
     # The merge was not successful, there are conflicts.
     MERGE_FAILED = 2
 
     # The merge succeeded but we could not push it to the target repository.
     PUSH_FAILED = 3
 
     # The specified target is not a head in the target repository.
     TARGET_IS_NOT_HEAD = 4
 
     # The source repository contains more branches than the target. Pushing
     # the merge will create additional branches in the target.
     HG_SOURCE_HAS_MORE_BRANCHES = 5
 
     # The target reference has multiple heads. That does not allow to correctly
     # identify the target location. This could only happen for mercurial
     # branches.
     HG_TARGET_HAS_MULTIPLE_HEADS = 6
 
     # The target repository is locked
     TARGET_IS_LOCKED = 7
 
     # Deprecated, use MISSING_TARGET_REF or MISSING_SOURCE_REF instead.
     # A involved commit could not be found.
     _DEPRECATED_MISSING_COMMIT = 8
 
     # The target repo reference is missing.
     MISSING_TARGET_REF = 9
 
     # The source repo reference is missing.
     MISSING_SOURCE_REF = 10
 
     # The merge was not successful, there are conflicts related to sub
     # repositories.
     SUBREPO_MERGE_FAILED = 11
 
 
 class UpdateFailureReason(object):
     """
     Enumeration with all the reasons why the pull request update could fail.
 
     DO NOT change the number of the reasons, as they may be stored in the
     database.
 
     Changing the name of a reason is acceptable and encouraged to deprecate old
     reasons.
     """
 
     # Everything went well.
     NONE = 0
 
     # An unexpected exception was raised. Check the logs for more details.
     UNKNOWN = 1
 
     # The pull request is up to date.
     NO_CHANGE = 2
 
     # The pull request has a reference type that is not supported for update.
     WRONG_REF_TYPE = 3
 
     # Update failed because the target reference is missing.
     MISSING_TARGET_REF = 4
 
     # Update failed because the source reference is missing.
     MISSING_SOURCE_REF = 5
 
 
 class MergeResponse(object):
 
     # uses .format(**metadata) for variables
     MERGE_STATUS_MESSAGES = {
         MergeFailureReason.NONE: lazy_ugettext(
             u'This pull request can be automatically merged.'),
         MergeFailureReason.UNKNOWN: lazy_ugettext(
             u'This pull request cannot be merged because of an unhandled exception. '
             u'{exception}'),
         MergeFailureReason.MERGE_FAILED: lazy_ugettext(
             u'This pull request cannot be merged because of merge conflicts.'),
         MergeFailureReason.PUSH_FAILED: lazy_ugettext(
             u'This pull request could not be merged because push to '
             u'target:`{target}@{merge_commit}` failed.'),
         MergeFailureReason.TARGET_IS_NOT_HEAD: lazy_ugettext(
             u'This pull request cannot be merged because the target '
             u'`{target_ref.name}` is not a head.'),
         MergeFailureReason.HG_SOURCE_HAS_MORE_BRANCHES: lazy_ugettext(
             u'This pull request cannot be merged because the source contains '
             u'more branches than the target.'),
         MergeFailureReason.HG_TARGET_HAS_MULTIPLE_HEADS: lazy_ugettext(
             u'This pull request cannot be merged because the target `{target_ref.name}` '
             u'has multiple heads: `{heads}`.'),
         MergeFailureReason.TARGET_IS_LOCKED: lazy_ugettext(
             u'This pull request cannot be merged because the target repository is '
             u'locked by {locked_by}.'),
 
         MergeFailureReason.MISSING_TARGET_REF: lazy_ugettext(
             u'This pull request cannot be merged because the target '
             u'reference `{target_ref.name}` is missing.'),
         MergeFailureReason.MISSING_SOURCE_REF: lazy_ugettext(
             u'This pull request cannot be merged because the source '
             u'reference `{source_ref.name}` is missing.'),
         MergeFailureReason.SUBREPO_MERGE_FAILED: lazy_ugettext(
             u'This pull request cannot be merged because of conflicts related '
             u'to sub repositories.'),
 
         # Deprecations
         MergeFailureReason._DEPRECATED_MISSING_COMMIT: lazy_ugettext(
             u'This pull request cannot be merged because the target or the '
             u'source reference is missing.'),
 
     }
 
     def __init__(self, possible, executed, merge_ref, failure_reason, metadata=None):
         self.possible = possible
         self.executed = executed
         self.merge_ref = merge_ref
         self.failure_reason = failure_reason
         self.metadata = metadata or {}
 
     def __repr__(self):
         return '<MergeResponse:{} {}>'.format(self.label, self.failure_reason)
 
     def __eq__(self, other):
         same_instance = isinstance(other, self.__class__)
         return same_instance \
                and self.possible == other.possible \
                and self.executed == other.executed \
                and self.failure_reason == other.failure_reason
 
     @property
     def label(self):
         label_dict = dict((v, k) for k, v in MergeFailureReason.__dict__.items() if
                           not k.startswith('_'))
         return label_dict.get(self.failure_reason)
 
     @property
     def merge_status_message(self):
         """
         Return a human friendly error message for the given merge status code.
         """
         msg = safe_unicode(self.MERGE_STATUS_MESSAGES[self.failure_reason])
         try:
             return msg.format(**self.metadata)
         except Exception:
             log.exception('Failed to format %s message', self)
             return msg
 
     def asdict(self):
         data = {}
         for k in ['possible', 'executed', 'merge_ref', 'failure_reason',
                   'merge_status_message']:
             data[k] = getattr(self, k)
         return data
 
 
 class BaseRepository(object):
     """
     Base Repository for final backends
 
     .. attribute:: DEFAULT_BRANCH_NAME
 
        name of default branch (i.e. "trunk" for svn, "master" for git etc.
 
     .. attribute:: commit_ids
 
        list of all available commit ids, in ascending order
 
     .. attribute:: path
 
        absolute path to the repository
 
     .. attribute:: bookmarks
 
        Mapping from name to :term:`Commit ID` of the bookmark. Empty in case
        there are no bookmarks or the backend implementation does not support
        bookmarks.
 
     .. attribute:: tags
 
        Mapping from name to :term:`Commit ID` of the tag.
 
     """
 
     DEFAULT_BRANCH_NAME = None
     DEFAULT_CONTACT = u"Unknown"
     DEFAULT_DESCRIPTION = u"unknown"
     EMPTY_COMMIT_ID = '0' * 40
 
     path = None
 
     _is_empty = None
     _commit_ids = {}
 
     def __init__(self, repo_path, config=None, create=False, **kwargs):
         """
         Initializes repository. Raises RepositoryError if repository could
         not be find at the given ``repo_path`` or directory at ``repo_path``
         exists and ``create`` is set to True.
 
         :param repo_path: local path of the repository
         :param config: repository configuration
         :param create=False: if set to True, would try to create repository.
         :param src_url=None: if set, should be proper url from which repository
           would be cloned; requires ``create`` parameter to be set to True -
           raises RepositoryError if src_url is set and create evaluates to
           False
         """
         raise NotImplementedError
 
     def __repr__(self):
         return '<%s at %s>' % (self.__class__.__name__, self.path)
 
     def __len__(self):
         return self.count()
 
     def __eq__(self, other):
         same_instance = isinstance(other, self.__class__)
         return same_instance and other.path == self.path
 
     def __ne__(self, other):
         return not self.__eq__(other)
 
     def get_create_shadow_cache_pr_path(self, db_repo):
         path = db_repo.cached_diffs_dir
         if not os.path.exists(path):
             os.makedirs(path, 0o755)
         return path
 
     @classmethod
     def get_default_config(cls, default=None):
         config = Config()
         if default and isinstance(default, list):
             for section, key, val in default:
                 config.set(section, key, val)
         return config
 
     @LazyProperty
     def _remote(self):
         raise NotImplementedError
 
     def _heads(self, branch=None):
         return []
 
     @LazyProperty
     def EMPTY_COMMIT(self):
         return EmptyCommit(self.EMPTY_COMMIT_ID)
 
     @LazyProperty
     def alias(self):
         for k, v in settings.BACKENDS.items():
             if v.split('.')[-1] == str(self.__class__.__name__):
                 return k
 
     @LazyProperty
     def name(self):
         return safe_unicode(os.path.basename(self.path))
 
     @LazyProperty
     def description(self):
         raise NotImplementedError
 
     def refs(self):
         """
         returns a `dict` with branches, bookmarks, tags, and closed_branches
         for this repository
         """
         return dict(
             branches=self.branches,
             branches_closed=self.branches_closed,
             tags=self.tags,
             bookmarks=self.bookmarks
         )
 
     @LazyProperty
     def branches(self):
         """
         A `dict` which maps branch names to commit ids.
         """
         raise NotImplementedError
 
     @LazyProperty
     def branches_closed(self):
         """
         A `dict` which maps tags names to commit ids.
         """
         raise NotImplementedError
 
     @LazyProperty
     def bookmarks(self):
         """
         A `dict` which maps tags names to commit ids.
         """
         raise NotImplementedError
 
     @LazyProperty
     def tags(self):
         """
         A `dict` which maps tags names to commit ids.
         """
         raise NotImplementedError
 
     @LazyProperty
     def size(self):
         """
         Returns combined size in bytes for all repository files
         """
         tip = self.get_commit()
         return tip.size
 
     def size_at_commit(self, commit_id):
         commit = self.get_commit(commit_id)
         return commit.size
 
     def _check_for_empty(self):
         no_commits = len(self._commit_ids) == 0
         if no_commits:
             # check on remote to be sure
             return self._remote.is_empty()
         else:
             return False
 
     def is_empty(self):
         if rhodecode.is_test:
             return self._check_for_empty()
 
         if self._is_empty is None:
             # cache empty for production, but not tests
             self._is_empty = self._check_for_empty()
 
         return self._is_empty
 
     @staticmethod
     def check_url(url, config):
         """
         Function will check given url and try to verify if it's a valid
         link.
         """
         raise NotImplementedError
 
     @staticmethod
     def is_valid_repository(path):
         """
         Check if given `path` contains a valid repository of this backend
         """
         raise NotImplementedError
 
     # ==========================================================================
     # COMMITS
     # ==========================================================================
 
     @CachedProperty
     def commit_ids(self):
         raise NotImplementedError
 
     def append_commit_id(self, commit_id):
         if commit_id not in self.commit_ids:
             self._rebuild_cache(self.commit_ids + [commit_id])
 
         # clear cache
         self._invalidate_prop_cache('commit_ids')
         self._is_empty = False
 
     def get_commit(self, commit_id=None, commit_idx=None, pre_load=None, translate_tag=None):
         """
         Returns instance of `BaseCommit` class. If `commit_id` and `commit_idx`
         are both None, most recent commit is returned.
 
         :param pre_load: Optional. List of commit attributes to load.
 
         :raises ``EmptyRepositoryError``: if there are no commits
         """
         raise NotImplementedError
 
     def __iter__(self):
         for commit_id in self.commit_ids:
             yield self.get_commit(commit_id=commit_id)
 
     def get_commits(
             self, start_id=None, end_id=None, start_date=None, end_date=None,
             branch_name=None, show_hidden=False, pre_load=None, translate_tags=None):
         """
         Returns iterator of `BaseCommit` objects from start to end
         not inclusive. This should behave just like a list, ie. end is not
         inclusive.
 
         :param start_id: None or str, must be a valid commit id
         :param end_id: None or str, must be a valid commit id
         :param start_date:
         :param end_date:
         :param branch_name:
         :param show_hidden:
         :param pre_load:
         :param translate_tags:
         """
         raise NotImplementedError
 
     def __getitem__(self, key):
         """
         Allows index based access to the commit objects of this repository.
         """
         pre_load = ["author", "branch", "date", "message", "parents"]
         if isinstance(key, slice):
             return self._get_range(key, pre_load)
         return self.get_commit(commit_idx=key, pre_load=pre_load)
 
     def _get_range(self, slice_obj, pre_load):
         for commit_id in self.commit_ids.__getitem__(slice_obj):
             yield self.get_commit(commit_id=commit_id, pre_load=pre_load)
 
     def count(self):
         return len(self.commit_ids)
 
     def tag(self, name, user, commit_id=None, message=None, date=None, **opts):
         """
         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
         """
         raise NotImplementedError
 
     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
         """
         raise NotImplementedError
 
     def get_diff(
             self, commit1, commit2, path=None, ignore_whitespace=False,
             context=3, path1=None):
         """
         Returns (git like) *diff*, as plain text. Shows changes introduced by
         `commit2` since `commit1`.
 
         :param commit1: Entry point from which diff is shown. Can be
           ``self.EMPTY_COMMIT`` - in this case, patch showing all
           the changes since empty state of the repository until `commit2`
         :param commit2: Until which commit changes should be shown.
         :param path: Can be set to a path of a file to create a diff of that
           file. If `path1` is also set, this value is only associated to
           `commit2`.
         :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``.
         :param path1: Can be set to a path to associate with `commit1`. This
           parameter works only for backends which support diff generation for
           different paths. Other backends will raise a `ValueError` if `path1`
           is set and has a different value than `path`.
         :param file_path: filter this diff by given path pattern
         """
         raise NotImplementedError
 
     def strip(self, commit_id, branch=None):
         """
         Strip given commit_id from the repository
         """
         raise NotImplementedError
 
     def get_common_ancestor(self, commit_id1, commit_id2, repo2):
         """
         Return a latest common ancestor commit if one exists for this repo
         `commit_id1` vs `commit_id2` from `repo2`.
 
         :param commit_id1: Commit it from this repository to use as a
            target for the comparison.
         :param commit_id2: Source commit id to use for comparison.
         :param repo2: Source repository to use for comparison.
         """
         raise NotImplementedError
 
     def compare(self, commit_id1, commit_id2, repo2, merge, pre_load=None):
         """
         Compare this repository's revision `commit_id1` with `commit_id2`.
 
         Returns a tuple(commits, ancestor) that would be merged from
         `commit_id2`. Doing a normal compare (``merge=False``), ``None``
         will be returned as ancestor.
 
         :param commit_id1: Commit it from this repository to use as a
            target for the comparison.
         :param commit_id2: Source commit id to use for comparison.
         :param repo2: Source repository to use for comparison.
         :param merge: If set to ``True`` will do a merge compare which also
            returns the common ancestor.
         :param pre_load: Optional. List of commit attributes to load.
         """
         raise NotImplementedError
 
     def merge(self, repo_id, workspace_id, target_ref, source_repo, source_ref,
               user_name='', user_email='', message='', dry_run=False,
               use_rebase=False, close_branch=False):
         """
         Merge the revisions specified in `source_ref` from `source_repo`
         onto the `target_ref` of this repository.
 
         `source_ref` and `target_ref` are named tupls with the following
         fields `type`, `name` and `commit_id`.
 
         Returns a MergeResponse named tuple with the following fields
         'possible', 'executed', 'source_commit', 'target_commit',
         'merge_commit'.
 
         :param repo_id: `repo_id` target repo id.
         :param workspace_id: `workspace_id` unique identifier.
         :param target_ref: `target_ref` points to the commit on top of which
             the `source_ref` should be merged.
         :param source_repo: The repository that contains the commits to be
             merged.
         :param source_ref: `source_ref` points to the topmost commit from
             the `source_repo` which should be merged.
         :param user_name: Merge commit `user_name`.
         :param user_email: Merge commit `user_email`.
         :param message: Merge commit `message`.
         :param dry_run: If `True` the merge will not take place.
         :param use_rebase: If `True` commits from the source will be rebased
             on top of the target instead of being merged.
         :param close_branch: If `True` branch will be close before merging it
         """
         if dry_run:
             message = message or settings.MERGE_DRY_RUN_MESSAGE
             user_email = user_email or settings.MERGE_DRY_RUN_EMAIL
             user_name = user_name or settings.MERGE_DRY_RUN_USER
         else:
             if not user_name:
                 raise ValueError('user_name cannot be empty')
             if not user_email:
                 raise ValueError('user_email cannot be empty')
             if not message:
                 raise ValueError('message cannot be empty')
 
         try:
             return self._merge_repo(
                 repo_id, workspace_id, target_ref, source_repo,
                 source_ref, message, user_name, user_email, dry_run=dry_run,
                 use_rebase=use_rebase, close_branch=close_branch)
         except RepositoryError as exc:
             log.exception('Unexpected failure when running merge, dry-run=%s', dry_run)
             return MergeResponse(
                 False, False, None, MergeFailureReason.UNKNOWN,
                 metadata={'exception': str(exc)})
 
     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):
         """Internal implementation of merge."""
         raise NotImplementedError
 
     def _maybe_prepare_merge_workspace(
             self, repo_id, workspace_id, target_ref, source_ref):
         """
         Create the merge workspace.
 
         :param workspace_id: `workspace_id` unique identifier.
         """
         raise NotImplementedError
 
-    def _get_legacy_shadow_repository_path(self, workspace_id):
+    @classmethod
+    def _get_legacy_shadow_repository_path(cls, repo_path, workspace_id):
         """
         Legacy version that was used before. We still need it for
         backward compat
         """
         return os.path.join(
-            os.path.dirname(self.path),
-            '.__shadow_%s_%s' % (os.path.basename(self.path), workspace_id))
+            os.path.dirname(repo_path),
+            '.__shadow_%s_%s' % (os.path.basename(repo_path), workspace_id))
 
-    def _get_shadow_repository_path(self, repo_id, workspace_id):
+    @classmethod
+    def _get_shadow_repository_path(cls, repo_path, repo_id, workspace_id):
         # The name of the shadow repository must start with '.', so it is
         # skipped by 'rhodecode.lib.utils.get_filesystem_repos'.
-        legacy_repository_path = self._get_legacy_shadow_repository_path(workspace_id)
+        legacy_repository_path = cls._get_legacy_shadow_repository_path(repo_path, workspace_id)
         if os.path.exists(legacy_repository_path):
             return legacy_repository_path
         else:
             return os.path.join(
-                os.path.dirname(self.path),
+                os.path.dirname(repo_path),
                 '.__shadow_repo_%s_%s' % (repo_id, workspace_id))
 
     def cleanup_merge_workspace(self, repo_id, workspace_id):
         """
         Remove merge workspace.
 
         This function MUST not fail in case there is no workspace associated to
         the given `workspace_id`.
 
         :param workspace_id: `workspace_id` unique identifier.
         """
-        shadow_repository_path = self._get_shadow_repository_path(repo_id, workspace_id)
+        shadow_repository_path = self._get_shadow_repository_path(
+            self.path, repo_id, workspace_id)
         shadow_repository_path_del = '{}.{}.delete'.format(
             shadow_repository_path, time.time())
 
         # move the shadow repo, so it never conflicts with the one used.
         # we use this method because shutil.rmtree had some edge case problems
         # removing symlinked repositories
         if not os.path.isdir(shadow_repository_path):
             return
 
         shutil.move(shadow_repository_path, shadow_repository_path_del)
         try:
             shutil.rmtree(shadow_repository_path_del, ignore_errors=False)
         except Exception:
             log.exception('Failed to gracefully remove shadow repo under %s',
                           shadow_repository_path_del)
             shutil.rmtree(shadow_repository_path_del, ignore_errors=True)
 
     # ========== #
     # COMMIT API #
     # ========== #
 
     @LazyProperty
     def in_memory_commit(self):
         """
         Returns :class:`InMemoryCommit` object for this repository.
         """
         raise NotImplementedError
 
     # ======================== #
     # UTILITIES FOR SUBCLASSES #
     # ======================== #
 
     def _validate_diff_commits(self, commit1, commit2):
         """
         Validates that the given commits are related to this repository.
 
         Intended as a utility for sub classes to have a consistent validation
         of input parameters in methods like :meth:`get_diff`.
         """
         self._validate_commit(commit1)
         self._validate_commit(commit2)
         if (isinstance(commit1, EmptyCommit) and
                 isinstance(commit2, EmptyCommit)):
             raise ValueError("Cannot compare two empty commits")
 
     def _validate_commit(self, commit):
         if not isinstance(commit, BaseCommit):
             raise TypeError(
                 "%s is not of type BaseCommit" % repr(commit))
         if commit.repository != self and not isinstance(commit, EmptyCommit):
             raise ValueError(
                 "Commit %s must be a valid commit from this repository %s, "
                 "related to this repository instead %s." %
                 (commit, self, commit.repository))
 
     def _validate_commit_id(self, commit_id):
         if not isinstance(commit_id, compat.string_types):
             raise TypeError("commit_id must be a string value")
 
     def _validate_commit_idx(self, commit_idx):
         if not isinstance(commit_idx, (int, long)):
             raise TypeError("commit_idx must be a numeric value")
 
     def _validate_branch_name(self, branch_name):
         if branch_name and branch_name not in self.branches_all:
             msg = ("Branch %s not found in %s" % (branch_name, self))
             raise BranchDoesNotExistError(msg)
 
     #
     # Supporting deprecated API parts
     # TODO: johbo: consider to move this into a mixin
     #
 
     @property
     def EMPTY_CHANGESET(self):
         warnings.warn(
             "Use EMPTY_COMMIT or EMPTY_COMMIT_ID instead", DeprecationWarning)
         return self.EMPTY_COMMIT_ID
 
     @property
     def revisions(self):
         warnings.warn("Use commits attribute instead", DeprecationWarning)
         return self.commit_ids
 
     @revisions.setter
     def revisions(self, value):
         warnings.warn("Use commits attribute instead", DeprecationWarning)
         self.commit_ids = value
 
     def get_changeset(self, revision=None, pre_load=None):
         warnings.warn("Use get_commit instead", DeprecationWarning)
         commit_id = None
         commit_idx = None
         if isinstance(revision, compat.string_types):
             commit_id = revision
         else:
             commit_idx = revision
         return self.get_commit(
             commit_id=commit_id, commit_idx=commit_idx, pre_load=pre_load)
 
     def get_changesets(
             self, start=None, end=None, start_date=None, end_date=None,
             branch_name=None, pre_load=None):
         warnings.warn("Use get_commits instead", DeprecationWarning)
         start_id = self._revision_to_commit(start)
         end_id = self._revision_to_commit(end)
         return self.get_commits(
             start_id=start_id, end_id=end_id, start_date=start_date,
             end_date=end_date, branch_name=branch_name, pre_load=pre_load)
 
     def _revision_to_commit(self, revision):
         """
         Translates a revision to a commit_id
 
         Helps to support the old changeset based API which allows to use
         commit ids and commit indices interchangeable.
         """
         if revision is None:
             return revision
 
         if isinstance(revision, compat.string_types):
             commit_id = revision
         else:
             commit_id = self.commit_ids[revision]
         return commit_id
 
     @property
     def in_memory_changeset(self):
         warnings.warn("Use in_memory_commit instead", DeprecationWarning)
         return self.in_memory_commit
 
     def get_path_permissions(self, username):
         """
         Returns a path permission checker or None if not supported
 
         :param username: session user name
         :return: an instance of BasePathPermissionChecker or None
         """
         return None
 
     def install_hooks(self, force=False):
         return self._remote.install_hooks(force)
 
     def get_hooks_info(self):
         return self._remote.get_hooks_info()
 
 
 class BaseCommit(object):
     """
     Each backend should implement it's commit representation.
 
     **Attributes**
 
         ``repository``
             repository object within which commit exists
 
         ``id``
             The commit id, may be ``raw_id`` or i.e. for mercurial's tip
             just ``tip``.
 
         ``raw_id``
             raw commit representation (i.e. full 40 length sha for git
             backend)
 
         ``short_id``
             shortened (if apply) version of ``raw_id``; it would be simple
             shortcut for ``raw_id[:12]`` for git/mercurial backends or same
             as ``raw_id`` for subversion
 
         ``idx``
             commit index
 
         ``files``
             list of ``FileNode`` (``Node`` with NodeKind.FILE) objects
 
         ``dirs``
             list of ``DirNode`` (``Node`` with NodeKind.DIR) objects
 
         ``nodes``
             combined list of ``Node`` objects
 
         ``author``
             author of the commit, as unicode
 
         ``message``
             message of the commit, as unicode
 
         ``parents``
             list of parent commits
 
     """
 
     branch = None
     """
     Depending on the backend this should be set to the branch name of the
     commit. Backends not supporting branches on commits should leave this
     value as ``None``.
     """
 
     _ARCHIVE_PREFIX_TEMPLATE = b'{repo_name}-{short_id}'
     """
     This template is used to generate a default prefix for repository archives
     if no prefix has been specified.
     """
 
     def __str__(self):
         return '<%s at %s:%s>' % (
             self.__class__.__name__, self.idx, self.short_id)
 
     def __repr__(self):
         return self.__str__()
 
     def __unicode__(self):
         return u'%s:%s' % (self.idx, self.short_id)
 
     def __eq__(self, other):
         same_instance = isinstance(other, self.__class__)
         return same_instance and self.raw_id == other.raw_id
 
     def __json__(self):
         parents = []
         try:
             for parent in self.parents:
                 parents.append({'raw_id': parent.raw_id})
         except NotImplementedError:
             # empty commit doesn't have parents implemented
             pass
 
         return {
             'short_id': self.short_id,
             'raw_id': self.raw_id,
             'revision': self.idx,
             'message': self.message,
             'date': self.date,
             'author': self.author,
             'parents': parents,
             'branch': self.branch
         }
 
     def __getstate__(self):
         d = self.__dict__.copy()
         d.pop('_remote', None)
         d.pop('repository', None)
         return d
 
     def _get_refs(self):
         return {
             'branches': [self.branch] if self.branch else [],
             'bookmarks': getattr(self, 'bookmarks', []),
             'tags': self.tags
         }
 
     @LazyProperty
     def last(self):
         """
         ``True`` if this is last commit in repository, ``False``
         otherwise; trying to access this attribute while there is no
         commits would raise `EmptyRepositoryError`
         """
         if self.repository is None:
             raise CommitError("Cannot check if it's most recent commit")
         return self.raw_id == self.repository.commit_ids[-1]
 
     @LazyProperty
     def parents(self):
         """
         Returns list of parent commits.
         """
         raise NotImplementedError
 
     @LazyProperty
     def first_parent(self):
         """
         Returns list of parent commits.
         """
         return self.parents[0] if self.parents else EmptyCommit()
 
     @property
     def merge(self):
         """
         Returns boolean if commit is a merge.
         """
         return len(self.parents) > 1
 
     @LazyProperty
     def children(self):
         """
         Returns list of child commits.
         """
         raise NotImplementedError
 
     @LazyProperty
     def id(self):
         """
         Returns string identifying this commit.
         """
         raise NotImplementedError
 
     @LazyProperty
     def raw_id(self):
         """
         Returns raw string identifying this commit.
         """
         raise NotImplementedError
 
     @LazyProperty
     def short_id(self):
         """
         Returns shortened version of ``raw_id`` attribute, as string,
         identifying this commit, useful for presentation to users.
         """
         raise NotImplementedError
 
     @LazyProperty
     def idx(self):
         """
         Returns integer identifying this commit.
         """
         raise NotImplementedError
 
     @LazyProperty
     def committer(self):
         """
         Returns committer for this commit
         """
         raise NotImplementedError
 
     @LazyProperty
     def committer_name(self):
         """
         Returns committer name for this commit
         """
 
         return author_name(self.committer)
 
     @LazyProperty
     def committer_email(self):
         """
         Returns committer email address for this commit
         """
 
         return author_email(self.committer)
 
     @LazyProperty
     def author(self):
         """
         Returns author for this commit
         """
 
         raise NotImplementedError
 
     @LazyProperty
     def author_name(self):
         """
         Returns author name for this commit
         """
 
         return author_name(self.author)
 
     @LazyProperty
     def author_email(self):
         """
         Returns author email address for this commit
         """
 
         return author_email(self.author)
 
     def get_file_mode(self, path):
         """
         Returns stat mode of the file at `path`.
         """
         raise NotImplementedError
 
     def is_link(self, path):
         """
         Returns ``True`` if given `path` is a symlink
         """
         raise NotImplementedError
 
     def is_node_binary(self, path):
         """
         Returns ``True`` is given path is a binary file
         """
         raise NotImplementedError
 
     def get_file_content(self, path):
         """
         Returns content of the file at the given `path`.
         """
         raise NotImplementedError
 
     def get_file_content_streamed(self, path):
         """
         returns a streaming response from vcsserver with file content
         """
         raise NotImplementedError
 
     def get_file_size(self, path):
         """
         Returns size of the file at the given `path`.
         """
         raise NotImplementedError
 
     def get_path_commit(self, path, pre_load=None):
         """
         Returns last commit of the file at the given `path`.
 
         :param pre_load: Optional. List of commit attributes to load.
         """
         commits = self.get_path_history(path, limit=1, pre_load=pre_load)
         if not commits:
             raise RepositoryError(
                 'Failed to fetch history for path {}. '
                 'Please check if such path exists in your repository'.format(
                     path))
         return commits[0]
 
     def get_path_history(self, path, limit=None, pre_load=None):
         """
         Returns history of file as reversed list of :class:`BaseCommit`
         objects for which file at given `path` has been modified.
 
         :param limit: Optional. Allows to limit the size of the returned
            history. This is intended as a hint to the underlying backend, so
            that it can apply optimizations depending on the limit.
         :param pre_load: Optional. List of commit attributes to load.
         """
         raise NotImplementedError
 
     def get_file_annotate(self, path, pre_load=None):
         """
         Returns a generator of four element tuples with
         lineno, sha, commit lazy loader and line
 
         :param pre_load: Optional. List of commit attributes to load.
         """
         raise NotImplementedError
 
     def get_nodes(self, path):
         """
         Returns combined ``DirNode`` and ``FileNode`` objects list representing
         state of commit at the given ``path``.
 
         :raises ``CommitError``: if node at the given ``path`` is not
           instance of ``DirNode``
         """
         raise NotImplementedError
 
     def get_node(self, path):
         """
         Returns ``Node`` object from the given ``path``.
 
         :raises ``NodeDoesNotExistError``: if there is no node at the given
           ``path``
         """
         raise NotImplementedError
 
     def get_largefile_node(self, path):
         """
         Returns the path to largefile from Mercurial/Git-lfs storage.
         or None if it's not a largefile node
         """
         return None
 
     def archive_repo(self, archive_dest_path, kind='tgz', subrepos=None,
                      prefix=None, write_metadata=False, mtime=None, archive_at_path='/'):
         """
         Creates an archive containing the contents of the repository.
 
         :param archive_dest_path: path to the file which to create the archive.
         :param kind: one of following: ``"tbz2"``, ``"tgz"``, ``"zip"``.
         :param prefix: name of root directory in archive.
             Default is repository name and commit's short_id joined with dash:
             ``"{repo_name}-{short_id}"``.
         :param write_metadata: write a metadata file into archive.
         :param mtime: custom modification time for archive creation, defaults
             to time.time() if not given.
         :param archive_at_path: pack files at this path (default '/')
 
         :raise VCSError: If prefix has a problem.
         """
         allowed_kinds = [x[0] for x in settings.ARCHIVE_SPECS]
         if kind not in allowed_kinds:
             raise ImproperArchiveTypeError(
                 'Archive kind (%s) not supported use one of %s' %
                 (kind, allowed_kinds))
 
         prefix = self._validate_archive_prefix(prefix)
 
         mtime = mtime is not None or time.mktime(self.date.timetuple())
 
         file_info = []
         cur_rev = self.repository.get_commit(commit_id=self.raw_id)
         for _r, _d, files in cur_rev.walk(archive_at_path):
             for f in files:
                 f_path = os.path.join(prefix, f.path)
                 file_info.append(
                     (f_path, f.mode, f.is_link(), f.raw_bytes))
 
         if write_metadata:
             metadata = [
                 ('repo_name', self.repository.name),
                 ('commit_id', self.raw_id),
                 ('mtime', mtime),
                 ('branch', self.branch),
                 ('tags', ','.join(self.tags)),
             ]
             meta = ["%s:%s" % (f_name, value) for f_name, value in metadata]
             file_info.append(('.archival.txt', 0o644, False, '\n'.join(meta)))
 
         connection.Hg.archive_repo(archive_dest_path, mtime, file_info, kind)
 
     def _validate_archive_prefix(self, prefix):
         if prefix is None:
             prefix = self._ARCHIVE_PREFIX_TEMPLATE.format(
                 repo_name=safe_str(self.repository.name),
                 short_id=self.short_id)
         elif not isinstance(prefix, str):
             raise ValueError("prefix not a bytes object: %s" % repr(prefix))
         elif prefix.startswith('/'):
             raise VCSError("Prefix cannot start with leading slash")
         elif prefix.strip() == '':
             raise VCSError("Prefix cannot be empty")
         return prefix
 
     @LazyProperty
     def root(self):
         """
         Returns ``RootNode`` object for this commit.
         """
         return self.get_node('')
 
     def next(self, branch=None):
         """
         Returns next commit from current, if branch is gives it will return
         next commit belonging to this branch
 
         :param branch: show commits within the given named branch
         """
         indexes = xrange(self.idx + 1, self.repository.count())
         return self._find_next(indexes, branch)
 
     def prev(self, branch=None):
         """
         Returns previous commit from current, if branch is gives it will
         return previous commit belonging to this branch
 
         :param branch: show commit within the given named branch
         """
         indexes = xrange(self.idx - 1, -1, -1)
         return self._find_next(indexes, branch)
 
     def _find_next(self, indexes, branch=None):
         if branch and self.branch != branch:
             raise VCSError('Branch option used on commit not belonging '
                            'to that branch')
 
         for next_idx in indexes:
             commit = self.repository.get_commit(commit_idx=next_idx)
             if branch and branch != commit.branch:
                 continue
             return commit
         raise CommitDoesNotExistError
 
     def diff(self, ignore_whitespace=True, context=3):
         """
         Returns a `Diff` object representing the change made by this commit.
         """
         parent = self.first_parent
         diff = self.repository.get_diff(
             parent, self,
             ignore_whitespace=ignore_whitespace,
             context=context)
         return diff
 
     @LazyProperty
     def added(self):
         """
         Returns list of added ``FileNode`` objects.
         """
         raise NotImplementedError
 
     @LazyProperty
     def changed(self):
         """
         Returns list of modified ``FileNode`` objects.
         """
         raise NotImplementedError
 
     @LazyProperty
     def removed(self):
         """
         Returns list of removed ``FileNode`` objects.
         """
         raise NotImplementedError
 
     @LazyProperty
     def size(self):
         """
         Returns total number of bytes from contents of all filenodes.
         """
         return sum((node.size for node in self.get_filenodes_generator()))
 
     def walk(self, topurl=''):
         """
         Similar to os.walk method. Insted of filesystem it walks through
         commit starting at given ``topurl``.  Returns generator of tuples
         (topnode, dirnodes, filenodes).
         """
         topnode = self.get_node(topurl)
         if not topnode.is_dir():
             return
         yield (topnode, topnode.dirs, topnode.files)
         for dirnode in topnode.dirs:
             for tup in self.walk(dirnode.path):
                 yield tup
 
     def get_filenodes_generator(self):
         """
         Returns generator that yields *all* file nodes.
         """
         for topnode, dirs, files in self.walk():
             for node in files:
                 yield node
 
     #
     # Utilities for sub classes to support consistent behavior
     #
 
     def no_node_at_path(self, path):
         return NodeDoesNotExistError(
             u"There is no file nor directory at the given path: "
             u"`%s` at commit %s" % (safe_unicode(path), self.short_id))
 
     def _fix_path(self, path):
         """
         Paths are stored without trailing slash so we need to get rid off it if
         needed.
         """
         return path.rstrip('/')
 
     #
     # Deprecated API based on changesets
     #
 
     @property
     def revision(self):
         warnings.warn("Use idx instead", DeprecationWarning)
         return self.idx
 
     @revision.setter
     def revision(self, value):
         warnings.warn("Use idx instead", DeprecationWarning)
         self.idx = value
 
     def get_file_changeset(self, path):
         warnings.warn("Use get_path_commit instead", DeprecationWarning)
         return self.get_path_commit(path)
 
 
 class BaseChangesetClass(type):
 
     def __instancecheck__(self, instance):
         return isinstance(instance, BaseCommit)
 
 
 class BaseChangeset(BaseCommit):
 
     __metaclass__ = BaseChangesetClass
 
     def __new__(cls, *args, **kwargs):
         warnings.warn(
             "Use BaseCommit instead of BaseChangeset", DeprecationWarning)
         return super(BaseChangeset, cls).__new__(cls, *args, **kwargs)
 
 
 class BaseInMemoryCommit(object):
     """
     Represents differences between repository's state (most recent head) and
     changes made *in place*.
 
     **Attributes**
 
         ``repository``
             repository object for this in-memory-commit
 
         ``added``
             list of ``FileNode`` objects marked as *added*
 
         ``changed``
             list of ``FileNode`` objects marked as *changed*
 
         ``removed``
             list of ``FileNode`` or ``RemovedFileNode`` objects marked to be
             *removed*
 
         ``parents``
             list of :class:`BaseCommit` instances representing parents of
             in-memory commit. Should always be 2-element sequence.
 
     """
 
     def __init__(self, repository):
         self.repository = repository
         self.added = []
         self.changed = []
         self.removed = []
         self.parents = []
 
     def add(self, *filenodes):
         """
         Marks given ``FileNode`` objects as *to be committed*.
 
         :raises ``NodeAlreadyExistsError``: if node with same path exists at
            latest commit
         :raises ``NodeAlreadyAddedError``: if node with same path is already
            marked as *added*
         """
         # Check if not already marked as *added* first
         for node in filenodes:
             if node.path in (n.path for n in self.added):
                 raise NodeAlreadyAddedError(
                     "Such FileNode %s is already marked for addition"
                     % node.path)
         for node in filenodes:
             self.added.append(node)
 
     def change(self, *filenodes):
         """
         Marks given ``FileNode`` objects to be *changed* in next commit.
 
         :raises ``EmptyRepositoryError``: if there are no commits yet
         :raises ``NodeAlreadyExistsError``: if node with same path is already
            marked to be *changed*
         :raises ``NodeAlreadyRemovedError``: if node with same path is already
            marked to be *removed*
         :raises ``NodeDoesNotExistError``: if node doesn't exist in latest
            commit
         :raises ``NodeNotChangedError``: if node hasn't really be changed
         """
         for node in filenodes:
             if node.path in (n.path for n in self.removed):
                 raise NodeAlreadyRemovedError(
                     "Node at %s is already marked as removed" % node.path)
         try:
             self.repository.get_commit()
         except EmptyRepositoryError:
             raise EmptyRepositoryError(
                 "Nothing to change - try to *add* new nodes rather than "
                 "changing them")
         for node in filenodes:
             if node.path in (n.path for n in self.changed):
                 raise NodeAlreadyChangedError(
                     "Node at '%s' is already marked as changed" % node.path)
             self.changed.append(node)
 
     def remove(self, *filenodes):
         """
         Marks given ``FileNode`` (or ``RemovedFileNode``) objects to be
         *removed* in next commit.
 
         :raises ``NodeAlreadyRemovedError``: if node has been already marked to
            be *removed*
         :raises ``NodeAlreadyChangedError``: if node has been already marked to
            be *changed*
         """
         for node in filenodes:
             if node.path in (n.path for n in self.removed):
                 raise NodeAlreadyRemovedError(
                     "Node is already marked to for removal at %s" % node.path)
             if node.path in (n.path for n in self.changed):
                 raise NodeAlreadyChangedError(
                     "Node is already marked to be changed at %s" % node.path)
             # We only mark node as *removed* - real removal is done by
             # commit method
             self.removed.append(node)
 
     def reset(self):
         """
         Resets this instance to initial state (cleans ``added``, ``changed``
         and ``removed`` lists).
         """
         self.added = []
         self.changed = []
         self.removed = []
         self.parents = []
 
     def get_ipaths(self):
         """
         Returns generator of paths from nodes marked as added, changed or
         removed.
         """
         for node in itertools.chain(self.added, self.changed, self.removed):
             yield node.path
 
     def get_paths(self):
         """
         Returns list of paths from nodes marked as added, changed or removed.
         """
         return list(self.get_ipaths())
 
     def check_integrity(self, parents=None):
         """
         Checks in-memory commit's integrity. Also, sets parents if not
         already set.
 
         :raises CommitError: if any error occurs (i.e.
           ``NodeDoesNotExistError``).
         """
         if not self.parents:
             parents = parents or []
             if len(parents) == 0:
                 try:
                     parents = [self.repository.get_commit(), None]
                 except EmptyRepositoryError:
                     parents = [None, None]
             elif len(parents) == 1:
                 parents += [None]
             self.parents = parents
 
         # Local parents, only if not None
         parents = [p for p in self.parents if p]
 
         # Check nodes marked as added
         for p in parents:
             for node in self.added:
                 try:
                     p.get_node(node.path)
                 except NodeDoesNotExistError:
                     pass
                 else:
                     raise NodeAlreadyExistsError(
                         "Node `%s` already exists at %s" % (node.path, p))
 
         # Check nodes marked as changed
         missing = set(self.changed)
         not_changed = set(self.changed)
         if self.changed and not parents:
             raise NodeDoesNotExistError(str(self.changed[0].path))
         for p in parents:
             for node in self.changed:
                 try:
                     old = p.get_node(node.path)
                     missing.remove(node)
                     # if content actually changed, remove node from not_changed
                     if old.content != node.content:
                         not_changed.remove(node)
                 except NodeDoesNotExistError:
                     pass
         if self.changed and missing:
             raise NodeDoesNotExistError(
                 "Node `%s` marked as modified but missing in parents: %s"
                 % (node.path, parents))
 
         if self.changed and not_changed:
             raise NodeNotChangedError(
                 "Node `%s` wasn't actually changed (parents: %s)"
                 % (not_changed.pop().path, parents))
 
         # Check nodes marked as removed
         if self.removed and not parents:
             raise NodeDoesNotExistError(
                 "Cannot remove node at %s as there "
                 "were no parents specified" % self.removed[0].path)
         really_removed = set()
         for p in parents:
             for node in self.removed:
                 try:
                     p.get_node(node.path)
                     really_removed.add(node)
                 except CommitError:
                     pass
         not_removed = set(self.removed) - really_removed
         if not_removed:
             # TODO: johbo: This code branch does not seem to be covered
             raise NodeDoesNotExistError(
                 "Cannot remove node at %s from "
                 "following parents: %s" % (not_removed, parents))
 
     def commit(self, message, author, parents=None, branch=None, date=None, **kwargs):
         """
         Performs in-memory commit (doesn't check workdir in any way) and
         returns newly created :class:`BaseCommit`. Updates repository's
         attribute `commits`.
 
         .. note::
 
            While overriding this method each backend's should call
            ``self.check_integrity(parents)`` in the first place.
 
         :param message: message of the commit
         :param author: full username, i.e. "Joe Doe <[e-mail]>"
         :param parents: single parent or sequence of parents from which commit
            would be derived
         :param date: ``datetime.datetime`` instance. Defaults to
            ``datetime.datetime.now()``.
         :param branch: branch name, as string. If none given, default backend's
            branch would be used.
 
         :raises ``CommitError``: if any error occurs while committing
         """
         raise NotImplementedError
 
 
 class BaseInMemoryChangesetClass(type):
 
     def __instancecheck__(self, instance):
         return isinstance(instance, BaseInMemoryCommit)
 
 
 class BaseInMemoryChangeset(BaseInMemoryCommit):
 
     __metaclass__ = BaseInMemoryChangesetClass
 
     def __new__(cls, *args, **kwargs):
         warnings.warn(
             "Use BaseCommit instead of BaseInMemoryCommit", DeprecationWarning)
         return super(BaseInMemoryChangeset, cls).__new__(cls, *args, **kwargs)
 
 
 class EmptyCommit(BaseCommit):
     """
     An dummy empty commit. It's possible to pass hash when creating
     an EmptyCommit
     """
 
     def __init__(
             self, commit_id=EMPTY_COMMIT_ID, repo=None, alias=None, idx=-1,
             message='', author='', date=None):
         self._empty_commit_id = commit_id
         # TODO: johbo: Solve idx parameter, default value does not make
         # too much sense
         self.idx = idx
         self.message = message
         self.author = author
         self.date = date or datetime.datetime.fromtimestamp(0)
         self.repository = repo
         self.alias = alias
 
     @LazyProperty
     def raw_id(self):
         """
         Returns raw string identifying this commit, useful for web
         representation.
         """
 
         return self._empty_commit_id
 
     @LazyProperty
     def branch(self):
         if self.alias:
             from rhodecode.lib.vcs.backends import get_backend
             return get_backend(self.alias).DEFAULT_BRANCH_NAME
 
     @LazyProperty
     def short_id(self):
         return self.raw_id[:12]
 
     @LazyProperty
     def id(self):
         return self.raw_id
 
     def get_path_commit(self, path):
         return self
 
     def get_file_content(self, path):
         return u''
 
     def get_file_content_streamed(self, path):
         yield self.get_file_content()
 
     def get_file_size(self, path):
         return 0
 
 
 class EmptyChangesetClass(type):
 
     def __instancecheck__(self, instance):
         return isinstance(instance, EmptyCommit)
 
 
 class EmptyChangeset(EmptyCommit):
 
     __metaclass__ = EmptyChangesetClass
 
     def __new__(cls, *args, **kwargs):
         warnings.warn(
             "Use EmptyCommit instead of EmptyChangeset", DeprecationWarning)
         return super(EmptyCommit, cls).__new__(cls, *args, **kwargs)
 
     def __init__(self, cs=EMPTY_COMMIT_ID, repo=None, requested_revision=None,
                  alias=None, revision=-1, message='', author='', date=None):
         if requested_revision is not None:
             warnings.warn(
                 "Parameter requested_revision not supported anymore",
                 DeprecationWarning)
         super(EmptyChangeset, self).__init__(
             commit_id=cs, repo=repo, alias=alias, idx=revision,
             message=message, author=author, date=date)
 
         @property
         def revision(self):
             warnings.warn("Use idx instead", DeprecationWarning)
             return self.idx
 
         @revision.setter
         def revision(self, value):
             warnings.warn("Use idx instead", DeprecationWarning)
             self.idx = value
 
 
 class EmptyRepository(BaseRepository):
     def __init__(self, repo_path=None, config=None, create=False, **kwargs):
         pass
 
     def get_diff(self, *args, **kwargs):
         from rhodecode.lib.vcs.backends.git.diff import GitDiff
         return GitDiff('')
 
 
 class CollectionGenerator(object):
 
     def __init__(self, repo, commit_ids, collection_size=None, pre_load=None, translate_tag=None):
         self.repo = repo
         self.commit_ids = commit_ids
         # TODO: (oliver) this isn't currently hooked up
         self.collection_size = None
         self.pre_load = pre_load
         self.translate_tag = translate_tag
 
     def __len__(self):
         if self.collection_size is not None:
             return self.collection_size
         return self.commit_ids.__len__()
 
     def __iter__(self):
         for commit_id in self.commit_ids:
             # TODO: johbo: Mercurial passes in commit indices or commit ids
             yield self._commit_factory(commit_id)
 
     def _commit_factory(self, commit_id):
         """
         Allows backends to override the way commits are generated.
         """
         return self.repo.get_commit(
             commit_id=commit_id, pre_load=self.pre_load,
             translate_tag=self.translate_tag)
 
     def __getslice__(self, i, j):
         """
         Returns an iterator of sliced repository
         """
         commit_ids = self.commit_ids[i:j]
         return self.__class__(
             self.repo, commit_ids, pre_load=self.pre_load,
             translate_tag=self.translate_tag)
 
     def __repr__(self):
         return '<CollectionGenerator[len:%s]>' % (self.__len__())
 
 
 class Config(object):
     """
     Represents the configuration for a repository.
 
     The API is inspired by :class:`ConfigParser.ConfigParser` from the
     standard library. It implements only the needed subset.
     """
 
     def __init__(self):
         self._values = {}
 
     def copy(self):
         clone = Config()
         for section, values in self._values.items():
             clone._values[section] = values.copy()
         return clone
 
     def __repr__(self):
         return '<Config(%s sections) at %s>' % (
             len(self._values), hex(id(self)))
 
     def items(self, section):
         return self._values.get(section, {}).iteritems()
 
     def get(self, section, option):
         return self._values.get(section, {}).get(option)
 
     def set(self, section, option, value):
         section_values = self._values.setdefault(section, {})
         section_values[option] = value
 
     def clear_section(self, section):
         self._values[section] = {}
 
     def serialize(self):
         """
         Creates a list of three tuples (section, key, value) representing
         this config object.
         """
         items = []
         for section in self._values:
             for option, value in self._values[section].items():
                 items.append(
                     (safe_str(section), safe_str(option), safe_str(value)))
         return items
 
 
 class Diff(object):
     """
     Represents a diff result from a repository backend.
 
     Subclasses have to provide a backend specific value for
     :attr:`_header_re` and :attr:`_meta_re`.
     """
     _meta_re = None
     _header_re = None
 
     def __init__(self, raw_diff):
         self.raw = raw_diff
 
     def chunks(self):
         """
         split the diff in chunks of separate --git a/file b/file chunks
         to make diffs consistent we must prepend with \n, and make sure
         we can detect last chunk as this was also has special rule
         """
 
         diff_parts = ('\n' + self.raw).split('\ndiff --git')
         header = diff_parts[0]
 
         if self._meta_re:
             match = self._meta_re.match(header)
 
         chunks = diff_parts[1:]
         total_chunks = len(chunks)
 
         return (
             DiffChunk(chunk, self, cur_chunk == total_chunks)
             for cur_chunk, chunk in enumerate(chunks, start=1))
 
 
 class DiffChunk(object):
 
     def __init__(self, chunk, diff, last_chunk):
         self._diff = diff
 
         # since we split by \ndiff --git that part is lost from original diff
         # we need to re-apply it at the end, EXCEPT ! if it's last chunk
         if not last_chunk:
             chunk += '\n'
 
         match = self._diff._header_re.match(chunk)
         self.header = match.groupdict()
         self.diff = chunk[match.end():]
         self.raw = chunk
 
 
 class BasePathPermissionChecker(object):
 
     @staticmethod
     def create_from_patterns(includes, excludes):
         if includes and '*' in includes and not excludes:
             return AllPathPermissionChecker()
         elif excludes and '*' in excludes:
             return NonePathPermissionChecker()
         else:
             return PatternPathPermissionChecker(includes, excludes)
 
     @property
     def has_full_access(self):
         raise NotImplemented()
 
     def has_access(self, path):
         raise NotImplemented()
 
 
 class AllPathPermissionChecker(BasePathPermissionChecker):
 
     @property
     def has_full_access(self):
         return True
 
     def has_access(self, path):
         return True
 
 
 class NonePathPermissionChecker(BasePathPermissionChecker):
 
     @property
     def has_full_access(self):
         return False
 
     def has_access(self, path):
         return False
 
 
 class PatternPathPermissionChecker(BasePathPermissionChecker):
 
     def __init__(self, includes, excludes):
         self.includes = includes
         self.excludes = excludes
         self.includes_re = [] if not includes else [
             re.compile(fnmatch.translate(pattern)) for pattern in includes]
         self.excludes_re = [] if not excludes else [
             re.compile(fnmatch.translate(pattern)) for pattern in excludes]
 
     @property
     def has_full_access(self):
         return '*' in self.includes and not self.excludes
 
     def has_access(self, path):
         for regex in self.excludes_re:
             if regex.match(path):
                 return False
         for regex in self.includes_re:
             if regex.match(path):
                 return True
         return False

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

@@ -1,1004 +1,1004 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 GIT repository module
 """
 
 import logging
 import os
 import re
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from rhodecode.lib.compat import OrderedDict
 from rhodecode.lib.datelib import (
     utcdate_fromtimestamp, makedate, date_astimestamp)
 from rhodecode.lib.utils import safe_unicode, safe_str
 from rhodecode.lib.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)
 
 
 SHA_PATTERN = re.compile(r'^[[0-9a-fA-F]{12}|[0-9a-fA-F]{40}]$')
 
 log = logging.getLogger(__name__)
 
 
 class GitRepository(BaseRepository):
     """
     Git repository backend.
     """
     DEFAULT_BRANCH_NAME = 'master'
 
     contact = BaseRepository.DEFAULT_CONTACT
 
     def __init__(self, repo_path, config=None, create=False, src_url=None,
                  do_workspace_checkout=False, with_wire=None, bare=False):
 
         self.path = safe_str(os.path.abspath(repo_path))
         self.config = config if config else self.get_default_config()
         self.with_wire = with_wire 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):
         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))
         if ((is_bstr and commit_id_or_idx.isdigit() and len(commit_id_or_idx) < 12)
             or isinstance(commit_id_or_idx, int) or is_null(commit_id_or_idx)):
             try:
                 commit_id_or_idx = self.commit_ids[int(commit_id_or_idx)]
             except Exception:
                 raise CommitDoesNotExistError(commit_missing_err)
 
         elif is_bstr:
             # Need to call remote to translate id for tagging scenario
             try:
                 remote_data = self._remote.get_object(commit_id_or_idx)
                 commit_id_or_idx = remote_data["commit_id"]
             except (CommitDoesNotExistError,):
                 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):
         """
         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)
 
         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):
         if commit_id1 == commit_id2:
             return commit_id1
 
         if self != repo2:
             commits = self._remote.get_missing_revs(
                 commit_id1, commit_id2, repo2.path)
             if commits:
                 commit = repo2.get_commit(commits[-1])
                 if commit.parents:
                     ancestor_id = commit.parents[0].raw_id
                 else:
                     ancestor_id = None
             else:
                 # no commits from other repo, ancestor_id is the commit_id2
                 ancestor_id = commit_id2
         else:
             output, __ = self.run_git_command(
                 ['merge-base', commit_id1, commit_id2])
             ancestor_id = re.findall(r'[0-9a-fA-F]{40}', output)[0]
 
         return ancestor_id
 
     def compare(self, commit_id1, commit_id2, repo2, merge, pre_load=None):
         repo1 = self
         ancestor_id = None
 
         if commit_id1 == commit_id2:
             commits = []
         elif repo1 != repo2:
             missing_ids = self._remote.get_missing_revs(commit_id1, commit_id2,
                                                         repo2.path)
             commits = [
                 repo2.get_commit(commit_id=commit_id, pre_load=pre_load)
                 for commit_id in reversed(missing_ids)]
         else:
             output, __ = repo1.run_git_command(
                 ['log', '--reverse', '--pretty=format: %H', '-s',
                  '%s..%s' % (commit_id1, commit_id2)])
             commits = [
                 repo1.get_commit(commit_id=commit_id, pre_load=pre_load)
                 for commit_id in re.findall(r'[0-9a-fA-F]{40}', output)]
 
         return commits
 
     @LazyProperty
     def in_memory_commit(self):
         """
         Returns ``GitInMemoryCommit`` object for this repository.
         """
         return GitInMemoryCommit(self)
 
     def pull(self, url, commit_ids=None, update_after=False):
         """
         Pull changes from external location. Pull is different in GIT
         that fetch since it's doing a checkout
 
         :param commit_ids: Optional. Can be set to a list of commit ids
            which shall be pulled from the other repository.
         """
         refs = None
         if commit_ids is not None:
             remote_refs = self._remote.get_remote_refs(url)
             refs = [ref for ref in remote_refs if remote_refs[ref] in commit_ids]
         self._remote.pull(url, refs=refs, update_after=update_after)
         self._remote.invalidate_vcs_cache()
 
     def fetch(self, url, commit_ids=None):
         """
         Fetch all git objects from external location.
         """
         self._remote.sync_fetch(url, refs=commit_ids)
         self._remote.invalidate_vcs_cache()
 
     def push(self, url):
         refs = None
         self._remote.sync_push(url, refs=refs)
 
     def set_refs(self, ref_name, commit_id):
         self._remote.set_refs(ref_name, commit_id)
         self._invalidate_prop_cache('_refs')
 
     def remove_ref(self, ref_name):
         self._remote.remove_ref(ref_name)
         self._invalidate_prop_cache('_refs')
 
     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 somehting more robust in this case.
             raise RepositoryError(
                 'Do not know how to merge into empty repositories yet')
 
         # N.B.(skreft): the --no-ff option is used to enforce the creation of a
         # commit message. We also specify the user who is doing the merge.
         cmd = ['-c', 'user.name="%s"' % safe_str(user_name),
                '-c', 'user.email=%s' % safe_str(user_email),
                'merge', '--no-ff', '-m', safe_str(merge_message)]
         cmd.extend(heads)
         try:
             output = self.run_git_command(cmd, fail_on_stderr=False)
         except RepositoryError:
             # Cleanup any merge leftovers
             self.run_git_command(['merge', '--abort'], fail_on_stderr=False)
             raise
 
     def _local_push(
             self, source_branch, repository_path, target_branch,
             enable_hooks=False, rc_scm_data=None):
         """
         Push the source_branch to the given repository and target_branch.
 
         Currently it if the target_branch is not master and the target repo is
         empty, the push will work, but then GitRepository won't be able to find
         the pushed branch or the commits. As the HEAD will be corrupted (i.e.,
         pointing to master, which does not exist).
 
         It does not run the hooks in the target repo.
         """
         # TODO(skreft): deal with the case in which the target repo is empty,
         # and the target_branch is not master.
         target_repo = GitRepository(repository_path)
         if (not target_repo.bare and
                 target_repo._current_branch() == target_branch):
             # Git prevents pushing to the checked out branch, so simulate it by
             # pulling into the target repository.
             target_repo._local_pull(self.path, source_branch)
         else:
             cmd = ['push', os.path.abspath(repository_path),
                    '%s:%s' % (source_branch, target_branch)]
             gitenv = {}
             if rc_scm_data:
                 gitenv.update({'RC_SCM_DATA': rc_scm_data})
 
             if not enable_hooks:
                 gitenv['RC_SKIP_HOOKS'] = '1'
             self.run_git_command(cmd, fail_on_stderr=False, extra_env=gitenv)
 
     def _get_new_pr_branch(self, source_branch, target_branch):
         prefix = 'pr_%s-%s_' % (source_branch, target_branch)
         pr_branches = []
         for branch in self.branches:
             if branch.startswith(prefix):
                 pr_branches.append(int(branch[len(prefix):]))
 
         if not pr_branches:
             branch_id = 0
         else:
             branch_id = max(pr_branches) + 1
 
         return '%s%d' % (prefix, branch_id)
 
     def _maybe_prepare_merge_workspace(
             self, repo_id, workspace_id, target_ref, source_ref):
         shadow_repository_path = self._get_shadow_repository_path(
-            repo_id, workspace_id)
+            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:
             log.exception('Failure when doing local merge on git shadow repo')
             merge_possible = False
             merge_failure_reason = MergeFailureReason.MERGE_FAILED
 
         if merge_possible and not dry_run:
             try:
                 shadow_repo._local_push(
                     pr_branch, self.path, target_ref.name, enable_hooks=True,
                     rc_scm_data=self.config.get('rhodecode', 'RC_SCM_DATA'))
                 merge_succeeded = True
             except RepositoryError:
                 log.exception(
                     'Failure when doing local push from the shadow '
                     'repository to the target repository at %s.', self.path)
                 merge_succeeded = False
                 merge_failure_reason = MergeFailureReason.PUSH_FAILED
                 metadata['target'] = 'git shadow repo'
                 metadata['merge_commit'] = pr_branch
         else:
             merge_succeeded = False
 
         return MergeResponse(
             merge_possible, merge_succeeded, merge_ref, merge_failure_reason,
             metadata=metadata)

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

@@ -1,952 +1,952 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 HG repository module
 """
 import os
 import logging
 import binascii
 import urllib
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from rhodecode.lib.compat import OrderedDict
 from rhodecode.lib.datelib import (
     date_to_timestamp_plus_offset, utcdate_fromtimestamp, makedate)
 from rhodecode.lib.utils import safe_unicode, safe_str
 from rhodecode.lib.utils2 import CachedProperty
 from rhodecode.lib.vcs import connection, exceptions
 from rhodecode.lib.vcs.backends.base import (
     BaseRepository, CollectionGenerator, Config, MergeResponse,
     MergeFailureReason, Reference, BasePathPermissionChecker)
 from rhodecode.lib.vcs.backends.hg.commit import MercurialCommit
 from rhodecode.lib.vcs.backends.hg.diff import MercurialDiff
 from rhodecode.lib.vcs.backends.hg.inmemory import MercurialInMemoryCommit
 from rhodecode.lib.vcs.exceptions import (
     EmptyRepositoryError, RepositoryError, TagAlreadyExistError,
     TagDoesNotExistError, CommitDoesNotExistError, SubrepoMergeError)
 from rhodecode.lib.vcs.compat import configparser
 
 hexlify = binascii.hexlify
 nullid = "\0" * 20
 
 log = logging.getLogger(__name__)
 
 
 class MercurialRepository(BaseRepository):
     """
     Mercurial repository backend
     """
     DEFAULT_BRANCH_NAME = 'default'
 
     def __init__(self, repo_path, config=None, create=False, src_url=None,
                  do_workspace_checkout=False, with_wire=None, bare=False):
         """
         Raises RepositoryError if repository could not be find at the given
         ``repo_path``.
 
         :param repo_path: local path of the repository
         :param config: config object containing the repo configuration
         :param create=False: if set to True, would try to create repository if
            it does not exist rather than raising exception
         :param src_url=None: would try to clone repository from given location
         :param do_workspace_checkout=False: sets update of working copy after
            making a clone
         :param bare: not used, compatible with other VCS
         """
 
         self.path = safe_str(os.path.abspath(repo_path))
         # mercurial since 4.4.X requires certain configuration to be present
         # because sometimes we init the repos with config we need to meet
         # special requirements
         self.config = config if config else self.get_default_config(
             default=[('extensions', 'largefiles', '1')])
         self.with_wire = with_wire or {"cache": False}  # default should not use cache
 
         self._init_repo(create, src_url, do_workspace_checkout)
 
         # caches
         self._commit_ids = {}
 
     @LazyProperty
     def _remote(self):
         repo_id = self.path
         return connection.Hg(self.path, repo_id, self.config, with_wire=self.with_wire)
 
     @CachedProperty
     def commit_ids(self):
         """
         Returns list of commit ids, in ascending order.  Being lazy
         attribute allows external tools to inject shas from cache.
         """
         commit_ids = self._get_all_commit_ids()
         self._rebuild_cache(commit_ids)
         return commit_ids
 
     def _rebuild_cache(self, commit_ids):
         self._commit_ids = dict((commit_id, index)
                                 for index, commit_id in enumerate(commit_ids))
 
     @CachedProperty
     def branches(self):
         return self._get_branches()
 
     @CachedProperty
     def branches_closed(self):
         return self._get_branches(active=False, closed=True)
 
     @CachedProperty
     def branches_all(self):
         all_branches = {}
         all_branches.update(self.branches)
         all_branches.update(self.branches_closed)
         return all_branches
 
     def _get_branches(self, active=True, closed=False):
         """
         Gets branches for this repository
         Returns only not closed active branches by default
 
         :param active: return also active branches
         :param closed: return also closed branches
 
         """
         if self.is_empty():
             return {}
 
         def get_name(ctx):
             return ctx[0]
 
         _branches = [(safe_unicode(n), hexlify(h),) for n, h in
                      self._remote.branches(active, closed).items()]
 
         return OrderedDict(sorted(_branches, key=get_name, reverse=False))
 
     @CachedProperty
     def tags(self):
         """
         Gets tags for this repository
         """
         return self._get_tags()
 
     def _get_tags(self):
         if self.is_empty():
             return {}
 
         def get_name(ctx):
             return ctx[0]
 
         _tags = [(safe_unicode(n), hexlify(h),) for n, h in
                  self._remote.tags().items()]
 
         return OrderedDict(sorted(_tags, key=get_name, reverse=True))
 
     def tag(self, name, user, commit_id=None, message=None, date=None, **kwargs):
         """
         Creates and returns a tag for the given ``commit_id``.
 
         :param name: name for new tag
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param commit_id: commit id for which new tag would be created
         :param message: message of the tag's commit
         :param date: date of tag's commit
 
         :raises TagAlreadyExistError: if tag with same name already exists
         """
         if name in self.tags:
             raise TagAlreadyExistError("Tag %s already exists" % name)
 
         commit = self.get_commit(commit_id=commit_id)
         local = kwargs.setdefault('local', False)
 
         if message is None:
             message = "Added tag %s for commit %s" % (name, commit.short_id)
 
         date, tz = date_to_timestamp_plus_offset(date)
 
         self._remote.tag(name, commit.raw_id, message, local, user, date, tz)
         self._remote.invalidate_vcs_cache()
 
         # Reinitialize tags
         self._invalidate_prop_cache('tags')
         tag_id = self.tags[name]
 
         return self.get_commit(commit_id=tag_id)
 
     def remove_tag(self, name, user, message=None, date=None):
         """
         Removes tag with the given `name`.
 
         :param name: name of the tag to be removed
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param message: message of the tag's removal commit
         :param date: date of tag's removal commit
 
         :raises TagDoesNotExistError: if tag with given name does not exists
         """
         if name not in self.tags:
             raise TagDoesNotExistError("Tag %s does not exist" % name)
 
         if message is None:
             message = "Removed tag %s" % name
         local = False
 
         date, tz = date_to_timestamp_plus_offset(date)
 
         self._remote.tag(name, nullid, message, local, user, date, tz)
         self._remote.invalidate_vcs_cache()
         self._invalidate_prop_cache('tags')
 
     @LazyProperty
     def bookmarks(self):
         """
         Gets bookmarks for this repository
         """
         return self._get_bookmarks()
 
     def _get_bookmarks(self):
         if self.is_empty():
             return {}
 
         def get_name(ctx):
             return ctx[0]
 
         _bookmarks = [
             (safe_unicode(n), hexlify(h)) for n, h in
             self._remote.bookmarks().items()]
 
         return OrderedDict(sorted(_bookmarks, key=get_name))
 
     def _get_all_commit_ids(self):
         return self._remote.get_all_commit_ids('visible')
 
     def get_diff(
             self, commit1, commit2, path='', ignore_whitespace=False,
             context=3, path1=None):
         """
         Returns (git like) *diff*, as plain text. Shows changes introduced by
         `commit2` since `commit1`.
 
         :param commit1: Entry point from which diff is shown. Can be
           ``self.EMPTY_COMMIT`` - in this case, patch showing all
           the changes since empty state of the repository until `commit2`
         :param commit2: Until which commit changes should be shown.
         :param ignore_whitespace: If set to ``True``, would not show whitespace
           changes. Defaults to ``False``.
         :param context: How many lines before/after changed lines should be
           shown. Defaults to ``3``.
         """
         self._validate_diff_commits(commit1, commit2)
         if path1 is not None and path1 != path:
             raise ValueError("Diff of two different paths not supported.")
 
         if path:
             file_filter = [self.path, path]
         else:
             file_filter = None
 
         diff = self._remote.diff(
             commit1.raw_id, commit2.raw_id, file_filter=file_filter,
             opt_git=True, opt_ignorews=ignore_whitespace,
             context=context)
         return MercurialDiff(diff)
 
     def strip(self, commit_id, branch=None):
         self._remote.strip(commit_id, update=False, backup="none")
 
         self._remote.invalidate_vcs_cache()
         # clear cache
         self._invalidate_prop_cache('commit_ids')
 
         return len(self.commit_ids)
 
     def verify(self):
         verify = self._remote.verify()
 
         self._remote.invalidate_vcs_cache()
         return verify
 
     def hg_update_cache(self):
         update_cache = self._remote.hg_update_cache()
 
         self._remote.invalidate_vcs_cache()
         return update_cache
 
     def get_common_ancestor(self, commit_id1, commit_id2, repo2):
         if commit_id1 == commit_id2:
             return commit_id1
 
         ancestors = self._remote.revs_from_revspec(
             "ancestor(id(%s), id(%s))", commit_id1, commit_id2,
             other_path=repo2.path)
         return repo2[ancestors[0]].raw_id if ancestors else None
 
     def compare(self, commit_id1, commit_id2, repo2, merge, pre_load=None):
         if commit_id1 == commit_id2:
             commits = []
         else:
             if merge:
                 indexes = self._remote.revs_from_revspec(
                     "ancestors(id(%s)) - ancestors(id(%s)) - id(%s)",
                     commit_id2, commit_id1, commit_id1, other_path=repo2.path)
             else:
                 indexes = self._remote.revs_from_revspec(
                     "id(%s)..id(%s) - id(%s)", commit_id1, commit_id2,
                     commit_id1, other_path=repo2.path)
 
             commits = [repo2.get_commit(commit_idx=idx, pre_load=pre_load)
                        for idx in indexes]
 
         return commits
 
     @staticmethod
     def check_url(url, config):
         """
         Function will check given url and try to verify if it's a valid
         link. Sometimes it may happened that mercurial will issue basic
         auth request that can cause whole API to hang when used from python
         or other external calls.
 
         On failures it'll raise urllib2.HTTPError, exception is also thrown
         when the return code is non 200
         """
         # check first if it's not an local url
         if os.path.isdir(url) or url.startswith('file:'):
             return True
 
         # Request the _remote to verify the url
         return connection.Hg.check_url(url, config.serialize())
 
     @staticmethod
     def is_valid_repository(path):
         return os.path.isdir(os.path.join(path, '.hg'))
 
     def _init_repo(self, create, src_url=None, do_workspace_checkout=False):
         """
         Function will check for mercurial repository in given path. If there
         is no repository in that path it will raise an exception unless
         `create` parameter is set to True - in that case repository would
         be created.
 
         If `src_url` is given, would try to clone repository from the
         location at given clone_point. Additionally it'll make update to
         working copy accordingly to `do_workspace_checkout` flag.
         """
         if create and os.path.exists(self.path):
             raise RepositoryError(
                 "Cannot create repository at %s, location already exist"
                 % self.path)
 
         if src_url:
             url = str(self._get_url(src_url))
             MercurialRepository.check_url(url, self.config)
 
             self._remote.clone(url, self.path, do_workspace_checkout)
 
             # Don't try to create if we've already cloned repo
             create = False
 
         if create:
             os.makedirs(self.path, mode=0o755)
         self._remote.localrepository(create)
 
     @LazyProperty
     def in_memory_commit(self):
         return MercurialInMemoryCommit(self)
 
     @LazyProperty
     def description(self):
         description = self._remote.get_config_value(
             'web', 'description', untrusted=True)
         return safe_unicode(description or self.DEFAULT_DESCRIPTION)
 
     @LazyProperty
     def contact(self):
         contact = (
             self._remote.get_config_value("web", "contact") or
             self._remote.get_config_value("ui", "username"))
         return safe_unicode(contact or self.DEFAULT_CONTACT)
 
     @LazyProperty
     def last_change(self):
         """
         Returns last change made on this repository as
         `datetime.datetime` object.
         """
         try:
             return self.get_commit().date
         except RepositoryError:
             tzoffset = makedate()[1]
             return utcdate_fromtimestamp(self._get_fs_mtime(), tzoffset)
 
     def _get_fs_mtime(self):
         # fallback to filesystem
         cl_path = os.path.join(self.path, '.hg', "00changelog.i")
         st_path = os.path.join(self.path, '.hg', "store")
         if os.path.exists(cl_path):
             return os.stat(cl_path).st_mtime
         else:
             return os.stat(st_path).st_mtime
 
     def _get_url(self, url):
         """
         Returns normalized url. If schema is not given, would fall
         to filesystem
         (``file:///``) schema.
         """
         url = url.encode('utf8')
         if url != 'default' and '://' not in url:
             url = "file:" + urllib.pathname2url(url)
         return url
 
     def get_hook_location(self):
         """
         returns absolute path to location where hooks are stored
         """
         return os.path.join(self.path, '.hg', '.hgrc')
 
     def get_commit(self, commit_id=None, commit_idx=None, pre_load=None, translate_tag=None):
         """
         Returns ``MercurialCommit`` object representing repository's
         commit at the given `commit_id` or `commit_idx`.
         """
         if self.is_empty():
             raise EmptyRepositoryError("There are no commits yet")
 
         if commit_id is not None:
             self._validate_commit_id(commit_id)
             try:
                 # we have cached idx, use it without contacting the remote
                 idx = self._commit_ids[commit_id]
                 return MercurialCommit(self, commit_id, idx, pre_load=pre_load)
             except KeyError:
                 pass
 
         elif commit_idx is not None:
             self._validate_commit_idx(commit_idx)
             try:
                 _commit_id = self.commit_ids[commit_idx]
                 if commit_idx < 0:
                     commit_idx = self.commit_ids.index(_commit_id)
 
                 return MercurialCommit(self, _commit_id, commit_idx, pre_load=pre_load)
             except IndexError:
                 commit_id = commit_idx
         else:
             commit_id = "tip"
 
         if isinstance(commit_id, unicode):
             commit_id = safe_str(commit_id)
 
         try:
             raw_id, idx = self._remote.lookup(commit_id, both=True)
         except CommitDoesNotExistError:
             msg = "Commit {} does not exist for `{}`".format(
                 *map(safe_str, [commit_id, self.name]))
             raise CommitDoesNotExistError(msg)
 
         return MercurialCommit(self, raw_id, idx, pre_load=pre_load)
 
     def get_commits(
             self, start_id=None, end_id=None, start_date=None, end_date=None,
             branch_name=None, show_hidden=False, pre_load=None, translate_tags=None):
         """
         Returns generator of ``MercurialCommit`` objects from start to end
         (both are inclusive)
 
         :param start_id: None, str(commit_id)
         :param end_id: None, str(commit_id)
         :param start_date: if specified, commits with commit date less than
           ``start_date`` would be filtered out from returned set
         :param end_date: if specified, commits with commit date greater than
           ``end_date`` would be filtered out from returned set
         :param branch_name: if specified, commits not reachable from given
           branch would be filtered out from returned set
         :param show_hidden: Show hidden commits such as obsolete or hidden from
             Mercurial evolve
         :raise BranchDoesNotExistError: If given ``branch_name`` does not
             exist.
         :raise CommitDoesNotExistError: If commit for given ``start`` or
           ``end`` could not be found.
         """
         # actually we should check now if it's not an empty repo
         if self.is_empty():
             raise EmptyRepositoryError("There are no commits yet")
         self._validate_branch_name(branch_name)
 
         branch_ancestors = False
         if start_id is not None:
             self._validate_commit_id(start_id)
             c_start = self.get_commit(commit_id=start_id)
             start_pos = self._commit_ids[c_start.raw_id]
         else:
             start_pos = None
 
         if end_id is not None:
             self._validate_commit_id(end_id)
             c_end = self.get_commit(commit_id=end_id)
             end_pos = max(0, self._commit_ids[c_end.raw_id])
         else:
             end_pos = None
 
         if None not in [start_id, end_id] and start_pos > end_pos:
             raise RepositoryError(
                 "Start commit '%s' cannot be after end commit '%s'" %
                 (start_id, end_id))
 
         if end_pos is not None:
             end_pos += 1
 
         commit_filter = []
 
         if branch_name and not branch_ancestors:
             commit_filter.append('branch("%s")' % (branch_name,))
         elif branch_name and branch_ancestors:
             commit_filter.append('ancestors(branch("%s"))' % (branch_name,))
 
         if start_date and not end_date:
             commit_filter.append('date(">%s")' % (start_date,))
         if end_date and not start_date:
             commit_filter.append('date("<%s")' % (end_date,))
         if start_date and end_date:
             commit_filter.append(
                 'date(">%s") and date("<%s")' % (start_date, end_date))
 
         if not show_hidden:
             commit_filter.append('not obsolete()')
             commit_filter.append('not hidden()')
 
         # TODO: johbo: Figure out a simpler way for this solution
         collection_generator = CollectionGenerator
         if commit_filter:
             commit_filter = ' and '.join(map(safe_str, commit_filter))
             revisions = self._remote.rev_range([commit_filter])
             collection_generator = MercurialIndexBasedCollectionGenerator
         else:
             revisions = self.commit_ids
 
         if start_pos or end_pos:
             revisions = revisions[start_pos:end_pos]
 
         return collection_generator(self, revisions, pre_load=pre_load)
 
     def pull(self, url, commit_ids=None):
         """
         Pull changes from external location.
 
         :param commit_ids: Optional. Can be set to a list of commit ids
            which shall be pulled from the other repository.
         """
         url = self._get_url(url)
         self._remote.pull(url, commit_ids=commit_ids)
         self._remote.invalidate_vcs_cache()
 
     def fetch(self, url, commit_ids=None):
         """
         Backward compatibility with GIT fetch==pull
         """
         return self.pull(url, commit_ids=commit_ids)
 
     def push(self, url):
         url = self._get_url(url)
         self._remote.sync_push(url)
 
     def _local_clone(self, clone_path):
         """
         Create a local clone of the current repo.
         """
         self._remote.clone(self.path, clone_path, update_after_clone=True,
                            hooks=False)
 
     def _update(self, revision, clean=False):
         """
         Update the working copy to the specified revision.
         """
         log.debug('Doing checkout to commit: `%s` for %s', revision, self)
         self._remote.update(revision, clean=clean)
 
     def _identify(self):
         """
         Return the current state of the working directory.
         """
         return self._remote.identify().strip().rstrip('+')
 
     def _heads(self, branch=None):
         """
         Return the commit ids of the repository heads.
         """
         return self._remote.heads(branch=branch).strip().split(' ')
 
     def _ancestor(self, revision1, revision2):
         """
         Return the common ancestor of the two revisions.
         """
         return self._remote.ancestor(revision1, revision2)
 
     def _local_push(
             self, revision, repository_path, push_branches=False,
             enable_hooks=False):
         """
         Push the given revision to the specified repository.
 
         :param push_branches: allow to create branches in the target repo.
         """
         self._remote.push(
             [revision], repository_path, hooks=enable_hooks,
             push_branches=push_branches)
 
     def _local_merge(self, target_ref, merge_message, user_name, user_email,
                      source_ref, use_rebase=False, dry_run=False):
         """
         Merge the given source_revision into the checked out revision.
 
         Returns the commit id of the merge and a boolean indicating if the
         commit needs to be pushed.
         """
         self._update(target_ref.commit_id, clean=True)
 
         ancestor = self._ancestor(target_ref.commit_id, source_ref.commit_id)
         is_the_same_branch = self._is_the_same_branch(target_ref, source_ref)
 
         if ancestor == source_ref.commit_id:
             # Nothing to do, the changes were already integrated
             return target_ref.commit_id, False
 
         elif ancestor == target_ref.commit_id and is_the_same_branch:
             # In this case we should force a commit message
             return source_ref.commit_id, True
 
         if use_rebase:
             try:
                 bookmark_name = 'rcbook%s%s' % (source_ref.commit_id,
                                                 target_ref.commit_id)
                 self.bookmark(bookmark_name, revision=source_ref.commit_id)
                 self._remote.rebase(
                     source=source_ref.commit_id, dest=target_ref.commit_id)
                 self._remote.invalidate_vcs_cache()
                 self._update(bookmark_name, clean=True)
                 return self._identify(), True
             except RepositoryError:
                 # The rebase-abort may raise another exception which 'hides'
                 # the original one, therefore we log it here.
                 log.exception('Error while rebasing shadow repo during merge.')
 
                 # Cleanup any rebase leftovers
                 self._remote.invalidate_vcs_cache()
                 self._remote.rebase(abort=True)
                 self._remote.invalidate_vcs_cache()
                 self._remote.update(clean=True)
                 raise
         else:
             try:
                 self._remote.merge(source_ref.commit_id)
                 self._remote.invalidate_vcs_cache()
                 self._remote.commit(
                     message=safe_str(merge_message),
                     username=safe_str('%s <%s>' % (user_name, user_email)))
                 self._remote.invalidate_vcs_cache()
                 return self._identify(), True
             except RepositoryError:
                 # Cleanup any merge leftovers
                 self._remote.update(clean=True)
                 raise
 
     def _local_close(self, target_ref, user_name, user_email,
                      source_ref, close_message=''):
         """
         Close the branch of the given source_revision
 
         Returns the commit id of the close and a boolean indicating if the
         commit needs to be pushed.
         """
         self._update(source_ref.commit_id)
         message = close_message or "Closing branch: `{}`".format(source_ref.name)
         try:
             self._remote.commit(
                 message=safe_str(message),
                 username=safe_str('%s <%s>' % (user_name, user_email)),
                 close_branch=True)
             self._remote.invalidate_vcs_cache()
             return self._identify(), True
         except RepositoryError:
             # Cleanup any commit leftovers
             self._remote.update(clean=True)
             raise
 
     def _is_the_same_branch(self, target_ref, source_ref):
         return (
             self._get_branch_name(target_ref) ==
             self._get_branch_name(source_ref))
 
     def _get_branch_name(self, ref):
         if ref.type == 'branch':
             return ref.name
         return self._remote.ctx_branch(ref.commit_id)
 
     def _maybe_prepare_merge_workspace(
             self, repo_id, workspace_id, unused_target_ref, unused_source_ref):
         shadow_repository_path = self._get_shadow_repository_path(
-            repo_id, workspace_id)
+            self.path, repo_id, workspace_id)
         if not os.path.exists(shadow_repository_path):
             self._local_clone(shadow_repository_path)
             log.debug(
                 'Prepared shadow repository in %s', shadow_repository_path)
 
         return shadow_repository_path
 
     def _merge_repo(self, repo_id, workspace_id, target_ref,
                     source_repo, source_ref, merge_message,
                     merger_name, merger_email, dry_run=False,
                     use_rebase=False, close_branch=False):
 
         log.debug('Executing merge_repo with %s strategy, dry_run mode:%s',
                   'rebase' if use_rebase else 'merge', dry_run)
         if target_ref.commit_id not in self._heads():
             return MergeResponse(
                 False, False, None, MergeFailureReason.TARGET_IS_NOT_HEAD,
                 metadata={'target_ref': target_ref})
 
         try:
             if target_ref.type == 'branch' and len(self._heads(target_ref.name)) != 1:
                 heads = '\n,'.join(self._heads(target_ref.name))
                 metadata = {
                     'target_ref': target_ref,
                     'source_ref': source_ref,
                     'heads': heads
                 }
                 return MergeResponse(
                     False, False, None,
                     MergeFailureReason.HG_TARGET_HAS_MULTIPLE_HEADS,
                     metadata=metadata)
         except CommitDoesNotExistError:
             log.exception('Failure when looking up branch heads on hg target')
             return MergeResponse(
                 False, False, None, MergeFailureReason.MISSING_TARGET_REF,
                 metadata={'target_ref': target_ref})
 
         shadow_repository_path = self._maybe_prepare_merge_workspace(
             repo_id, workspace_id, target_ref, source_ref)
         shadow_repo = self.get_shadow_instance(shadow_repository_path)
 
         log.debug('Pulling in target reference %s', target_ref)
         self._validate_pull_reference(target_ref)
         shadow_repo._local_pull(self.path, target_ref)
 
         try:
             log.debug('Pulling in source reference %s', source_ref)
             source_repo._validate_pull_reference(source_ref)
             shadow_repo._local_pull(source_repo.path, source_ref)
         except CommitDoesNotExistError:
             log.exception('Failure when doing local pull on hg shadow repo')
             return MergeResponse(
                 False, False, None, MergeFailureReason.MISSING_SOURCE_REF,
                 metadata={'source_ref': source_ref})
 
         merge_ref = None
         merge_commit_id = None
         close_commit_id = None
         merge_failure_reason = MergeFailureReason.NONE
         metadata = {}
 
         # enforce that close branch should be used only in case we source from
         # an actual Branch
         close_branch = close_branch and source_ref.type == 'branch'
 
         # don't allow to close branch if source and target are the same
         close_branch = close_branch and source_ref.name != target_ref.name
 
         needs_push_on_close = False
         if close_branch and not use_rebase and not dry_run:
             try:
                 close_commit_id, needs_push_on_close = shadow_repo._local_close(
                     target_ref, merger_name, merger_email, source_ref)
                 merge_possible = True
             except RepositoryError:
                 log.exception('Failure when doing close branch on '
                               'shadow repo: %s', shadow_repo)
                 merge_possible = False
                 merge_failure_reason = MergeFailureReason.MERGE_FAILED
         else:
             merge_possible = True
 
         needs_push = False
         if merge_possible:
             try:
                 merge_commit_id, needs_push = shadow_repo._local_merge(
                     target_ref, merge_message, merger_name, merger_email,
                     source_ref, use_rebase=use_rebase, dry_run=dry_run)
                 merge_possible = True
 
                 # read the state of the close action, if it
                 # maybe required a push
                 needs_push = needs_push or needs_push_on_close
 
                 # Set a bookmark pointing to the merge commit. This bookmark
                 # may be used to easily identify the last successful merge
                 # commit in the shadow repository.
                 shadow_repo.bookmark('pr-merge', revision=merge_commit_id)
                 merge_ref = Reference('book', 'pr-merge', merge_commit_id)
             except SubrepoMergeError:
                 log.exception(
                     'Subrepo merge error during local merge on hg shadow repo.')
                 merge_possible = False
                 merge_failure_reason = MergeFailureReason.SUBREPO_MERGE_FAILED
                 needs_push = False
             except RepositoryError:
                 log.exception('Failure when doing local merge on hg shadow repo')
                 merge_possible = False
                 merge_failure_reason = MergeFailureReason.MERGE_FAILED
                 needs_push = False
 
         if merge_possible and not dry_run:
             if needs_push:
                 # In case the target is a bookmark, update it, so after pushing
                 # the bookmarks is also updated in the target.
                 if target_ref.type == 'book':
                     shadow_repo.bookmark(
                         target_ref.name, revision=merge_commit_id)
                 try:
                     shadow_repo_with_hooks = self.get_shadow_instance(
                         shadow_repository_path,
                         enable_hooks=True)
                     # This is the actual merge action, we push from shadow
                     # into origin.
                     # Note: the push_branches option will push any new branch
                     # defined in the source repository to the target. This may
                     # be dangerous as branches are permanent in Mercurial.
                     # This feature was requested in issue #441.
                     shadow_repo_with_hooks._local_push(
                         merge_commit_id, self.path, push_branches=True,
                         enable_hooks=True)
 
                     # maybe we also need to push the close_commit_id
                     if close_commit_id:
                         shadow_repo_with_hooks._local_push(
                             close_commit_id, self.path, push_branches=True,
                             enable_hooks=True)
                     merge_succeeded = True
                 except RepositoryError:
                     log.exception(
                         'Failure when doing local push from the shadow '
                         'repository to the target repository at %s.', self.path)
                     merge_succeeded = False
                     merge_failure_reason = MergeFailureReason.PUSH_FAILED
                     metadata['target'] = 'hg shadow repo'
                     metadata['merge_commit'] = merge_commit_id
             else:
                 merge_succeeded = True
         else:
             merge_succeeded = False
 
         return MergeResponse(
             merge_possible, merge_succeeded, merge_ref, merge_failure_reason,
             metadata=metadata)
 
     def get_shadow_instance(self, shadow_repository_path, enable_hooks=False, cache=False):
         config = self.config.copy()
         if not enable_hooks:
             config.clear_section('hooks')
         return MercurialRepository(shadow_repository_path, config, with_wire={"cache": cache})
 
     def _validate_pull_reference(self, reference):
         if not (reference.name in self.bookmarks or
                 reference.name in self.branches or
                 self.get_commit(reference.commit_id)):
             raise CommitDoesNotExistError(
                 'Unknown branch, bookmark or commit id')
 
     def _local_pull(self, repository_path, reference):
         """
         Fetch a branch, bookmark or commit from a local repository.
         """
         repository_path = os.path.abspath(repository_path)
         if repository_path == self.path:
             raise ValueError('Cannot pull from the same repository')
 
         reference_type_to_option_name = {
             'book': 'bookmark',
             'branch': 'branch',
         }
         option_name = reference_type_to_option_name.get(
             reference.type, 'revision')
 
         if option_name == 'revision':
             ref = reference.commit_id
         else:
             ref = reference.name
 
         options = {option_name: [ref]}
         self._remote.pull_cmd(repository_path, hooks=False, **options)
         self._remote.invalidate_vcs_cache()
 
     def bookmark(self, bookmark, revision=None):
         if isinstance(bookmark, unicode):
             bookmark = safe_str(bookmark)
         self._remote.bookmark(bookmark, revision=revision)
         self._remote.invalidate_vcs_cache()
 
     def get_path_permissions(self, username):
         hgacl_file = os.path.join(self.path, '.hg/hgacl')
 
         def read_patterns(suffix):
             svalue = None
             for section, option in [
                 ('narrowacl', username + suffix),
                 ('narrowacl', 'default' + suffix),
                 ('narrowhgacl', username + suffix),
                 ('narrowhgacl', 'default' + suffix)
             ]:
                 try:
                     svalue = hgacl.get(section, option)
                     break  # stop at the first value we find
                 except configparser.NoOptionError:
                     pass
             if not svalue:
                 return None
             result = ['/']
             for pattern in svalue.split():
                 result.append(pattern)
                 if '*' not in pattern and '?' not in pattern:
                     result.append(pattern + '/*')
             return result
 
         if os.path.exists(hgacl_file):
             try:
                 hgacl = configparser.RawConfigParser()
                 hgacl.read(hgacl_file)
 
                 includes = read_patterns('.includes')
                 excludes = read_patterns('.excludes')
                 return BasePathPermissionChecker.create_from_patterns(
                     includes, excludes)
             except BaseException as e:
                 msg = 'Cannot read ACL settings from {} on {}: {}'.format(
                     hgacl_file, self.name, e)
                 raise exceptions.RepositoryRequirementError(msg)
         else:
             return None
 
 
 class MercurialIndexBasedCollectionGenerator(CollectionGenerator):
 
     def _commit_factory(self, commit_id):
         return self.repo.get_commit(
             commit_idx=commit_id, pre_load=self.pre_load)

rhodecode/model/db.py

@@ -1,5208 +1,5213 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 Database Models for RhodeCode Enterprise
 """
 
 import re
 import os
 import time
 import string
 import hashlib
 import logging
 import datetime
 import uuid
 import warnings
 import ipaddress
 import functools
 import traceback
 import collections
 
 from sqlalchemy import (
     or_, and_, not_, func, TypeDecorator, event,
     Index, Sequence, UniqueConstraint, ForeignKey, CheckConstraint, Column,
     Boolean, String, Unicode, UnicodeText, DateTime, Integer, LargeBinary,
     Text, Float, PickleType)
 from sqlalchemy.sql.expression import true, false, case
 from sqlalchemy.sql.functions import coalesce, count  # pragma: no cover
 from sqlalchemy.orm import (
     relationship, joinedload, class_mapper, validates, aliased)
 from sqlalchemy.ext.declarative import declared_attr
 from sqlalchemy.ext.hybrid import hybrid_property
 from sqlalchemy.exc import IntegrityError  # pragma: no cover
 from sqlalchemy.dialects.mysql import LONGTEXT
 from zope.cachedescriptors.property import Lazy as LazyProperty
 from pyramid import compat
 from pyramid.threadlocal import get_current_request
 from webhelpers.text import collapse, remove_formatting
 
 from rhodecode.translation import _
 from rhodecode.lib.vcs import get_vcs_instance
 from rhodecode.lib.vcs.backends.base import EmptyCommit, Reference
 from rhodecode.lib.utils2 import (
     str2bool, safe_str, get_commit_safe, safe_unicode, sha1_safe,
     time_to_datetime, aslist, Optional, safe_int, get_clone_url, AttributeDict,
     glob2re, StrictAttributeDict, cleaned_uri, datetime_to_time, OrderedDefaultDict)
 from rhodecode.lib.jsonalchemy import MutationObj, MutationList, JsonType, \
     JsonRaw
 from rhodecode.lib.ext_json import json
 from rhodecode.lib.caching_query import FromCache
 from rhodecode.lib.encrypt import AESCipher, validate_and_get_enc_data
 from rhodecode.lib.encrypt2 import Encryptor
 from rhodecode.model.meta import Base, Session
 
 URL_SEP = '/'
 log = logging.getLogger(__name__)
 
 # =============================================================================
 # BASE CLASSES
 # =============================================================================
 
 # this is propagated from .ini file rhodecode.encrypted_values.secret or
 # beaker.session.secret if first is not set.
 # and initialized at environment.py
 ENCRYPTION_KEY = None
 
 # used to sort permissions by types, '#' used here is not allowed to be in
 # usernames, and it's very early in sorted string.printable table.
 PERMISSION_TYPE_SORT = {
     'admin': '####',
     'write': '###',
     'read':  '##',
     'none':  '#',
 }
 
 
 def display_user_sort(obj):
     """
     Sort function used to sort permissions in .permissions() function of
     Repository, RepoGroup, UserGroup. Also it put the default user in front
     of all other resources
     """
 
     if obj.username == User.DEFAULT_USER:
         return '#####'
     prefix = PERMISSION_TYPE_SORT.get(obj.permission.split('.')[-1], '')
     return prefix + obj.username
 
 
 def display_user_group_sort(obj):
     """
     Sort function used to sort permissions in .permissions() function of
     Repository, RepoGroup, UserGroup. Also it put the default user in front
     of all other resources
     """
 
     prefix = PERMISSION_TYPE_SORT.get(obj.permission.split('.')[-1], '')
     return prefix + obj.users_group_name
 
 
 def _hash_key(k):
     return sha1_safe(k)
 
 
 def in_filter_generator(qry, items, limit=500):
     """
     Splits IN() into multiple with OR
     e.g.::
     cnt = Repository.query().filter(
         or_(
             *in_filter_generator(Repository.repo_id, range(100000))
         )).count()
     """
     if not items:
         # empty list will cause empty query which might cause security issues
         # this can lead to hidden unpleasant results
         items = [-1]
 
     parts = []
     for chunk in xrange(0, len(items), limit):
         parts.append(
             qry.in_(items[chunk: chunk + limit])
         )
 
     return parts
 
 
 base_table_args = {
     'extend_existing': True,
     'mysql_engine': 'InnoDB',
     'mysql_charset': 'utf8',
     'sqlite_autoincrement': True
 }
 
 
 class EncryptedTextValue(TypeDecorator):
     """
     Special column for encrypted long text data, use like::
 
         value = Column("encrypted_value", EncryptedValue(), nullable=False)
 
     This column is intelligent so if value is in unencrypted form it return
     unencrypted form, but on save it always encrypts
     """
     impl = Text
 
     def process_bind_param(self, value, dialect):
         """
         Setter for storing value
         """
         import rhodecode
         if not value:
             return value
 
         # protect against double encrypting if values is already encrypted
         if value.startswith('enc$aes$') \
                 or value.startswith('enc$aes_hmac$') \
                 or value.startswith('enc2$'):
             raise ValueError('value needs to be in unencrypted format, '
                              'ie. not starting with enc$ or enc2$')
 
         algo = rhodecode.CONFIG.get('rhodecode.encrypted_values.algorithm') or 'aes'
         if algo == 'aes':
             return 'enc$aes_hmac$%s' % AESCipher(ENCRYPTION_KEY, hmac=True).encrypt(value)
         elif algo == 'fernet':
             return Encryptor(ENCRYPTION_KEY).encrypt(value)
         else:
             ValueError('Bad encryption algorithm, should be fernet or aes, got: {}'.format(algo))
 
     def process_result_value(self, value, dialect):
         """
         Getter for retrieving value
         """
 
         import rhodecode
         if not value:
             return value
 
         algo = rhodecode.CONFIG.get('rhodecode.encrypted_values.algorithm') or 'aes'
         enc_strict_mode = str2bool(rhodecode.CONFIG.get('rhodecode.encrypted_values.strict') or True)
         if algo == 'aes':
             decrypted_data = validate_and_get_enc_data(value, ENCRYPTION_KEY, enc_strict_mode)
         elif algo == 'fernet':
             return Encryptor(ENCRYPTION_KEY).decrypt(value)
         else:
             ValueError('Bad encryption algorithm, should be fernet or aes, got: {}'.format(algo))
         return decrypted_data
 
 
 class BaseModel(object):
     """
     Base Model for all classes
     """
 
     @classmethod
     def _get_keys(cls):
         """return column names for this model """
         return class_mapper(cls).c.keys()
 
     def get_dict(self):
         """
         return dict with keys and values corresponding
         to this model data """
 
         d = {}
         for k in self._get_keys():
             d[k] = getattr(self, k)
 
         # also use __json__() if present to get additional fields
         _json_attr = getattr(self, '__json__', None)
         if _json_attr:
             # update with attributes from __json__
             if callable(_json_attr):
                 _json_attr = _json_attr()
             for k, val in _json_attr.iteritems():
                 d[k] = val
         return d
 
     def get_appstruct(self):
         """return list with keys and values tuples corresponding
         to this model data """
 
         lst = []
         for k in self._get_keys():
             lst.append((k, getattr(self, k),))
         return lst
 
     def populate_obj(self, populate_dict):
         """populate model with data from given populate_dict"""
 
         for k in self._get_keys():
             if k in populate_dict:
                 setattr(self, k, populate_dict[k])
 
     @classmethod
     def query(cls):
         return Session().query(cls)
 
     @classmethod
     def get(cls, id_):
         if id_:
             return cls.query().get(id_)
 
     @classmethod
     def get_or_404(cls, id_):
         from pyramid.httpexceptions import HTTPNotFound
 
         try:
             id_ = int(id_)
         except (TypeError, ValueError):
             raise HTTPNotFound()
 
         res = cls.query().get(id_)
         if not res:
             raise HTTPNotFound()
         return res
 
     @classmethod
     def getAll(cls):
         # deprecated and left for backward compatibility
         return cls.get_all()
 
     @classmethod
     def get_all(cls):
         return cls.query().all()
 
     @classmethod
     def delete(cls, id_):
         obj = cls.query().get(id_)
         Session().delete(obj)
 
     @classmethod
     def identity_cache(cls, session, attr_name, value):
         exist_in_session = []
         for (item_cls, pkey), instance in session.identity_map.items():
             if cls == item_cls and getattr(instance, attr_name) == value:
                 exist_in_session.append(instance)
         if exist_in_session:
             if len(exist_in_session) == 1:
                 return exist_in_session[0]
             log.exception(
                 'multiple objects with attr %s and '
                 'value %s found with same name: %r',
                 attr_name, value, exist_in_session)
 
     def __repr__(self):
         if hasattr(self, '__unicode__'):
             # python repr needs to return str
             try:
                 return safe_str(self.__unicode__())
             except UnicodeDecodeError:
                 pass
         return '<DB:%s>' % (self.__class__.__name__)
 
 
 class RhodeCodeSetting(Base, BaseModel):
     __tablename__ = 'rhodecode_settings'
     __table_args__ = (
         UniqueConstraint('app_settings_name'),
         base_table_args
     )
 
     SETTINGS_TYPES = {
         'str': safe_str,
         'int': safe_int,
         'unicode': safe_unicode,
         'bool': str2bool,
         'list': functools.partial(aslist, sep=',')
     }
     DEFAULT_UPDATE_URL = 'https://rhodecode.com/api/v1/info/versions'
     GLOBAL_CONF_KEY = 'app_settings'
 
     app_settings_id = Column("app_settings_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     app_settings_name = Column("app_settings_name", String(255), nullable=True, unique=None, default=None)
     _app_settings_value = Column("app_settings_value", String(4096), nullable=True, unique=None, default=None)
     _app_settings_type = Column("app_settings_type", String(255), nullable=True, unique=None, default=None)
 
     def __init__(self, key='', val='', type='unicode'):
         self.app_settings_name = key
         self.app_settings_type = type
         self.app_settings_value = val
 
     @validates('_app_settings_value')
     def validate_settings_value(self, key, val):
         assert type(val) == unicode
         return val
 
     @hybrid_property
     def app_settings_value(self):
         v = self._app_settings_value
         _type = self.app_settings_type
         if _type:
             _type = self.app_settings_type.split('.')[0]
             # decode the encrypted value
             if 'encrypted' in self.app_settings_type:
                 cipher = EncryptedTextValue()
                 v = safe_unicode(cipher.process_result_value(v, None))
 
         converter = self.SETTINGS_TYPES.get(_type) or \
             self.SETTINGS_TYPES['unicode']
         return converter(v)
 
     @app_settings_value.setter
     def app_settings_value(self, val):
         """
         Setter that will always make sure we use unicode in app_settings_value
 
         :param val:
         """
         val = safe_unicode(val)
         # encode the encrypted value
         if 'encrypted' in self.app_settings_type:
             cipher = EncryptedTextValue()
             val = safe_unicode(cipher.process_bind_param(val, None))
         self._app_settings_value = val
 
     @hybrid_property
     def app_settings_type(self):
         return self._app_settings_type
 
     @app_settings_type.setter
     def app_settings_type(self, val):
         if val.split('.')[0] not in self.SETTINGS_TYPES:
             raise Exception('type must be one of %s got %s'
                             % (self.SETTINGS_TYPES.keys(), val))
         self._app_settings_type = val
 
     @classmethod
     def get_by_prefix(cls, prefix):
         return RhodeCodeSetting.query()\
             .filter(RhodeCodeSetting.app_settings_name.startswith(prefix))\
             .all()
 
     def __unicode__(self):
         return u"<%s('%s:%s[%s]')>" % (
             self.__class__.__name__,
             self.app_settings_name, self.app_settings_value,
             self.app_settings_type
         )
 
 
 class RhodeCodeUi(Base, BaseModel):
     __tablename__ = 'rhodecode_ui'
     __table_args__ = (
         UniqueConstraint('ui_key'),
         base_table_args
     )
 
     HOOK_REPO_SIZE = 'changegroup.repo_size'
     # HG
     HOOK_PRE_PULL = 'preoutgoing.pre_pull'
     HOOK_PULL = 'outgoing.pull_logger'
     HOOK_PRE_PUSH = 'prechangegroup.pre_push'
     HOOK_PRETX_PUSH = 'pretxnchangegroup.pre_push'
     HOOK_PUSH = 'changegroup.push_logger'
     HOOK_PUSH_KEY = 'pushkey.key_push'
 
     HOOKS_BUILTIN = [
         HOOK_PRE_PULL,
         HOOK_PULL,
         HOOK_PRE_PUSH,
         HOOK_PRETX_PUSH,
         HOOK_PUSH,
         HOOK_PUSH_KEY,
     ]
 
     # TODO: johbo: Unify way how hooks are configured for git and hg,
     # git part is currently hardcoded.
 
     # SVN PATTERNS
     SVN_BRANCH_ID = 'vcs_svn_branch'
     SVN_TAG_ID = 'vcs_svn_tag'
 
     ui_id = Column(
         "ui_id", Integer(), nullable=False, unique=True, default=None,
         primary_key=True)
     ui_section = Column(
         "ui_section", String(255), nullable=True, unique=None, default=None)
     ui_key = Column(
         "ui_key", String(255), nullable=True, unique=None, default=None)
     ui_value = Column(
         "ui_value", String(255), nullable=True, unique=None, default=None)
     ui_active = Column(
         "ui_active", Boolean(), nullable=True, unique=None, default=True)
 
     def __repr__(self):
         return '<%s[%s]%s=>%s]>' % (self.__class__.__name__, self.ui_section,
                                     self.ui_key, self.ui_value)
 
 
 class RepoRhodeCodeSetting(Base, BaseModel):
     __tablename__ = 'repo_rhodecode_settings'
     __table_args__ = (
         UniqueConstraint(
             'app_settings_name', 'repository_id',
             name='uq_repo_rhodecode_setting_name_repo_id'),
         base_table_args
     )
 
     repository_id = Column(
         "repository_id", Integer(), ForeignKey('repositories.repo_id'),
         nullable=False)
     app_settings_id = Column(
         "app_settings_id", Integer(), nullable=False, unique=True,
         default=None, primary_key=True)
     app_settings_name = Column(
         "app_settings_name", String(255), nullable=True, unique=None,
         default=None)
     _app_settings_value = Column(
         "app_settings_value", String(4096), nullable=True, unique=None,
         default=None)
     _app_settings_type = Column(
         "app_settings_type", String(255), nullable=True, unique=None,
         default=None)
 
     repository = relationship('Repository')
 
     def __init__(self, repository_id, key='', val='', type='unicode'):
         self.repository_id = repository_id
         self.app_settings_name = key
         self.app_settings_type = type
         self.app_settings_value = val
 
     @validates('_app_settings_value')
     def validate_settings_value(self, key, val):
         assert type(val) == unicode
         return val
 
     @hybrid_property
     def app_settings_value(self):
         v = self._app_settings_value
         type_ = self.app_settings_type
         SETTINGS_TYPES = RhodeCodeSetting.SETTINGS_TYPES
         converter = SETTINGS_TYPES.get(type_) or SETTINGS_TYPES['unicode']
         return converter(v)
 
     @app_settings_value.setter
     def app_settings_value(self, val):
         """
         Setter that will always make sure we use unicode in app_settings_value
 
         :param val:
         """
         self._app_settings_value = safe_unicode(val)
 
     @hybrid_property
     def app_settings_type(self):
         return self._app_settings_type
 
     @app_settings_type.setter
     def app_settings_type(self, val):
         SETTINGS_TYPES = RhodeCodeSetting.SETTINGS_TYPES
         if val not in SETTINGS_TYPES:
             raise Exception('type must be one of %s got %s'
                             % (SETTINGS_TYPES.keys(), val))
         self._app_settings_type = val
 
     def __unicode__(self):
         return u"<%s('%s:%s:%s[%s]')>" % (
             self.__class__.__name__, self.repository.repo_name,
             self.app_settings_name, self.app_settings_value,
             self.app_settings_type
         )
 
 
 class RepoRhodeCodeUi(Base, BaseModel):
     __tablename__ = 'repo_rhodecode_ui'
     __table_args__ = (
         UniqueConstraint(
             'repository_id', 'ui_section', 'ui_key',
             name='uq_repo_rhodecode_ui_repository_id_section_key'),
         base_table_args
     )
 
     repository_id = Column(
         "repository_id", Integer(), ForeignKey('repositories.repo_id'),
         nullable=False)
     ui_id = Column(
         "ui_id", Integer(), nullable=False, unique=True, default=None,
         primary_key=True)
     ui_section = Column(
         "ui_section", String(255), nullable=True, unique=None, default=None)
     ui_key = Column(
         "ui_key", String(255), nullable=True, unique=None, default=None)
     ui_value = Column(
         "ui_value", String(255), nullable=True, unique=None, default=None)
     ui_active = Column(
         "ui_active", Boolean(), nullable=True, unique=None, default=True)
 
     repository = relationship('Repository')
 
     def __repr__(self):
         return '<%s[%s:%s]%s=>%s]>' % (
             self.__class__.__name__, self.repository.repo_name,
             self.ui_section, self.ui_key, self.ui_value)
 
 
 class User(Base, BaseModel):
     __tablename__ = 'users'
     __table_args__ = (
         UniqueConstraint('username'), UniqueConstraint('email'),
         Index('u_username_idx', 'username'),
         Index('u_email_idx', 'email'),
         base_table_args
     )
 
     DEFAULT_USER = 'default'
     DEFAULT_USER_EMAIL = '[e-mail]'
     DEFAULT_GRAVATAR_URL = 'https://secure.gravatar.com/avatar/{md5email}?d=identicon&s={size}'
 
     user_id = Column("user_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     username = Column("username", String(255), nullable=True, unique=None, default=None)
     password = Column("password", String(255), nullable=True, unique=None, default=None)
     active = Column("active", Boolean(), nullable=True, unique=None, default=True)
     admin = Column("admin", Boolean(), nullable=True, unique=None, default=False)
     name = Column("firstname", String(255), nullable=True, unique=None, default=None)
     lastname = Column("lastname", String(255), nullable=True, unique=None, default=None)
     _email = Column("email", String(255), nullable=True, unique=None, default=None)
     last_login = Column("last_login", DateTime(timezone=False), nullable=True, unique=None, default=None)
     last_activity = Column('last_activity', DateTime(timezone=False), nullable=True, unique=None, default=None)
 
     extern_type = Column("extern_type", String(255), nullable=True, unique=None, default=None)
     extern_name = Column("extern_name", String(255), nullable=True, unique=None, default=None)
     _api_key = Column("api_key", String(255), nullable=True, unique=None, default=None)
     inherit_default_permissions = Column("inherit_default_permissions", Boolean(), nullable=False, unique=None, default=True)
     created_on = Column('created_on', DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
     _user_data = Column("user_data", LargeBinary(), nullable=True)  # JSON data
 
     user_log = relationship('UserLog')
     user_perms = relationship('UserToPerm', primaryjoin="User.user_id==UserToPerm.user_id", cascade='all')
 
     repositories = relationship('Repository')
     repository_groups = relationship('RepoGroup')
     user_groups = relationship('UserGroup')
 
     user_followers = relationship('UserFollowing', primaryjoin='UserFollowing.follows_user_id==User.user_id', cascade='all')
     followings = relationship('UserFollowing', primaryjoin='UserFollowing.user_id==User.user_id', cascade='all')
 
     repo_to_perm = relationship('UserRepoToPerm', primaryjoin='UserRepoToPerm.user_id==User.user_id', cascade='all')
     repo_group_to_perm = relationship('UserRepoGroupToPerm', primaryjoin='UserRepoGroupToPerm.user_id==User.user_id', cascade='all')
     user_group_to_perm = relationship('UserUserGroupToPerm', primaryjoin='UserUserGroupToPerm.user_id==User.user_id', cascade='all')
 
     group_member = relationship('UserGroupMember', cascade='all')
 
     notifications = relationship('UserNotification', cascade='all')
     # notifications assigned to this user
     user_created_notifications = relationship('Notification', cascade='all')
     # comments created by this user
     user_comments = relationship('ChangesetComment', cascade='all')
     # user profile extra info
     user_emails = relationship('UserEmailMap', cascade='all')
     user_ip_map = relationship('UserIpMap', cascade='all')
     user_auth_tokens = relationship('UserApiKeys', cascade='all')
     user_ssh_keys = relationship('UserSshKeys', cascade='all')
 
     # gists
     user_gists = relationship('Gist', cascade='all')
     # user pull requests
     user_pull_requests = relationship('PullRequest', cascade='all')
     # external identities
     extenal_identities = relationship(
         'ExternalIdentity',
         primaryjoin="User.user_id==ExternalIdentity.local_user_id",
         cascade='all')
     # review rules
     user_review_rules = relationship('RepoReviewRuleUser', cascade='all')
 
     def __unicode__(self):
         return u"<%s('id:%s:%s')>" % (self.__class__.__name__,
                                       self.user_id, self.username)
 
     @hybrid_property
     def email(self):
         return self._email
 
     @email.setter
     def email(self, val):
         self._email = val.lower() if val else None
 
     @hybrid_property
     def first_name(self):
         from rhodecode.lib import helpers as h
         if self.name:
             return h.escape(self.name)
         return self.name
 
     @hybrid_property
     def last_name(self):
         from rhodecode.lib import helpers as h
         if self.lastname:
             return h.escape(self.lastname)
         return self.lastname
 
     @hybrid_property
     def api_key(self):
         """
         Fetch if exist an auth-token with role ALL connected to this user
         """
         user_auth_token = UserApiKeys.query()\
             .filter(UserApiKeys.user_id == self.user_id)\
             .filter(or_(UserApiKeys.expires == -1,
                             UserApiKeys.expires >= time.time()))\
             .filter(UserApiKeys.role == UserApiKeys.ROLE_ALL).first()
         if user_auth_token:
             user_auth_token = user_auth_token.api_key
 
         return user_auth_token
 
     @api_key.setter
     def api_key(self, val):
         # don't allow to set API key this is deprecated for now
         self._api_key = None
 
     @property
     def reviewer_pull_requests(self):
         return PullRequestReviewers.query() \
             .options(joinedload(PullRequestReviewers.pull_request)) \
             .filter(PullRequestReviewers.user_id == self.user_id) \
             .all()
 
     @property
     def firstname(self):
         # alias for future
         return self.name
 
     @property
     def emails(self):
         other = UserEmailMap.query()\
             .filter(UserEmailMap.user == self) \
             .order_by(UserEmailMap.email_id.asc()) \
             .all()
         return [self.email] + [x.email for x in other]
 
     @property
     def auth_tokens(self):
         auth_tokens = self.get_auth_tokens()
         return [x.api_key for x in auth_tokens]
 
     def get_auth_tokens(self):
         return UserApiKeys.query()\
             .filter(UserApiKeys.user == self)\
             .order_by(UserApiKeys.user_api_key_id.asc())\
             .all()
 
     @LazyProperty
     def feed_token(self):
         return self.get_feed_token()
 
     def get_feed_token(self, cache=True):
         feed_tokens = UserApiKeys.query()\
             .filter(UserApiKeys.user == self)\
             .filter(UserApiKeys.role == UserApiKeys.ROLE_FEED)
         if cache:
             feed_tokens = feed_tokens.options(
                 FromCache("sql_cache_short", "get_user_feed_token_%s" % self.user_id))
 
         feed_tokens = feed_tokens.all()
         if feed_tokens:
             return feed_tokens[0].api_key
         return 'NO_FEED_TOKEN_AVAILABLE'
 
     @classmethod
     def get(cls, user_id, cache=False):
         if not user_id:
             return
 
         user = cls.query()
         if cache:
             user = user.options(
                 FromCache("sql_cache_short", "get_users_%s" % user_id))
         return user.get(user_id)
 
     @classmethod
     def extra_valid_auth_tokens(cls, user, role=None):
         tokens = UserApiKeys.query().filter(UserApiKeys.user == user)\
                 .filter(or_(UserApiKeys.expires == -1,
                             UserApiKeys.expires >= time.time()))
         if role:
             tokens = tokens.filter(or_(UserApiKeys.role == role,
                                        UserApiKeys.role == UserApiKeys.ROLE_ALL))
         return tokens.all()
 
     def authenticate_by_token(self, auth_token, roles=None, scope_repo_id=None):
         from rhodecode.lib import auth
 
         log.debug('Trying to authenticate user: %s via auth-token, '
                   'and roles: %s', self, roles)
 
         if not auth_token:
             return False
 
         roles = (roles or []) + [UserApiKeys.ROLE_ALL]
         tokens_q = UserApiKeys.query()\
             .filter(UserApiKeys.user_id == self.user_id)\
             .filter(or_(UserApiKeys.expires == -1,
                         UserApiKeys.expires >= time.time()))
 
         tokens_q = tokens_q.filter(UserApiKeys.role.in_(roles))
 
         crypto_backend = auth.crypto_backend()
         enc_token_map = {}
         plain_token_map = {}
         for token in tokens_q:
             if token.api_key.startswith(crypto_backend.ENC_PREF):
                 enc_token_map[token.api_key] = token
             else:
                 plain_token_map[token.api_key] = token
         log.debug(
             'Found %s plain and %s encrypted user tokens to check for authentication',
             len(plain_token_map), len(enc_token_map))
 
         # plain token match comes first
         match = plain_token_map.get(auth_token)
 
         # check encrypted tokens now
         if not match:
             for token_hash, token in enc_token_map.items():
                 # NOTE(marcink): this is expensive to calculate, but most secure
                 if crypto_backend.hash_check(auth_token, token_hash):
                     match = token
                     break
 
         if match:
             log.debug('Found matching token %s', match)
             if match.repo_id:
                 log.debug('Found scope, checking for scope match of token %s', match)
                 if match.repo_id == scope_repo_id:
                     return True
                 else:
                     log.debug(
                         'AUTH_TOKEN: scope mismatch, token has a set repo scope: %s, '
                         'and calling scope is:%s, skipping further checks',
                          match.repo, scope_repo_id)
                     return False
             else:
                 return True
 
         return False
 
     @property
     def ip_addresses(self):
         ret = UserIpMap.query().filter(UserIpMap.user == self).all()
         return [x.ip_addr for x in ret]
 
     @property
     def username_and_name(self):
         return '%s (%s %s)' % (self.username, self.first_name, self.last_name)
 
     @property
     def username_or_name_or_email(self):
         full_name = self.full_name if self.full_name is not ' ' else None
         return self.username or full_name or self.email
 
     @property
     def full_name(self):
         return '%s %s' % (self.first_name, self.last_name)
 
     @property
     def full_name_or_username(self):
         return ('%s %s' % (self.first_name, self.last_name)
                 if (self.first_name and self.last_name) else self.username)
 
     @property
     def full_contact(self):
         return '%s %s <%s>' % (self.first_name, self.last_name, self.email)
 
     @property
     def short_contact(self):
         return '%s %s' % (self.first_name, self.last_name)
 
     @property
     def is_admin(self):
         return self.admin
 
     def AuthUser(self, **kwargs):
         """
         Returns instance of AuthUser for this user
         """
         from rhodecode.lib.auth import AuthUser
         return AuthUser(user_id=self.user_id, username=self.username, **kwargs)
 
     @hybrid_property
     def user_data(self):
         if not self._user_data:
             return {}
 
         try:
             return json.loads(self._user_data)
         except TypeError:
             return {}
 
     @user_data.setter
     def user_data(self, val):
         if not isinstance(val, dict):
             raise Exception('user_data must be dict, got %s' % type(val))
         try:
             self._user_data = json.dumps(val)
         except Exception:
             log.error(traceback.format_exc())
 
     @classmethod
     def get_by_username(cls, username, case_insensitive=False,
                         cache=False, identity_cache=False):
         session = Session()
 
         if case_insensitive:
             q = cls.query().filter(
                 func.lower(cls.username) == func.lower(username))
         else:
             q = cls.query().filter(cls.username == username)
 
         if cache:
             if identity_cache:
                 val = cls.identity_cache(session, 'username', username)
                 if val:
                     return val
             else:
                 cache_key = "get_user_by_name_%s" % _hash_key(username)
                 q = q.options(
                     FromCache("sql_cache_short", cache_key))
 
         return q.scalar()
 
     @classmethod
     def get_by_auth_token(cls, auth_token, cache=False):
         q = UserApiKeys.query()\
             .filter(UserApiKeys.api_key == auth_token)\
             .filter(or_(UserApiKeys.expires == -1,
                         UserApiKeys.expires >= time.time()))
         if cache:
             q = q.options(
                 FromCache("sql_cache_short", "get_auth_token_%s" % auth_token))
 
         match = q.first()
         if match:
             return match.user
 
     @classmethod
     def get_by_email(cls, email, case_insensitive=False, cache=False):
 
         if case_insensitive:
             q = cls.query().filter(func.lower(cls.email) == func.lower(email))
 
         else:
             q = cls.query().filter(cls.email == email)
 
         email_key = _hash_key(email)
         if cache:
             q = q.options(
                 FromCache("sql_cache_short", "get_email_key_%s" % email_key))
 
         ret = q.scalar()
         if ret is None:
             q = UserEmailMap.query()
             # try fetching in alternate email map
             if case_insensitive:
                 q = q.filter(func.lower(UserEmailMap.email) == func.lower(email))
             else:
                 q = q.filter(UserEmailMap.email == email)
             q = q.options(joinedload(UserEmailMap.user))
             if cache:
                 q = q.options(
                     FromCache("sql_cache_short", "get_email_map_key_%s" % email_key))
             ret = getattr(q.scalar(), 'user', None)
 
         return ret
 
     @classmethod
     def get_from_cs_author(cls, author):
         """
         Tries to get User objects out of commit author string
 
         :param author:
         """
         from rhodecode.lib.helpers import email, author_name
         # Valid email in the attribute passed, see if they're in the system
         _email = email(author)
         if _email:
             user = cls.get_by_email(_email, case_insensitive=True)
             if user:
                 return user
         # Maybe we can match by username?
         _author = author_name(author)
         user = cls.get_by_username(_author, case_insensitive=True)
         if user:
             return user
 
     def update_userdata(self, **kwargs):
         usr = self
         old = usr.user_data
         old.update(**kwargs)
         usr.user_data = old
         Session().add(usr)
         log.debug('updated userdata with ', kwargs)
 
     def update_lastlogin(self):
         """Update user lastlogin"""
         self.last_login = datetime.datetime.now()
         Session().add(self)
         log.debug('updated user %s lastlogin', self.username)
 
     def update_password(self, new_password):
         from rhodecode.lib.auth import get_crypt_password
 
         self.password = get_crypt_password(new_password)
         Session().add(self)
 
     @classmethod
     def get_first_super_admin(cls):
         user = User.query()\
             .filter(User.admin == true()) \
             .order_by(User.user_id.asc()) \
             .first()
 
         if user is None:
             raise Exception('FATAL: Missing administrative account!')
         return user
 
     @classmethod
     def get_all_super_admins(cls, only_active=False):
         """
         Returns all admin accounts sorted by username
         """
         qry = User.query().filter(User.admin == true()).order_by(User.username.asc())
         if only_active:
             qry = qry.filter(User.active == true())
         return qry.all()
 
     @classmethod
     def get_default_user(cls, cache=False, refresh=False):
         user = User.get_by_username(User.DEFAULT_USER, cache=cache)
         if user is None:
             raise Exception('FATAL: Missing default account!')
         if refresh:
             # The default user might be based on outdated state which
             # has been loaded from the cache.
             # A call to refresh() ensures that the
             # latest state from the database is used.
             Session().refresh(user)
         return user
 
     def _get_default_perms(self, user, suffix=''):
         from rhodecode.model.permission import PermissionModel
         return PermissionModel().get_default_perms(user.user_perms, suffix)
 
     def get_default_perms(self, suffix=''):
         return self._get_default_perms(self, suffix)
 
     def get_api_data(self, include_secrets=False, details='full'):
         """
         Common function for generating user related data for API
 
         :param include_secrets: By default secrets in the API data will be replaced
            by a placeholder value to prevent exposing this data by accident. In case
            this data shall be exposed, set this flag to ``True``.
 
         :param details: details can be 'basic|full' basic gives only a subset of
            the available user information that includes user_id, name and emails.
         """
         user = self
         user_data = self.user_data
         data = {
             'user_id': user.user_id,
             'username': user.username,
             'firstname': user.name,
             'lastname': user.lastname,
             'email': user.email,
             'emails': user.emails,
         }
         if details == 'basic':
             return data
 
         auth_token_length = 40
         auth_token_replacement = '*' * auth_token_length
 
         extras = {
             'auth_tokens': [auth_token_replacement],
             'active': user.active,
             'admin': user.admin,
             'extern_type': user.extern_type,
             'extern_name': user.extern_name,
             'last_login': user.last_login,
             'last_activity': user.last_activity,
             'ip_addresses': user.ip_addresses,
             'language': user_data.get('language')
         }
         data.update(extras)
 
         if include_secrets:
             data['auth_tokens'] = user.auth_tokens
         return data
 
     def __json__(self):
         data = {
             'full_name': self.full_name,
             'full_name_or_username': self.full_name_or_username,
             'short_contact': self.short_contact,
             'full_contact': self.full_contact,
         }
         data.update(self.get_api_data())
         return data
 
 
 class UserApiKeys(Base, BaseModel):
     __tablename__ = 'user_api_keys'
     __table_args__ = (
         Index('uak_api_key_idx', 'api_key', unique=True),
         Index('uak_api_key_expires_idx', 'api_key', 'expires'),
         base_table_args
     )
     __mapper_args__ = {}
 
     # ApiKey role
     ROLE_ALL = 'token_role_all'
     ROLE_HTTP = 'token_role_http'
     ROLE_VCS = 'token_role_vcs'
     ROLE_API = 'token_role_api'
     ROLE_FEED = 'token_role_feed'
     ROLE_PASSWORD_RESET = 'token_password_reset'
 
     ROLES = [ROLE_ALL, ROLE_HTTP, ROLE_VCS, ROLE_API, ROLE_FEED]
 
     user_api_key_id = Column("user_api_key_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=True, unique=None, default=None)
     api_key = Column("api_key", String(255), nullable=False, unique=True)
     description = Column('description', UnicodeText().with_variant(UnicodeText(1024), 'mysql'))
     expires = Column('expires', Float(53), nullable=False)
     role = Column('role', String(255), nullable=True)
     created_on = Column('created_on', DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
 
     # scope columns
     repo_id = Column(
         'repo_id', Integer(), ForeignKey('repositories.repo_id'),
         nullable=True, unique=None, default=None)
     repo = relationship('Repository', lazy='joined')
 
     repo_group_id = Column(
         'repo_group_id', Integer(), ForeignKey('groups.group_id'),
         nullable=True, unique=None, default=None)
     repo_group = relationship('RepoGroup', lazy='joined')
 
     user = relationship('User', lazy='joined')
 
     def __unicode__(self):
         return u"<%s('%s')>" % (self.__class__.__name__, self.role)
 
     def __json__(self):
         data = {
             'auth_token': self.api_key,
             'role': self.role,
             'scope': self.scope_humanized,
             'expired': self.expired
         }
         return data
 
     def get_api_data(self, include_secrets=False):
         data = self.__json__()
         if include_secrets:
             return data
         else:
             data['auth_token'] = self.token_obfuscated
             return data
 
     @hybrid_property
     def description_safe(self):
         from rhodecode.lib import helpers as h
         return h.escape(self.description)
 
     @property
     def expired(self):
         if self.expires == -1:
             return False
         return time.time() > self.expires
 
     @classmethod
     def _get_role_name(cls, role):
         return {
             cls.ROLE_ALL: _('all'),
             cls.ROLE_HTTP: _('http/web interface'),
             cls.ROLE_VCS: _('vcs (git/hg/svn protocol)'),
             cls.ROLE_API: _('api calls'),
             cls.ROLE_FEED: _('feed access'),
         }.get(role, role)
 
     @property
     def role_humanized(self):
         return self._get_role_name(self.role)
 
     def _get_scope(self):
         if self.repo:
             return 'Repository: {}'.format(self.repo.repo_name)
         if self.repo_group:
             return 'RepositoryGroup: {} (recursive)'.format(self.repo_group.group_name)
         return 'Global'
 
     @property
     def scope_humanized(self):
         return self._get_scope()
 
     @property
     def token_obfuscated(self):
         if self.api_key:
             return self.api_key[:4] + "****"
 
 
 class UserEmailMap(Base, BaseModel):
     __tablename__ = 'user_email_map'
     __table_args__ = (
         Index('uem_email_idx', 'email'),
         UniqueConstraint('email'),
         base_table_args
     )
     __mapper_args__ = {}
 
     email_id = Column("email_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=True, unique=None, default=None)
     _email = Column("email", String(255), nullable=True, unique=False, default=None)
     user = relationship('User', lazy='joined')
 
     @validates('_email')
     def validate_email(self, key, email):
         # check if this email is not main one
         main_email = Session().query(User).filter(User.email == email).scalar()
         if main_email is not None:
             raise AttributeError('email %s is present is user table' % email)
         return email
 
     @hybrid_property
     def email(self):
         return self._email
 
     @email.setter
     def email(self, val):
         self._email = val.lower() if val else None
 
 
 class UserIpMap(Base, BaseModel):
     __tablename__ = 'user_ip_map'
     __table_args__ = (
         UniqueConstraint('user_id', 'ip_addr'),
         base_table_args
     )
     __mapper_args__ = {}
 
     ip_id = Column("ip_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=True, unique=None, default=None)
     ip_addr = Column("ip_addr", String(255), nullable=True, unique=False, default=None)
     active = Column("active", Boolean(), nullable=True, unique=None, default=True)
     description = Column("description", String(10000), nullable=True, unique=None, default=None)
     user = relationship('User', lazy='joined')
 
     @hybrid_property
     def description_safe(self):
         from rhodecode.lib import helpers as h
         return h.escape(self.description)
 
     @classmethod
     def _get_ip_range(cls, ip_addr):
         net = ipaddress.ip_network(safe_unicode(ip_addr), strict=False)
         return [str(net.network_address), str(net.broadcast_address)]
 
     def __json__(self):
         return {
           'ip_addr': self.ip_addr,
           'ip_range': self._get_ip_range(self.ip_addr),
         }
 
     def __unicode__(self):
         return u"<%s('user_id:%s=>%s')>" % (self.__class__.__name__,
                                             self.user_id, self.ip_addr)
 
 
 class UserSshKeys(Base, BaseModel):
     __tablename__ = 'user_ssh_keys'
     __table_args__ = (
         Index('usk_ssh_key_fingerprint_idx', 'ssh_key_fingerprint'),
 
         UniqueConstraint('ssh_key_fingerprint'),
 
         base_table_args
     )
     __mapper_args__ = {}
 
     ssh_key_id = Column('ssh_key_id', Integer(), nullable=False, unique=True, default=None, primary_key=True)
     ssh_key_data = Column('ssh_key_data', String(10240), nullable=False, unique=None, default=None)
     ssh_key_fingerprint = Column('ssh_key_fingerprint', String(255), nullable=False, unique=None, default=None)
 
     description = Column('description', UnicodeText().with_variant(UnicodeText(1024), 'mysql'))
 
     created_on = Column('created_on', DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
     accessed_on = Column('accessed_on', DateTime(timezone=False), nullable=True, default=None)
     user_id = Column('user_id', Integer(), ForeignKey('users.user_id'), nullable=True, unique=None, default=None)
 
     user = relationship('User', lazy='joined')
 
     def __json__(self):
         data = {
             'ssh_fingerprint': self.ssh_key_fingerprint,
             'description': self.description,
             'created_on': self.created_on
         }
         return data
 
     def get_api_data(self):
         data = self.__json__()
         return data
 
 
 class UserLog(Base, BaseModel):
     __tablename__ = 'user_logs'
     __table_args__ = (
         base_table_args,
     )
 
     VERSION_1 = 'v1'
     VERSION_2 = 'v2'
     VERSIONS = [VERSION_1, VERSION_2]
 
     user_log_id = Column("user_log_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id',ondelete='SET NULL'), nullable=True, unique=None, default=None)
     username = Column("username", String(255), nullable=True, unique=None, default=None)
     repository_id = Column("repository_id", Integer(), ForeignKey('repositories.repo_id', ondelete='SET NULL'), nullable=True, unique=None, default=None)
     repository_name = Column("repository_name", String(255), nullable=True, unique=None, default=None)
     user_ip = Column("user_ip", String(255), nullable=True, unique=None, default=None)
     action = Column("action", Text().with_variant(Text(1200000), 'mysql'), nullable=True, unique=None, default=None)
     action_date = Column("action_date", DateTime(timezone=False), nullable=True, unique=None, default=None)
 
     version = Column("version", String(255), nullable=True, default=VERSION_1)
     user_data = Column('user_data_json', MutationObj.as_mutable(JsonType(dialect_map=dict(mysql=LONGTEXT()))))
     action_data = Column('action_data_json', MutationObj.as_mutable(JsonType(dialect_map=dict(mysql=LONGTEXT()))))
 
     def __unicode__(self):
         return u"<%s('id:%s:%s')>" % (
             self.__class__.__name__, self.repository_name, self.action)
 
     def __json__(self):
         return {
             'user_id': self.user_id,
             'username': self.username,
             'repository_id': self.repository_id,
             'repository_name': self.repository_name,
             'user_ip': self.user_ip,
             'action_date': self.action_date,
             'action': self.action,
         }
 
     @hybrid_property
     def entry_id(self):
         return self.user_log_id
 
     @property
     def action_as_day(self):
         return datetime.date(*self.action_date.timetuple()[:3])
 
     user = relationship('User')
     repository = relationship('Repository', cascade='')
 
 
 class UserGroup(Base, BaseModel):
     __tablename__ = 'users_groups'
     __table_args__ = (
         base_table_args,
     )
 
     users_group_id = Column("users_group_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     users_group_name = Column("users_group_name", String(255), nullable=False, unique=True, default=None)
     user_group_description = Column("user_group_description", String(10000), nullable=True, unique=None, default=None)
     users_group_active = Column("users_group_active", Boolean(), nullable=True, unique=None, default=None)
     inherit_default_permissions = Column("users_group_inherit_default_permissions", Boolean(), nullable=False, unique=None, default=True)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=False, unique=False, default=None)
     created_on = Column('created_on', DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
     _group_data = Column("group_data", LargeBinary(), nullable=True)  # JSON data
 
     members = relationship('UserGroupMember', cascade="all, delete, delete-orphan", lazy="joined")
     users_group_to_perm = relationship('UserGroupToPerm', cascade='all')
     users_group_repo_to_perm = relationship('UserGroupRepoToPerm', cascade='all')
     users_group_repo_group_to_perm = relationship('UserGroupRepoGroupToPerm', cascade='all')
     user_user_group_to_perm = relationship('UserUserGroupToPerm', cascade='all')
     user_group_user_group_to_perm = relationship('UserGroupUserGroupToPerm ', primaryjoin="UserGroupUserGroupToPerm.target_user_group_id==UserGroup.users_group_id", cascade='all')
 
     user_group_review_rules = relationship('RepoReviewRuleUserGroup', cascade='all')
     user = relationship('User', primaryjoin="User.user_id==UserGroup.user_id")
 
     @classmethod
     def _load_group_data(cls, column):
         if not column:
             return {}
 
         try:
             return json.loads(column) or {}
         except TypeError:
             return {}
 
     @hybrid_property
     def description_safe(self):
         from rhodecode.lib import helpers as h
         return h.escape(self.user_group_description)
 
     @hybrid_property
     def group_data(self):
         return self._load_group_data(self._group_data)
 
     @group_data.expression
     def group_data(self, **kwargs):
         return self._group_data
 
     @group_data.setter
     def group_data(self, val):
         try:
             self._group_data = json.dumps(val)
         except Exception:
             log.error(traceback.format_exc())
 
     @classmethod
     def _load_sync(cls, group_data):
         if group_data:
             return group_data.get('extern_type')
 
     @property
     def sync(self):
         return self._load_sync(self.group_data)
 
     def __unicode__(self):
         return u"<%s('id:%s:%s')>" % (self.__class__.__name__,
                                       self.users_group_id,
                                       self.users_group_name)
 
     @classmethod
     def get_by_group_name(cls, group_name, cache=False,
                           case_insensitive=False):
         if case_insensitive:
             q = cls.query().filter(func.lower(cls.users_group_name) ==
                                    func.lower(group_name))
 
         else:
             q = cls.query().filter(cls.users_group_name == group_name)
         if cache:
             q = q.options(
                 FromCache("sql_cache_short", "get_group_%s" % _hash_key(group_name)))
         return q.scalar()
 
     @classmethod
     def get(cls, user_group_id, cache=False):
         if not user_group_id:
             return
 
         user_group = cls.query()
         if cache:
             user_group = user_group.options(
                 FromCache("sql_cache_short", "get_users_group_%s" % user_group_id))
         return user_group.get(user_group_id)
 
     def permissions(self, with_admins=True, with_owner=True,
                     expand_from_user_groups=False):
         """
         Permissions for user groups
         """
         _admin_perm = 'usergroup.admin'
 
         owner_row = []
         if with_owner:
             usr = AttributeDict(self.user.get_dict())
             usr.owner_row = True
             usr.permission = _admin_perm
             owner_row.append(usr)
 
         super_admin_ids = []
         super_admin_rows = []
         if with_admins:
             for usr in User.get_all_super_admins():
                 super_admin_ids.append(usr.user_id)
                 # if this admin is also owner, don't double the record
                 if usr.user_id == owner_row[0].user_id:
                     owner_row[0].admin_row = True
                 else:
                     usr = AttributeDict(usr.get_dict())
                     usr.admin_row = True
                     usr.permission = _admin_perm
                     super_admin_rows.append(usr)
 
         q = UserUserGroupToPerm.query().filter(UserUserGroupToPerm.user_group == self)
         q = q.options(joinedload(UserUserGroupToPerm.user_group),
                       joinedload(UserUserGroupToPerm.user),
                       joinedload(UserUserGroupToPerm.permission),)
 
         # get owners and admins and permissions. We do a trick of re-writing
         # objects from sqlalchemy to named-tuples due to sqlalchemy session
         # has a global reference and changing one object propagates to all
         # others. This means if admin is also an owner admin_row that change
         # would propagate to both objects
         perm_rows = []
         for _usr in q.all():
             usr = AttributeDict(_usr.user.get_dict())
             # if this user is also owner/admin, mark as duplicate record
             if usr.user_id == owner_row[0].user_id or usr.user_id in super_admin_ids:
                 usr.duplicate_perm = True
             usr.permission = _usr.permission.permission_name
             perm_rows.append(usr)
 
         # filter the perm rows by 'default' first and then sort them by
         # admin,write,read,none permissions sorted again alphabetically in
         # each group
         perm_rows = sorted(perm_rows, key=display_user_sort)
 
         user_groups_rows = []
         if expand_from_user_groups:
             for ug in self.permission_user_groups(with_members=True):
                 for user_data in ug.members:
                     user_groups_rows.append(user_data)
 
         return super_admin_rows + owner_row + perm_rows + user_groups_rows
 
     def permission_user_groups(self, with_members=False):
         q = UserGroupUserGroupToPerm.query()\
             .filter(UserGroupUserGroupToPerm.target_user_group == self)
         q = q.options(joinedload(UserGroupUserGroupToPerm.user_group),
                       joinedload(UserGroupUserGroupToPerm.target_user_group),
                       joinedload(UserGroupUserGroupToPerm.permission),)
 
         perm_rows = []
         for _user_group in q.all():
             entry = AttributeDict(_user_group.user_group.get_dict())
             entry.permission = _user_group.permission.permission_name
             if with_members:
                 entry.members = [x.user.get_dict()
                                  for x in _user_group.user_group.members]
             perm_rows.append(entry)
 
         perm_rows = sorted(perm_rows, key=display_user_group_sort)
         return perm_rows
 
     def _get_default_perms(self, user_group, suffix=''):
         from rhodecode.model.permission import PermissionModel
         return PermissionModel().get_default_perms(user_group.users_group_to_perm, suffix)
 
     def get_default_perms(self, suffix=''):
         return self._get_default_perms(self, suffix)
 
     def get_api_data(self, with_group_members=True, include_secrets=False):
         """
         :param include_secrets: See :meth:`User.get_api_data`, this parameter is
            basically forwarded.
 
         """
         user_group = self
         data = {
             'users_group_id': user_group.users_group_id,
             'group_name': user_group.users_group_name,
             'group_description': user_group.user_group_description,
             'active': user_group.users_group_active,
             'owner': user_group.user.username,
             'sync': user_group.sync,
             'owner_email': user_group.user.email,
         }
 
         if with_group_members:
             users = []
             for user in user_group.members:
                 user = user.user
                 users.append(user.get_api_data(include_secrets=include_secrets))
             data['users'] = users
 
         return data
 
 
 class UserGroupMember(Base, BaseModel):
     __tablename__ = 'users_groups_members'
     __table_args__ = (
         base_table_args,
     )
 
     users_group_member_id = Column("users_group_member_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     users_group_id = Column("users_group_id", Integer(), ForeignKey('users_groups.users_group_id'), nullable=False, unique=None, default=None)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=False, unique=None, default=None)
 
     user = relationship('User', lazy='joined')
     users_group = relationship('UserGroup')
 
     def __init__(self, gr_id='', u_id=''):
         self.users_group_id = gr_id
         self.user_id = u_id
 
 
 class RepositoryField(Base, BaseModel):
     __tablename__ = 'repositories_fields'
     __table_args__ = (
         UniqueConstraint('repository_id', 'field_key'),  # no-multi field
         base_table_args,
     )
 
     PREFIX = 'ex_'  # prefix used in form to not conflict with already existing fields
 
     repo_field_id = Column("repo_field_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     repository_id = Column("repository_id", Integer(), ForeignKey('repositories.repo_id'), nullable=False, unique=None, default=None)
     field_key = Column("field_key", String(250))
     field_label = Column("field_label", String(1024), nullable=False)
     field_value = Column("field_value", String(10000), nullable=False)
     field_desc = Column("field_desc", String(1024), nullable=False)
     field_type = Column("field_type", String(255), nullable=False, unique=None)
     created_on = Column('created_on', DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
 
     repository = relationship('Repository')
 
     @property
     def field_key_prefixed(self):
         return 'ex_%s' % self.field_key
 
     @classmethod
     def un_prefix_key(cls, key):
         if key.startswith(cls.PREFIX):
             return key[len(cls.PREFIX):]
         return key
 
     @classmethod
     def get_by_key_name(cls, key, repo):
         row = cls.query()\
                 .filter(cls.repository == repo)\
                 .filter(cls.field_key == key).scalar()
         return row
 
 
 class Repository(Base, BaseModel):
     __tablename__ = 'repositories'
     __table_args__ = (
         Index('r_repo_name_idx', 'repo_name', mysql_length=255),
         base_table_args,
     )
     DEFAULT_CLONE_URI = '{scheme}://{user}@{netloc}/{repo}'
     DEFAULT_CLONE_URI_ID = '{scheme}://{user}@{netloc}/_{repoid}'
     DEFAULT_CLONE_URI_SSH = 'ssh://{sys_user}@{hostname}/{repo}'
 
     STATE_CREATED = 'repo_state_created'
     STATE_PENDING = 'repo_state_pending'
     STATE_ERROR = 'repo_state_error'
 
     LOCK_AUTOMATIC = 'lock_auto'
     LOCK_API = 'lock_api'
     LOCK_WEB = 'lock_web'
     LOCK_PULL = 'lock_pull'
 
     NAME_SEP = URL_SEP
 
     repo_id = Column(
         "repo_id", Integer(), nullable=False, unique=True, default=None,
         primary_key=True)
     _repo_name = Column(
         "repo_name", Text(), nullable=False, default=None)
     _repo_name_hash = Column(
         "repo_name_hash", String(255), nullable=False, unique=True)
     repo_state = Column("repo_state", String(255), nullable=True)
 
     clone_uri = Column(
         "clone_uri", EncryptedTextValue(), nullable=True, unique=False,
         default=None)
     push_uri = Column(
         "push_uri", EncryptedTextValue(), nullable=True, unique=False,
         default=None)
     repo_type = Column(
         "repo_type", String(255), nullable=False, unique=False, default=None)
     user_id = Column(
         "user_id", Integer(), ForeignKey('users.user_id'), nullable=False,
         unique=False, default=None)
     private = Column(
         "private", Boolean(), nullable=True, unique=None, default=None)
     archived = Column(
         "archived", Boolean(), nullable=True, unique=None, default=None)
     enable_statistics = Column(
         "statistics", Boolean(), nullable=True, unique=None, default=True)
     enable_downloads = Column(
         "downloads", Boolean(), nullable=True, unique=None, default=True)
     description = Column(
         "description", String(10000), nullable=True, unique=None, default=None)
     created_on = Column(
         'created_on', DateTime(timezone=False), nullable=True, unique=None,
         default=datetime.datetime.now)
     updated_on = Column(
         'updated_on', DateTime(timezone=False), nullable=True, unique=None,
         default=datetime.datetime.now)
     _landing_revision = Column(
         "landing_revision", String(255), nullable=False, unique=False,
         default=None)
     enable_locking = Column(
         "enable_locking", Boolean(), nullable=False, unique=None,
         default=False)
     _locked = Column(
         "locked", String(255), nullable=True, unique=False, default=None)
     _changeset_cache = Column(
         "changeset_cache", LargeBinary(), nullable=True)  # JSON data
 
     fork_id = Column(
         "fork_id", Integer(), ForeignKey('repositories.repo_id'),
         nullable=True, unique=False, default=None)
     group_id = Column(
         "group_id", Integer(), ForeignKey('groups.group_id'), nullable=True,
         unique=False, default=None)
 
     user = relationship('User', lazy='joined')
     fork = relationship('Repository', remote_side=repo_id, lazy='joined')
     group = relationship('RepoGroup', lazy='joined')
     repo_to_perm = relationship(
         'UserRepoToPerm', cascade='all',
         order_by='UserRepoToPerm.repo_to_perm_id')
     users_group_to_perm = relationship('UserGroupRepoToPerm', cascade='all')
     stats = relationship('Statistics', cascade='all', uselist=False)
 
     followers = relationship(
         'UserFollowing',
         primaryjoin='UserFollowing.follows_repo_id==Repository.repo_id',
         cascade='all')
     extra_fields = relationship(
         'RepositoryField', cascade="all, delete, delete-orphan")
     logs = relationship('UserLog')
     comments = relationship(
         'ChangesetComment', cascade="all, delete, delete-orphan")
     pull_requests_source = relationship(
         'PullRequest',
         primaryjoin='PullRequest.source_repo_id==Repository.repo_id',
         cascade="all, delete, delete-orphan")
     pull_requests_target = relationship(
         'PullRequest',
         primaryjoin='PullRequest.target_repo_id==Repository.repo_id',
         cascade="all, delete, delete-orphan")
     ui = relationship('RepoRhodeCodeUi', cascade="all")
     settings = relationship('RepoRhodeCodeSetting', cascade="all")
     integrations = relationship('Integration', cascade="all, delete, delete-orphan")
 
     scoped_tokens = relationship('UserApiKeys', cascade="all")
 
     artifacts = relationship('FileStore', cascade="all")
 
     def __unicode__(self):
         return u"<%s('%s:%s')>" % (self.__class__.__name__, self.repo_id,
                                    safe_unicode(self.repo_name))
 
     @hybrid_property
     def description_safe(self):
         from rhodecode.lib import helpers as h
         return h.escape(self.description)
 
     @hybrid_property
     def landing_rev(self):
         # always should return [rev_type, rev]
         if self._landing_revision:
             _rev_info = self._landing_revision.split(':')
             if len(_rev_info) < 2:
                 _rev_info.insert(0, 'rev')
             return [_rev_info[0], _rev_info[1]]
         return [None, None]
 
     @landing_rev.setter
     def landing_rev(self, val):
         if ':' not in val:
             raise ValueError('value must be delimited with `:` and consist '
                              'of <rev_type>:<rev>, got %s instead' % val)
         self._landing_revision = val
 
     @hybrid_property
     def locked(self):
         if self._locked:
             user_id, timelocked, reason = self._locked.split(':')
             lock_values = int(user_id), timelocked, reason
         else:
             lock_values = [None, None, None]
         return lock_values
 
     @locked.setter
     def locked(self, val):
         if val and isinstance(val, (list, tuple)):
             self._locked = ':'.join(map(str, val))
         else:
             self._locked = None
 
     @hybrid_property
     def changeset_cache(self):
         from rhodecode.lib.vcs.backends.base import EmptyCommit
         dummy = EmptyCommit().__json__()
         if not self._changeset_cache:
             dummy['source_repo_id'] = self.repo_id
             return json.loads(json.dumps(dummy))
 
         try:
             return json.loads(self._changeset_cache)
         except TypeError:
             return dummy
         except Exception:
             log.error(traceback.format_exc())
             return dummy
 
     @changeset_cache.setter
     def changeset_cache(self, val):
         try:
             self._changeset_cache = json.dumps(val)
         except Exception:
             log.error(traceback.format_exc())
 
     @hybrid_property
     def repo_name(self):
         return self._repo_name
 
     @repo_name.setter
     def repo_name(self, value):
         self._repo_name = value
         self._repo_name_hash = hashlib.sha1(safe_str(value)).hexdigest()
 
     @classmethod
     def normalize_repo_name(cls, repo_name):
         """
         Normalizes os specific repo_name to the format internally stored inside
         database using URL_SEP
 
         :param cls:
         :param repo_name:
         """
         return cls.NAME_SEP.join(repo_name.split(os.sep))
 
     @classmethod
     def get_by_repo_name(cls, repo_name, cache=False, identity_cache=False):
         session = Session()
         q = session.query(cls).filter(cls.repo_name == repo_name)
 
         if cache:
             if identity_cache:
                 val = cls.identity_cache(session, 'repo_name', repo_name)
                 if val:
                     return val
             else:
                 cache_key = "get_repo_by_name_%s" % _hash_key(repo_name)
                 q = q.options(
                     FromCache("sql_cache_short", cache_key))
 
         return q.scalar()
 
     @classmethod
     def get_by_id_or_repo_name(cls, repoid):
         if isinstance(repoid, (int, long)):
             try:
                 repo = cls.get(repoid)
             except ValueError:
                 repo = None
         else:
             repo = cls.get_by_repo_name(repoid)
         return repo
 
     @classmethod
     def get_by_full_path(cls, repo_full_path):
         repo_name = repo_full_path.split(cls.base_path(), 1)[-1]
         repo_name = cls.normalize_repo_name(repo_name)
         return cls.get_by_repo_name(repo_name.strip(URL_SEP))
 
     @classmethod
     def get_repo_forks(cls, repo_id):
         return cls.query().filter(Repository.fork_id == repo_id)
 
     @classmethod
     def base_path(cls):
         """
         Returns base path when all repos are stored
 
         :param cls:
         """
         q = Session().query(RhodeCodeUi)\
             .filter(RhodeCodeUi.ui_key == cls.NAME_SEP)
         q = q.options(FromCache("sql_cache_short", "repository_repo_path"))
         return q.one().ui_value
 
     @classmethod
     def get_all_repos(cls, user_id=Optional(None), group_id=Optional(None),
                       case_insensitive=True, archived=False):
         q = Repository.query()
 
         if not archived:
             q = q.filter(Repository.archived.isnot(true()))
 
         if not isinstance(user_id, Optional):
             q = q.filter(Repository.user_id == user_id)
 
         if not isinstance(group_id, Optional):
             q = q.filter(Repository.group_id == group_id)
 
         if case_insensitive:
             q = q.order_by(func.lower(Repository.repo_name))
         else:
             q = q.order_by(Repository.repo_name)
 
         return q.all()
 
     @property
     def repo_uid(self):
         return '_{}'.format(self.repo_id)
 
     @property
     def forks(self):
         """
         Return forks of this repo
         """
         return Repository.get_repo_forks(self.repo_id)
 
     @property
     def parent(self):
         """
         Returns fork parent
         """
         return self.fork
 
     @property
     def just_name(self):
         return self.repo_name.split(self.NAME_SEP)[-1]
 
     @property
     def groups_with_parents(self):
         groups = []
         if self.group is None:
             return groups
 
         cur_gr = self.group
         groups.insert(0, cur_gr)
         while 1:
             gr = getattr(cur_gr, 'parent_group', None)
             cur_gr = cur_gr.parent_group
             if gr is None:
                 break
             groups.insert(0, gr)
 
         return groups
 
     @property
     def groups_and_repo(self):
         return self.groups_with_parents, self
 
     @LazyProperty
     def repo_path(self):
         """
         Returns base full path for that repository means where it actually
         exists on a filesystem
         """
         q = Session().query(RhodeCodeUi).filter(
             RhodeCodeUi.ui_key == self.NAME_SEP)
         q = q.options(FromCache("sql_cache_short", "repository_repo_path"))
         return q.one().ui_value
 
     @property
     def repo_full_path(self):
         p = [self.repo_path]
         # we need to split the name by / since this is how we store the
         # names in the database, but that eventually needs to be converted
         # into a valid system path
         p += self.repo_name.split(self.NAME_SEP)
         return os.path.join(*map(safe_unicode, p))
 
     @property
     def cache_keys(self):
         """
         Returns associated cache keys for that repo
         """
         invalidation_namespace = CacheKey.REPO_INVALIDATION_NAMESPACE.format(
             repo_id=self.repo_id)
         return CacheKey.query()\
             .filter(CacheKey.cache_args == invalidation_namespace)\
             .order_by(CacheKey.cache_key)\
             .all()
 
     @property
     def cached_diffs_relative_dir(self):
         """
         Return a relative to the repository store path of cached diffs
         used for safe display for users, who shouldn't know the absolute store
         path
         """
         return os.path.join(
             os.path.dirname(self.repo_name),
             self.cached_diffs_dir.split(os.path.sep)[-1])
 
     @property
     def cached_diffs_dir(self):
         path = self.repo_full_path
         return os.path.join(
             os.path.dirname(path),
             '.__shadow_diff_cache_repo_{}'.format(self.repo_id))
 
     def cached_diffs(self):
         diff_cache_dir = self.cached_diffs_dir
         if os.path.isdir(diff_cache_dir):
             return os.listdir(diff_cache_dir)
         return []
 
     def shadow_repos(self):
         shadow_repos_pattern = '.__shadow_repo_{}'.format(self.repo_id)
         return [
             x for x in os.listdir(os.path.dirname(self.repo_full_path))
             if x.startswith(shadow_repos_pattern)]
 
     def get_new_name(self, repo_name):
         """
         returns new full repository name based on assigned group and new new
 
         :param group_name:
         """
         path_prefix = self.group.full_path_splitted if self.group else []
         return self.NAME_SEP.join(path_prefix + [repo_name])
 
     @property
     def _config(self):
         """
         Returns db based config object.
         """
         from rhodecode.lib.utils import make_db_config
         return make_db_config(clear_session=False, repo=self)
 
     def permissions(self, with_admins=True, with_owner=True,
                     expand_from_user_groups=False):
         """
         Permissions for repositories
         """
         _admin_perm = 'repository.admin'
 
         owner_row = []
         if with_owner:
             usr = AttributeDict(self.user.get_dict())
             usr.owner_row = True
             usr.permission = _admin_perm
             usr.permission_id = None
             owner_row.append(usr)
 
         super_admin_ids = []
         super_admin_rows = []
         if with_admins:
             for usr in User.get_all_super_admins():
                 super_admin_ids.append(usr.user_id)
                 # if this admin is also owner, don't double the record
                 if usr.user_id == owner_row[0].user_id:
                     owner_row[0].admin_row = True
                 else:
                     usr = AttributeDict(usr.get_dict())
                     usr.admin_row = True
                     usr.permission = _admin_perm
                     usr.permission_id = None
                     super_admin_rows.append(usr)
 
         q = UserRepoToPerm.query().filter(UserRepoToPerm.repository == self)
         q = q.options(joinedload(UserRepoToPerm.repository),
                       joinedload(UserRepoToPerm.user),
                       joinedload(UserRepoToPerm.permission),)
 
         # get owners and admins and permissions. We do a trick of re-writing
         # objects from sqlalchemy to named-tuples due to sqlalchemy session
         # has a global reference and changing one object propagates to all
         # others. This means if admin is also an owner admin_row that change
         # would propagate to both objects
         perm_rows = []
         for _usr in q.all():
             usr = AttributeDict(_usr.user.get_dict())
             # if this user is also owner/admin, mark as duplicate record
             if usr.user_id == owner_row[0].user_id or usr.user_id in super_admin_ids:
                 usr.duplicate_perm = True
             # also check if this permission is maybe used by branch_permissions
             if _usr.branch_perm_entry:
                 usr.branch_rules = [x.branch_rule_id for x in _usr.branch_perm_entry]
 
             usr.permission = _usr.permission.permission_name
             usr.permission_id = _usr.repo_to_perm_id
             perm_rows.append(usr)
 
         # filter the perm rows by 'default' first and then sort them by
         # admin,write,read,none permissions sorted again alphabetically in
         # each group
         perm_rows = sorted(perm_rows, key=display_user_sort)
 
         user_groups_rows = []
         if expand_from_user_groups:
             for ug in self.permission_user_groups(with_members=True):
                 for user_data in ug.members:
                     user_groups_rows.append(user_data)
 
         return super_admin_rows + owner_row + perm_rows + user_groups_rows
 
     def permission_user_groups(self, with_members=True):
         q = UserGroupRepoToPerm.query()\
             .filter(UserGroupRepoToPerm.repository == self)
         q = q.options(joinedload(UserGroupRepoToPerm.repository),
                       joinedload(UserGroupRepoToPerm.users_group),
                       joinedload(UserGroupRepoToPerm.permission),)
 
         perm_rows = []
         for _user_group in q.all():
             entry = AttributeDict(_user_group.users_group.get_dict())
             entry.permission = _user_group.permission.permission_name
             if with_members:
                 entry.members = [x.user.get_dict()
                                  for x in _user_group.users_group.members]
             perm_rows.append(entry)
 
         perm_rows = sorted(perm_rows, key=display_user_group_sort)
         return perm_rows
 
     def get_api_data(self, include_secrets=False):
         """
         Common function for generating repo api data
 
         :param include_secrets: See :meth:`User.get_api_data`.
 
         """
         # TODO: mikhail: Here there is an anti-pattern, we probably need to
         # move this methods on models level.
         from rhodecode.model.settings import SettingsModel
         from rhodecode.model.repo import RepoModel
 
         repo = self
         _user_id, _time, _reason = self.locked
 
         data = {
             'repo_id': repo.repo_id,
             'repo_name': repo.repo_name,
             'repo_type': repo.repo_type,
             'clone_uri': repo.clone_uri or '',
             'push_uri': repo.push_uri or '',
             'url': RepoModel().get_url(self),
             'private': repo.private,
             'created_on': repo.created_on,
             'description': repo.description_safe,
             'landing_rev': repo.landing_rev,
             'owner': repo.user.username,
             'fork_of': repo.fork.repo_name if repo.fork else None,
             'fork_of_id': repo.fork.repo_id if repo.fork else None,
             'enable_statistics': repo.enable_statistics,
             'enable_locking': repo.enable_locking,
             'enable_downloads': repo.enable_downloads,
             'last_changeset': repo.changeset_cache,
             'locked_by': User.get(_user_id).get_api_data(
                 include_secrets=include_secrets) if _user_id else None,
             'locked_date': time_to_datetime(_time) if _time else None,
             'lock_reason': _reason if _reason else None,
         }
 
         # TODO: mikhail: should be per-repo settings here
         rc_config = SettingsModel().get_all_settings()
         repository_fields = str2bool(
             rc_config.get('rhodecode_repository_fields'))
         if repository_fields:
             for f in self.extra_fields:
                 data[f.field_key_prefixed] = f.field_value
 
         return data
 
     @classmethod
     def lock(cls, repo, user_id, lock_time=None, lock_reason=None):
         if not lock_time:
             lock_time = time.time()
         if not lock_reason:
             lock_reason = cls.LOCK_AUTOMATIC
         repo.locked = [user_id, lock_time, lock_reason]
         Session().add(repo)
         Session().commit()
 
     @classmethod
     def unlock(cls, repo):
         repo.locked = None
         Session().add(repo)
         Session().commit()
 
     @classmethod
     def getlock(cls, repo):
         return repo.locked
 
     def is_user_lock(self, user_id):
         if self.lock[0]:
             lock_user_id = safe_int(self.lock[0])
             user_id = safe_int(user_id)
             # both are ints, and they are equal
             return all([lock_user_id, user_id]) and lock_user_id == user_id
 
         return False
 
     def get_locking_state(self, action, user_id, only_when_enabled=True):
         """
         Checks locking on this repository, if locking is enabled and lock is
         present returns a tuple of make_lock, locked, locked_by.
         make_lock can have 3 states None (do nothing) True, make lock
         False release lock, This value is later propagated to hooks, which
         do the locking. Think about this as signals passed to hooks what to do.
 
         """
         # TODO: johbo: This is part of the business logic and should be moved
         # into the RepositoryModel.
 
         if action not in ('push', 'pull'):
             raise ValueError("Invalid action value: %s" % repr(action))
 
         # defines if locked error should be thrown to user
         currently_locked = False
         # defines if new lock should be made, tri-state
         make_lock = None
         repo = self
         user = User.get(user_id)
 
         lock_info = repo.locked
 
         if repo and (repo.enable_locking or not only_when_enabled):
             if action == 'push':
                 # check if it's already locked !, if it is compare users
                 locked_by_user_id = lock_info[0]
                 if user.user_id == locked_by_user_id:
                     log.debug(
                         'Got `push` action from user %s, now unlocking', user)
                     # unlock if we have push from user who locked
                     make_lock = False
                 else:
                     # we're not the same user who locked, ban with
                     # code defined in settings (default is 423 HTTP Locked) !
                     log.debug('Repo %s is currently locked by %s', repo, user)
                     currently_locked = True
             elif action == 'pull':
                 # [0] user [1] date
                 if lock_info[0] and lock_info[1]:
                     log.debug('Repo %s is currently locked by %s', repo, user)
                     currently_locked = True
                 else:
                     log.debug('Setting lock on repo %s by %s', repo, user)
                     make_lock = True
 
         else:
             log.debug('Repository %s do not have locking enabled', repo)
 
         log.debug('FINAL locking values make_lock:%s,locked:%s,locked_by:%s',
                   make_lock, currently_locked, lock_info)
 
         from rhodecode.lib.auth import HasRepoPermissionAny
         perm_check = HasRepoPermissionAny('repository.write', 'repository.admin')
         if make_lock and not perm_check(repo_name=repo.repo_name, user=user):
             # if we don't have at least write permission we cannot make a lock
             log.debug('lock state reset back to FALSE due to lack '
                       'of at least read permission')
             make_lock = False
 
         return make_lock, currently_locked, lock_info
 
     @property
     def last_commit_cache_update_diff(self):
         return time.time() - (safe_int(self.changeset_cache.get('updated_on')) or 0)
 
     @property
     def last_commit_change(self):
         from rhodecode.lib.vcs.utils.helpers import parse_datetime
         empty_date = datetime.datetime.fromtimestamp(0)
         date_latest = self.changeset_cache.get('date', empty_date)
         try:
             return parse_datetime(date_latest)
         except Exception:
             return empty_date
 
     @property
     def last_db_change(self):
         return self.updated_on
 
     @property
     def clone_uri_hidden(self):
         clone_uri = self.clone_uri
         if clone_uri:
             import urlobject
             url_obj = urlobject.URLObject(cleaned_uri(clone_uri))
             if url_obj.password:
                 clone_uri = url_obj.with_password('*****')
         return clone_uri
 
     @property
     def push_uri_hidden(self):
         push_uri = self.push_uri
         if push_uri:
             import urlobject
             url_obj = urlobject.URLObject(cleaned_uri(push_uri))
             if url_obj.password:
                 push_uri = url_obj.with_password('*****')
         return push_uri
 
     def clone_url(self, **override):
         from rhodecode.model.settings import SettingsModel
 
         uri_tmpl = None
         if 'with_id' in override:
             uri_tmpl = self.DEFAULT_CLONE_URI_ID
             del override['with_id']
 
         if 'uri_tmpl' in override:
             uri_tmpl = override['uri_tmpl']
             del override['uri_tmpl']
 
         ssh = False
         if 'ssh' in override:
             ssh = True
             del override['ssh']
 
         # we didn't override our tmpl from **overrides
         request = get_current_request()
         if not uri_tmpl:
             if hasattr(request, 'call_context') and hasattr(request.call_context, 'rc_config'):
                 rc_config = request.call_context.rc_config
             else:
                 rc_config = SettingsModel().get_all_settings(cache=True)
             if ssh:
                 uri_tmpl = rc_config.get(
                     'rhodecode_clone_uri_ssh_tmpl') or self.DEFAULT_CLONE_URI_SSH
             else:
                 uri_tmpl = rc_config.get(
                     'rhodecode_clone_uri_tmpl') or self.DEFAULT_CLONE_URI
 
         return get_clone_url(request=request,
                              uri_tmpl=uri_tmpl,
                              repo_name=self.repo_name,
                              repo_id=self.repo_id, **override)
 
     def set_state(self, state):
         self.repo_state = state
         Session().add(self)
     #==========================================================================
     # SCM PROPERTIES
     #==========================================================================
 
     def get_commit(self, commit_id=None, commit_idx=None, pre_load=None):
         return get_commit_safe(
             self.scm_instance(), commit_id, commit_idx, pre_load=pre_load)
 
     def get_changeset(self, rev=None, pre_load=None):
         warnings.warn("Use get_commit", DeprecationWarning)
         commit_id = None
         commit_idx = None
         if isinstance(rev, compat.string_types):
             commit_id = rev
         else:
             commit_idx = rev
         return self.get_commit(commit_id=commit_id, commit_idx=commit_idx,
                                pre_load=pre_load)
 
     def get_landing_commit(self):
         """
         Returns landing commit, or if that doesn't exist returns the tip
         """
         _rev_type, _rev = self.landing_rev
         commit = self.get_commit(_rev)
         if isinstance(commit, EmptyCommit):
             return self.get_commit()
         return commit
 
     def update_commit_cache(self, cs_cache=None, config=None):
         """
         Update cache of last commit for repository, keys should be::
 
             source_repo_id
             short_id
             raw_id
             revision
             parents
             message
             date
             author
             updated_on
 
         """
         from rhodecode.lib.vcs.backends.base import BaseChangeset
         if cs_cache is None:
             # use no-cache version here
             scm_repo = self.scm_instance(cache=False, config=config)
 
             empty = scm_repo is None or scm_repo.is_empty()
             if not empty:
                 cs_cache = scm_repo.get_commit(
                     pre_load=["author", "date", "message", "parents", "branch"])
             else:
                 cs_cache = EmptyCommit()
 
         if isinstance(cs_cache, BaseChangeset):
             cs_cache = cs_cache.__json__()
 
         def is_outdated(new_cs_cache):
             if (new_cs_cache['raw_id'] != self.changeset_cache['raw_id'] or
                 new_cs_cache['revision'] != self.changeset_cache['revision']):
                 return True
             return False
 
         # check if we have maybe already latest cached revision
         if is_outdated(cs_cache) or not self.changeset_cache:
             _default = datetime.datetime.utcnow()
             last_change = cs_cache.get('date') or _default
             # we check if last update is newer than the new value
             # if yes, we use the current timestamp instead. Imagine you get
             # old commit pushed 1y ago, we'd set last update 1y to ago.
             last_change_timestamp = datetime_to_time(last_change)
             current_timestamp = datetime_to_time(last_change)
             if last_change_timestamp > current_timestamp:
                 cs_cache['date'] = _default
 
             cs_cache['updated_on'] = time.time()
             self.changeset_cache = cs_cache
             Session().add(self)
             Session().commit()
 
             log.debug('updated repo %s with new commit cache %s',
                       self.repo_name, cs_cache)
         else:
             cs_cache = self.changeset_cache
             cs_cache['updated_on'] = time.time()
             self.changeset_cache = cs_cache
             Session().add(self)
             Session().commit()
 
             log.debug('Skipping update_commit_cache for repo:`%s` '
                       'commit already with latest changes', self.repo_name)
 
     @property
     def tip(self):
         return self.get_commit('tip')
 
     @property
     def author(self):
         return self.tip.author
 
     @property
     def last_change(self):
         return self.scm_instance().last_change
 
     def get_comments(self, revisions=None):
         """
         Returns comments for this repository grouped by revisions
 
         :param revisions: filter query by revisions only
         """
         cmts = ChangesetComment.query()\
             .filter(ChangesetComment.repo == self)
         if revisions:
             cmts = cmts.filter(ChangesetComment.revision.in_(revisions))
         grouped = collections.defaultdict(list)
         for cmt in cmts.all():
             grouped[cmt.revision].append(cmt)
         return grouped
 
     def statuses(self, revisions=None):
         """
         Returns statuses for this repository
 
         :param revisions: list of revisions to get statuses for
         """
         statuses = ChangesetStatus.query()\
             .filter(ChangesetStatus.repo == self)\
             .filter(ChangesetStatus.version == 0)
 
         if revisions:
             # Try doing the filtering in chunks to avoid hitting limits
             size = 500
             status_results = []
             for chunk in xrange(0, len(revisions), size):
                 status_results += statuses.filter(
                     ChangesetStatus.revision.in_(
                         revisions[chunk: chunk+size])
                 ).all()
         else:
             status_results = statuses.all()
 
         grouped = {}
 
         # maybe we have open new pullrequest without a status?
         stat = ChangesetStatus.STATUS_UNDER_REVIEW
         status_lbl = ChangesetStatus.get_status_lbl(stat)
         for pr in PullRequest.query().filter(PullRequest.source_repo == self).all():
             for rev in pr.revisions:
                 pr_id = pr.pull_request_id
                 pr_repo = pr.target_repo.repo_name
                 grouped[rev] = [stat, status_lbl, pr_id, pr_repo]
 
         for stat in status_results:
             pr_id = pr_repo = None
             if stat.pull_request:
                 pr_id = stat.pull_request.pull_request_id
                 pr_repo = stat.pull_request.target_repo.repo_name
             grouped[stat.revision] = [str(stat.status), stat.status_lbl,
                                       pr_id, pr_repo]
         return grouped
 
     # ==========================================================================
     # SCM CACHE INSTANCE
     # ==========================================================================
 
     def scm_instance(self, **kwargs):
         import rhodecode
 
         # Passing a config will not hit the cache currently only used
         # for repo2dbmapper
         config = kwargs.pop('config', None)
         cache = kwargs.pop('cache', None)
         vcs_full_cache = kwargs.pop('vcs_full_cache', None)
         if vcs_full_cache is not None:
             # allows override global config
             full_cache = vcs_full_cache
         else:
             full_cache = str2bool(rhodecode.CONFIG.get('vcs_full_cache'))
         # if cache is NOT defined use default global, else we have a full
         # control over cache behaviour
         if cache is None and full_cache and not config:
             log.debug('Initializing pure cached instance for %s', self.repo_path)
             return self._get_instance_cached()
 
         # cache here is sent to the "vcs server"
         return self._get_instance(cache=bool(cache), config=config)
 
     def _get_instance_cached(self):
         from rhodecode.lib import rc_cache
 
         cache_namespace_uid = 'cache_repo_instance.{}'.format(self.repo_id)
         invalidation_namespace = CacheKey.REPO_INVALIDATION_NAMESPACE.format(
             repo_id=self.repo_id)
         region = rc_cache.get_or_create_region('cache_repo_longterm', cache_namespace_uid)
 
         @region.conditional_cache_on_arguments(namespace=cache_namespace_uid)
         def get_instance_cached(repo_id, context_id, _cache_state_uid):
             return self._get_instance(repo_state_uid=_cache_state_uid)
 
         # we must use thread scoped cache here,
         # because each thread of gevent needs it's own not shared connection and cache
         # we also alter `args` so the cache key is individual for every green thread.
         inv_context_manager = rc_cache.InvalidationContext(
             uid=cache_namespace_uid, invalidation_namespace=invalidation_namespace,
             thread_scoped=True)
         with inv_context_manager as invalidation_context:
             cache_state_uid = invalidation_context.cache_data['cache_state_uid']
             args = (self.repo_id, inv_context_manager.cache_key, cache_state_uid)
 
             # re-compute and store cache if we get invalidate signal
             if invalidation_context.should_invalidate():
                 instance = get_instance_cached.refresh(*args)
             else:
                 instance = get_instance_cached(*args)
 
             log.debug('Repo instance fetched in %.4fs', inv_context_manager.compute_time)
             return instance
 
     def _get_instance(self, cache=True, config=None, repo_state_uid=None):
         log.debug('Initializing %s instance `%s` with cache flag set to: %s',
                   self.repo_type, self.repo_path, cache)
         config = config or self._config
         custom_wire = {
             'cache': cache,  # controls the vcs.remote cache
             'repo_state_uid': repo_state_uid
         }
         repo = get_vcs_instance(
             repo_path=safe_str(self.repo_full_path),
             config=config,
             with_wire=custom_wire,
             create=False,
             _vcs_alias=self.repo_type)
         if repo is not None:
             repo.count()  # cache rebuild
         return repo
 
+    def get_shadow_repository_path(self, workspace_id):
+        from rhodecode.lib.vcs.backends.base import BaseRepository
+        shadow_repo_path = BaseRepository._get_shadow_repository_path(
+            self.repo_full_path, self.repo_id, workspace_id)
+        return shadow_repo_path
+
     def __json__(self):
         return {'landing_rev': self.landing_rev}
 
     def get_dict(self):
 
         # Since we transformed `repo_name` to a hybrid property, we need to
         # keep compatibility with the code which uses `repo_name` field.
 
         result = super(Repository, self).get_dict()
         result['repo_name'] = result.pop('_repo_name', None)
         return result
 
 
 class RepoGroup(Base, BaseModel):
     __tablename__ = 'groups'
     __table_args__ = (
         UniqueConstraint('group_name', 'group_parent_id'),
         base_table_args,
     )
     __mapper_args__ = {'order_by': 'group_name'}
 
     CHOICES_SEPARATOR = '/'  # used to generate select2 choices for nested groups
 
     group_id = Column("group_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     _group_name = Column("group_name", String(255), nullable=False, unique=True, default=None)
     group_name_hash = Column("repo_group_name_hash", String(1024), nullable=False, unique=False)
     group_parent_id = Column("group_parent_id", Integer(), ForeignKey('groups.group_id'), nullable=True, unique=None, default=None)
     group_description = Column("group_description", String(10000), nullable=True, unique=None, default=None)
     enable_locking = Column("enable_locking", Boolean(), nullable=False, unique=None, default=False)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=False, unique=False, default=None)
     created_on = Column('created_on', DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
     updated_on = Column('updated_on', DateTime(timezone=False), nullable=True, unique=None, default=datetime.datetime.now)
     personal = Column('personal', Boolean(), nullable=True, unique=None, default=None)
     _changeset_cache = Column(
         "changeset_cache", LargeBinary(), nullable=True)  # JSON data
 
     repo_group_to_perm = relationship('UserRepoGroupToPerm', cascade='all', order_by='UserRepoGroupToPerm.group_to_perm_id')
     users_group_to_perm = relationship('UserGroupRepoGroupToPerm', cascade='all')
     parent_group = relationship('RepoGroup', remote_side=group_id)
     user = relationship('User')
     integrations = relationship('Integration', cascade="all, delete, delete-orphan")
 
     def __init__(self, group_name='', parent_group=None):
         self.group_name = group_name
         self.parent_group = parent_group
 
     def __unicode__(self):
         return u"<%s('id:%s:%s')>" % (
             self.__class__.__name__, self.group_id, self.group_name)
 
     @hybrid_property
     def group_name(self):
         return self._group_name
 
     @group_name.setter
     def group_name(self, value):
         self._group_name = value
         self.group_name_hash = self.hash_repo_group_name(value)
 
     @hybrid_property
     def changeset_cache(self):
         from rhodecode.lib.vcs.backends.base import EmptyCommit
         dummy = EmptyCommit().__json__()
         if not self._changeset_cache:
             dummy['source_repo_id'] = ''
             return json.loads(json.dumps(dummy))
 
         try:
             return json.loads(self._changeset_cache)
         except TypeError:
             return dummy
         except Exception:
             log.error(traceback.format_exc())
             return dummy
 
     @changeset_cache.setter
     def changeset_cache(self, val):
         try:
             self._changeset_cache = json.dumps(val)
         except Exception:
             log.error(traceback.format_exc())
 
     @validates('group_parent_id')
     def validate_group_parent_id(self, key, val):
         """
         Check cycle references for a parent group to self
         """
         if self.group_id and val:
             assert val != self.group_id
 
         return val
 
     @hybrid_property
     def description_safe(self):
         from rhodecode.lib import helpers as h
         return h.escape(self.group_description)
 
     @classmethod
     def hash_repo_group_name(cls, repo_group_name):
         val = remove_formatting(repo_group_name)
         val = safe_str(val).lower()
         chars = []
         for c in val:
             if c not in string.ascii_letters:
                 c = str(ord(c))
             chars.append(c)
 
         return ''.join(chars)
 
     @classmethod
     def _generate_choice(cls, repo_group):
         from webhelpers.html import literal as _literal
         _name = lambda k: _literal(cls.CHOICES_SEPARATOR.join(k))
         return repo_group.group_id, _name(repo_group.full_path_splitted)
 
     @classmethod
     def groups_choices(cls, groups=None, show_empty_group=True):
         if not groups:
             groups = cls.query().all()
 
         repo_groups = []
         if show_empty_group:
             repo_groups = [(-1, u'-- %s --' % _('No parent'))]
 
         repo_groups.extend([cls._generate_choice(x) for x in groups])
 
         repo_groups = sorted(
             repo_groups, key=lambda t: t[1].split(cls.CHOICES_SEPARATOR)[0])
         return repo_groups
 
     @classmethod
     def url_sep(cls):
         return URL_SEP
 
     @classmethod
     def get_by_group_name(cls, group_name, cache=False, case_insensitive=False):
         if case_insensitive:
             gr = cls.query().filter(func.lower(cls.group_name)
                                     == func.lower(group_name))
         else:
             gr = cls.query().filter(cls.group_name == group_name)
         if cache:
             name_key = _hash_key(group_name)
             gr = gr.options(
                 FromCache("sql_cache_short", "get_group_%s" % name_key))
         return gr.scalar()
 
     @classmethod
     def get_user_personal_repo_group(cls, user_id):
         user = User.get(user_id)
         if user.username == User.DEFAULT_USER:
             return None
 
         return cls.query()\
             .filter(cls.personal == true()) \
             .filter(cls.user == user) \
             .order_by(cls.group_id.asc()) \
             .first()
 
     @classmethod
     def get_all_repo_groups(cls, user_id=Optional(None), group_id=Optional(None),
                             case_insensitive=True):
         q = RepoGroup.query()
 
         if not isinstance(user_id, Optional):
             q = q.filter(RepoGroup.user_id == user_id)
 
         if not isinstance(group_id, Optional):
             q = q.filter(RepoGroup.group_parent_id == group_id)
 
         if case_insensitive:
             q = q.order_by(func.lower(RepoGroup.group_name))
         else:
             q = q.order_by(RepoGroup.group_name)
         return q.all()
 
     @property
     def parents(self, parents_recursion_limit = 10):
         groups = []
         if self.parent_group is None:
             return groups
         cur_gr = self.parent_group
         groups.insert(0, cur_gr)
         cnt = 0
         while 1:
             cnt += 1
             gr = getattr(cur_gr, 'parent_group', None)
             cur_gr = cur_gr.parent_group
             if gr is None:
                 break
             if cnt == parents_recursion_limit:
                 # this will prevent accidental infinit loops
                 log.error('more than %s parents found for group %s, stopping '
                           'recursive parent fetching', parents_recursion_limit, self)
                 break
 
             groups.insert(0, gr)
         return groups
 
     @property
     def last_commit_cache_update_diff(self):
         return time.time() - (safe_int(self.changeset_cache.get('updated_on')) or 0)
 
     @property
     def last_commit_change(self):
         from rhodecode.lib.vcs.utils.helpers import parse_datetime
         empty_date = datetime.datetime.fromtimestamp(0)
         date_latest = self.changeset_cache.get('date', empty_date)
         try:
             return parse_datetime(date_latest)
         except Exception:
             return empty_date
 
     @property
     def last_db_change(self):
         return self.updated_on
 
     @property
     def children(self):
         return RepoGroup.query().filter(RepoGroup.parent_group == self)
 
     @property
     def name(self):
         return self.group_name.split(RepoGroup.url_sep())[-1]
 
     @property
     def full_path(self):
         return self.group_name
 
     @property
     def full_path_splitted(self):
         return self.group_name.split(RepoGroup.url_sep())
 
     @property
     def repositories(self):
         return Repository.query()\
                 .filter(Repository.group == self)\
                 .order_by(Repository.repo_name)
 
     @property
     def repositories_recursive_count(self):
         cnt = self.repositories.count()
 
         def children_count(group):
             cnt = 0
             for child in group.children:
                 cnt += child.repositories.count()
                 cnt += children_count(child)
             return cnt
 
         return cnt + children_count(self)
 
     def _recursive_objects(self, include_repos=True, include_groups=True):
         all_ = []
 
         def _get_members(root_gr):
             if include_repos:
                 for r in root_gr.repositories:
                     all_.append(r)
             childs = root_gr.children.all()
             if childs:
                 for gr in childs:
                     if include_groups:
                         all_.append(gr)
                     _get_members(gr)
 
         root_group = []
         if include_groups:
             root_group = [self]
 
         _get_members(self)
         return root_group + all_
 
     def recursive_groups_and_repos(self):
         """
         Recursive return all groups, with repositories in those groups
         """
         return self._recursive_objects()
 
     def recursive_groups(self):
         """
         Returns all children groups for this group including children of children
         """
         return self._recursive_objects(include_repos=False)
 
     def recursive_repos(self):
         """
         Returns all children repositories for this group
         """
         return self._recursive_objects(include_groups=False)
 
     def get_new_name(self, group_name):
         """
         returns new full group name based on parent and new name
 
         :param group_name:
         """
         path_prefix = (self.parent_group.full_path_splitted if
                        self.parent_group else [])
         return RepoGroup.url_sep().join(path_prefix + [group_name])
 
     def update_commit_cache(self, config=None):
         """
         Update cache of last changeset for newest repository inside this group, keys should be::
 
             source_repo_id
             short_id
             raw_id
             revision
             parents
             message
             date
             author
 
         """
         from rhodecode.lib.vcs.utils.helpers import parse_datetime
 
         def repo_groups_and_repos():
             all_entries = OrderedDefaultDict(list)
 
             def _get_members(root_gr, pos=0):
 
                 for repo in root_gr.repositories:
                     all_entries[root_gr].append(repo)
 
                 # fill in all parent positions
                 for parent_group in root_gr.parents:
                     all_entries[parent_group].extend(all_entries[root_gr])
 
                 children_groups = root_gr.children.all()
                 if children_groups:
                     for cnt, gr in enumerate(children_groups, 1):
                         _get_members(gr, pos=pos+cnt)
 
             _get_members(root_gr=self)
             return all_entries
 
         empty_date = datetime.datetime.fromtimestamp(0)
         for repo_group, repos in repo_groups_and_repos().items():
 
             latest_repo_cs_cache = {}
             for repo in repos:
                 repo_cs_cache = repo.changeset_cache
                 date_latest = latest_repo_cs_cache.get('date', empty_date)
                 date_current = repo_cs_cache.get('date', empty_date)
                 current_timestamp = datetime_to_time(parse_datetime(date_latest))
                 if current_timestamp < datetime_to_time(parse_datetime(date_current)):
                     latest_repo_cs_cache = repo_cs_cache
                     latest_repo_cs_cache['source_repo_id'] = repo.repo_id
 
             latest_repo_cs_cache['updated_on'] = time.time()
             repo_group.changeset_cache = latest_repo_cs_cache
             Session().add(repo_group)
             Session().commit()
 
             log.debug('updated repo group %s with new commit cache %s',
                       repo_group.group_name, latest_repo_cs_cache)
 
     def permissions(self, with_admins=True, with_owner=True,
                     expand_from_user_groups=False):
         """
         Permissions for repository groups
         """
         _admin_perm = 'group.admin'
 
         owner_row = []
         if with_owner:
             usr = AttributeDict(self.user.get_dict())
             usr.owner_row = True
             usr.permission = _admin_perm
             owner_row.append(usr)
 
         super_admin_ids = []
         super_admin_rows = []
         if with_admins:
             for usr in User.get_all_super_admins():
                 super_admin_ids.append(usr.user_id)
                 # if this admin is also owner, don't double the record
                 if usr.user_id == owner_row[0].user_id:
                     owner_row[0].admin_row = True
                 else:
                     usr = AttributeDict(usr.get_dict())
                     usr.admin_row = True
                     usr.permission = _admin_perm
                     super_admin_rows.append(usr)
 
         q = UserRepoGroupToPerm.query().filter(UserRepoGroupToPerm.group == self)
         q = q.options(joinedload(UserRepoGroupToPerm.group),
                       joinedload(UserRepoGroupToPerm.user),
                       joinedload(UserRepoGroupToPerm.permission),)
 
         # get owners and admins and permissions. We do a trick of re-writing
         # objects from sqlalchemy to named-tuples due to sqlalchemy session
         # has a global reference and changing one object propagates to all
         # others. This means if admin is also an owner admin_row that change
         # would propagate to both objects
         perm_rows = []
         for _usr in q.all():
             usr = AttributeDict(_usr.user.get_dict())
             # if this user is also owner/admin, mark as duplicate record
             if usr.user_id == owner_row[0].user_id or usr.user_id in super_admin_ids:
                 usr.duplicate_perm = True
             usr.permission = _usr.permission.permission_name
             perm_rows.append(usr)
 
         # filter the perm rows by 'default' first and then sort them by
         # admin,write,read,none permissions sorted again alphabetically in
         # each group
         perm_rows = sorted(perm_rows, key=display_user_sort)
 
         user_groups_rows = []
         if expand_from_user_groups:
             for ug in self.permission_user_groups(with_members=True):
                 for user_data in ug.members:
                     user_groups_rows.append(user_data)
 
         return super_admin_rows + owner_row + perm_rows + user_groups_rows
 
     def permission_user_groups(self, with_members=False):
         q = UserGroupRepoGroupToPerm.query()\
             .filter(UserGroupRepoGroupToPerm.group == self)
         q = q.options(joinedload(UserGroupRepoGroupToPerm.group),
                       joinedload(UserGroupRepoGroupToPerm.users_group),
                       joinedload(UserGroupRepoGroupToPerm.permission),)
 
         perm_rows = []
         for _user_group in q.all():
             entry = AttributeDict(_user_group.users_group.get_dict())
             entry.permission = _user_group.permission.permission_name
             if with_members:
                 entry.members = [x.user.get_dict()
                                  for x in _user_group.users_group.members]
             perm_rows.append(entry)
 
         perm_rows = sorted(perm_rows, key=display_user_group_sort)
         return perm_rows
 
     def get_api_data(self):
         """
         Common function for generating api data
 
         """
         group = self
         data = {
             'group_id': group.group_id,
             'group_name': group.group_name,
             'group_description': group.description_safe,
             'parent_group': group.parent_group.group_name if group.parent_group else None,
             'repositories': [x.repo_name for x in group.repositories],
             'owner': group.user.username,
         }
         return data
 
     def get_dict(self):
         # Since we transformed `group_name` to a hybrid property, we need to
         # keep compatibility with the code which uses `group_name` field.
         result = super(RepoGroup, self).get_dict()
         result['group_name'] = result.pop('_group_name', None)
         return result
 
 
 class Permission(Base, BaseModel):
     __tablename__ = 'permissions'
     __table_args__ = (
         Index('p_perm_name_idx', 'permission_name'),
         base_table_args,
     )
 
     PERMS = [
         ('hg.admin', _('RhodeCode Super Administrator')),
 
         ('repository.none', _('Repository no access')),
         ('repository.read', _('Repository read access')),
         ('repository.write', _('Repository write access')),
         ('repository.admin', _('Repository admin access')),
 
         ('group.none', _('Repository group no access')),
         ('group.read', _('Repository group read access')),
         ('group.write', _('Repository group write access')),
         ('group.admin', _('Repository group admin access')),
 
         ('usergroup.none', _('User group no access')),
         ('usergroup.read', _('User group read access')),
         ('usergroup.write', _('User group write access')),
         ('usergroup.admin', _('User group admin access')),
 
         ('branch.none', _('Branch no permissions')),
         ('branch.merge', _('Branch access by web merge')),
         ('branch.push', _('Branch access by push')),
         ('branch.push_force', _('Branch access by push with force')),
 
         ('hg.repogroup.create.false', _('Repository Group creation disabled')),
         ('hg.repogroup.create.true', _('Repository Group creation enabled')),
 
         ('hg.usergroup.create.false', _('User Group creation disabled')),
         ('hg.usergroup.create.true', _('User Group creation enabled')),
 
         ('hg.create.none', _('Repository creation disabled')),
         ('hg.create.repository', _('Repository creation enabled')),
         ('hg.create.write_on_repogroup.true', _('Repository creation enabled with write permission to a repository group')),
         ('hg.create.write_on_repogroup.false', _('Repository creation disabled with write permission to a repository group')),
 
         ('hg.fork.none', _('Repository forking disabled')),
         ('hg.fork.repository', _('Repository forking enabled')),
 
         ('hg.register.none', _('Registration disabled')),
         ('hg.register.manual_activate', _('User Registration with manual account activation')),
         ('hg.register.auto_activate', _('User Registration with automatic account activation')),
 
         ('hg.password_reset.enabled', _('Password reset enabled')),
         ('hg.password_reset.hidden', _('Password reset hidden')),
         ('hg.password_reset.disabled', _('Password reset disabled')),
 
         ('hg.extern_activate.manual', _('Manual activation of external account')),
         ('hg.extern_activate.auto', _('Automatic activation of external account')),
 
         ('hg.inherit_default_perms.false', _('Inherit object permissions from default user disabled')),
         ('hg.inherit_default_perms.true', _('Inherit object permissions from default user enabled')),
     ]
 
     # definition of system default permissions for DEFAULT user, created on
     # system setup
     DEFAULT_USER_PERMISSIONS = [
         # object perms
         'repository.read',
         'group.read',
         'usergroup.read',
         # branch, for backward compat we need same value as before so forced pushed
         'branch.push_force',
         # global
         'hg.create.repository',
         'hg.repogroup.create.false',
         'hg.usergroup.create.false',
         'hg.create.write_on_repogroup.true',
         'hg.fork.repository',
         'hg.register.manual_activate',
         'hg.password_reset.enabled',
         'hg.extern_activate.auto',
         'hg.inherit_default_perms.true',
     ]
 
     # defines which permissions are more important higher the more important
     # Weight defines which permissions are more important.
     # The higher number the more important.
     PERM_WEIGHTS = {
         'repository.none': 0,
         'repository.read': 1,
         'repository.write': 3,
         'repository.admin': 4,
 
         'group.none': 0,
         'group.read': 1,
         'group.write': 3,
         'group.admin': 4,
 
         'usergroup.none': 0,
         'usergroup.read': 1,
         'usergroup.write': 3,
         'usergroup.admin': 4,
 
         'branch.none': 0,
         'branch.merge': 1,
         'branch.push': 3,
         'branch.push_force': 4,
 
         'hg.repogroup.create.false': 0,
         'hg.repogroup.create.true': 1,
 
         'hg.usergroup.create.false': 0,
         'hg.usergroup.create.true': 1,
 
         'hg.fork.none': 0,
         'hg.fork.repository': 1,
         'hg.create.none': 0,
         'hg.create.repository': 1
     }
 
     permission_id = Column("permission_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     permission_name = Column("permission_name", String(255), nullable=True, unique=None, default=None)
     permission_longname = Column("permission_longname", String(255), nullable=True, unique=None, default=None)
 
     def __unicode__(self):
         return u"<%s('%s:%s')>" % (
             self.__class__.__name__, self.permission_id, self.permission_name
         )
 
     @classmethod
     def get_by_key(cls, key):
         return cls.query().filter(cls.permission_name == key).scalar()
 
     @classmethod
     def get_default_repo_perms(cls, user_id, repo_id=None):
         q = Session().query(UserRepoToPerm, Repository, Permission)\
             .join((Permission, UserRepoToPerm.permission_id == Permission.permission_id))\
             .join((Repository, UserRepoToPerm.repository_id == Repository.repo_id))\
             .filter(UserRepoToPerm.user_id == user_id)
         if repo_id:
             q = q.filter(UserRepoToPerm.repository_id == repo_id)
         return q.all()
 
     @classmethod
     def get_default_repo_branch_perms(cls, user_id, repo_id=None):
         q = Session().query(UserToRepoBranchPermission, UserRepoToPerm, Permission) \
             .join(
                 Permission,
                 UserToRepoBranchPermission.permission_id == Permission.permission_id) \
             .join(
                 UserRepoToPerm,
                 UserToRepoBranchPermission.rule_to_perm_id == UserRepoToPerm.repo_to_perm_id) \
             .filter(UserRepoToPerm.user_id == user_id)
 
         if repo_id:
             q = q.filter(UserToRepoBranchPermission.repository_id == repo_id)
         return q.order_by(UserToRepoBranchPermission.rule_order).all()
 
     @classmethod
     def get_default_repo_perms_from_user_group(cls, user_id, repo_id=None):
         q = Session().query(UserGroupRepoToPerm, Repository, Permission)\
             .join(
                 Permission,
                 UserGroupRepoToPerm.permission_id == Permission.permission_id)\
             .join(
                 Repository,
                 UserGroupRepoToPerm.repository_id == Repository.repo_id)\
             .join(
                 UserGroup,
                 UserGroupRepoToPerm.users_group_id ==
                 UserGroup.users_group_id)\
             .join(
                 UserGroupMember,
                 UserGroupRepoToPerm.users_group_id ==
                 UserGroupMember.users_group_id)\
             .filter(
                 UserGroupMember.user_id == user_id,
                 UserGroup.users_group_active == true())
         if repo_id:
             q = q.filter(UserGroupRepoToPerm.repository_id == repo_id)
         return q.all()
 
     @classmethod
     def get_default_repo_branch_perms_from_user_group(cls, user_id, repo_id=None):
         q = Session().query(UserGroupToRepoBranchPermission, UserGroupRepoToPerm, Permission) \
             .join(
                 Permission,
                 UserGroupToRepoBranchPermission.permission_id == Permission.permission_id) \
             .join(
                 UserGroupRepoToPerm,
                 UserGroupToRepoBranchPermission.rule_to_perm_id == UserGroupRepoToPerm.users_group_to_perm_id) \
             .join(
                 UserGroup,
                 UserGroupRepoToPerm.users_group_id == UserGroup.users_group_id) \
             .join(
                 UserGroupMember,
                 UserGroupRepoToPerm.users_group_id == UserGroupMember.users_group_id) \
             .filter(
                 UserGroupMember.user_id == user_id,
                 UserGroup.users_group_active == true())
 
         if repo_id:
             q = q.filter(UserGroupToRepoBranchPermission.repository_id == repo_id)
         return q.order_by(UserGroupToRepoBranchPermission.rule_order).all()
 
     @classmethod
     def get_default_group_perms(cls, user_id, repo_group_id=None):
         q = Session().query(UserRepoGroupToPerm, RepoGroup, Permission)\
             .join(
                 Permission,
                 UserRepoGroupToPerm.permission_id == Permission.permission_id)\
             .join(
                 RepoGroup,
                 UserRepoGroupToPerm.group_id == RepoGroup.group_id)\
             .filter(UserRepoGroupToPerm.user_id == user_id)
         if repo_group_id:
             q = q.filter(UserRepoGroupToPerm.group_id == repo_group_id)
         return q.all()
 
     @classmethod
     def get_default_group_perms_from_user_group(
             cls, user_id, repo_group_id=None):
         q = Session().query(UserGroupRepoGroupToPerm, RepoGroup, Permission)\
             .join(
                 Permission,
                 UserGroupRepoGroupToPerm.permission_id ==
                 Permission.permission_id)\
             .join(
                 RepoGroup,
                 UserGroupRepoGroupToPerm.group_id == RepoGroup.group_id)\
             .join(
                 UserGroup,
                 UserGroupRepoGroupToPerm.users_group_id ==
                 UserGroup.users_group_id)\
             .join(
                 UserGroupMember,
                 UserGroupRepoGroupToPerm.users_group_id ==
                 UserGroupMember.users_group_id)\
             .filter(
                 UserGroupMember.user_id == user_id,
                 UserGroup.users_group_active == true())
         if repo_group_id:
             q = q.filter(UserGroupRepoGroupToPerm.group_id == repo_group_id)
         return q.all()
 
     @classmethod
     def get_default_user_group_perms(cls, user_id, user_group_id=None):
         q = Session().query(UserUserGroupToPerm, UserGroup, Permission)\
             .join((Permission, UserUserGroupToPerm.permission_id == Permission.permission_id))\
             .join((UserGroup, UserUserGroupToPerm.user_group_id == UserGroup.users_group_id))\
             .filter(UserUserGroupToPerm.user_id == user_id)
         if user_group_id:
             q = q.filter(UserUserGroupToPerm.user_group_id == user_group_id)
         return q.all()
 
     @classmethod
     def get_default_user_group_perms_from_user_group(
             cls, user_id, user_group_id=None):
         TargetUserGroup = aliased(UserGroup, name='target_user_group')
         q = Session().query(UserGroupUserGroupToPerm, UserGroup, Permission)\
             .join(
                 Permission,
                 UserGroupUserGroupToPerm.permission_id ==
                 Permission.permission_id)\
             .join(
                 TargetUserGroup,
                 UserGroupUserGroupToPerm.target_user_group_id ==
                 TargetUserGroup.users_group_id)\
             .join(
                 UserGroup,
                 UserGroupUserGroupToPerm.user_group_id ==
                 UserGroup.users_group_id)\
             .join(
                 UserGroupMember,
                 UserGroupUserGroupToPerm.user_group_id ==
                 UserGroupMember.users_group_id)\
             .filter(
                 UserGroupMember.user_id == user_id,
                 UserGroup.users_group_active == true())
         if user_group_id:
             q = q.filter(
                 UserGroupUserGroupToPerm.user_group_id == user_group_id)
 
         return q.all()
 
 
 class UserRepoToPerm(Base, BaseModel):
     __tablename__ = 'repo_to_perm'
     __table_args__ = (
         UniqueConstraint('user_id', 'repository_id', 'permission_id'),
         base_table_args
     )
 
     repo_to_perm_id = Column("repo_to_perm_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=False, unique=None, default=None)
     permission_id = Column("permission_id", Integer(), ForeignKey('permissions.permission_id'), nullable=False, unique=None, default=None)
     repository_id = Column("repository_id", Integer(), ForeignKey('repositories.repo_id'), nullable=False, unique=None, default=None)
 
     user = relationship('User')
     repository = relationship('Repository')
     permission = relationship('Permission')
 
     branch_perm_entry = relationship('UserToRepoBranchPermission', cascade="all, delete, delete-orphan", lazy='joined')
 
     @classmethod
     def create(cls, user, repository, permission):
         n = cls()
         n.user = user
         n.repository = repository
         n.permission = permission
         Session().add(n)
         return n
 
     def __unicode__(self):
         return u'<%s => %s >' % (self.user, self.repository)
 
 
 class UserUserGroupToPerm(Base, BaseModel):
     __tablename__ = 'user_user_group_to_perm'
     __table_args__ = (
         UniqueConstraint('user_id', 'user_group_id', 'permission_id'),
         base_table_args
     )
 
     user_user_group_to_perm_id = Column("user_user_group_to_perm_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=False, unique=None, default=None)
     permission_id = Column("permission_id", Integer(), ForeignKey('permissions.permission_id'), nullable=False, unique=None, default=None)
     user_group_id = Column("user_group_id", Integer(), ForeignKey('users_groups.users_group_id'), nullable=False, unique=None, default=None)
 
     user = relationship('User')
     user_group = relationship('UserGroup')
     permission = relationship('Permission')
 
     @classmethod
     def create(cls, user, user_group, permission):
         n = cls()
         n.user = user
         n.user_group = user_group
         n.permission = permission
         Session().add(n)
         return n
 
     def __unicode__(self):
         return u'<%s => %s >' % (self.user, self.user_group)
 
 
 class UserToPerm(Base, BaseModel):
     __tablename__ = 'user_to_perm'
     __table_args__ = (
         UniqueConstraint('user_id', 'permission_id'),
         base_table_args
     )
 
     user_to_perm_id = Column("user_to_perm_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=False, unique=None, default=None)
     permission_id = Column("permission_id", Integer(), ForeignKey('permissions.permission_id'), nullable=False, unique=None, default=None)
 
     user = relationship('User')
     permission = relationship('Permission', lazy='joined')
 
     def __unicode__(self):
         return u'<%s => %s >' % (self.user, self.permission)
 
 
 class UserGroupRepoToPerm(Base, BaseModel):
     __tablename__ = 'users_group_repo_to_perm'
     __table_args__ = (
         UniqueConstraint('repository_id', 'users_group_id', 'permission_id'),
         base_table_args
     )
 
     users_group_to_perm_id = Column("users_group_to_perm_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     users_group_id = Column("users_group_id", Integer(), ForeignKey('users_groups.users_group_id'), nullable=False, unique=None, default=None)
     permission_id = Column("permission_id", Integer(), ForeignKey('permissions.permission_id'), nullable=False, unique=None, default=None)
     repository_id = Column("repository_id", Integer(), ForeignKey('repositories.repo_id'), nullable=False, unique=None, default=None)
 
     users_group = relationship('UserGroup')
     permission = relationship('Permission')
     repository = relationship('Repository')
     user_group_branch_perms = relationship('UserGroupToRepoBranchPermission', cascade='all')
 
     @classmethod
     def create(cls, users_group, repository, permission):
         n = cls()
         n.users_group = users_group
         n.repository = repository
         n.permission = permission
         Session().add(n)
         return n
 
     def __unicode__(self):
         return u'<UserGroupRepoToPerm:%s => %s >' % (self.users_group, self.repository)
 
 
 class UserGroupUserGroupToPerm(Base, BaseModel):
     __tablename__ = 'user_group_user_group_to_perm'
     __table_args__ = (
         UniqueConstraint('target_user_group_id', 'user_group_id', 'permission_id'),
         CheckConstraint('target_user_group_id != user_group_id'),
         base_table_args
     )
 
     user_group_user_group_to_perm_id = Column("user_group_user_group_to_perm_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     target_user_group_id = Column("target_user_group_id", Integer(), ForeignKey('users_groups.users_group_id'), nullable=False, unique=None, default=None)
     permission_id = Column("permission_id", Integer(), ForeignKey('permissions.permission_id'), nullable=False, unique=None, default=None)
     user_group_id = Column("user_group_id", Integer(), ForeignKey('users_groups.users_group_id'), nullable=False, unique=None, default=None)
 
     target_user_group = relationship('UserGroup', primaryjoin='UserGroupUserGroupToPerm.target_user_group_id==UserGroup.users_group_id')
     user_group = relationship('UserGroup', primaryjoin='UserGroupUserGroupToPerm.user_group_id==UserGroup.users_group_id')
     permission = relationship('Permission')
 
     @classmethod
     def create(cls, target_user_group, user_group, permission):
         n = cls()
         n.target_user_group = target_user_group
         n.user_group = user_group
         n.permission = permission
         Session().add(n)
         return n
 
     def __unicode__(self):
         return u'<UserGroupUserGroup:%s => %s >' % (self.target_user_group, self.user_group)
 
 
 class UserGroupToPerm(Base, BaseModel):
     __tablename__ = 'users_group_to_perm'
     __table_args__ = (
         UniqueConstraint('users_group_id', 'permission_id',),
         base_table_args
     )
 
     users_group_to_perm_id = Column("users_group_to_perm_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     users_group_id = Column("users_group_id", Integer(), ForeignKey('users_groups.users_group_id'), nullable=False, unique=None, default=None)
     permission_id = Column("permission_id", Integer(), ForeignKey('permissions.permission_id'), nullable=False, unique=None, default=None)
 
     users_group = relationship('UserGroup')
     permission = relationship('Permission')
 
 
 class UserRepoGroupToPerm(Base, BaseModel):
     __tablename__ = 'user_repo_group_to_perm'
     __table_args__ = (
         UniqueConstraint('user_id', 'group_id', 'permission_id'),
         base_table_args
     )
 
     group_to_perm_id = Column("group_to_perm_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=False, unique=None, default=None)
     group_id = Column("group_id", Integer(), ForeignKey('groups.group_id'), nullable=False, unique=None, default=None)
     permission_id = Column("permission_id", Integer(), ForeignKey('permissions.permission_id'), nullable=False, unique=None, default=None)
 
     user = relationship('User')
     group = relationship('RepoGroup')
     permission = relationship('Permission')
 
     @classmethod
     def create(cls, user, repository_group, permission):
         n = cls()
         n.user = user
         n.group = repository_group
         n.permission = permission
         Session().add(n)
         return n
 
 
 class UserGroupRepoGroupToPerm(Base, BaseModel):
     __tablename__ = 'users_group_repo_group_to_perm'
     __table_args__ = (
         UniqueConstraint('users_group_id', 'group_id'),
         base_table_args
     )
 
     users_group_repo_group_to_perm_id = Column("users_group_repo_group_to_perm_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     users_group_id = Column("users_group_id", Integer(), ForeignKey('users_groups.users_group_id'), nullable=False, unique=None, default=None)
     group_id = Column("group_id", Integer(), ForeignKey('groups.group_id'), nullable=False, unique=None, default=None)
     permission_id = Column("permission_id", Integer(), ForeignKey('permissions.permission_id'), nullable=False, unique=None, default=None)
 
     users_group = relationship('UserGroup')
     permission = relationship('Permission')
     group = relationship('RepoGroup')
 
     @classmethod
     def create(cls, user_group, repository_group, permission):
         n = cls()
         n.users_group = user_group
         n.group = repository_group
         n.permission = permission
         Session().add(n)
         return n
 
     def __unicode__(self):
         return u'<UserGroupRepoGroupToPerm:%s => %s >' % (self.users_group, self.group)
 
 
 class Statistics(Base, BaseModel):
     __tablename__ = 'statistics'
     __table_args__ = (
          base_table_args
     )
 
     stat_id = Column("stat_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     repository_id = Column("repository_id", Integer(), ForeignKey('repositories.repo_id'), nullable=False, unique=True, default=None)
     stat_on_revision = Column("stat_on_revision", Integer(), nullable=False)
     commit_activity = Column("commit_activity", LargeBinary(1000000), nullable=False)#JSON data
     commit_activity_combined = Column("commit_activity_combined", LargeBinary(), nullable=False)#JSON data
     languages = Column("languages", LargeBinary(1000000), nullable=False)#JSON data
 
     repository = relationship('Repository', single_parent=True)
 
 
 class UserFollowing(Base, BaseModel):
     __tablename__ = 'user_followings'
     __table_args__ = (
         UniqueConstraint('user_id', 'follows_repository_id'),
         UniqueConstraint('user_id', 'follows_user_id'),
         base_table_args
     )
 
     user_following_id = Column("user_following_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=False, unique=None, default=None)
     follows_repo_id = Column("follows_repository_id", Integer(), ForeignKey('repositories.repo_id'), nullable=True, unique=None, default=None)
     follows_user_id = Column("follows_user_id", Integer(), ForeignKey('users.user_id'), nullable=True, unique=None, default=None)
     follows_from = Column('follows_from', DateTime(timezone=False), nullable=True, unique=None, default=datetime.datetime.now)
 
     user = relationship('User', primaryjoin='User.user_id==UserFollowing.user_id')
 
     follows_user = relationship('User', primaryjoin='User.user_id==UserFollowing.follows_user_id')
     follows_repository = relationship('Repository', order_by='Repository.repo_name')
 
     @classmethod
     def get_repo_followers(cls, repo_id):
         return cls.query().filter(cls.follows_repo_id == repo_id)
 
 
 class CacheKey(Base, BaseModel):
     __tablename__ = 'cache_invalidation'
     __table_args__ = (
         UniqueConstraint('cache_key'),
         Index('key_idx', 'cache_key'),
         base_table_args,
     )
 
     CACHE_TYPE_FEED = 'FEED'
 
     # namespaces used to register process/thread aware caches
     REPO_INVALIDATION_NAMESPACE = 'repo_cache:{repo_id}'
     SETTINGS_INVALIDATION_NAMESPACE = 'system_settings'
 
     cache_id = Column("cache_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     cache_key = Column("cache_key", String(255), nullable=True, unique=None, default=None)
     cache_args = Column("cache_args", String(255), nullable=True, unique=None, default=None)
     cache_state_uid = Column("cache_state_uid", String(255), nullable=True, unique=None, default=None)
     cache_active = Column("cache_active", Boolean(), nullable=True, unique=None, default=False)
 
     def __init__(self, cache_key, cache_args='', cache_state_uid=None):
         self.cache_key = cache_key
         self.cache_args = cache_args
         self.cache_active = False
         # first key should be same for all entries, since all workers should share it
         self.cache_state_uid = cache_state_uid or self.generate_new_state_uid()
 
     def __unicode__(self):
         return u"<%s('%s:%s[%s]')>" % (
             self.__class__.__name__,
             self.cache_id, self.cache_key, self.cache_active)
 
     def _cache_key_partition(self):
         prefix, repo_name, suffix = self.cache_key.partition(self.cache_args)
         return prefix, repo_name, suffix
 
     def get_prefix(self):
         """
         Try to extract prefix from existing cache key. The key could consist
         of prefix, repo_name, suffix
         """
         # this returns prefix, repo_name, suffix
         return self._cache_key_partition()[0]
 
     def get_suffix(self):
         """
         get suffix that might have been used in _get_cache_key to
         generate self.cache_key. Only used for informational purposes
         in repo_edit.mako.
         """
         # prefix, repo_name, suffix
         return self._cache_key_partition()[2]
 
     @classmethod
     def generate_new_state_uid(cls, based_on=None):
         if based_on:
             return str(uuid.uuid5(uuid.NAMESPACE_URL, safe_str(based_on)))
         else:
             return str(uuid.uuid4())
 
     @classmethod
     def delete_all_cache(cls):
         """
         Delete all cache keys from database.
         Should only be run when all instances are down and all entries
         thus stale.
         """
         cls.query().delete()
         Session().commit()
 
     @classmethod
     def set_invalidate(cls, cache_uid, delete=False):
         """
         Mark all caches of a repo as invalid in the database.
         """
 
         try:
             qry = Session().query(cls).filter(cls.cache_args == cache_uid)
             if delete:
                 qry.delete()
                 log.debug('cache objects deleted for cache args %s',
                           safe_str(cache_uid))
             else:
                 qry.update({"cache_active": False,
                             "cache_state_uid": cls.generate_new_state_uid()})
                 log.debug('cache objects marked as invalid for cache args %s',
                           safe_str(cache_uid))
 
             Session().commit()
         except Exception:
             log.exception(
                 'Cache key invalidation failed for cache args %s',
                 safe_str(cache_uid))
             Session().rollback()
 
     @classmethod
     def get_active_cache(cls, cache_key):
         inv_obj = cls.query().filter(cls.cache_key == cache_key).scalar()
         if inv_obj:
             return inv_obj
         return None
 
     @classmethod
     def get_namespace_map(cls, namespace):
         return {
             x.cache_key: x
             for x in cls.query().filter(cls.cache_args == namespace)}
 
 
 class ChangesetComment(Base, BaseModel):
     __tablename__ = 'changeset_comments'
     __table_args__ = (
         Index('cc_revision_idx', 'revision'),
         base_table_args,
     )
 
     COMMENT_OUTDATED = u'comment_outdated'
     COMMENT_TYPE_NOTE = u'note'
     COMMENT_TYPE_TODO = u'todo'
     COMMENT_TYPES = [COMMENT_TYPE_NOTE, COMMENT_TYPE_TODO]
 
     comment_id = Column('comment_id', Integer(), nullable=False, primary_key=True)
     repo_id = Column('repo_id', Integer(), ForeignKey('repositories.repo_id'), nullable=False)
     revision = Column('revision', String(40), nullable=True)
     pull_request_id = Column("pull_request_id", Integer(), ForeignKey('pull_requests.pull_request_id'), nullable=True)
     pull_request_version_id = Column("pull_request_version_id", Integer(), ForeignKey('pull_request_versions.pull_request_version_id'), nullable=True)
     line_no = Column('line_no', Unicode(10), nullable=True)
     hl_lines = Column('hl_lines', Unicode(512), nullable=True)
     f_path = Column('f_path', Unicode(1000), nullable=True)
     user_id = Column('user_id', Integer(), ForeignKey('users.user_id'), nullable=False)
     text = Column('text', UnicodeText().with_variant(UnicodeText(25000), 'mysql'), nullable=False)
     created_on = Column('created_on', DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
     modified_at = Column('modified_at', DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
     renderer = Column('renderer', Unicode(64), nullable=True)
     display_state = Column('display_state',  Unicode(128), nullable=True)
 
     comment_type = Column('comment_type',  Unicode(128), nullable=True, default=COMMENT_TYPE_NOTE)
     resolved_comment_id = Column('resolved_comment_id', Integer(), ForeignKey('changeset_comments.comment_id'), nullable=True)
 
     resolved_comment = relationship('ChangesetComment', remote_side=comment_id, back_populates='resolved_by')
     resolved_by = relationship('ChangesetComment', back_populates='resolved_comment')
 
     author = relationship('User', lazy='joined')
     repo = relationship('Repository')
     status_change = relationship('ChangesetStatus', cascade="all, delete, delete-orphan", lazy='joined')
     pull_request = relationship('PullRequest', lazy='joined')
     pull_request_version = relationship('PullRequestVersion')
 
     @classmethod
     def get_users(cls, revision=None, pull_request_id=None):
         """
         Returns user associated with this ChangesetComment. ie those
         who actually commented
 
         :param cls:
         :param revision:
         """
         q = Session().query(User)\
                 .join(ChangesetComment.author)
         if revision:
             q = q.filter(cls.revision == revision)
         elif pull_request_id:
             q = q.filter(cls.pull_request_id == pull_request_id)
         return q.all()
 
     @classmethod
     def get_index_from_version(cls, pr_version, versions):
         num_versions = [x.pull_request_version_id for x in versions]
         try:
             return num_versions.index(pr_version) +1
         except (IndexError, ValueError):
             return
 
     @property
     def outdated(self):
         return self.display_state == self.COMMENT_OUTDATED
 
     def outdated_at_version(self, version):
         """
         Checks if comment is outdated for given pull request version
         """
         return self.outdated and self.pull_request_version_id != version
 
     def older_than_version(self, version):
         """
         Checks if comment is made from previous version than given
         """
         if version is None:
             return self.pull_request_version_id is not None
 
         return self.pull_request_version_id < version
 
     @property
     def resolved(self):
         return self.resolved_by[0] if self.resolved_by else None
 
     @property
     def is_todo(self):
         return self.comment_type == self.COMMENT_TYPE_TODO
 
     @property
     def is_inline(self):
         return self.line_no and self.f_path
 
     def get_index_version(self, versions):
         return self.get_index_from_version(
             self.pull_request_version_id, versions)
 
     def __repr__(self):
         if self.comment_id:
             return '<DB:Comment #%s>' % self.comment_id
         else:
             return '<DB:Comment at %#x>' % id(self)
 
     def get_api_data(self):
         comment = self
         data = {
             'comment_id': comment.comment_id,
             'comment_type': comment.comment_type,
             'comment_text': comment.text,
             'comment_status': comment.status_change,
             'comment_f_path': comment.f_path,
             'comment_lineno': comment.line_no,
             'comment_author': comment.author,
             'comment_created_on': comment.created_on,
             'comment_resolved_by': self.resolved
         }
         return data
 
     def __json__(self):
         data = dict()
         data.update(self.get_api_data())
         return data
 
 
 class ChangesetStatus(Base, BaseModel):
     __tablename__ = 'changeset_statuses'
     __table_args__ = (
         Index('cs_revision_idx', 'revision'),
         Index('cs_version_idx', 'version'),
         UniqueConstraint('repo_id', 'revision', 'version'),
         base_table_args
     )
 
     STATUS_NOT_REVIEWED = DEFAULT = 'not_reviewed'
     STATUS_APPROVED = 'approved'
     STATUS_REJECTED = 'rejected'
     STATUS_UNDER_REVIEW = 'under_review'
 
     STATUSES = [
         (STATUS_NOT_REVIEWED, _("Not Reviewed")),  # (no icon) and default
         (STATUS_APPROVED, _("Approved")),
         (STATUS_REJECTED, _("Rejected")),
         (STATUS_UNDER_REVIEW, _("Under Review")),
     ]
 
     changeset_status_id = Column('changeset_status_id', Integer(), nullable=False, primary_key=True)
     repo_id = Column('repo_id', Integer(), ForeignKey('repositories.repo_id'), nullable=False)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=False, unique=None)
     revision = Column('revision', String(40), nullable=False)
     status = Column('status', String(128), nullable=False, default=DEFAULT)
     changeset_comment_id = Column('changeset_comment_id', Integer(), ForeignKey('changeset_comments.comment_id'))
     modified_at = Column('modified_at', DateTime(), nullable=False, default=datetime.datetime.now)
     version = Column('version', Integer(), nullable=False, default=0)
     pull_request_id = Column("pull_request_id", Integer(), ForeignKey('pull_requests.pull_request_id'), nullable=True)
 
     author = relationship('User', lazy='joined')
     repo = relationship('Repository')
     comment = relationship('ChangesetComment', lazy='joined')
     pull_request = relationship('PullRequest', lazy='joined')
 
     def __unicode__(self):
         return u"<%s('%s[v%s]:%s')>" % (
             self.__class__.__name__,
             self.status, self.version, self.author
         )
 
     @classmethod
     def get_status_lbl(cls, value):
         return dict(cls.STATUSES).get(value)
 
     @property
     def status_lbl(self):
         return ChangesetStatus.get_status_lbl(self.status)
 
     def get_api_data(self):
         status = self
         data = {
             'status_id': status.changeset_status_id,
             'status': status.status,
         }
         return data
 
     def __json__(self):
         data = dict()
         data.update(self.get_api_data())
         return data
 
 
 class _SetState(object):
     """
     Context processor allowing changing state for sensitive operation such as
     pull request update or merge
     """
 
     def __init__(self, pull_request, pr_state, back_state=None):
         self._pr = pull_request
         self._org_state = back_state or pull_request.pull_request_state
         self._pr_state = pr_state
         self._current_state = None
 
     def __enter__(self):
         log.debug('StateLock: entering set state context, setting state to: `%s`',
                   self._pr_state)
         self.set_pr_state(self._pr_state)
         return self
 
     def __exit__(self, exc_type, exc_val, exc_tb):
         if exc_val is not None:
             log.error(traceback.format_exc(exc_tb))
             return None
 
         self.set_pr_state(self._org_state)
         log.debug('StateLock: exiting set state context, setting state to: `%s`',
                   self._org_state)
     @property
     def state(self):
         return self._current_state
 
     def set_pr_state(self, pr_state):
         try:
             self._pr.pull_request_state = pr_state
             Session().add(self._pr)
             Session().commit()
             self._current_state = pr_state
         except Exception:
             log.exception('Failed to set PullRequest %s state to %s', self._pr, pr_state)
             raise
 
 class _PullRequestBase(BaseModel):
     """
     Common attributes of pull request and version entries.
     """
 
     # .status values
     STATUS_NEW = u'new'
     STATUS_OPEN = u'open'
     STATUS_CLOSED = u'closed'
 
     # available states
     STATE_CREATING = u'creating'
     STATE_UPDATING = u'updating'
     STATE_MERGING = u'merging'
     STATE_CREATED = u'created'
 
     title = Column('title', Unicode(255), nullable=True)
     description = Column(
         'description', UnicodeText().with_variant(UnicodeText(10240), 'mysql'),
         nullable=True)
     description_renderer = Column('description_renderer', Unicode(64), nullable=True)
 
     # new/open/closed status of pull request (not approve/reject/etc)
     status = Column('status', Unicode(255), nullable=False, default=STATUS_NEW)
     created_on = Column(
         'created_on', DateTime(timezone=False), nullable=False,
         default=datetime.datetime.now)
     updated_on = Column(
         'updated_on', DateTime(timezone=False), nullable=False,
         default=datetime.datetime.now)
 
     pull_request_state = Column("pull_request_state", String(255), nullable=True)
 
     @declared_attr
     def user_id(cls):
         return Column(
             "user_id", Integer(), ForeignKey('users.user_id'), nullable=False,
             unique=None)
 
     # 500 revisions max
     _revisions = Column(
         'revisions', UnicodeText().with_variant(UnicodeText(20500), 'mysql'))
 
     @declared_attr
     def source_repo_id(cls):
         # TODO: dan: rename column to source_repo_id
         return Column(
             'org_repo_id', Integer(), ForeignKey('repositories.repo_id'),
             nullable=False)
 
     _source_ref = Column('org_ref', Unicode(255), nullable=False)
 
     @hybrid_property
     def source_ref(self):
         return self._source_ref
 
     @source_ref.setter
     def source_ref(self, val):
         parts = (val or '').split(':')
         if len(parts) != 3:
             raise ValueError(
                 'Invalid reference format given: {}, expected X:Y:Z'.format(val))
         self._source_ref = safe_unicode(val)
 
     _target_ref = Column('other_ref', Unicode(255), nullable=False)
 
     @hybrid_property
     def target_ref(self):
         return self._target_ref
 
     @target_ref.setter
     def target_ref(self, val):
         parts = (val or '').split(':')
         if len(parts) != 3:
             raise ValueError(
                 'Invalid reference format given: {}, expected X:Y:Z'.format(val))
         self._target_ref = safe_unicode(val)
 
     @declared_attr
     def target_repo_id(cls):
         # TODO: dan: rename column to target_repo_id
         return Column(
             'other_repo_id', Integer(), ForeignKey('repositories.repo_id'),
             nullable=False)
 
     _shadow_merge_ref = Column('shadow_merge_ref', Unicode(255), nullable=True)
 
     # TODO: dan: rename column to last_merge_source_rev
     _last_merge_source_rev = Column(
         'last_merge_org_rev', String(40), nullable=True)
     # TODO: dan: rename column to last_merge_target_rev
     _last_merge_target_rev = Column(
         'last_merge_other_rev', String(40), nullable=True)
     _last_merge_status = Column('merge_status', Integer(), nullable=True)
     merge_rev = Column('merge_rev', String(40), nullable=True)
 
     reviewer_data = Column(
         'reviewer_data_json', MutationObj.as_mutable(
             JsonType(dialect_map=dict(mysql=UnicodeText(16384)))))
 
     @property
     def reviewer_data_json(self):
         return json.dumps(self.reviewer_data)
 
     @hybrid_property
     def description_safe(self):
         from rhodecode.lib import helpers as h
         return h.escape(self.description)
 
     @hybrid_property
     def revisions(self):
         return self._revisions.split(':') if self._revisions else []
 
     @revisions.setter
     def revisions(self, val):
         self._revisions = ':'.join(val)
 
     @hybrid_property
     def last_merge_status(self):
         return safe_int(self._last_merge_status)
 
     @last_merge_status.setter
     def last_merge_status(self, val):
         self._last_merge_status = val
 
     @declared_attr
     def author(cls):
         return relationship('User', lazy='joined')
 
     @declared_attr
     def source_repo(cls):
         return relationship(
             'Repository',
             primaryjoin='%s.source_repo_id==Repository.repo_id' % cls.__name__)
 
     @property
     def source_ref_parts(self):
         return self.unicode_to_reference(self.source_ref)
 
     @declared_attr
     def target_repo(cls):
         return relationship(
             'Repository',
             primaryjoin='%s.target_repo_id==Repository.repo_id' % cls.__name__)
 
     @property
     def target_ref_parts(self):
         return self.unicode_to_reference(self.target_ref)
 
     @property
     def shadow_merge_ref(self):
         return self.unicode_to_reference(self._shadow_merge_ref)
 
     @shadow_merge_ref.setter
     def shadow_merge_ref(self, ref):
         self._shadow_merge_ref = self.reference_to_unicode(ref)
 
     @staticmethod
     def unicode_to_reference(raw):
         """
         Convert a unicode (or string) to a reference object.
         If unicode evaluates to False it returns None.
         """
         if raw:
             refs = raw.split(':')
             return Reference(*refs)
         else:
             return None
 
     @staticmethod
     def reference_to_unicode(ref):
         """
         Convert a reference object to unicode.
         If reference is None it returns None.
         """
         if ref:
             return u':'.join(ref)
         else:
             return None
 
     def get_api_data(self, with_merge_state=True):
         from rhodecode.model.pull_request import PullRequestModel
 
         pull_request = self
         if with_merge_state:
             merge_status = PullRequestModel().merge_status(pull_request)
             merge_state = {
                 'status': merge_status[0],
                 'message': safe_unicode(merge_status[1]),
             }
         else:
             merge_state = {'status': 'not_available',
                            'message': 'not_available'}
 
         merge_data = {
             'clone_url': PullRequestModel().get_shadow_clone_url(pull_request),
             'reference': (
                 pull_request.shadow_merge_ref._asdict()
                 if pull_request.shadow_merge_ref else None),
         }
 
         data = {
             'pull_request_id': pull_request.pull_request_id,
             'url': PullRequestModel().get_url(pull_request),
             'title': pull_request.title,
             'description': pull_request.description,
             'status': pull_request.status,
             'state': pull_request.pull_request_state,
             'created_on': pull_request.created_on,
             'updated_on': pull_request.updated_on,
             'commit_ids': pull_request.revisions,
             'review_status': pull_request.calculated_review_status(),
             'mergeable': merge_state,
             'source': {
                 'clone_url': pull_request.source_repo.clone_url(),
                 'repository': pull_request.source_repo.repo_name,
                 'reference': {
                     'name': pull_request.source_ref_parts.name,
                     'type': pull_request.source_ref_parts.type,
                     'commit_id': pull_request.source_ref_parts.commit_id,
                 },
             },
             'target': {
                 'clone_url': pull_request.target_repo.clone_url(),
                 'repository': pull_request.target_repo.repo_name,
                 'reference': {
                     'name': pull_request.target_ref_parts.name,
                     'type': pull_request.target_ref_parts.type,
                     'commit_id': pull_request.target_ref_parts.commit_id,
                 },
             },
             'merge': merge_data,
             'author': pull_request.author.get_api_data(include_secrets=False,
                                                        details='basic'),
             'reviewers': [
                 {
                     'user': reviewer.get_api_data(include_secrets=False,
                                                   details='basic'),
                     'reasons': reasons,
                     'review_status': st[0][1].status if st else 'not_reviewed',
                 }
                 for obj, reviewer, reasons, mandatory, st in
                 pull_request.reviewers_statuses()
             ]
         }
 
         return data
 
     def set_state(self, pull_request_state, final_state=None):
         """
         # goes from initial state to updating to initial state.
         # initial state can be changed by specifying back_state=
         with pull_request_obj.set_state(PullRequest.STATE_UPDATING):
            pull_request.merge()
 
         :param pull_request_state:
         :param final_state:
 
         """
 
         return _SetState(self, pull_request_state, back_state=final_state)
 
 
 class PullRequest(Base, _PullRequestBase):
     __tablename__ = 'pull_requests'
     __table_args__ = (
         base_table_args,
     )
 
     pull_request_id = Column(
         'pull_request_id', Integer(), nullable=False, primary_key=True)
 
     def __repr__(self):
         if self.pull_request_id:
             return '<DB:PullRequest #%s>' % self.pull_request_id
         else:
             return '<DB:PullRequest at %#x>' % id(self)
 
     reviewers = relationship('PullRequestReviewers',
                              cascade="all, delete, delete-orphan")
     statuses = relationship('ChangesetStatus',
                             cascade="all, delete, delete-orphan")
     comments = relationship('ChangesetComment',
                             cascade="all, delete, delete-orphan")
     versions = relationship('PullRequestVersion',
                             cascade="all, delete, delete-orphan",
                             lazy='dynamic')
 
     @classmethod
     def get_pr_display_object(cls, pull_request_obj, org_pull_request_obj,
                               internal_methods=None):
 
         class PullRequestDisplay(object):
             """
             Special object wrapper for showing PullRequest data via Versions
             It mimics PR object as close as possible. This is read only object
             just for display
             """
 
             def __init__(self, attrs, internal=None):
                 self.attrs = attrs
                 # internal have priority over the given ones via attrs
                 self.internal = internal or ['versions']
 
             def __getattr__(self, item):
                 if item in self.internal:
                     return getattr(self, item)
                 try:
                     return self.attrs[item]
                 except KeyError:
                     raise AttributeError(
                         '%s object has no attribute %s' % (self, item))
 
             def __repr__(self):
                 return '<DB:PullRequestDisplay #%s>' % self.attrs.get('pull_request_id')
 
             def versions(self):
                 return pull_request_obj.versions.order_by(
                     PullRequestVersion.pull_request_version_id).all()
 
             def is_closed(self):
                 return pull_request_obj.is_closed()
 
             @property
             def pull_request_version_id(self):
                 return getattr(pull_request_obj, 'pull_request_version_id', None)
 
         attrs = StrictAttributeDict(pull_request_obj.get_api_data(with_merge_state=False))
 
         attrs.author = StrictAttributeDict(
             pull_request_obj.author.get_api_data())
         if pull_request_obj.target_repo:
             attrs.target_repo = StrictAttributeDict(
                 pull_request_obj.target_repo.get_api_data())
             attrs.target_repo.clone_url = pull_request_obj.target_repo.clone_url
 
         if pull_request_obj.source_repo:
             attrs.source_repo = StrictAttributeDict(
                 pull_request_obj.source_repo.get_api_data())
             attrs.source_repo.clone_url = pull_request_obj.source_repo.clone_url
 
         attrs.source_ref_parts = pull_request_obj.source_ref_parts
         attrs.target_ref_parts = pull_request_obj.target_ref_parts
         attrs.revisions = pull_request_obj.revisions
 
         attrs.shadow_merge_ref = org_pull_request_obj.shadow_merge_ref
         attrs.reviewer_data = org_pull_request_obj.reviewer_data
         attrs.reviewer_data_json = org_pull_request_obj.reviewer_data_json
 
         return PullRequestDisplay(attrs, internal=internal_methods)
 
     def is_closed(self):
         return self.status == self.STATUS_CLOSED
 
     def __json__(self):
         return {
             'revisions': self.revisions,
         }
 
     def calculated_review_status(self):
         from rhodecode.model.changeset_status import ChangesetStatusModel
         return ChangesetStatusModel().calculated_review_status(self)
 
     def reviewers_statuses(self):
         from rhodecode.model.changeset_status import ChangesetStatusModel
         return ChangesetStatusModel().reviewers_statuses(self)
 
     @property
     def workspace_id(self):
         from rhodecode.model.pull_request import PullRequestModel
         return PullRequestModel()._workspace_id(self)
 
     def get_shadow_repo(self):
         workspace_id = self.workspace_id
+        shadow_repository_path = self.target_repo.get_shadow_repository_path(workspace_id)
+        if os.path.isdir(shadow_repository_path):
         vcs_obj = self.target_repo.scm_instance()
-        shadow_repository_path = vcs_obj._get_shadow_repository_path(
-            self.target_repo.repo_id, workspace_id)
-        if os.path.isdir(shadow_repository_path):
             return vcs_obj.get_shadow_instance(shadow_repository_path)
 
 
 class PullRequestVersion(Base, _PullRequestBase):
     __tablename__ = 'pull_request_versions'
     __table_args__ = (
         base_table_args,
     )
 
     pull_request_version_id = Column(
         'pull_request_version_id', Integer(), nullable=False, primary_key=True)
     pull_request_id = Column(
         'pull_request_id', Integer(),
         ForeignKey('pull_requests.pull_request_id'), nullable=False)
     pull_request = relationship('PullRequest')
 
     def __repr__(self):
         if self.pull_request_version_id:
             return '<DB:PullRequestVersion #%s>' % self.pull_request_version_id
         else:
             return '<DB:PullRequestVersion at %#x>' % id(self)
 
     @property
     def reviewers(self):
         return self.pull_request.reviewers
 
     @property
     def versions(self):
         return self.pull_request.versions
 
     def is_closed(self):
         # calculate from original
         return self.pull_request.status == self.STATUS_CLOSED
 
     def calculated_review_status(self):
         return self.pull_request.calculated_review_status()
 
     def reviewers_statuses(self):
         return self.pull_request.reviewers_statuses()
 
 
 class PullRequestReviewers(Base, BaseModel):
     __tablename__ = 'pull_request_reviewers'
     __table_args__ = (
         base_table_args,
     )
 
     @hybrid_property
     def reasons(self):
         if not self._reasons:
             return []
         return self._reasons
 
     @reasons.setter
     def reasons(self, val):
         val = val or []
         if any(not isinstance(x, compat.string_types) for x in val):
             raise Exception('invalid reasons type, must be list of strings')
         self._reasons = val
 
     pull_requests_reviewers_id = Column(
         'pull_requests_reviewers_id', Integer(), nullable=False,
         primary_key=True)
     pull_request_id = Column(
         "pull_request_id", Integer(),
         ForeignKey('pull_requests.pull_request_id'), nullable=False)
     user_id = Column(
         "user_id", Integer(), ForeignKey('users.user_id'), nullable=True)
     _reasons = Column(
         'reason', MutationList.as_mutable(
             JsonType('list', dialect_map=dict(mysql=UnicodeText(16384)))))
 
     mandatory = Column("mandatory", Boolean(), nullable=False, default=False)
     user = relationship('User')
     pull_request = relationship('PullRequest')
 
     rule_data = Column(
         'rule_data_json',
         JsonType(dialect_map=dict(mysql=UnicodeText(16384))))
 
     def rule_user_group_data(self):
         """
         Returns the voting user group rule data for this reviewer
         """
 
         if self.rule_data and 'vote_rule' in self.rule_data:
             user_group_data = {}
             if 'rule_user_group_entry_id' in self.rule_data:
                 # means a group with voting rules !
                 user_group_data['id'] = self.rule_data['rule_user_group_entry_id']
                 user_group_data['name'] = self.rule_data['rule_name']
                 user_group_data['vote_rule'] = self.rule_data['vote_rule']
 
             return user_group_data
 
     def __unicode__(self):
         return u"<%s('id:%s')>" % (self.__class__.__name__,
                                    self.pull_requests_reviewers_id)
 
 
 class Notification(Base, BaseModel):
     __tablename__ = 'notifications'
     __table_args__ = (
         Index('notification_type_idx', 'type'),
         base_table_args,
     )
 
     TYPE_CHANGESET_COMMENT = u'cs_comment'
     TYPE_MESSAGE = u'message'
     TYPE_MENTION = u'mention'
     TYPE_REGISTRATION = u'registration'
     TYPE_PULL_REQUEST = u'pull_request'
     TYPE_PULL_REQUEST_COMMENT = u'pull_request_comment'
 
     notification_id = Column('notification_id', Integer(), nullable=False, primary_key=True)
     subject = Column('subject', Unicode(512), nullable=True)
     body = Column('body', UnicodeText().with_variant(UnicodeText(50000), 'mysql'), nullable=True)
     created_by = Column("created_by", Integer(), ForeignKey('users.user_id'), nullable=True)
     created_on = Column('created_on', DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
     type_ = Column('type', Unicode(255))
 
     created_by_user = relationship('User')
     notifications_to_users = relationship('UserNotification', lazy='joined',
                                           cascade="all, delete, delete-orphan")
 
     @property
     def recipients(self):
         return [x.user for x in UserNotification.query()\
                 .filter(UserNotification.notification == self)\
                 .order_by(UserNotification.user_id.asc()).all()]
 
     @classmethod
     def create(cls, created_by, subject, body, recipients, type_=None):
         if type_ is None:
             type_ = Notification.TYPE_MESSAGE
 
         notification = cls()
         notification.created_by_user = created_by
         notification.subject = subject
         notification.body = body
         notification.type_ = type_
         notification.created_on = datetime.datetime.now()
 
         # For each recipient link the created notification to his account
         for u in recipients:
             assoc = UserNotification()
             assoc.user_id = u.user_id
             assoc.notification = notification
 
             # if created_by is inside recipients mark his notification
             # as read
             if u.user_id == created_by.user_id:
                 assoc.read = True
             Session().add(assoc)
 
         Session().add(notification)
 
         return notification
 
 
 class UserNotification(Base, BaseModel):
     __tablename__ = 'user_to_notification'
     __table_args__ = (
         UniqueConstraint('user_id', 'notification_id'),
         base_table_args
     )
 
     user_id = Column('user_id', Integer(), ForeignKey('users.user_id'), primary_key=True)
     notification_id = Column("notification_id", Integer(), ForeignKey('notifications.notification_id'), primary_key=True)
     read = Column('read', Boolean, default=False)
     sent_on = Column('sent_on', DateTime(timezone=False), nullable=True, unique=None)
 
     user = relationship('User', lazy="joined")
     notification = relationship('Notification', lazy="joined",
                                 order_by=lambda: Notification.created_on.desc(),)
 
     def mark_as_read(self):
         self.read = True
         Session().add(self)
 
 
 class Gist(Base, BaseModel):
     __tablename__ = 'gists'
     __table_args__ = (
         Index('g_gist_access_id_idx', 'gist_access_id'),
         Index('g_created_on_idx', 'created_on'),
         base_table_args
     )
 
     GIST_PUBLIC = u'public'
     GIST_PRIVATE = u'private'
     DEFAULT_FILENAME = u'gistfile1.txt'
 
     ACL_LEVEL_PUBLIC = u'acl_public'
     ACL_LEVEL_PRIVATE = u'acl_private'
 
     gist_id = Column('gist_id', Integer(), primary_key=True)
     gist_access_id = Column('gist_access_id', Unicode(250))
     gist_description = Column('gist_description', UnicodeText().with_variant(UnicodeText(1024), 'mysql'))
     gist_owner = Column('user_id', Integer(), ForeignKey('users.user_id'), nullable=True)
     gist_expires = Column('gist_expires', Float(53), nullable=False)
     gist_type = Column('gist_type', Unicode(128), nullable=False)
     created_on = Column('created_on', DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
     modified_at = Column('modified_at', DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
     acl_level = Column('acl_level', Unicode(128), nullable=True)
 
     owner = relationship('User')
 
     def __repr__(self):
         return '<Gist:[%s]%s>' % (self.gist_type, self.gist_access_id)
 
     @hybrid_property
     def description_safe(self):
         from rhodecode.lib import helpers as h
         return h.escape(self.gist_description)
 
     @classmethod
     def get_or_404(cls, id_):
         from pyramid.httpexceptions import HTTPNotFound
 
         res = cls.query().filter(cls.gist_access_id == id_).scalar()
         if not res:
             raise HTTPNotFound()
         return res
 
     @classmethod
     def get_by_access_id(cls, gist_access_id):
         return cls.query().filter(cls.gist_access_id == gist_access_id).scalar()
 
     def gist_url(self):
         from rhodecode.model.gist import GistModel
         return GistModel().get_url(self)
 
     @classmethod
     def base_path(cls):
         """
         Returns base path when all gists are stored
 
         :param cls:
         """
         from rhodecode.model.gist import GIST_STORE_LOC
         q = Session().query(RhodeCodeUi)\
             .filter(RhodeCodeUi.ui_key == URL_SEP)
         q = q.options(FromCache("sql_cache_short", "repository_repo_path"))
         return os.path.join(q.one().ui_value, GIST_STORE_LOC)
 
     def get_api_data(self):
         """
         Common function for generating gist related data for API
         """
         gist = self
         data = {
             'gist_id': gist.gist_id,
             'type': gist.gist_type,
             'access_id': gist.gist_access_id,
             'description': gist.gist_description,
             'url': gist.gist_url(),
             'expires': gist.gist_expires,
             'created_on': gist.created_on,
             'modified_at': gist.modified_at,
             'content': None,
             'acl_level': gist.acl_level,
         }
         return data
 
     def __json__(self):
         data = dict(
         )
         data.update(self.get_api_data())
         return data
     # SCM functions
 
     def scm_instance(self, **kwargs):
         """
         Get an instance of VCS Repository
 
         :param kwargs:
         """
         from rhodecode.model.gist import GistModel
         full_repo_path = os.path.join(self.base_path(), self.gist_access_id)
         return get_vcs_instance(
             repo_path=safe_str(full_repo_path), create=False,
             _vcs_alias=GistModel.vcs_backend)
 
 
 class ExternalIdentity(Base, BaseModel):
     __tablename__ = 'external_identities'
     __table_args__ = (
         Index('local_user_id_idx', 'local_user_id'),
         Index('external_id_idx', 'external_id'),
         base_table_args
     )
 
     external_id = Column('external_id', Unicode(255), default=u'', primary_key=True)
     external_username = Column('external_username', Unicode(1024), default=u'')
     local_user_id = Column('local_user_id', Integer(), ForeignKey('users.user_id'), primary_key=True)
     provider_name = Column('provider_name', Unicode(255), default=u'', primary_key=True)
     access_token = Column('access_token', String(1024), default=u'')
     alt_token = Column('alt_token', String(1024), default=u'')
     token_secret = Column('token_secret', String(1024), default=u'')
 
     @classmethod
     def by_external_id_and_provider(cls, external_id, provider_name, local_user_id=None):
         """
         Returns ExternalIdentity instance based on search params
 
         :param external_id:
         :param provider_name:
         :return: ExternalIdentity
         """
         query = cls.query()
         query = query.filter(cls.external_id == external_id)
         query = query.filter(cls.provider_name == provider_name)
         if local_user_id:
             query = query.filter(cls.local_user_id == local_user_id)
         return query.first()
 
     @classmethod
     def user_by_external_id_and_provider(cls, external_id, provider_name):
         """
         Returns User instance based on search params
 
         :param external_id:
         :param provider_name:
         :return: User
         """
         query = User.query()
         query = query.filter(cls.external_id == external_id)
         query = query.filter(cls.provider_name == provider_name)
         query = query.filter(User.user_id == cls.local_user_id)
         return query.first()
 
     @classmethod
     def by_local_user_id(cls, local_user_id):
         """
         Returns all tokens for user
 
         :param local_user_id:
         :return: ExternalIdentity
         """
         query = cls.query()
         query = query.filter(cls.local_user_id == local_user_id)
         return query
 
     @classmethod
     def load_provider_plugin(cls, plugin_id):
         from rhodecode.authentication.base import loadplugin
         _plugin_id = 'egg:rhodecode-enterprise-ee#{}'.format(plugin_id)
         auth_plugin = loadplugin(_plugin_id)
         return auth_plugin
 
 
 class Integration(Base, BaseModel):
     __tablename__ = 'integrations'
     __table_args__ = (
         base_table_args
     )
 
     integration_id = Column('integration_id', Integer(), primary_key=True)
     integration_type = Column('integration_type', String(255))
     enabled = Column('enabled', Boolean(), nullable=False)
     name = Column('name', String(255), nullable=False)
     child_repos_only = Column('child_repos_only', Boolean(), nullable=False,
         default=False)
 
     settings = Column(
         'settings_json', MutationObj.as_mutable(
             JsonType(dialect_map=dict(mysql=UnicodeText(16384)))))
     repo_id = Column(
         'repo_id', Integer(), ForeignKey('repositories.repo_id'),
         nullable=True, unique=None, default=None)
     repo = relationship('Repository', lazy='joined')
 
     repo_group_id = Column(
         'repo_group_id', Integer(), ForeignKey('groups.group_id'),
         nullable=True, unique=None, default=None)
     repo_group = relationship('RepoGroup', lazy='joined')
 
     @property
     def scope(self):
         if self.repo:
             return repr(self.repo)
         if self.repo_group:
             if self.child_repos_only:
                 return repr(self.repo_group) + ' (child repos only)'
             else:
                 return repr(self.repo_group) + ' (recursive)'
         if self.child_repos_only:
             return 'root_repos'
         return 'global'
 
     def __repr__(self):
         return '<Integration(%r, %r)>' % (self.integration_type, self.scope)
 
 
 class RepoReviewRuleUser(Base, BaseModel):
     __tablename__ = 'repo_review_rules_users'
     __table_args__ = (
         base_table_args
     )
 
     repo_review_rule_user_id = Column('repo_review_rule_user_id', Integer(), primary_key=True)
     repo_review_rule_id = Column("repo_review_rule_id", Integer(), ForeignKey('repo_review_rules.repo_review_rule_id'))
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=False)
     mandatory = Column("mandatory", Boolean(), nullable=False, default=False)
     user = relationship('User')
 
     def rule_data(self):
         return {
             'mandatory': self.mandatory
         }
 
 
 class RepoReviewRuleUserGroup(Base, BaseModel):
     __tablename__ = 'repo_review_rules_users_groups'
     __table_args__ = (
         base_table_args
     )
 
     VOTE_RULE_ALL = -1
 
     repo_review_rule_users_group_id = Column('repo_review_rule_users_group_id', Integer(), primary_key=True)
     repo_review_rule_id = Column("repo_review_rule_id", Integer(), ForeignKey('repo_review_rules.repo_review_rule_id'))
     users_group_id = Column("users_group_id", Integer(),ForeignKey('users_groups.users_group_id'), nullable=False)
     mandatory = Column("mandatory", Boolean(), nullable=False, default=False)
     vote_rule = Column("vote_rule", Integer(), nullable=True, default=VOTE_RULE_ALL)
     users_group = relationship('UserGroup')
 
     def rule_data(self):
         return {
             'mandatory': self.mandatory,
             'vote_rule': self.vote_rule
         }
 
     @property
     def vote_rule_label(self):
         if not self.vote_rule or self.vote_rule == self.VOTE_RULE_ALL:
             return 'all must vote'
         else:
             return 'min. vote {}'.format(self.vote_rule)
 
 
 class RepoReviewRule(Base, BaseModel):
     __tablename__ = 'repo_review_rules'
     __table_args__ = (
         base_table_args
     )
 
     repo_review_rule_id = Column(
         'repo_review_rule_id', Integer(), primary_key=True)
     repo_id = Column(
         "repo_id", Integer(), ForeignKey('repositories.repo_id'))
     repo = relationship('Repository', backref='review_rules')
 
     review_rule_name = Column('review_rule_name', String(255))
     _branch_pattern = Column("branch_pattern", UnicodeText().with_variant(UnicodeText(255), 'mysql'), default=u'*')  # glob
     _target_branch_pattern = Column("target_branch_pattern", UnicodeText().with_variant(UnicodeText(255), 'mysql'), default=u'*')  # glob
     _file_pattern = Column("file_pattern", UnicodeText().with_variant(UnicodeText(255), 'mysql'), default=u'*')  # glob
 
     use_authors_for_review = Column("use_authors_for_review", Boolean(), nullable=False, default=False)
     forbid_author_to_review = Column("forbid_author_to_review", Boolean(), nullable=False, default=False)
     forbid_commit_author_to_review = Column("forbid_commit_author_to_review", Boolean(), nullable=False, default=False)
     forbid_adding_reviewers = Column("forbid_adding_reviewers", Boolean(), nullable=False, default=False)
 
     rule_users = relationship('RepoReviewRuleUser')
     rule_user_groups = relationship('RepoReviewRuleUserGroup')
 
     def _validate_pattern(self, value):
         re.compile('^' + glob2re(value) + '$')
 
     @hybrid_property
     def source_branch_pattern(self):
         return self._branch_pattern or '*'
 
     @source_branch_pattern.setter
     def source_branch_pattern(self, value):
         self._validate_pattern(value)
         self._branch_pattern = value or '*'
 
     @hybrid_property
     def target_branch_pattern(self):
         return self._target_branch_pattern or '*'
 
     @target_branch_pattern.setter
     def target_branch_pattern(self, value):
         self._validate_pattern(value)
         self._target_branch_pattern = value or '*'
 
     @hybrid_property
     def file_pattern(self):
         return self._file_pattern or '*'
 
     @file_pattern.setter
     def file_pattern(self, value):
         self._validate_pattern(value)
         self._file_pattern = value or '*'
 
     def matches(self, source_branch, target_branch, files_changed):
         """
         Check if this review rule matches a branch/files in a pull request
 
         :param source_branch: source branch name for the commit
         :param target_branch: target branch name for the commit
         :param files_changed: list of file paths changed in the pull request
         """
 
         source_branch = source_branch or ''
         target_branch = target_branch or ''
         files_changed = files_changed or []
 
         branch_matches = True
         if source_branch or target_branch:
             if self.source_branch_pattern == '*':
                 source_branch_match = True
             else:
                 if self.source_branch_pattern.startswith('re:'):
                     source_pattern = self.source_branch_pattern[3:]
                 else:
                     source_pattern = '^' + glob2re(self.source_branch_pattern) + '$'
                 source_branch_regex = re.compile(source_pattern)
                 source_branch_match = bool(source_branch_regex.search(source_branch))
             if self.target_branch_pattern == '*':
                 target_branch_match = True
             else:
                 if self.target_branch_pattern.startswith('re:'):
                     target_pattern = self.target_branch_pattern[3:]
                 else:
                     target_pattern = '^' + glob2re(self.target_branch_pattern) + '$'
                 target_branch_regex = re.compile(target_pattern)
                 target_branch_match = bool(target_branch_regex.search(target_branch))
 
             branch_matches = source_branch_match and target_branch_match
 
         files_matches = True
         if self.file_pattern != '*':
             files_matches = False
             if self.file_pattern.startswith('re:'):
                 file_pattern = self.file_pattern[3:]
             else:
                 file_pattern = glob2re(self.file_pattern)
             file_regex = re.compile(file_pattern)
             for filename in files_changed:
                 if file_regex.search(filename):
                     files_matches = True
                     break
 
         return branch_matches and files_matches
 
     @property
     def review_users(self):
         """ Returns the users which this rule applies to """
 
         users = collections.OrderedDict()
 
         for rule_user in self.rule_users:
             if rule_user.user.active:
                 if rule_user.user not in users:
                     users[rule_user.user.username] = {
                         'user': rule_user.user,
                         'source': 'user',
                         'source_data': {},
                         'data': rule_user.rule_data()
                     }
 
         for rule_user_group in self.rule_user_groups:
             source_data = {
                 'user_group_id': rule_user_group.users_group.users_group_id,
                 'name': rule_user_group.users_group.users_group_name,
                 'members': len(rule_user_group.users_group.members)
             }
             for member in rule_user_group.users_group.members:
                 if member.user.active:
                     key = member.user.username
                     if key in users:
                         # skip this member as we have him already
                         # this prevents from override the "first" matched
                         # users with duplicates in multiple groups
                         continue
 
                     users[key] = {
                         'user': member.user,
                         'source': 'user_group',
                         'source_data': source_data,
                         'data': rule_user_group.rule_data()
                     }
 
         return users
 
     def user_group_vote_rule(self, user_id):
 
         rules = []
         if not self.rule_user_groups:
             return rules
 
         for user_group in self.rule_user_groups:
             user_group_members = [x.user_id for x in user_group.users_group.members]
             if user_id in user_group_members:
                 rules.append(user_group)
         return rules
 
     def __repr__(self):
         return '<RepoReviewerRule(id=%r, repo=%r)>' % (
             self.repo_review_rule_id, self.repo)
 
 
 class ScheduleEntry(Base, BaseModel):
     __tablename__ = 'schedule_entries'
     __table_args__ = (
         UniqueConstraint('schedule_name', name='s_schedule_name_idx'),
         UniqueConstraint('task_uid', name='s_task_uid_idx'),
         base_table_args,
     )
 
     schedule_types = ['crontab', 'timedelta', 'integer']
     schedule_entry_id = Column('schedule_entry_id', Integer(), primary_key=True)
 
     schedule_name = Column("schedule_name", String(255), nullable=False, unique=None, default=None)
     schedule_description = Column("schedule_description", String(10000), nullable=True, unique=None, default=None)
     schedule_enabled = Column("schedule_enabled", Boolean(), nullable=False, unique=None, default=True)
 
     _schedule_type = Column("schedule_type", String(255), nullable=False, unique=None, default=None)
     schedule_definition = Column('schedule_definition_json', MutationObj.as_mutable(JsonType(default=lambda: "", dialect_map=dict(mysql=LONGTEXT()))))
 
     schedule_last_run = Column('schedule_last_run', DateTime(timezone=False), nullable=True, unique=None, default=None)
     schedule_total_run_count = Column('schedule_total_run_count', Integer(), nullable=True, unique=None, default=0)
 
     # task
     task_uid = Column("task_uid", String(255), nullable=False, unique=None, default=None)
     task_dot_notation = Column("task_dot_notation", String(4096), nullable=False, unique=None, default=None)
     task_args = Column('task_args_json', MutationObj.as_mutable(JsonType(default=list, dialect_map=dict(mysql=LONGTEXT()))))
     task_kwargs = Column('task_kwargs_json', MutationObj.as_mutable(JsonType(default=dict, dialect_map=dict(mysql=LONGTEXT()))))
 
     created_on = Column('created_on', DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
     updated_on = Column('updated_on', DateTime(timezone=False), nullable=True, unique=None, default=None)
 
     @hybrid_property
     def schedule_type(self):
         return self._schedule_type
 
     @schedule_type.setter
     def schedule_type(self, val):
         if val not in self.schedule_types:
             raise ValueError('Value must be on of `{}` and got `{}`'.format(
                 val, self.schedule_type))
 
         self._schedule_type = val
 
     @classmethod
     def get_uid(cls, obj):
         args = obj.task_args
         kwargs = obj.task_kwargs
         if isinstance(args, JsonRaw):
             try:
                 args = json.loads(args)
             except ValueError:
                 args = tuple()
 
         if isinstance(kwargs, JsonRaw):
             try:
                 kwargs = json.loads(kwargs)
             except ValueError:
                 kwargs = dict()
 
         dot_notation = obj.task_dot_notation
         val = '.'.join(map(safe_str, [
             sorted(dot_notation), args, sorted(kwargs.items())]))
         return hashlib.sha1(val).hexdigest()
 
     @classmethod
     def get_by_schedule_name(cls, schedule_name):
         return cls.query().filter(cls.schedule_name == schedule_name).scalar()
 
     @classmethod
     def get_by_schedule_id(cls, schedule_id):
         return cls.query().filter(cls.schedule_entry_id == schedule_id).scalar()
 
     @property
     def task(self):
         return self.task_dot_notation
 
     @property
     def schedule(self):
         from rhodecode.lib.celerylib.utils import raw_2_schedule
         schedule = raw_2_schedule(self.schedule_definition, self.schedule_type)
         return schedule
 
     @property
     def args(self):
         try:
             return list(self.task_args or [])
         except ValueError:
             return list()
 
     @property
     def kwargs(self):
         try:
             return dict(self.task_kwargs or {})
         except ValueError:
             return dict()
 
     def _as_raw(self, val):
         if hasattr(val, 'de_coerce'):
             val = val.de_coerce()
             if val:
                 val = json.dumps(val)
 
         return val
 
     @property
     def schedule_definition_raw(self):
         return self._as_raw(self.schedule_definition)
 
     @property
     def args_raw(self):
         return self._as_raw(self.task_args)
 
     @property
     def kwargs_raw(self):
         return self._as_raw(self.task_kwargs)
 
     def __repr__(self):
         return '<DB:ScheduleEntry({}:{})>'.format(
             self.schedule_entry_id, self.schedule_name)
 
 
 @event.listens_for(ScheduleEntry, 'before_update')
 def update_task_uid(mapper, connection, target):
     target.task_uid = ScheduleEntry.get_uid(target)
 
 
 @event.listens_for(ScheduleEntry, 'before_insert')
 def set_task_uid(mapper, connection, target):
     target.task_uid = ScheduleEntry.get_uid(target)
 
 
 class _BaseBranchPerms(BaseModel):
     @classmethod
     def compute_hash(cls, value):
         return sha1_safe(value)
 
     @hybrid_property
     def branch_pattern(self):
         return self._branch_pattern or '*'
 
     @hybrid_property
     def branch_hash(self):
         return self._branch_hash
 
     def _validate_glob(self, value):
         re.compile('^' + glob2re(value) + '$')
 
     @branch_pattern.setter
     def branch_pattern(self, value):
         self._validate_glob(value)
         self._branch_pattern = value or '*'
         # set the Hash when setting the branch pattern
         self._branch_hash = self.compute_hash(self._branch_pattern)
 
     def matches(self, branch):
         """
         Check if this the branch matches entry
 
         :param branch: branch name for the commit
         """
 
         branch = branch or ''
 
         branch_matches = True
         if branch:
             branch_regex = re.compile('^' + glob2re(self.branch_pattern) + '$')
             branch_matches = bool(branch_regex.search(branch))
 
         return branch_matches
 
 
 class UserToRepoBranchPermission(Base, _BaseBranchPerms):
     __tablename__ = 'user_to_repo_branch_permissions'
     __table_args__ = (
         {'extend_existing': True, 'mysql_engine': 'InnoDB',
          'mysql_charset': 'utf8', 'sqlite_autoincrement': True,}
     )
 
     branch_rule_id = Column('branch_rule_id', Integer(), primary_key=True)
 
     repository_id = Column('repository_id', Integer(), ForeignKey('repositories.repo_id'), nullable=False, unique=None, default=None)
     repo = relationship('Repository', backref='user_branch_perms')
 
     permission_id = Column('permission_id', Integer(), ForeignKey('permissions.permission_id'), nullable=False, unique=None, default=None)
     permission = relationship('Permission')
 
     rule_to_perm_id = Column('rule_to_perm_id', Integer(), ForeignKey('repo_to_perm.repo_to_perm_id'), nullable=False, unique=None, default=None)
     user_repo_to_perm = relationship('UserRepoToPerm')
 
     rule_order = Column('rule_order', Integer(), nullable=False)
     _branch_pattern = Column('branch_pattern', UnicodeText().with_variant(UnicodeText(2048), 'mysql'), default=u'*')  # glob
     _branch_hash = Column('branch_hash', UnicodeText().with_variant(UnicodeText(2048), 'mysql'))
 
     def __unicode__(self):
         return u'<UserBranchPermission(%s => %r)>' % (
             self.user_repo_to_perm, self.branch_pattern)
 
 
 class UserGroupToRepoBranchPermission(Base, _BaseBranchPerms):
     __tablename__ = 'user_group_to_repo_branch_permissions'
     __table_args__ = (
         {'extend_existing': True, 'mysql_engine': 'InnoDB',
          'mysql_charset': 'utf8', 'sqlite_autoincrement': True,}
     )
 
     branch_rule_id = Column('branch_rule_id', Integer(), primary_key=True)
 
     repository_id = Column('repository_id', Integer(), ForeignKey('repositories.repo_id'), nullable=False, unique=None, default=None)
     repo = relationship('Repository', backref='user_group_branch_perms')
 
     permission_id = Column('permission_id', Integer(), ForeignKey('permissions.permission_id'), nullable=False, unique=None, default=None)
     permission = relationship('Permission')
 
     rule_to_perm_id = Column('rule_to_perm_id', Integer(), ForeignKey('users_group_repo_to_perm.users_group_to_perm_id'), nullable=False, unique=None, default=None)
     user_group_repo_to_perm = relationship('UserGroupRepoToPerm')
 
     rule_order = Column('rule_order', Integer(), nullable=False)
     _branch_pattern = Column('branch_pattern', UnicodeText().with_variant(UnicodeText(2048), 'mysql'), default=u'*')  # glob
     _branch_hash = Column('branch_hash', UnicodeText().with_variant(UnicodeText(2048), 'mysql'))
 
     def __unicode__(self):
         return u'<UserBranchPermission(%s => %r)>' % (
             self.user_group_repo_to_perm, self.branch_pattern)
 
 
 class UserBookmark(Base, BaseModel):
     __tablename__ = 'user_bookmarks'
     __table_args__ = (
         UniqueConstraint('user_id', 'bookmark_repo_id'),
         UniqueConstraint('user_id', 'bookmark_repo_group_id'),
         UniqueConstraint('user_id', 'bookmark_position'),
         base_table_args
     )
 
     user_bookmark_id = Column("user_bookmark_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
     user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=False, unique=None, default=None)
     position = Column("bookmark_position", Integer(), nullable=False)
     title = Column("bookmark_title", String(255), nullable=True, unique=None, default=None)
     redirect_url = Column("bookmark_redirect_url", String(10240), nullable=True, unique=None, default=None)
     created_on = Column("created_on", DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
 
     bookmark_repo_id = Column("bookmark_repo_id", Integer(), ForeignKey("repositories.repo_id"), nullable=True, unique=None, default=None)
     bookmark_repo_group_id = Column("bookmark_repo_group_id", Integer(), ForeignKey("groups.group_id"), nullable=True, unique=None, default=None)
 
     user = relationship("User")
 
     repository = relationship("Repository")
     repository_group = relationship("RepoGroup")
 
     @classmethod
     def get_by_position_for_user(cls, position, user_id):
         return cls.query() \
             .filter(UserBookmark.user_id == user_id) \
             .filter(UserBookmark.position == position).scalar()
 
     @classmethod
     def get_bookmarks_for_user(cls, user_id):
         return cls.query() \
             .filter(UserBookmark.user_id == user_id) \
             .options(joinedload(UserBookmark.repository)) \
             .options(joinedload(UserBookmark.repository_group)) \
             .order_by(UserBookmark.position.asc()) \
             .all()
 
     def __unicode__(self):
         return u'<UserBookmark(%d @ %r)>' % (self.position, self.redirect_url)
 
 
 class FileStore(Base, BaseModel):
     __tablename__ = 'file_store'
     __table_args__ = (
         base_table_args
     )
 
     file_store_id = Column('file_store_id', Integer(), primary_key=True)
     file_uid = Column('file_uid', String(1024), nullable=False)
     file_display_name = Column('file_display_name', UnicodeText().with_variant(UnicodeText(2048), 'mysql'), nullable=True)
     file_description = Column('file_description', UnicodeText().with_variant(UnicodeText(10240), 'mysql'), nullable=True)
     file_org_name = Column('file_org_name', UnicodeText().with_variant(UnicodeText(10240), 'mysql'), nullable=False)
 
     # sha256 hash
     file_hash = Column('file_hash', String(512), nullable=False)
     file_size = Column('file_size', Integer(), nullable=False)
 
     created_on = Column('created_on', DateTime(timezone=False), nullable=False, default=datetime.datetime.now)
     accessed_on = Column('accessed_on', DateTime(timezone=False), nullable=True)
     accessed_count = Column('accessed_count', Integer(), default=0)
 
     enabled = Column('enabled', Boolean(), nullable=False, default=True)
 
     # if repo/repo_group reference is set, check for permissions
     check_acl = Column('check_acl', Boolean(), nullable=False, default=True)
 
     user_id = Column('user_id', Integer(), ForeignKey('users.user_id'), nullable=False)
     upload_user = relationship('User', lazy='joined', primaryjoin='User.user_id==FileStore.user_id')
 
     # scope limited to user, which requester have access to
     scope_user_id = Column(
         'scope_user_id', Integer(), ForeignKey('users.user_id'),
         nullable=True, unique=None, default=None)
     user = relationship('User', lazy='joined', primaryjoin='User.user_id==FileStore.scope_user_id')
 
     # scope limited to user group, which requester have access to
     scope_user_group_id = Column(
         'scope_user_group_id', Integer(), ForeignKey('users_groups.users_group_id'),
         nullable=True, unique=None, default=None)
     user_group = relationship('UserGroup', lazy='joined')
 
     # scope limited to repo, which requester have access to
     scope_repo_id = Column(
         'scope_repo_id', Integer(), ForeignKey('repositories.repo_id'),
         nullable=True, unique=None, default=None)
     repo = relationship('Repository', lazy='joined')
 
     # scope limited to repo group, which requester have access to
     scope_repo_group_id = Column(
         'scope_repo_group_id', Integer(), ForeignKey('groups.group_id'),
         nullable=True, unique=None, default=None)
     repo_group = relationship('RepoGroup', lazy='joined')
 
     @classmethod
     def create(cls, file_uid, filename, file_hash, file_size, file_display_name='',
                file_description='', enabled=True, check_acl=True, user_id=None,
                scope_user_id=None, scope_repo_id=None, scope_repo_group_id=None):
 
         store_entry = FileStore()
         store_entry.file_uid = file_uid
         store_entry.file_display_name = file_display_name
         store_entry.file_org_name = filename
         store_entry.file_size = file_size
         store_entry.file_hash = file_hash
         store_entry.file_description = file_description
 
         store_entry.check_acl = check_acl
         store_entry.enabled = enabled
 
         store_entry.user_id = user_id
         store_entry.scope_user_id = scope_user_id
         store_entry.scope_repo_id = scope_repo_id
         store_entry.scope_repo_group_id = scope_repo_group_id
         return store_entry
 
     @classmethod
     def bump_access_counter(cls, file_uid, commit=True):
         FileStore().query()\
             .filter(FileStore.file_uid == file_uid)\
             .update({FileStore.accessed_count: (FileStore.accessed_count + 1),
                      FileStore.accessed_on: datetime.datetime.now()})
         if commit:
             Session().commit()
 
     def __repr__(self):
         return '<FileStore({})>'.format(self.file_store_id)
 
 
 class DbMigrateVersion(Base, BaseModel):
     __tablename__ = 'db_migrate_version'
     __table_args__ = (
         base_table_args,
     )
 
     repository_id = Column('repository_id', String(250), primary_key=True)
     repository_path = Column('repository_path', Text)
     version = Column('version', Integer)
 
     @classmethod
     def set_version(cls, version):
         """
         Helper for forcing a different version, usually for debugging purposes via ishell.
         """
         ver = DbMigrateVersion.query().first()
         ver.version = version
         Session().commit()
 
 
 class DbSession(Base, BaseModel):
     __tablename__ = 'db_session'
     __table_args__ = (
         base_table_args,
     )
 
     def __repr__(self):
         return '<DB:DbSession({})>'.format(self.id)
 
     id = Column('id', Integer())
     namespace = Column('namespace', String(255), primary_key=True)
     accessed = Column('accessed', DateTime, nullable=False)
     created = Column('created', DateTime, nullable=False)
     data = Column('data', PickleType, nullable=False)

rhodecode/tests/lib/middleware/test_simplevcs.py

@@ -1,475 +1,473 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import base64
 
 import mock
 import pytest
 
 from rhodecode.lib.utils2 import AttributeDict
 from rhodecode.tests.utils import CustomTestApp
 
 from rhodecode.lib.caching_query import FromCache
 from rhodecode.lib.hooks_daemon import DummyHooksCallbackDaemon
 from rhodecode.lib.middleware import simplevcs
 from rhodecode.lib.middleware.https_fixup import HttpsFixup
 from rhodecode.lib.middleware.utils import scm_app_http
 from rhodecode.model.db import User, _hash_key
 from rhodecode.model.meta import Session
 from rhodecode.tests import (
     HG_REPO, TEST_USER_ADMIN_LOGIN, TEST_USER_ADMIN_PASS)
 from rhodecode.tests.lib.middleware import mock_scm_app
 
 
 class StubVCSController(simplevcs.SimpleVCS):
 
     SCM = 'hg'
     stub_response_body = tuple()
 
     def __init__(self, *args, **kwargs):
         super(StubVCSController, self).__init__(*args, **kwargs)
         self._action = 'pull'
         self._is_shadow_repo_dir = True
         self._name = HG_REPO
         self.set_repo_names(None)
 
     @property
     def is_shadow_repo_dir(self):
         return self._is_shadow_repo_dir
 
     def _get_repository_name(self, environ):
         return self._name
 
     def _get_action(self, environ):
         return self._action
 
     def _create_wsgi_app(self, repo_path, repo_name, config):
         def fake_app(environ, start_response):
             headers = [
                 ('Http-Accept', 'application/mercurial')
             ]
             start_response('200 OK', headers)
             return self.stub_response_body
         return fake_app
 
     def _create_config(self, extras, repo_name, scheme='http'):
         return None
 
 
 @pytest.fixture
 def vcscontroller(baseapp, config_stub, request_stub):
     config_stub.testing_securitypolicy()
     config_stub.include('rhodecode.authentication')
     config_stub.include('rhodecode.authentication.plugins.auth_rhodecode')
     config_stub.include('rhodecode.authentication.plugins.auth_token')
 
     controller = StubVCSController(
         baseapp.config.get_settings(), request_stub.registry)
     app = HttpsFixup(controller, baseapp.config.get_settings())
     app = CustomTestApp(app)
 
     _remove_default_user_from_query_cache()
 
     # Sanity checks that things are set up correctly
     app.get('/' + HG_REPO, status=200)
 
     app.controller = controller
     return app
 
 
 def _remove_default_user_from_query_cache():
     user = User.get_default_user(cache=True)
     query = Session().query(User).filter(User.username == user.username)
     query = query.options(
         FromCache("sql_cache_short", "get_user_%s" % _hash_key(user.username)))
     query.invalidate()
     Session().expire(user)
 
 
 def test_handles_exceptions_during_permissions_checks(
         vcscontroller, disable_anonymous_user):
     user_and_pass = '%s:%s' % (TEST_USER_ADMIN_LOGIN, TEST_USER_ADMIN_PASS)
     auth_password = base64.encodestring(user_and_pass).strip()
     extra_environ = {
         'AUTH_TYPE': 'Basic',
         'HTTP_AUTHORIZATION': 'Basic %s' % auth_password,
         'REMOTE_USER': TEST_USER_ADMIN_LOGIN,
     }
 
     # Verify that things are hooked up correctly
     vcscontroller.get('/', status=200, extra_environ=extra_environ)
 
     # Simulate trouble during permission checks
     with mock.patch('rhodecode.model.db.User.get_by_username',
                     side_effect=Exception) as get_user:
         # Verify that a correct 500 is returned and check that the expected
         # code path was hit.
         vcscontroller.get('/', status=500, extra_environ=extra_environ)
         assert get_user.called
 
 
 def test_returns_forbidden_if_no_anonymous_access(
         vcscontroller, disable_anonymous_user):
     vcscontroller.get('/', status=401)
 
 
 class StubFailVCSController(simplevcs.SimpleVCS):
     def _handle_request(self, environ, start_response):
         raise Exception("BOOM")
 
 
 @pytest.fixture(scope='module')
 def fail_controller(baseapp):
     controller = StubFailVCSController(
         baseapp.config.get_settings(), baseapp.config)
     controller = HttpsFixup(controller, baseapp.config.get_settings())
     controller = CustomTestApp(controller)
     return controller
 
 
 def test_handles_exceptions_as_internal_server_error(fail_controller):
     fail_controller.get('/', status=500)
 
 
 def test_provides_traceback_for_appenlight(fail_controller):
     response = fail_controller.get(
         '/', status=500, extra_environ={'appenlight.client': 'fake'})
     assert 'appenlight.__traceback' in response.request.environ
 
 
 def test_provides_utils_scm_app_as_scm_app_by_default(baseapp, request_stub):
     controller = StubVCSController(baseapp.config.get_settings(), request_stub.registry)
     assert controller.scm_app is scm_app_http
 
 
 def test_allows_to_override_scm_app_via_config(baseapp, request_stub):
     config = baseapp.config.get_settings().copy()
     config['vcs.scm_app_implementation'] = (
         'rhodecode.tests.lib.middleware.mock_scm_app')
     controller = StubVCSController(config, request_stub.registry)
     assert controller.scm_app is mock_scm_app
 
 
 @pytest.mark.parametrize('query_string, expected', [
     ('cmd=stub_command', True),
     ('cmd=listkeys', False),
 ])
 def test_should_check_locking(query_string, expected):
     result = simplevcs._should_check_locking(query_string)
     assert result == expected
 
 
 class TestShadowRepoRegularExpression(object):
     pr_segment = 'pull-request'
     shadow_segment = 'repository'
 
     @pytest.mark.parametrize('url, expected', [
         # repo with/without groups
         ('My-Repo/{pr_segment}/1/{shadow_segment}', True),
         ('Group/My-Repo/{pr_segment}/2/{shadow_segment}', True),
         ('Group/Sub-Group/My-Repo/{pr_segment}/3/{shadow_segment}', True),
         ('Group/Sub-Group1/Sub-Group2/My-Repo/{pr_segment}/3/{shadow_segment}', True),
 
         # pull request ID
         ('MyRepo/{pr_segment}/1/{shadow_segment}', True),
         ('MyRepo/{pr_segment}/1234567890/{shadow_segment}', True),
         ('MyRepo/{pr_segment}/-1/{shadow_segment}', False),
         ('MyRepo/{pr_segment}/invalid/{shadow_segment}', False),
 
         # unicode
         (u'Sp€çîál-Repö/{pr_segment}/1/{shadow_segment}', True),
         (u'Sp€çîál-Gröüp/Sp€çîál-Repö/{pr_segment}/1/{shadow_segment}', True),
 
         # trailing/leading slash
         ('/My-Repo/{pr_segment}/1/{shadow_segment}', False),
         ('My-Repo/{pr_segment}/1/{shadow_segment}/', False),
         ('/My-Repo/{pr_segment}/1/{shadow_segment}/', False),
 
         # misc
         ('My-Repo/{pr_segment}/1/{shadow_segment}/extra', False),
         ('My-Repo/{pr_segment}/1/{shadow_segment}extra', False),
     ])
     def test_shadow_repo_regular_expression(self, url, expected):
         from rhodecode.lib.middleware.simplevcs import SimpleVCS
         url = url.format(
             pr_segment=self.pr_segment,
             shadow_segment=self.shadow_segment)
         match_obj = SimpleVCS.shadow_repo_re.match(url)
         assert (match_obj is not None) == expected
 
 
 @pytest.mark.backends('git', 'hg')
 class TestShadowRepoExposure(object):
 
     def test_pull_on_shadow_repo_propagates_to_wsgi_app(
             self, baseapp, request_stub):
         """
         Check that a pull action to a shadow repo is propagated to the
         underlying wsgi app.
         """
         controller = StubVCSController(
             baseapp.config.get_settings(), request_stub.registry)
         controller._check_ssl = mock.Mock()
         controller.is_shadow_repo = True
         controller._action = 'pull'
         controller._is_shadow_repo_dir = True
         controller.stub_response_body = 'dummy body value'
         controller._get_default_cache_ttl = mock.Mock(
             return_value=(False, 0))
 
         environ_stub = {
             'HTTP_HOST': 'test.example.com',
             'HTTP_ACCEPT': 'application/mercurial',
             'REQUEST_METHOD': 'GET',
             'wsgi.url_scheme': 'http',
         }
 
         response = controller(environ_stub, mock.Mock())
         response_body = ''.join(response)
 
         # Assert that we got the response from the wsgi app.
         assert response_body == controller.stub_response_body
 
     def test_pull_on_shadow_repo_that_is_missing(self, baseapp, request_stub):
         """
         Check that a pull action to a shadow repo is propagated to the
         underlying wsgi app.
         """
         controller = StubVCSController(
             baseapp.config.get_settings(), request_stub.registry)
         controller._check_ssl = mock.Mock()
         controller.is_shadow_repo = True
         controller._action = 'pull'
         controller._is_shadow_repo_dir = False
         controller.stub_response_body = 'dummy body value'
         environ_stub = {
             'HTTP_HOST': 'test.example.com',
             'HTTP_ACCEPT': 'application/mercurial',
             'REQUEST_METHOD': 'GET',
             'wsgi.url_scheme': 'http',
         }
 
         response = controller(environ_stub, mock.Mock())
         response_body = ''.join(response)
 
         # Assert that we got the response from the wsgi app.
         assert '404 Not Found' in response_body
 
     def test_push_on_shadow_repo_raises(self, baseapp, request_stub):
         """
         Check that a push action to a shadow repo is aborted.
         """
         controller = StubVCSController(
             baseapp.config.get_settings(), request_stub.registry)
         controller._check_ssl = mock.Mock()
         controller.is_shadow_repo = True
         controller._action = 'push'
         controller.stub_response_body = 'dummy body value'
         environ_stub = {
             'HTTP_HOST': 'test.example.com',
             'HTTP_ACCEPT': 'application/mercurial',
             'REQUEST_METHOD': 'GET',
             'wsgi.url_scheme': 'http',
         }
 
         response = controller(environ_stub, mock.Mock())
         response_body = ''.join(response)
 
         assert response_body != controller.stub_response_body
         # Assert that a 406 error is returned.
         assert '406 Not Acceptable' in response_body
 
     def test_set_repo_names_no_shadow(self, baseapp, request_stub):
         """
         Check that the set_repo_names method sets all names to the one returned
         by the _get_repository_name method on a request to a non shadow repo.
         """
         environ_stub = {}
         controller = StubVCSController(
             baseapp.config.get_settings(), request_stub.registry)
         controller._name = 'RepoGroup/MyRepo'
         controller.set_repo_names(environ_stub)
         assert not controller.is_shadow_repo
         assert (controller.url_repo_name ==
                 controller.acl_repo_name ==
                 controller.vcs_repo_name ==
                 controller._get_repository_name(environ_stub))
 
     def test_set_repo_names_with_shadow(
             self, baseapp, pr_util, config_stub, request_stub):
         """
         Check that the set_repo_names method sets correct names on a request
         to a shadow repo.
         """
         from rhodecode.model.pull_request import PullRequestModel
 
         pull_request = pr_util.create_pull_request()
         shadow_url = '{target}/{pr_segment}/{pr_id}/{shadow_segment}'.format(
             target=pull_request.target_repo.repo_name,
             pr_id=pull_request.pull_request_id,
             pr_segment=TestShadowRepoRegularExpression.pr_segment,
             shadow_segment=TestShadowRepoRegularExpression.shadow_segment)
         controller = StubVCSController(
             baseapp.config.get_settings(), request_stub.registry)
         controller._name = shadow_url
         controller.set_repo_names({})
 
         # Get file system path to shadow repo for assertions.
         workspace_id = PullRequestModel()._workspace_id(pull_request)
-        target_vcs = pull_request.target_repo.scm_instance()
-        vcs_repo_name = target_vcs._get_shadow_repository_path(
-            pull_request.target_repo.repo_id, workspace_id)
+        vcs_repo_name = pull_request.target_repo.get_shadow_repository_path(workspace_id)
 
         assert controller.vcs_repo_name == vcs_repo_name
         assert controller.url_repo_name == shadow_url
         assert controller.acl_repo_name == pull_request.target_repo.repo_name
         assert controller.is_shadow_repo
 
     def test_set_repo_names_with_shadow_but_missing_pr(
             self, baseapp, pr_util, config_stub, request_stub):
         """
         Checks that the set_repo_names method enforces matching target repos
         and pull request IDs.
         """
         pull_request = pr_util.create_pull_request()
         shadow_url = '{target}/{pr_segment}/{pr_id}/{shadow_segment}'.format(
             target=pull_request.target_repo.repo_name,
             pr_id=999999999,
             pr_segment=TestShadowRepoRegularExpression.pr_segment,
             shadow_segment=TestShadowRepoRegularExpression.shadow_segment)
         controller = StubVCSController(
             baseapp.config.get_settings(), request_stub.registry)
         controller._name = shadow_url
         controller.set_repo_names({})
 
         assert not controller.is_shadow_repo
         assert (controller.url_repo_name ==
                 controller.acl_repo_name ==
                 controller.vcs_repo_name)
 
 
 @pytest.mark.usefixtures('baseapp')
 class TestGenerateVcsResponse(object):
 
     def test_ensures_that_start_response_is_called_early_enough(self):
         self.call_controller_with_response_body(iter(['a', 'b']))
         assert self.start_response.called
 
     def test_invalidates_cache_after_body_is_consumed(self):
         result = self.call_controller_with_response_body(iter(['a', 'b']))
         assert not self.was_cache_invalidated()
         # Consume the result
         list(result)
         assert self.was_cache_invalidated()
 
     def test_raises_unknown_exceptions(self):
         result = self.call_controller_with_response_body(
             self.raise_result_iter(vcs_kind='unknown'))
         with pytest.raises(Exception):
             list(result)
 
     def test_prepare_callback_daemon_is_called(self):
         def side_effect(extras, environ, action, txn_id=None):
             return DummyHooksCallbackDaemon(), extras
 
         prepare_patcher = mock.patch.object(
             StubVCSController, '_prepare_callback_daemon')
         with prepare_patcher as prepare_mock:
             prepare_mock.side_effect = side_effect
             self.call_controller_with_response_body(iter(['a', 'b']))
         assert prepare_mock.called
         assert prepare_mock.call_count == 1
 
     def call_controller_with_response_body(self, response_body):
         settings = {
             'base_path': 'fake_base_path',
             'vcs.hooks.protocol': 'http',
             'vcs.hooks.direct_calls': False,
         }
         registry = AttributeDict()
         controller = StubVCSController(settings, registry)
         controller._invalidate_cache = mock.Mock()
         controller.stub_response_body = response_body
         self.start_response = mock.Mock()
         result = controller._generate_vcs_response(
             environ={}, start_response=self.start_response,
             repo_path='fake_repo_path',
             extras={}, action='push')
         self.controller = controller
         return result
 
     def raise_result_iter(self, vcs_kind='repo_locked'):
         """
         Simulates an exception due to a vcs raised exception if kind vcs_kind
         """
         raise self.vcs_exception(vcs_kind=vcs_kind)
         yield "never_reached"
 
     def vcs_exception(self, vcs_kind='repo_locked'):
         locked_exception = Exception('TEST_MESSAGE')
         locked_exception._vcs_kind = vcs_kind
         return locked_exception
 
     def was_cache_invalidated(self):
         return self.controller._invalidate_cache.called
 
 
 class TestInitializeGenerator(object):
 
     def test_drains_first_element(self):
         gen = self.factory(['__init__', 1, 2])
         result = list(gen)
         assert result == [1, 2]
 
     @pytest.mark.parametrize('values', [
         [],
         [1, 2],
     ])
     def test_raises_value_error(self, values):
         with pytest.raises(ValueError):
             self.factory(values)
 
     @simplevcs.initialize_generator
     def factory(self, iterable):
         for elem in iterable:
             yield elem
 
 
 class TestPrepareHooksDaemon(object):
     def test_calls_imported_prepare_callback_daemon(self, app_settings, request_stub):
         expected_extras = {'extra1': 'value1'}
         daemon = DummyHooksCallbackDaemon()
 
         controller = StubVCSController(app_settings, request_stub.registry)
         prepare_patcher = mock.patch.object(
             simplevcs, 'prepare_callback_daemon',
             return_value=(daemon, expected_extras))
         with prepare_patcher as prepare_mock:
             callback_daemon, extras = controller._prepare_callback_daemon(
                 expected_extras.copy(), {}, 'push')
         prepare_mock.assert_called_once_with(
             expected_extras,
             protocol=app_settings['vcs.hooks.protocol'],
             host=app_settings['vcs.hooks.host'],
             txn_id=None,
             use_direct_calls=app_settings['vcs.hooks.direct_calls'])
 
         assert callback_daemon == daemon
         assert extras == extras