statsd/audit-logs: cleanup push/pull user agent code.

- add normalization
- add user-agent tag into statsd call so we can track protocol/vcs type usage
- internal merge also sends in repo type

rhodecode/apps/ssh_support/lib/backends/base.py

@@ -1,162 +1,163 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2016-2020 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import os
 import sys
 import json
 import logging
 
 from rhodecode.lib.hooks_daemon import prepare_callback_daemon
 from rhodecode.lib.vcs.conf import settings as vcs_settings
 from rhodecode.model.scm import ScmModel
 
 log = logging.getLogger(__name__)
 
 
 class VcsServer(object):
+    repo_user_agent = None  # set in child classes
     _path = None  # set executable path for hg/git/svn binary
     backend = None  # set in child classes
     tunnel = None  # subprocess handling tunnel
     write_perms = ['repository.admin', 'repository.write']
     read_perms = ['repository.read', 'repository.admin', 'repository.write']
 
     def __init__(self, user, user_permissions, config, env):
         self.user = user
         self.user_permissions = user_permissions
         self.config = config
         self.env = env
         self.stdin = sys.stdin
 
         self.repo_name = None
         self.repo_mode = None
         self.store = ''
         self.ini_path = ''
 
     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 has_write_perm(self):
         permission = self.user_permissions.get(self.repo_name)
         if permission in ['repository.write', 'repository.admin']:
             return True
 
         return False
 
     def _check_permissions(self, action):
         permission = self.user_permissions.get(self.repo_name)
         log.debug('permission for %s on %s are: %s',
                   self.user, self.repo_name, permission)
 
         if not permission:
             log.error('user `%s` permissions to repo:%s are empty. Forbidding access.',
                       self.user, self.repo_name)
             return -2
 
         if action == 'pull':
             if permission in self.read_perms:
                 log.info(
                     'READ Permissions for User "%s" detected to repo "%s"!',
                     self.user, self.repo_name)
                 return 0
         else:
             if permission in self.write_perms:
                 log.info(
                     'WRITE, or Higher Permissions for User "%s" detected to repo "%s"!',
                     self.user, self.repo_name)
                 return 0
 
         log.error('Cannot properly fetch or verify user `%s` permissions. '
                   'Permissions: %s, vcs action: %s',
                   self.user, permission, action)
         return -2
 
     def update_environment(self, action, extras=None):
 
         scm_data = {
             'ip': os.environ['SSH_CLIENT'].split()[0],
             'username': self.user.username,
             'user_id': self.user.user_id,
             'action': action,
             'repository': self.repo_name,
             'scm': self.backend,
             'config': self.ini_path,
             'repo_store': self.store,
             'make_lock': None,
             'locked_by': [None, None],
             'server_url': None,
-            'user_agent': 'ssh-user-agent',
+            'user_agent': '{}/ssh-user-agent'.format(self.repo_user_agent),
             'hooks': ['push', 'pull'],
             'hooks_module': 'rhodecode.lib.hooks_daemon',
             'is_shadow_repo': False,
             'detect_force_push': False,
             'check_branch_perms': False,
 
             'SSH': True,
             'SSH_PERMISSIONS': self.user_permissions.get(self.repo_name),
         }
         if extras:
             scm_data.update(extras)
         os.putenv("RC_SCM_DATA", json.dumps(scm_data))
 
     def get_root_store(self):
         root_store = self.store
         if not root_store.endswith('/'):
             # always append trailing slash
             root_store = root_store + '/'
         return root_store
 
     def _handle_tunnel(self, extras):
         # pre-auth
         action = 'pull'
         exit_code = self._check_permissions(action)
         if exit_code:
             return exit_code, False
 
         req = self.env['request']
         server_url = req.host_url + req.script_name
         extras['server_url'] = server_url
 
         log.debug('Using %s binaries from path %s', self.backend, self._path)
         exit_code = self.tunnel.run(extras)
 
         return exit_code, action == "push"
 
     def run(self, tunnel_extras=None):
         tunnel_extras = tunnel_extras or {}
         extras = {}
         extras.update(tunnel_extras)
 
         callback_daemon, extras = prepare_callback_daemon(
             extras, protocol=vcs_settings.HOOKS_PROTOCOL,
             host=vcs_settings.HOOKS_HOST,
             use_direct_calls=False)
 
         with callback_daemon:
             try:
                 return self._handle_tunnel(extras)
             finally:
                 log.debug('Running cleanup with cache invalidation')
                 if self.repo_name:
                     self._invalidate_cache(self.repo_name)

rhodecode/apps/ssh_support/lib/backends/git.py

@@ -1,74 +1,75 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2016-2020 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import os
 import sys
 import logging
 
 from .base import VcsServer
 
 log = logging.getLogger(__name__)
 
 
 class GitTunnelWrapper(object):
     process = None
 
     def __init__(self, server):
         self.server = server
         self.stdin = sys.stdin
         self.stdout = sys.stdout
 
     def create_hooks_env(self):
         pass
 
     def command(self):
         root = self.server.get_root_store()
         command = "cd {root}; {git_path} {mode} '{root}{repo_name}'".format(
             root=root, git_path=self.server.git_path,
             mode=self.server.repo_mode, repo_name=self.server.repo_name)
         log.debug("Final CMD: %s", command)
         return command
 
     def run(self, extras):
         action = "push" if self.server.repo_mode == "receive-pack" else "pull"
         exit_code = self.server._check_permissions(action)
         if exit_code:
             return exit_code
 
         self.server.update_environment(action=action, extras=extras)
         self.create_hooks_env()
         return os.system(self.command())
 
 
 class GitServer(VcsServer):
     backend = 'git'
+    repo_user_agent = 'git'
 
     def __init__(self, store, ini_path, repo_name, repo_mode,
                  user, user_permissions, config, env):
         super(GitServer, self).\
             __init__(user, user_permissions, config, env)
 
         self.store = store
         self.ini_path = ini_path
         self.repo_name = repo_name
         self._path = self.git_path = config.get('app:main', 'ssh.executable.git')
 
         self.repo_mode = repo_mode
         self.tunnel = GitTunnelWrapper(server=self)

rhodecode/apps/ssh_support/lib/backends/hg.py

@@ -1,147 +1,148 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2016-2020 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import os
 import sys
 import logging
 import tempfile
 import textwrap
 import collections
 from .base import VcsServer
 from rhodecode.model.db import RhodeCodeUi
 from rhodecode.model.settings import VcsSettingsModel
 
 log = logging.getLogger(__name__)
 
 
 class MercurialTunnelWrapper(object):
     process = None
 
     def __init__(self, server):
         self.server = server
         self.stdin = sys.stdin
         self.stdout = sys.stdout
         self.hooks_env_fd, self.hooks_env_path = tempfile.mkstemp(prefix='hgrc_rhodecode_')
 
     def create_hooks_env(self):
         repo_name = self.server.repo_name
         hg_flags = self.server.config_to_hgrc(repo_name)
 
         content = textwrap.dedent(
             '''
             # RhodeCode SSH hooks version=2.0.0
             {custom}
             '''
         ).format(custom='\n'.join(hg_flags))
 
         root = self.server.get_root_store()
         hgrc_custom = os.path.join(root, repo_name, '.hg', 'hgrc_rhodecode')
         hgrc_main = os.path.join(root, repo_name, '.hg', 'hgrc')
 
         # cleanup custom hgrc file
         if os.path.isfile(hgrc_custom):
             with open(hgrc_custom, 'wb') as f:
                 f.write('')
             log.debug('Cleanup custom hgrc file under %s', hgrc_custom)
 
         # write temp
         with os.fdopen(self.hooks_env_fd, 'w') as hooks_env_file:
             hooks_env_file.write(content)
 
         return self.hooks_env_path
 
     def remove_configs(self):
         os.remove(self.hooks_env_path)
 
     def command(self, hgrc_path):
         root = self.server.get_root_store()
 
         command = (
             "cd {root}; HGRCPATH={hgrc} {hg_path} -R {root}{repo_name} "
             "serve --stdio".format(
                 root=root, hg_path=self.server.hg_path,
                 repo_name=self.server.repo_name, hgrc=hgrc_path))
         log.debug("Final CMD: %s", command)
         return command
 
     def run(self, extras):
         # at this point we cannot tell, we do further ACL checks
         # inside the hooks
         action = '?'
         # permissions are check via `pre_push_ssh_auth` hook
         self.server.update_environment(action=action, extras=extras)
         custom_hgrc_file = self.create_hooks_env()
 
         try:
             return os.system(self.command(custom_hgrc_file))
         finally:
             self.remove_configs()
 
 
 class MercurialServer(VcsServer):
     backend = 'hg'
+    repo_user_agent = 'mercurial'
     cli_flags = ['phases', 'largefiles', 'extensions', 'experimental', 'hooks']
 
     def __init__(self, store, ini_path, repo_name, user, user_permissions, config, env):
         super(MercurialServer, self).__init__(user, user_permissions, config, env)
 
         self.store = store
         self.ini_path = ini_path
         self.repo_name = repo_name
         self._path = self.hg_path = config.get('app:main', 'ssh.executable.hg')
         self.tunnel = MercurialTunnelWrapper(server=self)
 
     def config_to_hgrc(self, repo_name):
         ui_sections = collections.defaultdict(list)
         ui = VcsSettingsModel(repo=repo_name).get_ui_settings(section=None, key=None)
 
         # write default hooks
         default_hooks = [
             ('pretxnchangegroup.ssh_auth', 'python:vcsserver.hooks.pre_push_ssh_auth'),
             ('pretxnchangegroup.ssh', 'python:vcsserver.hooks.pre_push_ssh'),
             ('changegroup.ssh', 'python:vcsserver.hooks.post_push_ssh'),
 
             ('preoutgoing.ssh', 'python:vcsserver.hooks.pre_pull_ssh'),
             ('outgoing.ssh', 'python:vcsserver.hooks.post_pull_ssh'),
         ]
 
         for k, v in default_hooks:
             ui_sections['hooks'].append((k, v))
 
         for entry in ui:
             if not entry.active:
                 continue
             sec = entry.section
             key = entry.key
 
             if sec in self.cli_flags:
                 # we want only custom hooks, so we skip builtins
                 if sec == 'hooks' and key in RhodeCodeUi.HOOKS_BUILTIN:
                     continue
 
                 ui_sections[sec].append([key, entry.value])
 
         flags = []
         for _sec, key_val in ui_sections.items():
             flags.append(' ')
             flags.append('[{}]'.format(_sec))
             for key, val in key_val:
                 flags.append('{}= {}'.format(key, val))
         return flags

rhodecode/apps/ssh_support/lib/backends/svn.py

@@ -1,257 +1,258 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2016-2020 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import os
 import re
 import sys
 import logging
 import signal
 import tempfile
 from subprocess import Popen, PIPE
 import urlparse
 
 from .base import VcsServer
 
 log = logging.getLogger(__name__)
 
 
 class SubversionTunnelWrapper(object):
     process = None
 
     def __init__(self, server):
         self.server = server
         self.timeout = 30
         self.stdin = sys.stdin
         self.stdout = sys.stdout
         self.svn_conf_fd, self.svn_conf_path = tempfile.mkstemp()
         self.hooks_env_fd, self.hooks_env_path = tempfile.mkstemp()
 
         self.read_only = True  # flag that we set to make the hooks readonly
 
     def create_svn_config(self):
         content = (
             '[general]\n'
             'hooks-env = {}\n').format(self.hooks_env_path)
         with os.fdopen(self.svn_conf_fd, 'w') as config_file:
             config_file.write(content)
 
     def create_hooks_env(self):
         content = (
             '[default]\n'
             'LANG = en_US.UTF-8\n')
         if self.read_only:
             content += 'SSH_READ_ONLY = 1\n'
         with os.fdopen(self.hooks_env_fd, 'w') as hooks_env_file:
             hooks_env_file.write(content)
 
     def remove_configs(self):
         os.remove(self.svn_conf_path)
         os.remove(self.hooks_env_path)
 
     def command(self):
         root = self.server.get_root_store()
         username = self.server.user.username
 
         command = [
             self.server.svn_path, '-t',
             '--config-file', self.svn_conf_path,
             '--tunnel-user', username,
             '-r', root]
         log.debug("Final CMD: %s", ' '.join(command))
         return command
 
     def start(self):
         command = self.command()
         self.process = Popen(' '.join(command), stdin=PIPE, shell=True)
 
     def sync(self):
         while self.process.poll() is None:
             next_byte = self.stdin.read(1)
             if not next_byte:
                 break
             self.process.stdin.write(next_byte)
         self.remove_configs()
 
     @property
     def return_code(self):
         return self.process.returncode
 
     def get_first_client_response(self):
         signal.signal(signal.SIGALRM, self.interrupt)
         signal.alarm(self.timeout)
         first_response = self._read_first_client_response()
         signal.alarm(0)
         return (self._parse_first_client_response(first_response)
                 if first_response else None)
 
     def patch_first_client_response(self, response, **kwargs):
         self.create_hooks_env()
         data = response.copy()
         data.update(kwargs)
         data['url'] = self._svn_string(data['url'])
         data['ra_client'] = self._svn_string(data['ra_client'])
         data['client'] = data['client'] or ''
         buffer_ = (
             "( {version} ( {capabilities} ) {url}{ra_client}"
             "( {client}) ) ".format(**data))
         self.process.stdin.write(buffer_)
 
     def fail(self, message):
         print("( failure ( ( 210005 {message} 0: 0 ) ) )".format(
             message=self._svn_string(message)))
         self.remove_configs()
         self.process.kill()
         return 1
 
     def interrupt(self, signum, frame):
         self.fail("Exited by timeout")
 
     def _svn_string(self, str_):
         if not str_:
             return ''
         return '{length}:{string} '.format(length=len(str_), string=str_)
 
     def _read_first_client_response(self):
         buffer_ = ""
         brackets_stack = []
         while True:
             next_byte = self.stdin.read(1)
             buffer_ += next_byte
             if next_byte == "(":
                 brackets_stack.append(next_byte)
             elif next_byte == ")":
                 brackets_stack.pop()
             elif next_byte == " " and not brackets_stack:
                 break
 
         return buffer_
 
     def _parse_first_client_response(self, buffer_):
         """
         According to the Subversion RA protocol, the first request
         should look like:
 
         ( version:number ( cap:word ... ) url:string ? ra-client:string
            ( ? client:string ) )
 
         Please check https://svn.apache.org/repos/asf/subversion/trunk/subversion/libsvn_ra_svn/protocol
         """
         version_re = r'(?P<version>\d+)'
         capabilities_re = r'\(\s(?P<capabilities>[\w\d\-\ ]+)\s\)'
         url_re = r'\d+\:(?P<url>[\W\w]+)'
         ra_client_re = r'(\d+\:(?P<ra_client>[\W\w]+)\s)'
         client_re = r'(\d+\:(?P<client>[\W\w]+)\s)*'
         regex = re.compile(
             r'^\(\s{version}\s{capabilities}\s{url}\s{ra_client}'
             r'\(\s{client}\)\s\)\s*$'.format(
                 version=version_re, capabilities=capabilities_re,
                 url=url_re, ra_client=ra_client_re, client=client_re))
         matcher = regex.match(buffer_)
 
         return matcher.groupdict() if matcher else None
 
     def _match_repo_name(self, url):
         """
         Given an server url, try to match it against ALL known repository names.
         This handles a tricky SVN case for SSH and subdir commits.
         E.g if our repo name is my-svn-repo, a svn commit on file in a subdir would
         result in the url with this subdir added.
         """
         # case 1 direct match, we don't do any "heavy" lookups
         if url in self.server.user_permissions:
             return url
 
         log.debug('Extracting repository name from subdir path %s', url)
         # case 2 we check all permissions, and match closes possible case...
         # NOTE(dan): In this case we only know that url has a subdir parts, it's safe
         # to assume that it will have the repo name as prefix, we ensure the prefix
         # for similar repositories isn't matched by adding a /
         # e.g subgroup/repo-name/ and subgroup/repo-name-1/ would work correct.
         for repo_name in self.server.user_permissions:
             repo_name_prefix = repo_name + '/'
             if url.startswith(repo_name_prefix):
                 log.debug('Found prefix %s match, returning proper repository name',
                           repo_name_prefix)
                 return repo_name
 
         return
 
     def run(self, extras):
         action = 'pull'
         self.create_svn_config()
         self.start()
 
         first_response = self.get_first_client_response()
         if not first_response:
             return self.fail("Repository name cannot be extracted")
 
         url_parts = urlparse.urlparse(first_response['url'])
 
         self.server.repo_name = self._match_repo_name(url_parts.path.strip('/'))
 
         exit_code = self.server._check_permissions(action)
         if exit_code:
             return exit_code
 
         # set the readonly flag to False if we have proper permissions
         if self.server.has_write_perm():
             self.read_only = False
         self.server.update_environment(action=action, extras=extras)
 
         self.patch_first_client_response(first_response)
         self.sync()
         return self.return_code
 
 
 class SubversionServer(VcsServer):
     backend = 'svn'
+    repo_user_agent = 'svn'
 
     def __init__(self, store, ini_path, repo_name,
                  user, user_permissions, config, env):
         super(SubversionServer, self)\
             .__init__(user, user_permissions, config, env)
         self.store = store
         self.ini_path = ini_path
         # NOTE(dan): repo_name at this point is empty,
         # this is set later in .run() based from parsed input stream
         self.repo_name = repo_name
         self._path = self.svn_path = config.get('app:main', 'ssh.executable.svn')
 
         self.tunnel = SubversionTunnelWrapper(server=self)
 
     def _handle_tunnel(self, extras):
 
         # pre-auth
         action = 'pull'
         # Special case for SVN, we extract repo name at later stage
         # exit_code = self._check_permissions(action)
         # if exit_code:
         #     return exit_code, False
 
         req = self.env['request']
         server_url = req.host_url + req.script_name
         extras['server_url'] = server_url
 
         log.debug('Using %s binaries from path %s', self.backend, self._path)
         exit_code = self.tunnel.run(extras)
 
         return exit_code, action == "push"
 
 

rhodecode/apps/ssh_support/tests/test_server_git.py

@@ -1,156 +1,156 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2016-2020 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import json
 import os
 
 import mock
 import pytest
 
 from rhodecode.apps.ssh_support.lib.backends.git import GitServer
 from rhodecode.apps.ssh_support.tests.conftest import plain_dummy_env, plain_dummy_user
 
 
 class GitServerCreator(object):
     root = '/tmp/repo/path/'
     git_path = '/usr/local/bin/git'
     config_data = {
         'app:main': {
             'ssh.executable.git': git_path,
             'vcs.hooks.protocol': 'http',
         }
     }
     repo_name = 'test_git'
     repo_mode = 'receive-pack'
     user = plain_dummy_user()
 
     def __init__(self):
         def config_get(part, key):
             return self.config_data.get(part, {}).get(key)
         self.config_mock = mock.Mock()
         self.config_mock.get = mock.Mock(side_effect=config_get)
 
     def create(self, **kwargs):
         parameters = {
             'store': self.root,
             'ini_path': '',
             'user': self.user,
             'repo_name': self.repo_name,
             'repo_mode': self.repo_mode,
             'user_permissions': {
                 self.repo_name: 'repository.admin'
             },
             'config': self.config_mock,
             'env': plain_dummy_env()
         }
         parameters.update(kwargs)
         server = GitServer(**parameters)
         return server
 
 
 @pytest.fixture()
 def git_server(app):
     return GitServerCreator()
 
 
 class TestGitServer(object):
 
     def test_command(self, git_server):
         server = git_server.create()
         expected_command = (
             'cd {root}; {git_path} {repo_mode} \'{root}{repo_name}\''.format(
                 root=git_server.root, git_path=git_server.git_path,
                 repo_mode=git_server.repo_mode, repo_name=git_server.repo_name)
         )
         assert expected_command == server.tunnel.command()
 
     @pytest.mark.parametrize('permissions, action, code', [
         ({}, 'pull', -2),
         ({'test_git': 'repository.read'}, 'pull', 0),
         ({'test_git': 'repository.read'}, 'push', -2),
         ({'test_git': 'repository.write'}, 'push', 0),
         ({'test_git': 'repository.admin'}, 'push', 0),
 
     ])
     def test_permission_checks(self, git_server, permissions, action, code):
         server = git_server.create(user_permissions=permissions)
         result = server._check_permissions(action)
         assert result is code
 
     @pytest.mark.parametrize('permissions, value', [
         ({}, False),
         ({'test_git': 'repository.read'}, False),
         ({'test_git': 'repository.write'}, True),
         ({'test_git': 'repository.admin'}, True),
 
     ])
     def test_has_write_permissions(self, git_server, permissions, value):
         server = git_server.create(user_permissions=permissions)
         result = server.has_write_perm()
         assert result is value
 
     def test_run_returns_executes_command(self, git_server):
         server = git_server.create()
         from rhodecode.apps.ssh_support.lib.backends.git import GitTunnelWrapper
 
         os.environ['SSH_CLIENT'] = '127.0.0.1'
         with mock.patch.object(GitTunnelWrapper, 'create_hooks_env') as _patch:
             _patch.return_value = 0
             with mock.patch.object(GitTunnelWrapper, 'command', return_value='date'):
                 exit_code = server.run()
 
         assert exit_code == (0, False)
 
     @pytest.mark.parametrize(
         'repo_mode, action', [
             ['receive-pack', 'push'],
             ['upload-pack', 'pull']
         ])
     def test_update_environment(self, git_server, repo_mode, action):
         server = git_server.create(repo_mode=repo_mode)
         store = server.store
 
         with mock.patch('os.environ', {'SSH_CLIENT': '10.10.10.10 b'}):
             with mock.patch('os.putenv') as putenv_mock:
                 server.update_environment(action)
 
         expected_data = {
             'username': git_server.user.username,
             'user_id': git_server.user.user_id,
             'scm': 'git',
             'repository': git_server.repo_name,
             'make_lock': None,
             'action': action,
             'ip': '10.10.10.10',
             'locked_by': [None, None],
             'config': '',
             'repo_store': store,
             'server_url': None,
             'hooks': ['push', 'pull'],
             'is_shadow_repo': False,
             'hooks_module': 'rhodecode.lib.hooks_daemon',
             'check_branch_perms': False,
             'detect_force_push': False,
-            'user_agent': u'ssh-user-agent',
+            'user_agent': u'git/ssh-user-agent',
             'SSH': True,
             'SSH_PERMISSIONS': 'repository.admin',
         }
         args, kwargs = putenv_mock.call_args
         assert json.loads(args[1]) == expected_data

rhodecode/lib/hooks_base.py

@@ -1,535 +1,538 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2013-2020 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 
 """
 Set of hooks run by RhodeCode Enterprise
 """
 
 import os
 import logging
 
 import rhodecode
 from rhodecode import events
 from rhodecode.lib import helpers as h
 from rhodecode.lib import audit_logger
-from rhodecode.lib.utils2 import safe_str
+from rhodecode.lib.utils2 import safe_str, user_agent_normalizer
 from rhodecode.lib.exceptions import (
     HTTPLockedRC, HTTPBranchProtected, UserCreationError)
 from rhodecode.model.db import Repository, User
 from rhodecode.lib.statsd_client import StatsdClient
 
 log = logging.getLogger(__name__)
 
 
 class HookResponse(object):
     def __init__(self, status, output):
         self.status = status
         self.output = output
 
     def __add__(self, other):
         other_status = getattr(other, 'status', 0)
         new_status = max(self.status, other_status)
         other_output = getattr(other, 'output', '')
         new_output = self.output + other_output
 
         return HookResponse(new_status, new_output)
 
     def __bool__(self):
         return self.status == 0
 
 
 def is_shadow_repo(extras):
     """
     Returns ``True`` if this is an action executed against a shadow repository.
     """
     return extras['is_shadow_repo']
 
 
 def _get_scm_size(alias, root_path):
 
     if not alias.startswith('.'):
         alias += '.'
 
     size_scm, size_root = 0, 0
     for path, unused_dirs, files in os.walk(safe_str(root_path)):
         if path.find(alias) != -1:
             for f in files:
                 try:
                     size_scm += os.path.getsize(os.path.join(path, f))
                 except OSError:
                     pass
         else:
             for f in files:
                 try:
                     size_root += os.path.getsize(os.path.join(path, f))
                 except OSError:
                     pass
 
     size_scm_f = h.format_byte_size_binary(size_scm)
     size_root_f = h.format_byte_size_binary(size_root)
     size_total_f = h.format_byte_size_binary(size_root + size_scm)
 
     return size_scm_f, size_root_f, size_total_f
 
 
 # actual hooks called by Mercurial internally, and GIT by our Python Hooks
 def repo_size(extras):
     """Present size of repository after push."""
     repo = Repository.get_by_repo_name(extras.repository)
     vcs_part = safe_str(u'.%s' % repo.repo_type)
     size_vcs, size_root, size_total = _get_scm_size(vcs_part,
                                                     repo.repo_full_path)
     msg = ('Repository `%s` size summary %s:%s repo:%s total:%s\n'
            % (repo.repo_name, vcs_part, size_vcs, size_root, size_total))
     return HookResponse(0, msg)
 
 
 def pre_push(extras):
     """
     Hook executed before pushing code.
 
     It bans pushing when the repository is locked.
     """
 
     user = User.get_by_username(extras.username)
     output = ''
     if extras.locked_by[0] and user.user_id != int(extras.locked_by[0]):
         locked_by = User.get(extras.locked_by[0]).username
         reason = extras.locked_by[2]
         # this exception is interpreted in git/hg middlewares and based
         # on that proper return code is server to client
         _http_ret = HTTPLockedRC(
             _locked_by_explanation(extras.repository, locked_by, reason))
         if str(_http_ret.code).startswith('2'):
             # 2xx Codes don't raise exceptions
             output = _http_ret.title
         else:
             raise _http_ret
 
     hook_response = ''
     if not is_shadow_repo(extras):
         if extras.commit_ids and extras.check_branch_perms:
 
             auth_user = user.AuthUser()
             repo = Repository.get_by_repo_name(extras.repository)
             affected_branches = []
             if repo.repo_type == 'hg':
                 for entry in extras.commit_ids:
                     if entry['type'] == 'branch':
                         is_forced = bool(entry['multiple_heads'])
                         affected_branches.append([entry['name'], is_forced])
             elif repo.repo_type == 'git':
                 for entry in extras.commit_ids:
                     if entry['type'] == 'heads':
                         is_forced = bool(entry['pruned_sha'])
                         affected_branches.append([entry['name'], is_forced])
 
             for branch_name, is_forced in affected_branches:
 
                 rule, branch_perm = auth_user.get_rule_and_branch_permission(
                     extras.repository, branch_name)
                 if not branch_perm:
                     # no branch permission found for this branch, just keep checking
                     continue
 
                 if branch_perm == 'branch.push_force':
                     continue
                 elif branch_perm == 'branch.push' and is_forced is False:
                     continue
                 elif branch_perm == 'branch.push' and is_forced is True:
                     halt_message = 'Branch `{}` changes rejected by rule {}. ' \
                                    'FORCE PUSH FORBIDDEN.'.format(branch_name, rule)
                 else:
                     halt_message = 'Branch `{}` changes rejected by rule {}.'.format(
                         branch_name, rule)
 
                 if halt_message:
                     _http_ret = HTTPBranchProtected(halt_message)
                     raise _http_ret
 
         # Propagate to external components. This is done after checking the
         # lock, for consistent behavior.
         hook_response = pre_push_extension(
             repo_store_path=Repository.base_path(), **extras)
         events.trigger(events.RepoPrePushEvent(
             repo_name=extras.repository, extras=extras))
 
     return HookResponse(0, output) + hook_response
 
 
 def pre_pull(extras):
     """
     Hook executed before pulling the code.
 
     It bans pulling when the repository is locked.
     """
 
     output = ''
     if extras.locked_by[0]:
         locked_by = User.get(extras.locked_by[0]).username
         reason = extras.locked_by[2]
         # this exception is interpreted in git/hg middlewares and based
         # on that proper return code is server to client
         _http_ret = HTTPLockedRC(
             _locked_by_explanation(extras.repository, locked_by, reason))
         if str(_http_ret.code).startswith('2'):
             # 2xx Codes don't raise exceptions
             output = _http_ret.title
         else:
             raise _http_ret
 
     # Propagate to external components. This is done after checking the
     # lock, for consistent behavior.
     hook_response = ''
     if not is_shadow_repo(extras):
         extras.hook_type = extras.hook_type or 'pre_pull'
         hook_response = pre_pull_extension(
             repo_store_path=Repository.base_path(), **extras)
         events.trigger(events.RepoPrePullEvent(
             repo_name=extras.repository, extras=extras))
 
     return HookResponse(0, output) + hook_response
 
 
 def post_pull(extras):
     """Hook executed after client pulls the code."""
 
     audit_user = audit_logger.UserWrap(
         username=extras.username,
         ip_addr=extras.ip)
     repo = audit_logger.RepoWrap(repo_name=extras.repository)
     audit_logger.store(
         'user.pull', action_data={'user_agent': extras.user_agent},
         user=audit_user, repo=repo, commit=True)
 
     statsd = StatsdClient.statsd
     if statsd:
-        statsd.incr('rhodecode_pull_total')
-
+        statsd.incr('rhodecode_pull_total', tags=[
+            'user-agent:{}'.format(user_agent_normalizer(extras.user_agent)),
+        ])
     output = ''
     # make lock is a tri state False, True, None. We only make lock on True
     if extras.make_lock is True and not is_shadow_repo(extras):
         user = User.get_by_username(extras.username)
         Repository.lock(Repository.get_by_repo_name(extras.repository),
                         user.user_id,
                         lock_reason=Repository.LOCK_PULL)
         msg = 'Made lock on repo `%s`' % (extras.repository,)
         output += msg
 
     if extras.locked_by[0]:
         locked_by = User.get(extras.locked_by[0]).username
         reason = extras.locked_by[2]
         _http_ret = HTTPLockedRC(
             _locked_by_explanation(extras.repository, locked_by, reason))
         if str(_http_ret.code).startswith('2'):
             # 2xx Codes don't raise exceptions
             output += _http_ret.title
 
     # Propagate to external components.
     hook_response = ''
     if not is_shadow_repo(extras):
         extras.hook_type = extras.hook_type or 'post_pull'
         hook_response = post_pull_extension(
             repo_store_path=Repository.base_path(), **extras)
         events.trigger(events.RepoPullEvent(
             repo_name=extras.repository, extras=extras))
 
     return HookResponse(0, output) + hook_response
 
 
 def post_push(extras):
     """Hook executed after user pushes to the repository."""
     commit_ids = extras.commit_ids
 
     # log the push call
     audit_user = audit_logger.UserWrap(
         username=extras.username, ip_addr=extras.ip)
     repo = audit_logger.RepoWrap(repo_name=extras.repository)
     audit_logger.store(
         'user.push', action_data={
             'user_agent': extras.user_agent,
             'commit_ids': commit_ids[:400]},
         user=audit_user, repo=repo, commit=True)
 
     statsd = StatsdClient.statsd
     if statsd:
-        statsd.incr('rhodecode_push_total')
+        statsd.incr('rhodecode_push_total', tags=[
+            'user-agent:{}'.format(user_agent_normalizer(extras.user_agent)),
+        ])
 
     # Propagate to external components.
     output = ''
     # make lock is a tri state False, True, None. We only release lock on False
     if extras.make_lock is False and not is_shadow_repo(extras):
         Repository.unlock(Repository.get_by_repo_name(extras.repository))
         msg = 'Released lock on repo `{}`\n'.format(safe_str(extras.repository))
         output += msg
 
     if extras.locked_by[0]:
         locked_by = User.get(extras.locked_by[0]).username
         reason = extras.locked_by[2]
         _http_ret = HTTPLockedRC(
             _locked_by_explanation(extras.repository, locked_by, reason))
         # TODO: johbo: if not?
         if str(_http_ret.code).startswith('2'):
             # 2xx Codes don't raise exceptions
             output += _http_ret.title
 
     if extras.new_refs:
         tmpl = '{}/{}/pull-request/new?{{ref_type}}={{ref_name}}'.format(
             safe_str(extras.server_url), safe_str(extras.repository))
 
         for branch_name in extras.new_refs['branches']:
             output += 'RhodeCode: open pull request link: {}\n'.format(
                 tmpl.format(ref_type='branch', ref_name=safe_str(branch_name)))
 
         for book_name in extras.new_refs['bookmarks']:
             output += 'RhodeCode: open pull request link: {}\n'.format(
                 tmpl.format(ref_type='bookmark', ref_name=safe_str(book_name)))
 
     hook_response = ''
     if not is_shadow_repo(extras):
         hook_response = post_push_extension(
             repo_store_path=Repository.base_path(),
             **extras)
         events.trigger(events.RepoPushEvent(
             repo_name=extras.repository, pushed_commit_ids=commit_ids, extras=extras))
 
     output += 'RhodeCode: push completed\n'
     return HookResponse(0, output) + hook_response
 
 
 def _locked_by_explanation(repo_name, user_name, reason):
     message = (
         'Repository `%s` locked by user `%s`. Reason:`%s`'
         % (repo_name, user_name, reason))
     return message
 
 
 def check_allowed_create_user(user_dict, created_by, **kwargs):
     # pre create hooks
     if pre_create_user.is_active():
         hook_result = pre_create_user(created_by=created_by, **user_dict)
         allowed = hook_result.status == 0
         if not allowed:
             reason = hook_result.output
             raise UserCreationError(reason)
 
 
 class ExtensionCallback(object):
     """
     Forwards a given call to rcextensions, sanitizes keyword arguments.
 
     Does check if there is an extension active for that hook. If it is
     there, it will forward all `kwargs_keys` keyword arguments to the
     extension callback.
     """
 
     def __init__(self, hook_name, kwargs_keys):
         self._hook_name = hook_name
         self._kwargs_keys = set(kwargs_keys)
 
     def __call__(self, *args, **kwargs):
         log.debug('Calling extension callback for `%s`', self._hook_name)
         callback = self._get_callback()
         if not callback:
             log.debug('extension callback `%s` not found, skipping...', self._hook_name)
             return
 
         kwargs_to_pass = {}
         for key in self._kwargs_keys:
             try:
                 kwargs_to_pass[key] = kwargs[key]
             except KeyError:
                 log.error('Failed to fetch %s key from given kwargs. '
                           'Expected keys: %s', key, self._kwargs_keys)
                 raise
 
         # backward compat for removed api_key for old hooks. This was it works
         # with older rcextensions that require api_key present
         if self._hook_name in ['CREATE_USER_HOOK', 'DELETE_USER_HOOK']:
             kwargs_to_pass['api_key'] = '_DEPRECATED_'
         return callback(**kwargs_to_pass)
 
     def is_active(self):
         return hasattr(rhodecode.EXTENSIONS, self._hook_name)
 
     def _get_callback(self):
         return getattr(rhodecode.EXTENSIONS, self._hook_name, None)
 
 
 pre_pull_extension = ExtensionCallback(
     hook_name='PRE_PULL_HOOK',
     kwargs_keys=(
         'server_url', 'config', 'scm', 'username', 'ip', 'action',
         'repository', 'hook_type', 'user_agent', 'repo_store_path',))
 
 
 post_pull_extension = ExtensionCallback(
     hook_name='PULL_HOOK',
     kwargs_keys=(
         'server_url', 'config', 'scm', 'username', 'ip', 'action',
         'repository', 'hook_type', 'user_agent', 'repo_store_path',))
 
 
 pre_push_extension = ExtensionCallback(
     hook_name='PRE_PUSH_HOOK',
     kwargs_keys=(
         'server_url', 'config', 'scm', 'username', 'ip', 'action',
         'repository', 'repo_store_path', 'commit_ids', 'hook_type', 'user_agent',))
 
 
 post_push_extension = ExtensionCallback(
     hook_name='PUSH_HOOK',
     kwargs_keys=(
         'server_url', 'config', 'scm', 'username', 'ip', 'action',
         'repository', 'repo_store_path', 'commit_ids', 'hook_type', 'user_agent',))
 
 
 pre_create_user = ExtensionCallback(
     hook_name='PRE_CREATE_USER_HOOK',
     kwargs_keys=(
         'username', 'password', 'email', 'firstname', 'lastname', 'active',
         'admin', 'created_by'))
 
 
 create_pull_request = ExtensionCallback(
     hook_name='CREATE_PULL_REQUEST',
     kwargs_keys=(
         'server_url', 'config', 'scm', 'username', 'ip', 'action',
         'repository', 'pull_request_id', 'url', 'title', 'description',
         'status', 'created_on', 'updated_on', 'commit_ids', 'review_status',
         'mergeable', 'source', 'target', 'author', 'reviewers'))
 
 
 merge_pull_request = ExtensionCallback(
     hook_name='MERGE_PULL_REQUEST',
     kwargs_keys=(
         'server_url', 'config', 'scm', 'username', 'ip', 'action',
         'repository', 'pull_request_id', 'url', 'title', 'description',
         'status', 'created_on', 'updated_on', 'commit_ids', 'review_status',
         'mergeable', 'source', 'target', 'author', 'reviewers'))
 
 
 close_pull_request = ExtensionCallback(
     hook_name='CLOSE_PULL_REQUEST',
     kwargs_keys=(
         'server_url', 'config', 'scm', 'username', 'ip', 'action',
         'repository', 'pull_request_id', 'url', 'title', 'description',
         'status', 'created_on', 'updated_on', 'commit_ids', 'review_status',
         'mergeable', 'source', 'target', 'author', 'reviewers'))
 
 
 review_pull_request = ExtensionCallback(
     hook_name='REVIEW_PULL_REQUEST',
     kwargs_keys=(
         'server_url', 'config', 'scm', 'username', 'ip', 'action',
         'repository', 'pull_request_id', 'url', 'title', 'description',
         'status', 'created_on', 'updated_on', 'commit_ids', 'review_status',
         'mergeable', 'source', 'target', 'author', 'reviewers'))
 
 
 comment_pull_request = ExtensionCallback(
     hook_name='COMMENT_PULL_REQUEST',
     kwargs_keys=(
         'server_url', 'config', 'scm', 'username', 'ip', 'action',
         'repository', 'pull_request_id', 'url', 'title', 'description',
         'status', 'comment', 'created_on', 'updated_on', 'commit_ids', 'review_status',
         'mergeable', 'source', 'target', 'author', 'reviewers'))
 
 
 comment_edit_pull_request = ExtensionCallback(
     hook_name='COMMENT_EDIT_PULL_REQUEST',
     kwargs_keys=(
         'server_url', 'config', 'scm', 'username', 'ip', 'action',
         'repository', 'pull_request_id', 'url', 'title', 'description',
         'status', 'comment', 'created_on', 'updated_on', 'commit_ids', 'review_status',
         'mergeable', 'source', 'target', 'author', 'reviewers'))
 
 
 update_pull_request = ExtensionCallback(
     hook_name='UPDATE_PULL_REQUEST',
     kwargs_keys=(
         'server_url', 'config', 'scm', 'username', 'ip', 'action',
         'repository', 'pull_request_id', 'url', 'title', 'description',
         'status', 'created_on', 'updated_on', 'commit_ids', 'review_status',
         'mergeable', 'source', 'target', 'author', 'reviewers'))
 
 
 create_user = ExtensionCallback(
     hook_name='CREATE_USER_HOOK',
     kwargs_keys=(
         'username', 'full_name_or_username', 'full_contact', 'user_id',
         'name', 'firstname', 'short_contact', 'admin', 'lastname',
         'ip_addresses', 'extern_type', 'extern_name',
         'email', 'api_keys', 'last_login',
         'full_name', 'active', 'password', 'emails',
         'inherit_default_permissions', 'created_by', 'created_on'))
 
 
 delete_user = ExtensionCallback(
     hook_name='DELETE_USER_HOOK',
     kwargs_keys=(
         'username', 'full_name_or_username', 'full_contact', 'user_id',
         'name', 'firstname', 'short_contact', 'admin', 'lastname',
         'ip_addresses',
         'email', 'last_login',
         'full_name', 'active', 'password', 'emails',
         'inherit_default_permissions', 'deleted_by'))
 
 
 create_repository = ExtensionCallback(
     hook_name='CREATE_REPO_HOOK',
     kwargs_keys=(
         'repo_name', 'repo_type', 'description', 'private', 'created_on',
         'enable_downloads', 'repo_id', 'user_id', 'enable_statistics',
         'clone_uri', 'fork_id', 'group_id', 'created_by'))
 
 
 delete_repository = ExtensionCallback(
     hook_name='DELETE_REPO_HOOK',
     kwargs_keys=(
         'repo_name', 'repo_type', 'description', 'private', 'created_on',
         'enable_downloads', 'repo_id', 'user_id', 'enable_statistics',
         'clone_uri', 'fork_id', 'group_id', 'deleted_by', 'deleted_on'))
 
 
 comment_commit_repository = ExtensionCallback(
     hook_name='COMMENT_COMMIT_REPO_HOOK',
     kwargs_keys=(
         'repo_name', 'repo_type', 'description', 'private', 'created_on',
         'enable_downloads', 'repo_id', 'user_id', 'enable_statistics',
         'clone_uri', 'fork_id', 'group_id',
         'repository', 'created_by', 'comment', 'commit'))
 
 comment_edit_commit_repository = ExtensionCallback(
     hook_name='COMMENT_EDIT_COMMIT_REPO_HOOK',
     kwargs_keys=(
         'repo_name', 'repo_type', 'description', 'private', 'created_on',
         'enable_downloads', 'repo_id', 'user_id', 'enable_statistics',
         'clone_uri', 'fork_id', 'group_id',
         'repository', 'created_by', 'comment', 'commit'))
 
 
 create_repository_group = ExtensionCallback(
     hook_name='CREATE_REPO_GROUP_HOOK',
     kwargs_keys=(
         'group_name', 'group_parent_id', 'group_description',
         'group_id', 'user_id', 'created_by', 'created_on',
         'enable_locking'))

rhodecode/lib/utils2.py

@@ -1,1148 +1,1165 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2011-2020 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 
 """
 Some simple helper functions
 """
 
 import collections
 import datetime
 import dateutil.relativedelta
 import hashlib
 import logging
 import re
 import sys
 import time
 import urllib
 import urlobject
 import uuid
 import getpass
 from functools import update_wrapper, partial, wraps
 
 import pygments.lexers
 import sqlalchemy
 import sqlalchemy.engine.url
 import sqlalchemy.exc
 import sqlalchemy.sql
 import webob
 import pyramid.threadlocal
 from pyramid import compat
 from pyramid.settings import asbool
 
 import rhodecode
 from rhodecode.translation import _, _pluralize
 
 
 def md5(s):
     return hashlib.md5(s).hexdigest()
 
 
 def md5_safe(s):
     return md5(safe_str(s))
 
 
 def sha1(s):
     return hashlib.sha1(s).hexdigest()
 
 
 def sha1_safe(s):
     return sha1(safe_str(s))
 
 
 def __get_lem(extra_mapping=None):
     """
     Get language extension map based on what's inside pygments lexers
     """
     d = collections.defaultdict(lambda: [])
 
     def __clean(s):
         s = s.lstrip('*')
         s = s.lstrip('.')
 
         if s.find('[') != -1:
             exts = []
             start, stop = s.find('['), s.find(']')
 
             for suffix in s[start + 1:stop]:
                 exts.append(s[:s.find('[')] + suffix)
             return [e.lower() for e in exts]
         else:
             return [s.lower()]
 
     for lx, t in sorted(pygments.lexers.LEXERS.items()):
         m = map(__clean, t[-2])
         if m:
             m = reduce(lambda x, y: x + y, m)
             for ext in m:
                 desc = lx.replace('Lexer', '')
                 d[ext].append(desc)
 
     data = dict(d)
 
     extra_mapping = extra_mapping or {}
     if extra_mapping:
         for k, v in extra_mapping.items():
             if k not in data:
                 # register new mapping2lexer
                 data[k] = [v]
 
     return data
 
 
 def str2bool(_str):
     """
     returns True/False value from given string, it tries to translate the
     string into boolean
 
     :param _str: string value to translate into boolean
     :rtype: boolean
     :returns: boolean from given string
     """
     if _str is None:
         return False
     if _str in (True, False):
         return _str
     _str = str(_str).strip().lower()
     return _str in ('t', 'true', 'y', 'yes', 'on', '1')
 
 
 def aslist(obj, sep=None, strip=True):
     """
     Returns given string separated by sep as list
 
     :param obj:
     :param sep:
     :param strip:
     """
     if isinstance(obj, (basestring,)):
         lst = obj.split(sep)
         if strip:
             lst = [v.strip() for v in lst]
         return lst
     elif isinstance(obj, (list, tuple)):
         return obj
     elif obj is None:
         return []
     else:
         return [obj]
 
 
 def convert_line_endings(line, mode):
     """
     Converts a given line  "line end" accordingly to given mode
 
     Available modes are::
         0 - Unix
         1 - Mac
         2 - DOS
 
     :param line: given line to convert
     :param mode: mode to convert to
     :rtype: str
     :return: converted line according to mode
     """
     if mode == 0:
         line = line.replace('\r\n', '\n')
         line = line.replace('\r', '\n')
     elif mode == 1:
         line = line.replace('\r\n', '\r')
         line = line.replace('\n', '\r')
     elif mode == 2:
         line = re.sub('\r(?!\n)|(?<!\r)\n', '\r\n', line)
     return line
 
 
 def detect_mode(line, default):
     """
     Detects line break for given line, if line break couldn't be found
     given default value is returned
 
     :param line: str line
     :param default: default
     :rtype: int
     :return: value of line end on of 0 - Unix, 1 - Mac, 2 - DOS
     """
     if line.endswith('\r\n'):
         return 2
     elif line.endswith('\n'):
         return 0
     elif line.endswith('\r'):
         return 1
     else:
         return default
 
 
 def safe_int(val, default=None):
     """
     Returns int() of val if val is not convertable to int use default
     instead
 
     :param val:
     :param default:
     """
 
     try:
         val = int(val)
     except (ValueError, TypeError):
         val = default
 
     return val
 
 
 def safe_unicode(str_, from_encoding=None, use_chardet=False):
     """
     safe unicode function. Does few trick to turn str_ into unicode
 
     In case of UnicodeDecode error, we try to return it with encoding detected
     by chardet library if it fails fallback to unicode with errors replaced
 
     :param str_: string to decode
     :rtype: unicode
     :returns: unicode object
     """
     if isinstance(str_, unicode):
         return str_
 
     if not from_encoding:
         DEFAULT_ENCODINGS = aslist(rhodecode.CONFIG.get('default_encoding',
                                                         'utf8'), sep=',')
         from_encoding = DEFAULT_ENCODINGS
 
     if not isinstance(from_encoding, (list, tuple)):
         from_encoding = [from_encoding]
 
     try:
         return unicode(str_)
     except UnicodeDecodeError:
         pass
 
     for enc in from_encoding:
         try:
             return unicode(str_, enc)
         except UnicodeDecodeError:
             pass
 
     if use_chardet:
         try:
             import chardet
             encoding = chardet.detect(str_)['encoding']
             if encoding is None:
                 raise Exception()
             return str_.decode(encoding)
         except (ImportError, UnicodeDecodeError, Exception):
             return unicode(str_, from_encoding[0], 'replace')
     else:
         return unicode(str_, from_encoding[0], 'replace')
 
 def safe_str(unicode_, to_encoding=None, use_chardet=False):
     """
     safe str function. Does few trick to turn unicode_ into string
 
     In case of UnicodeEncodeError, we try to return it with encoding detected
     by chardet library if it fails fallback to string with errors replaced
 
     :param unicode_: unicode to encode
     :rtype: str
     :returns: str object
     """
 
     # if it's not basestr cast to str
     if not isinstance(unicode_, compat.string_types):
         return str(unicode_)
 
     if isinstance(unicode_, str):
         return unicode_
 
     if not to_encoding:
         DEFAULT_ENCODINGS = aslist(rhodecode.CONFIG.get('default_encoding',
                                                         'utf8'), sep=',')
         to_encoding = DEFAULT_ENCODINGS
 
     if not isinstance(to_encoding, (list, tuple)):
         to_encoding = [to_encoding]
 
     for enc in to_encoding:
         try:
             return unicode_.encode(enc)
         except UnicodeEncodeError:
             pass
 
     if use_chardet:
         try:
             import chardet
             encoding = chardet.detect(unicode_)['encoding']
             if encoding is None:
                 raise UnicodeEncodeError()
 
             return unicode_.encode(encoding)
         except (ImportError, UnicodeEncodeError):
             return unicode_.encode(to_encoding[0], 'replace')
     else:
         return unicode_.encode(to_encoding[0], 'replace')
 
 
 def remove_suffix(s, suffix):
     if s.endswith(suffix):
         s = s[:-1 * len(suffix)]
     return s
 
 
 def remove_prefix(s, prefix):
     if s.startswith(prefix):
         s = s[len(prefix):]
     return s
 
 
 def find_calling_context(ignore_modules=None):
     """
     Look through the calling stack and return the frame which called
     this function and is part of core module ( ie. rhodecode.* )
 
     :param ignore_modules: list of modules to ignore eg. ['rhodecode.lib']
     """
 
     ignore_modules = ignore_modules or []
 
     f = sys._getframe(2)
     while f.f_back is not None:
         name = f.f_globals.get('__name__')
         if name and name.startswith(__name__.split('.')[0]):
             if name not in ignore_modules:
                 return f
         f = f.f_back
     return None
 
 
 def ping_connection(connection, branch):
     if branch:
         # "branch" refers to a sub-connection of a connection,
         # we don't want to bother pinging on these.
         return
 
     # turn off "close with result".  This flag is only used with
     # "connectionless" execution, otherwise will be False in any case
     save_should_close_with_result = connection.should_close_with_result
     connection.should_close_with_result = False
 
     try:
         # run a SELECT 1.   use a core select() so that
         # the SELECT of a scalar value without a table is
         # appropriately formatted for the backend
         connection.scalar(sqlalchemy.sql.select([1]))
     except sqlalchemy.exc.DBAPIError as err:
         # catch SQLAlchemy's DBAPIError, which is a wrapper
         # for the DBAPI's exception.  It includes a .connection_invalidated
         # attribute which specifies if this connection is a "disconnect"
         # condition, which is based on inspection of the original exception
         # by the dialect in use.
         if err.connection_invalidated:
             # run the same SELECT again - the connection will re-validate
             # itself and establish a new connection.  The disconnect detection
             # here also causes the whole connection pool to be invalidated
             # so that all stale connections are discarded.
             connection.scalar(sqlalchemy.sql.select([1]))
         else:
             raise
     finally:
         # restore "close with result"
         connection.should_close_with_result = save_should_close_with_result
 
 
 def engine_from_config(configuration, prefix='sqlalchemy.', **kwargs):
     """Custom engine_from_config functions."""
     log = logging.getLogger('sqlalchemy.engine')
     use_ping_connection = asbool(configuration.pop('sqlalchemy.db1.ping_connection', None))
     debug = asbool(configuration.pop('sqlalchemy.db1.debug_query', None))
 
     engine = sqlalchemy.engine_from_config(configuration, prefix, **kwargs)
 
     def color_sql(sql):
         color_seq = '\033[1;33m'  # This is yellow: code 33
         normal = '\x1b[0m'
         return ''.join([color_seq, sql, normal])
 
     if use_ping_connection:
         log.debug('Adding ping_connection on the engine config.')
         sqlalchemy.event.listen(engine, "engine_connect", ping_connection)
 
     if debug:
         # attach events only for debug configuration
         def before_cursor_execute(conn, cursor, statement,
                                   parameters, context, executemany):
             setattr(conn, 'query_start_time', time.time())
             log.info(color_sql(">>>>> STARTING QUERY >>>>>"))
             calling_context = find_calling_context(ignore_modules=[
                 'rhodecode.lib.caching_query',
                 'rhodecode.model.settings',
             ])
             if calling_context:
                 log.info(color_sql('call context %s:%s' % (
                     calling_context.f_code.co_filename,
                     calling_context.f_lineno,
                 )))
 
         def after_cursor_execute(conn, cursor, statement,
                                  parameters, context, executemany):
             delattr(conn, 'query_start_time')
 
         sqlalchemy.event.listen(engine, "before_cursor_execute", before_cursor_execute)
         sqlalchemy.event.listen(engine, "after_cursor_execute", after_cursor_execute)
 
     return engine
 
 
 def get_encryption_key(config):
     secret = config.get('rhodecode.encrypted_values.secret')
     default = config['beaker.session.secret']
     return secret or default
 
 
 def age(prevdate, now=None, show_short_version=False, show_suffix=True,
         short_format=False):
     """
     Turns a datetime into an age string.
     If show_short_version is True, this generates a shorter string with
     an approximate age; ex. '1 day ago', rather than '1 day and 23 hours ago'.
 
     * IMPORTANT*
     Code of this function is written in special way so it's easier to
     backport it to javascript. If you mean to update it, please also update
     `jquery.timeago-extension.js` file
 
     :param prevdate: datetime object
     :param now: get current time, if not define we use
         `datetime.datetime.now()`
     :param show_short_version: if it should approximate the date and
         return a shorter string
     :param show_suffix:
     :param short_format: show short format, eg 2D instead of 2 days
     :rtype: unicode
     :returns: unicode words describing age
     """
 
     def _get_relative_delta(now, prevdate):
         base = dateutil.relativedelta.relativedelta(now, prevdate)
         return {
             'year': base.years,
             'month': base.months,
             'day': base.days,
             'hour': base.hours,
             'minute': base.minutes,
             'second': base.seconds,
         }
 
     def _is_leap_year(year):
         return year % 4 == 0 and (year % 100 != 0 or year % 400 == 0)
 
     def get_month(prevdate):
         return prevdate.month
 
     def get_year(prevdate):
         return prevdate.year
 
     now = now or datetime.datetime.now()
     order = ['year', 'month', 'day', 'hour', 'minute', 'second']
     deltas = {}
     future = False
 
     if prevdate > now:
         now_old = now
         now = prevdate
         prevdate = now_old
         future = True
     if future:
         prevdate = prevdate.replace(microsecond=0)
     # Get date parts deltas
     for part in order:
         rel_delta = _get_relative_delta(now, prevdate)
         deltas[part] = rel_delta[part]
 
     # Fix negative offsets (there is 1 second between 10:59:59 and 11:00:00,
     # not 1 hour, -59 minutes and -59 seconds)
     offsets = [[5, 60], [4, 60], [3, 24]]
     for element in offsets:  # seconds, minutes, hours
         num = element[0]
         length = element[1]
 
         part = order[num]
         carry_part = order[num - 1]
 
         if deltas[part] < 0:
             deltas[part] += length
             deltas[carry_part] -= 1
 
     # Same thing for days except that the increment depends on the (variable)
     # number of days in the month
     month_lengths = [31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31]
     if deltas['day'] < 0:
         if get_month(prevdate) == 2 and _is_leap_year(get_year(prevdate)):
             deltas['day'] += 29
         else:
             deltas['day'] += month_lengths[get_month(prevdate) - 1]
 
         deltas['month'] -= 1
 
     if deltas['month'] < 0:
         deltas['month'] += 12
         deltas['year'] -= 1
 
     # Format the result
     if short_format:
         fmt_funcs = {
             'year': lambda d: u'%dy' % d,
             'month': lambda d: u'%dm' % d,
             'day': lambda d: u'%dd' % d,
             'hour': lambda d: u'%dh' % d,
             'minute': lambda d: u'%dmin' % d,
             'second': lambda d: u'%dsec' % d,
         }
     else:
         fmt_funcs = {
             'year': lambda d: _pluralize(u'${num} year', u'${num} years', d, mapping={'num': d}).interpolate(),
             'month': lambda d: _pluralize(u'${num} month', u'${num} months', d, mapping={'num': d}).interpolate(),
             'day': lambda d: _pluralize(u'${num} day', u'${num} days', d, mapping={'num': d}).interpolate(),
             'hour': lambda d: _pluralize(u'${num} hour', u'${num} hours', d, mapping={'num': d}).interpolate(),
             'minute': lambda d: _pluralize(u'${num} minute', u'${num} minutes', d, mapping={'num': d}).interpolate(),
             'second': lambda d: _pluralize(u'${num} second', u'${num} seconds', d, mapping={'num': d}).interpolate(),
         }
 
     i = 0
     for part in order:
         value = deltas[part]
         if value != 0:
 
             if i < 5:
                 sub_part = order[i + 1]
                 sub_value = deltas[sub_part]
             else:
                 sub_value = 0
 
             if sub_value == 0 or show_short_version:
                 _val = fmt_funcs[part](value)
                 if future:
                     if show_suffix:
                         return _(u'in ${ago}', mapping={'ago': _val})
                     else:
                         return _(_val)
 
                 else:
                     if show_suffix:
                         return _(u'${ago} ago', mapping={'ago': _val})
                     else:
                         return _(_val)
 
             val = fmt_funcs[part](value)
             val_detail = fmt_funcs[sub_part](sub_value)
             mapping = {'val': val, 'detail': val_detail}
 
             if short_format:
                 datetime_tmpl = _(u'${val}, ${detail}', mapping=mapping)
                 if show_suffix:
                     datetime_tmpl = _(u'${val}, ${detail} ago', mapping=mapping)
                     if future:
                         datetime_tmpl = _(u'in ${val}, ${detail}', mapping=mapping)
             else:
                 datetime_tmpl = _(u'${val} and ${detail}', mapping=mapping)
                 if show_suffix:
                     datetime_tmpl = _(u'${val} and ${detail} ago', mapping=mapping)
                     if future:
                         datetime_tmpl = _(u'in ${val} and ${detail}', mapping=mapping)
 
             return datetime_tmpl
         i += 1
     return _(u'just now')
 
 
 def age_from_seconds(seconds):
     seconds = safe_int(seconds) or 0
     prevdate = time_to_datetime(time.time() + seconds)
     return age(prevdate, show_suffix=False, show_short_version=True)
 
 
 def cleaned_uri(uri):
     """
     Quotes '[' and ']' from uri if there is only one of them.
         according to RFC3986 we cannot use such chars in uri
     :param uri:
     :return: uri without this chars
     """
     return urllib.quote(uri, safe='@$:/')
 
 
 def credentials_filter(uri):
     """
     Returns a url with removed credentials
 
     :param uri:
     """
     import urlobject
     if isinstance(uri, rhodecode.lib.encrypt.InvalidDecryptedValue):
         return 'InvalidDecryptionKey'
 
     url_obj = urlobject.URLObject(cleaned_uri(uri))
     url_obj = url_obj.without_password().without_username()
 
     return url_obj
 
 
 def get_host_info(request):
     """
     Generate host info, to obtain full url e.g https://server.com
     use this
     `{scheme}://{netloc}`
     """
     if not request:
         return {}
 
     qualified_home_url = request.route_url('home')
     parsed_url = urlobject.URLObject(qualified_home_url)
     decoded_path = safe_unicode(urllib.unquote(parsed_url.path.rstrip('/')))
 
     return {
         'scheme': parsed_url.scheme,
         'netloc': parsed_url.netloc+decoded_path,
         'hostname': parsed_url.hostname,
     }
 
 
 def get_clone_url(request, uri_tmpl, repo_name, repo_id, repo_type, **override):
     qualified_home_url = request.route_url('home')
     parsed_url = urlobject.URLObject(qualified_home_url)
     decoded_path = safe_unicode(urllib.unquote(parsed_url.path.rstrip('/')))
 
     args = {
         'scheme': parsed_url.scheme,
         'user': '',
         'sys_user': getpass.getuser(),
         # path if we use proxy-prefix
         'netloc': parsed_url.netloc+decoded_path,
         'hostname': parsed_url.hostname,
         'prefix': decoded_path,
         'repo': repo_name,
         'repoid': str(repo_id),
         'repo_type': repo_type
     }
     args.update(override)
     args['user'] = urllib.quote(safe_str(args['user']))
 
     for k, v in args.items():
         uri_tmpl = uri_tmpl.replace('{%s}' % k, v)
 
     # special case for SVN clone url
     if repo_type == 'svn':
         uri_tmpl = uri_tmpl.replace('ssh://', 'svn+ssh://')
 
     # remove leading @ sign if it's present. Case of empty user
     url_obj = urlobject.URLObject(uri_tmpl)
     url = url_obj.with_netloc(url_obj.netloc.lstrip('@'))
 
     return safe_unicode(url)
 
 
 def get_commit_safe(repo, commit_id=None, commit_idx=None, pre_load=None,
                     maybe_unreachable=False, reference_obj=None):
     """
     Safe version of get_commit if this commit doesn't exists for a
     repository it returns a Dummy one instead
 
     :param repo: repository instance
     :param commit_id: commit id as str
     :param commit_idx: numeric commit index
     :param pre_load: optional list of commit attributes to load
     :param maybe_unreachable: translate unreachable commits on git repos
     :param reference_obj: explicitly search via a reference obj in git. E.g "branch:123" would mean branch "123"
     """
     # TODO(skreft): remove these circular imports
     from rhodecode.lib.vcs.backends.base import BaseRepository, EmptyCommit
     from rhodecode.lib.vcs.exceptions import RepositoryError
     if not isinstance(repo, BaseRepository):
         raise Exception('You must pass an Repository '
                         'object as first argument got %s', type(repo))
 
     try:
         commit = repo.get_commit(
             commit_id=commit_id, commit_idx=commit_idx, pre_load=pre_load,
             maybe_unreachable=maybe_unreachable, reference_obj=reference_obj)
     except (RepositoryError, LookupError):
         commit = EmptyCommit()
     return commit
 
 
 def datetime_to_time(dt):
     if dt:
         return time.mktime(dt.timetuple())
 
 
 def time_to_datetime(tm):
     if tm:
         if isinstance(tm, compat.string_types):
             try:
                 tm = float(tm)
             except ValueError:
                 return
         return datetime.datetime.fromtimestamp(tm)
 
 
 def time_to_utcdatetime(tm):
     if tm:
         if isinstance(tm, compat.string_types):
             try:
                 tm = float(tm)
             except ValueError:
                 return
         return datetime.datetime.utcfromtimestamp(tm)
 
 
 MENTIONS_REGEX = re.compile(
     # ^@ or @ without any special chars in front
     r'(?:^@|[^a-zA-Z0-9\-\_\.]@)'
     # main body starts with letter, then can be . - _
     r'([a-zA-Z0-9]{1}[a-zA-Z0-9\-\_\.]+)',
     re.VERBOSE | re.MULTILINE)
 
 
 def extract_mentioned_users(s):
     """
     Returns unique usernames from given string s that have @mention
 
     :param s: string to get mentions
     """
     usrs = set()
     for username in MENTIONS_REGEX.findall(s):
         usrs.add(username)
 
     return sorted(list(usrs), key=lambda k: k.lower())
 
 
 class AttributeDictBase(dict):
     def __getstate__(self):
         odict = self.__dict__  # get attribute dictionary
         return odict
 
     def __setstate__(self, dict):
         self.__dict__ = dict
 
     __setattr__ = dict.__setitem__
     __delattr__ = dict.__delitem__
 
 
 class StrictAttributeDict(AttributeDictBase):
     """
     Strict Version of Attribute dict which raises an Attribute error when
     requested attribute is not set
     """
     def __getattr__(self, attr):
         try:
             return self[attr]
         except KeyError:
             raise AttributeError('%s object has no attribute %s' % (
                 self.__class__, attr))
 
 
 class AttributeDict(AttributeDictBase):
     def __getattr__(self, attr):
         return self.get(attr, None)
 
 
 
 class OrderedDefaultDict(collections.OrderedDict, collections.defaultdict):
     def __init__(self, default_factory=None, *args, **kwargs):
         # in python3 you can omit the args to super
         super(OrderedDefaultDict, self).__init__(*args, **kwargs)
         self.default_factory = default_factory
 
 
 def fix_PATH(os_=None):
     """
     Get current active python path, and append it to PATH variable to fix
     issues of subprocess calls and different python versions
     """
     if os_ is None:
         import os
     else:
         os = os_
 
     cur_path = os.path.split(sys.executable)[0]
     if not os.environ['PATH'].startswith(cur_path):
         os.environ['PATH'] = '%s:%s' % (cur_path, os.environ['PATH'])
 
 
 def obfuscate_url_pw(engine):
     _url = engine or ''
     try:
         _url = sqlalchemy.engine.url.make_url(engine)
         if _url.password:
             _url.password = 'XXXXX'
     except Exception:
         pass
     return unicode(_url)
 
 
 def get_server_url(environ):
     req = webob.Request(environ)
     return req.host_url + req.script_name
 
 
 def unique_id(hexlen=32):
     alphabet = "23456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghjklmnpqrstuvwxyz"
     return suuid(truncate_to=hexlen, alphabet=alphabet)
 
 
 def suuid(url=None, truncate_to=22, alphabet=None):
     """
     Generate and return a short URL safe UUID.
 
     If the url parameter is provided, set the namespace to the provided
     URL and generate a UUID.
 
     :param url to get the uuid for
     :truncate_to: truncate the basic 22 UUID to shorter version
 
     The IDs won't be universally unique any longer, but the probability of
     a collision will still be very low.
     """
     # Define our alphabet.
     _ALPHABET = alphabet or "23456789ABCDEFGHJKLMNPQRSTUVWXYZ"
 
     # If no URL is given, generate a random UUID.
     if url is None:
         unique_id = uuid.uuid4().int
     else:
         unique_id = uuid.uuid3(uuid.NAMESPACE_URL, url).int
 
     alphabet_length = len(_ALPHABET)
     output = []
     while unique_id > 0:
         digit = unique_id % alphabet_length
         output.append(_ALPHABET[digit])
         unique_id = int(unique_id / alphabet_length)
     return "".join(output)[:truncate_to]
 
 
 def get_current_rhodecode_user(request=None):
     """
     Gets rhodecode user from request
     """
     pyramid_request = request or pyramid.threadlocal.get_current_request()
 
     # web case
     if pyramid_request and hasattr(pyramid_request, 'user'):
         return pyramid_request.user
 
     # api case
     if pyramid_request and hasattr(pyramid_request, 'rpc_user'):
         return pyramid_request.rpc_user
 
     return None
 
 
 def action_logger_generic(action, namespace=''):
     """
     A generic logger for actions useful to the system overview, tries to find
     an acting user for the context of the call otherwise reports unknown user
 
     :param action: logging message eg 'comment 5 deleted'
     :param type: string
 
     :param namespace: namespace of the logging message eg. 'repo.comments'
     :param type: string
 
     """
 
     logger_name = 'rhodecode.actions'
 
     if namespace:
         logger_name += '.' + namespace
 
     log = logging.getLogger(logger_name)
 
     # get a user if we can
     user = get_current_rhodecode_user()
 
     logfunc = log.info
 
     if not user:
         user = '<unknown user>'
         logfunc = log.warning
 
     logfunc('Logging action by {}: {}'.format(user, action))
 
 
 def escape_split(text, sep=',', maxsplit=-1):
     r"""
     Allows for escaping of the separator: e.g. arg='foo\, bar'
 
     It should be noted that the way bash et. al. do command line parsing, those
     single quotes are required.
     """
     escaped_sep = r'\%s' % sep
 
     if escaped_sep not in text:
         return text.split(sep, maxsplit)
 
     before, _mid, after = text.partition(escaped_sep)
     startlist = before.split(sep, maxsplit)  # a regular split is fine here
     unfinished = startlist[-1]
     startlist = startlist[:-1]
 
     # recurse because there may be more escaped separators
     endlist = escape_split(after, sep, maxsplit)
 
     # finish building the escaped value. we use endlist[0] becaue the first
     # part of the string sent in recursion is the rest of the escaped value.
     unfinished += sep + endlist[0]
 
     return startlist + [unfinished] + endlist[1:]  # put together all the parts
 
 
 class OptionalAttr(object):
     """
     Special Optional Option that defines other attribute. Example::
 
         def test(apiuser, userid=Optional(OAttr('apiuser')):
             user = Optional.extract(userid)
             # calls
 
     """
 
     def __init__(self, attr_name):
         self.attr_name = attr_name
 
     def __repr__(self):
         return '<OptionalAttr:%s>' % self.attr_name
 
     def __call__(self):
         return self
 
 
 # alias
 OAttr = OptionalAttr
 
 
 class Optional(object):
     """
     Defines an optional parameter::
 
         param = param.getval() if isinstance(param, Optional) else param
         param = param() if isinstance(param, Optional) else param
 
     is equivalent of::
 
         param = Optional.extract(param)
 
     """
 
     def __init__(self, type_):
         self.type_ = type_
 
     def __repr__(self):
         return '<Optional:%s>' % self.type_.__repr__()
 
     def __call__(self):
         return self.getval()
 
     def getval(self):
         """
         returns value from this Optional instance
         """
         if isinstance(self.type_, OAttr):
             # use params name
             return self.type_.attr_name
         return self.type_
 
     @classmethod
     def extract(cls, val):
         """
         Extracts value from Optional() instance
 
         :param val:
         :return: original value if it's not Optional instance else
             value of instance
         """
         if isinstance(val, cls):
             return val.getval()
         return val
 
 
 def glob2re(pat):
     """
     Translate a shell PATTERN to a regular expression.
 
     There is no way to quote meta-characters.
     """
 
     i, n = 0, len(pat)
     res = ''
     while i < n:
         c = pat[i]
         i = i+1
         if c == '*':
             #res = res + '.*'
             res = res + '[^/]*'
         elif c == '?':
             #res = res + '.'
             res = res + '[^/]'
         elif c == '[':
             j = i
             if j < n and pat[j] == '!':
                 j = j+1
             if j < n and pat[j] == ']':
                 j = j+1
             while j < n and pat[j] != ']':
                 j = j+1
             if j >= n:
                 res = res + '\\['
             else:
                 stuff = pat[i:j].replace('\\','\\\\')
                 i = j+1
                 if stuff[0] == '!':
                     stuff = '^' + stuff[1:]
                 elif stuff[0] == '^':
                     stuff = '\\' + stuff
                 res = '%s[%s]' % (res, stuff)
         else:
             res = res + re.escape(c)
     return res + '\Z(?ms)'
 
 
 def parse_byte_string(size_str):
     match = re.match(r'(\d+)(MB|KB)', size_str, re.IGNORECASE)
     if not match:
         raise ValueError('Given size:%s is invalid, please make sure '
                          'to use format of <num>(MB|KB)' % size_str)
 
     _parts = match.groups()
     num, type_ = _parts
     return long(num) * {'mb': 1024*1024, 'kb': 1024}[type_.lower()]
 
 
 class CachedProperty(object):
     """
     Lazy Attributes. With option to invalidate the cache by running a method
 
     >>> class Foo(object):
     ...
     ...    @CachedProperty
     ...    def heavy_func(self):
     ...        return 'super-calculation'
     ...
     ... foo = Foo()
     ... foo.heavy_func() # first computation
     ... foo.heavy_func() # fetch from cache
     ... foo._invalidate_prop_cache('heavy_func')
 
     # at this point calling foo.heavy_func() will be re-computed
     """
 
     def __init__(self, func, func_name=None):
 
         if func_name is None:
             func_name = func.__name__
         self.data = (func, func_name)
         update_wrapper(self, func)
 
     def __get__(self, inst, class_):
         if inst is None:
             return self
 
         func, func_name = self.data
         value = func(inst)
         inst.__dict__[func_name] = value
         if '_invalidate_prop_cache' not in inst.__dict__:
             inst.__dict__['_invalidate_prop_cache'] = partial(
                 self._invalidate_prop_cache, inst)
         return value
 
     def _invalidate_prop_cache(self, inst, name):
         inst.__dict__.pop(name, None)
 
 
 def retry(func=None, exception=Exception, n_tries=5, delay=5, backoff=1, logger=True):
     """
     Retry decorator with exponential backoff.
 
     Parameters
     ----------
     func : typing.Callable, optional
         Callable on which the decorator is applied, by default None
     exception : Exception or tuple of Exceptions, optional
         Exception(s) that invoke retry, by default Exception
     n_tries : int, optional
         Number of tries before giving up, by default 5
     delay : int, optional
         Initial delay between retries in seconds, by default 5
     backoff : int, optional
         Backoff multiplier e.g. value of 2 will double the delay, by default 1
     logger : bool, optional
         Option to log or print, by default False
 
     Returns
     -------
     typing.Callable
         Decorated callable that calls itself when exception(s) occur.
 
     Examples
     --------
     >>> import random
     >>> @retry(exception=Exception, n_tries=3)
     ... def test_random(text):
     ...    x = random.random()
     ...    if x < 0.5:
     ...        raise Exception("Fail")
     ...    else:
     ...        print("Success: ", text)
     >>> test_random("It works!")
     """
 
     if func is None:
         return partial(
             retry,
             exception=exception,
             n_tries=n_tries,
             delay=delay,
             backoff=backoff,
             logger=logger,
         )
 
     @wraps(func)
     def wrapper(*args, **kwargs):
         _n_tries, n_delay = n_tries, delay
         log = logging.getLogger('rhodecode.retry')
 
         while _n_tries > 1:
             try:
                 return func(*args, **kwargs)
             except exception as e:
                 e_details = repr(e)
                 msg = "Exception on calling func {func}: {e}, " \
                       "Retrying in {n_delay} seconds..."\
                     .format(func=func, e=e_details, n_delay=n_delay)
                 if logger:
                     log.warning(msg)
                 else:
                     print(msg)
                 time.sleep(n_delay)
                 _n_tries -= 1
                 n_delay *= backoff
 
         return func(*args, **kwargs)
 
     return wrapper
+
+
+def user_agent_normalizer(user_agent_raw):
+    log = logging.getLogger('rhodecode.user_agent_normalizer')
+    ua = (user_agent_raw or '').strip().lower()
+
+    try:
+        if 'mercurial/proto-1.0' in ua:
+            ua = ua.replace('mercurial/proto-1.0', '')
+            ua = ua.replace('(', '').replace(')', '').strip()
+            ua = ua.replace('mercurial ', 'mercurial/')
+        elif ua.startswith('git'):
+            pass
+    except Exception:
+        log.exception('Failed to parse scm user-agent')
+
+    return ua

rhodecode/model/pull_request.py

@@ -1,2380 +1,2381 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2012-2020 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 
 """
 pull request model for RhodeCode
 """
 
 
 import json
 import logging
 import os
 
 import datetime
 import urllib
 import collections
 
 from pyramid import compat
 from pyramid.threadlocal import get_current_request
 
 from rhodecode.lib.vcs.nodes import FileNode
 from rhodecode.translation import lazy_ugettext
 from rhodecode.lib import helpers as h, hooks_utils, diffs
 from rhodecode.lib import audit_logger
 from rhodecode.lib.compat import OrderedDict
 from rhodecode.lib.hooks_daemon import prepare_callback_daemon
 from rhodecode.lib.markup_renderer import (
     DEFAULT_COMMENTS_RENDERER, RstTemplateRenderer)
 from rhodecode.lib.utils2 import (
     safe_unicode, safe_str, md5_safe, AttributeDict, safe_int,
     get_current_rhodecode_user)
 from rhodecode.lib.vcs.backends.base import (
     Reference, MergeResponse, MergeFailureReason, UpdateFailureReason,
     TargetRefMissing, SourceRefMissing)
 from rhodecode.lib.vcs.conf import settings as vcs_settings
 from rhodecode.lib.vcs.exceptions import (
     CommitDoesNotExistError, EmptyRepositoryError)
 from rhodecode.model import BaseModel
 from rhodecode.model.changeset_status import ChangesetStatusModel
 from rhodecode.model.comment import CommentsModel
 from rhodecode.model.db import (
     aliased, null, lazyload, and_, or_, func, String, cast, PullRequest, PullRequestReviewers, ChangesetStatus,
     PullRequestVersion, ChangesetComment, Repository, RepoReviewRule, User)
 from rhodecode.model.meta import Session
 from rhodecode.model.notification import NotificationModel, \
     EmailNotificationModel
 from rhodecode.model.scm import ScmModel
 from rhodecode.model.settings import VcsSettingsModel
 
 
 log = logging.getLogger(__name__)
 
 
 # Data structure to hold the response data when updating commits during a pull
 # request update.
 class UpdateResponse(object):
 
     def __init__(self, executed, reason, new, old, common_ancestor_id,
                  commit_changes, source_changed, target_changed):
 
         self.executed = executed
         self.reason = reason
         self.new = new
         self.old = old
         self.common_ancestor_id = common_ancestor_id
         self.changes = commit_changes
         self.source_changed = source_changed
         self.target_changed = target_changed
 
 
 def get_diff_info(
         source_repo, source_ref, target_repo, target_ref, get_authors=False,
         get_commit_authors=True):
     """
     Calculates detailed diff information for usage in preview of creation of a pull-request.
     This is also used for default reviewers logic
     """
 
     source_scm = source_repo.scm_instance()
     target_scm = target_repo.scm_instance()
 
     ancestor_id = target_scm.get_common_ancestor(target_ref, source_ref, source_scm)
     if not ancestor_id:
         raise ValueError(
             'cannot calculate diff info without a common ancestor. '
             'Make sure both repositories are related, and have a common forking commit.')
 
     # case here is that want a simple diff without incoming commits,
     # previewing what will be merged based only on commits in the source.
     log.debug('Using ancestor %s as source_ref instead of %s',
               ancestor_id, source_ref)
 
     # source of changes now is the common ancestor
     source_commit = source_scm.get_commit(commit_id=ancestor_id)
     # target commit becomes the source ref as it is the last commit
     # for diff generation this logic gives proper diff
     target_commit = source_scm.get_commit(commit_id=source_ref)
 
     vcs_diff = \
         source_scm.get_diff(commit1=source_commit, commit2=target_commit,
                             ignore_whitespace=False, context=3)
 
     diff_processor = diffs.DiffProcessor(
         vcs_diff, format='newdiff', diff_limit=None,
         file_limit=None, show_full_diff=True)
 
     _parsed = diff_processor.prepare()
 
     all_files = []
     all_files_changes = []
     changed_lines = {}
     stats = [0, 0]
     for f in _parsed:
         all_files.append(f['filename'])
         all_files_changes.append({
             'filename': f['filename'],
             'stats': f['stats']
         })
         stats[0] += f['stats']['added']
         stats[1] += f['stats']['deleted']
 
         changed_lines[f['filename']] = []
         if len(f['chunks']) < 2:
             continue
         # first line is "context" information
         for chunks in f['chunks'][1:]:
             for chunk in chunks['lines']:
                 if chunk['action'] not in ('del', 'mod'):
                     continue
                 changed_lines[f['filename']].append(chunk['old_lineno'])
 
     commit_authors = []
     user_counts = {}
     email_counts = {}
     author_counts = {}
     _commit_cache = {}
 
     commits = []
     if get_commit_authors:
         log.debug('Obtaining commit authors from set of commits')
         _compare_data = target_scm.compare(
             target_ref, source_ref, source_scm, merge=True,
             pre_load=["author", "date", "message"]
         )
 
         for commit in _compare_data:
             # NOTE(marcink): we serialize here, so we don't produce more vcsserver calls on data returned
             # at this function which is later called via JSON serialization
             serialized_commit = dict(
                 author=commit.author,
                 date=commit.date,
                 message=commit.message,
                 commit_id=commit.raw_id,
                 raw_id=commit.raw_id
             )
             commits.append(serialized_commit)
             user = User.get_from_cs_author(serialized_commit['author'])
             if user and user not in commit_authors:
                 commit_authors.append(user)
 
     # lines
     if get_authors:
         log.debug('Calculating authors of changed files')
         target_commit = source_repo.get_commit(ancestor_id)
 
         for fname, lines in changed_lines.items():
 
             try:
                 node = target_commit.get_node(fname, pre_load=["is_binary"])
             except Exception:
                 log.exception("Failed to load node with path %s", fname)
                 continue
 
             if not isinstance(node, FileNode):
                 continue
 
             # NOTE(marcink): for binary node we don't do annotation, just use last author
             if node.is_binary:
                 author = node.last_commit.author
                 email = node.last_commit.author_email
 
                 user = User.get_from_cs_author(author)
                 if user:
                     user_counts[user.user_id] = user_counts.get(user.user_id, 0) + 1
                 author_counts[author] = author_counts.get(author, 0) + 1
                 email_counts[email] = email_counts.get(email, 0) + 1
 
                 continue
 
             for annotation in node.annotate:
                 line_no, commit_id, get_commit_func, line_text = annotation
                 if line_no in lines:
                     if commit_id not in _commit_cache:
                         _commit_cache[commit_id] = get_commit_func()
                     commit = _commit_cache[commit_id]
                     author = commit.author
                     email = commit.author_email
                     user = User.get_from_cs_author(author)
                     if user:
                         user_counts[user.user_id] = user_counts.get(user.user_id, 0) + 1
                     author_counts[author] = author_counts.get(author, 0) + 1
                     email_counts[email] = email_counts.get(email, 0) + 1
 
     log.debug('Default reviewers processing finished')
 
     return {
         'commits': commits,
         'files': all_files_changes,
         'stats': stats,
         'ancestor': ancestor_id,
         # original authors of modified files
         'original_authors': {
             'users': user_counts,
             'authors': author_counts,
             'emails': email_counts,
         },
         'commit_authors': commit_authors
     }
 
 
 class PullRequestModel(BaseModel):
 
     cls = PullRequest
 
     DIFF_CONTEXT = diffs.DEFAULT_CONTEXT
 
     UPDATE_STATUS_MESSAGES = {
         UpdateFailureReason.NONE: lazy_ugettext(
             'Pull request update successful.'),
         UpdateFailureReason.UNKNOWN: lazy_ugettext(
             'Pull request update failed because of an unknown error.'),
         UpdateFailureReason.NO_CHANGE: lazy_ugettext(
             'No update needed because the source and target have not changed.'),
         UpdateFailureReason.WRONG_REF_TYPE: lazy_ugettext(
             'Pull request cannot be updated because the reference type is '
             'not supported for an update. Only Branch, Tag or Bookmark is allowed.'),
         UpdateFailureReason.MISSING_TARGET_REF: lazy_ugettext(
             'This pull request cannot be updated because the target '
             'reference is missing.'),
         UpdateFailureReason.MISSING_SOURCE_REF: lazy_ugettext(
             'This pull request cannot be updated because the source '
             'reference is missing.'),
     }
     REF_TYPES = ['bookmark', 'book', 'tag', 'branch']
     UPDATABLE_REF_TYPES = ['bookmark', 'book', 'branch']
 
     def __get_pull_request(self, pull_request):
         return self._get_instance((
             PullRequest, PullRequestVersion), pull_request)
 
     def _check_perms(self, perms, pull_request, user, api=False):
         if not api:
             return h.HasRepoPermissionAny(*perms)(
                 user=user, repo_name=pull_request.target_repo.repo_name)
         else:
             return h.HasRepoPermissionAnyApi(*perms)(
                 user=user, repo_name=pull_request.target_repo.repo_name)
 
     def check_user_read(self, pull_request, user, api=False):
         _perms = ('repository.admin', 'repository.write', 'repository.read',)
         return self._check_perms(_perms, pull_request, user, api)
 
     def check_user_merge(self, pull_request, user, api=False):
         _perms = ('repository.admin', 'repository.write', 'hg.admin',)
         return self._check_perms(_perms, pull_request, user, api)
 
     def check_user_update(self, pull_request, user, api=False):
         owner = user.user_id == pull_request.user_id
         return self.check_user_merge(pull_request, user, api) or owner
 
     def check_user_delete(self, pull_request, user):
         owner = user.user_id == pull_request.user_id
         _perms = ('repository.admin',)
         return self._check_perms(_perms, pull_request, user) or owner
 
     def is_user_reviewer(self, pull_request, user):
         return user.user_id in [
             x.user_id for x in
             pull_request.get_pull_request_reviewers(PullRequestReviewers.ROLE_REVIEWER)
             if x.user
         ]
 
     def check_user_change_status(self, pull_request, user, api=False):
         return self.check_user_update(pull_request, user, api) \
                or self.is_user_reviewer(pull_request, user)
 
     def check_user_comment(self, pull_request, user):
         owner = user.user_id == pull_request.user_id
         return self.check_user_read(pull_request, user) or owner
 
     def get(self, pull_request):
         return self.__get_pull_request(pull_request)
 
     def _prepare_get_all_query(self, repo_name, search_q=None, source=False,
                                statuses=None, opened_by=None, order_by=None,
                                order_dir='desc', only_created=False):
         repo = None
         if repo_name:
             repo = self._get_repo(repo_name)
 
         q = PullRequest.query()
 
         if search_q:
             like_expression = u'%{}%'.format(safe_unicode(search_q))
             q = q.join(User, User.user_id == PullRequest.user_id)
             q = q.filter(or_(
                 cast(PullRequest.pull_request_id, String).ilike(like_expression),
                 User.username.ilike(like_expression),
                 PullRequest.title.ilike(like_expression),
                 PullRequest.description.ilike(like_expression),
             ))
 
         # source or target
         if repo and source:
             q = q.filter(PullRequest.source_repo == repo)
         elif repo:
             q = q.filter(PullRequest.target_repo == repo)
 
         # closed,opened
         if statuses:
             q = q.filter(PullRequest.status.in_(statuses))
 
         # opened by filter
         if opened_by:
             q = q.filter(PullRequest.user_id.in_(opened_by))
 
         # only get those that are in "created" state
         if only_created:
             q = q.filter(PullRequest.pull_request_state == PullRequest.STATE_CREATED)
 
         order_map = {
             'name_raw': PullRequest.pull_request_id,
             'id': PullRequest.pull_request_id,
             'title': PullRequest.title,
             'updated_on_raw': PullRequest.updated_on,
             'target_repo': PullRequest.target_repo_id
         }
         if order_by and order_by in order_map:
             if order_dir == 'asc':
                 q = q.order_by(order_map[order_by].asc())
             else:
                 q = q.order_by(order_map[order_by].desc())
 
         return q
 
     def count_all(self, repo_name, search_q=None, source=False, statuses=None,
                   opened_by=None):
         """
         Count the number of pull requests for a specific repository.
 
         :param repo_name: target or source repo
         :param search_q: filter by text
         :param source: boolean flag to specify if repo_name refers to source
         :param statuses: list of pull request statuses
         :param opened_by: author user of the pull request
         :returns: int number of pull requests
         """
         q = self._prepare_get_all_query(
             repo_name, search_q=search_q, source=source, statuses=statuses,
             opened_by=opened_by)
 
         return q.count()
 
     def get_all(self, repo_name, search_q=None, source=False, statuses=None,
                 opened_by=None, offset=0, length=None, order_by=None, order_dir='desc'):
         """
         Get all pull requests for a specific repository.
 
         :param repo_name: target or source repo
         :param search_q: filter by text
         :param source: boolean flag to specify if repo_name refers to source
         :param statuses: list of pull request statuses
         :param opened_by: author user of the pull request
         :param offset: pagination offset
         :param length: length of returned list
         :param order_by: order of the returned list
         :param order_dir: 'asc' or 'desc' ordering direction
         :returns: list of pull requests
         """
         q = self._prepare_get_all_query(
             repo_name, search_q=search_q, source=source, statuses=statuses,
             opened_by=opened_by, order_by=order_by, order_dir=order_dir)
 
         if length:
             pull_requests = q.limit(length).offset(offset).all()
         else:
             pull_requests = q.all()
 
         return pull_requests
 
     def count_awaiting_review(self, repo_name, search_q=None, statuses=None):
         """
         Count the number of pull requests for a specific repository that are
         awaiting review.
 
         :param repo_name: target or source repo
         :param search_q: filter by text
         :param statuses: list of pull request statuses
         :returns: int number of pull requests
         """
         pull_requests = self.get_awaiting_review(
             repo_name, search_q=search_q, statuses=statuses)
 
         return len(pull_requests)
 
     def get_awaiting_review(self, repo_name, search_q=None, statuses=None,
                             offset=0, length=None, order_by=None, order_dir='desc'):
         """
         Get all pull requests for a specific repository that are awaiting
         review.
 
         :param repo_name: target or source repo
         :param search_q: filter by text
         :param statuses: list of pull request statuses
         :param offset: pagination offset
         :param length: length of returned list
         :param order_by: order of the returned list
         :param order_dir: 'asc' or 'desc' ordering direction
         :returns: list of pull requests
         """
         pull_requests = self.get_all(
             repo_name, search_q=search_q, statuses=statuses,
             order_by=order_by, order_dir=order_dir)
 
         _filtered_pull_requests = []
         for pr in pull_requests:
             status = pr.calculated_review_status()
             if status in [ChangesetStatus.STATUS_NOT_REVIEWED,
                           ChangesetStatus.STATUS_UNDER_REVIEW]:
                 _filtered_pull_requests.append(pr)
         if length:
             return _filtered_pull_requests[offset:offset+length]
         else:
             return _filtered_pull_requests
 
     def _prepare_awaiting_my_review_review_query(
             self, repo_name, user_id, search_q=None, statuses=None,
             order_by=None, order_dir='desc'):
 
         for_review_statuses = [
             ChangesetStatus.STATUS_UNDER_REVIEW, ChangesetStatus.STATUS_NOT_REVIEWED
         ]
 
         pull_request_alias = aliased(PullRequest)
         status_alias = aliased(ChangesetStatus)
         reviewers_alias = aliased(PullRequestReviewers)
         repo_alias = aliased(Repository)
 
         last_ver_subq = Session()\
             .query(func.min(ChangesetStatus.version)) \
             .filter(ChangesetStatus.pull_request_id == reviewers_alias.pull_request_id)\
             .filter(ChangesetStatus.user_id == reviewers_alias.user_id) \
             .subquery()
 
         q = Session().query(pull_request_alias) \
             .options(lazyload(pull_request_alias.author)) \
             .join(reviewers_alias,
                   reviewers_alias.pull_request_id == pull_request_alias.pull_request_id) \
             .join(repo_alias,
                   repo_alias.repo_id == pull_request_alias.target_repo_id) \
             .outerjoin(status_alias,
                        and_(status_alias.user_id == reviewers_alias.user_id,
                             status_alias.pull_request_id == reviewers_alias.pull_request_id)) \
             .filter(or_(status_alias.version == null(),
                         status_alias.version == last_ver_subq)) \
             .filter(reviewers_alias.user_id == user_id) \
             .filter(repo_alias.repo_name == repo_name) \
             .filter(or_(status_alias.status == null(), status_alias.status.in_(for_review_statuses))) \
             .group_by(pull_request_alias)
 
         # closed,opened
         if statuses:
             q = q.filter(pull_request_alias.status.in_(statuses))
 
         if search_q:
             like_expression = u'%{}%'.format(safe_unicode(search_q))
             q = q.join(User, User.user_id == pull_request_alias.user_id)
             q = q.filter(or_(
                 cast(pull_request_alias.pull_request_id, String).ilike(like_expression),
                 User.username.ilike(like_expression),
                 pull_request_alias.title.ilike(like_expression),
                 pull_request_alias.description.ilike(like_expression),
             ))
 
         order_map = {
             'name_raw': pull_request_alias.pull_request_id,
             'title': pull_request_alias.title,
             'updated_on_raw': pull_request_alias.updated_on,
             'target_repo': pull_request_alias.target_repo_id
         }
         if order_by and order_by in order_map:
             if order_dir == 'asc':
                 q = q.order_by(order_map[order_by].asc())
             else:
                 q = q.order_by(order_map[order_by].desc())
 
         return q
 
     def count_awaiting_my_review(self, repo_name, user_id, search_q=None, statuses=None):
         """
         Count the number of pull requests for a specific repository that are
         awaiting review from a specific user.
 
         :param repo_name: target or source repo
         :param user_id: reviewer user of the pull request
         :param search_q: filter by text
         :param statuses: list of pull request statuses
         :returns: int number of pull requests
         """
         q = self._prepare_awaiting_my_review_review_query(
             repo_name, user_id, search_q=search_q, statuses=statuses)
         return q.count()
 
     def get_awaiting_my_review(self, repo_name, user_id, search_q=None, statuses=None,
                                offset=0, length=None, order_by=None, order_dir='desc'):
         """
         Get all pull requests for a specific repository that are awaiting
         review from a specific user.
 
         :param repo_name: target or source repo
         :param user_id: reviewer user of the pull request
         :param search_q: filter by text
         :param statuses: list of pull request statuses
         :param offset: pagination offset
         :param length: length of returned list
         :param order_by: order of the returned list
         :param order_dir: 'asc' or 'desc' ordering direction
         :returns: list of pull requests
         """
 
         q = self._prepare_awaiting_my_review_review_query(
             repo_name, user_id, search_q=search_q, statuses=statuses,
             order_by=order_by, order_dir=order_dir)
 
         if length:
             pull_requests = q.limit(length).offset(offset).all()
         else:
             pull_requests = q.all()
 
         return pull_requests
 
     def _prepare_im_participating_query(self, user_id=None, statuses=None, query='',
                                         order_by=None, order_dir='desc'):
         """
         return a query of pull-requests user is an creator, or he's added as a reviewer
         """
         q = PullRequest.query()
         if user_id:
             reviewers_subquery = Session().query(
                 PullRequestReviewers.pull_request_id).filter(
                 PullRequestReviewers.user_id == user_id).subquery()
             user_filter = or_(
                 PullRequest.user_id == user_id,
                 PullRequest.pull_request_id.in_(reviewers_subquery)
             )
             q = PullRequest.query().filter(user_filter)
 
         # closed,opened
         if statuses:
             q = q.filter(PullRequest.status.in_(statuses))
 
         if query:
             like_expression = u'%{}%'.format(safe_unicode(query))
             q = q.join(User, User.user_id == PullRequest.user_id)
             q = q.filter(or_(
                 cast(PullRequest.pull_request_id, String).ilike(like_expression),
                 User.username.ilike(like_expression),
                 PullRequest.title.ilike(like_expression),
                 PullRequest.description.ilike(like_expression),
             ))
 
         order_map = {
             'name_raw': PullRequest.pull_request_id,
             'title': PullRequest.title,
             'updated_on_raw': PullRequest.updated_on,
             'target_repo': PullRequest.target_repo_id
         }
         if order_by and order_by in order_map:
             if order_dir == 'asc':
                 q = q.order_by(order_map[order_by].asc())
             else:
                 q = q.order_by(order_map[order_by].desc())
 
         return q
 
     def count_im_participating_in(self, user_id=None, statuses=None, query=''):
         q = self._prepare_im_participating_query(user_id, statuses=statuses, query=query)
         return q.count()
 
     def get_im_participating_in(
             self, user_id=None, statuses=None, query='', offset=0,
             length=None, order_by=None, order_dir='desc'):
         """
         Get all Pull requests that i'm participating in as a reviewer, or i have opened
         """
 
         q = self._prepare_im_participating_query(
             user_id, statuses=statuses, query=query, order_by=order_by,
             order_dir=order_dir)
 
         if length:
             pull_requests = q.limit(length).offset(offset).all()
         else:
             pull_requests = q.all()
 
         return pull_requests
 
     def _prepare_participating_in_for_review_query(
             self, user_id, statuses=None, query='', order_by=None, order_dir='desc'):
 
         for_review_statuses = [
             ChangesetStatus.STATUS_UNDER_REVIEW, ChangesetStatus.STATUS_NOT_REVIEWED
         ]
 
         pull_request_alias = aliased(PullRequest)
         status_alias = aliased(ChangesetStatus)
         reviewers_alias = aliased(PullRequestReviewers)
 
         last_ver_subq = Session()\
             .query(func.min(ChangesetStatus.version)) \
             .filter(ChangesetStatus.pull_request_id == reviewers_alias.pull_request_id)\
             .filter(ChangesetStatus.user_id == reviewers_alias.user_id) \
             .subquery()
 
         q = Session().query(pull_request_alias) \
             .options(lazyload(pull_request_alias.author)) \
             .join(reviewers_alias,
                   reviewers_alias.pull_request_id == pull_request_alias.pull_request_id) \
             .outerjoin(status_alias,
                        and_(status_alias.user_id == reviewers_alias.user_id,
                             status_alias.pull_request_id == reviewers_alias.pull_request_id)) \
             .filter(or_(status_alias.version == null(),
                         status_alias.version == last_ver_subq)) \
             .filter(reviewers_alias.user_id == user_id) \
             .filter(or_(status_alias.status == null(), status_alias.status.in_(for_review_statuses))) \
             .group_by(pull_request_alias)
 
         # closed,opened
         if statuses:
             q = q.filter(pull_request_alias.status.in_(statuses))
 
         if query:
             like_expression = u'%{}%'.format(safe_unicode(query))
             q = q.join(User, User.user_id == pull_request_alias.user_id)
             q = q.filter(or_(
                 cast(pull_request_alias.pull_request_id, String).ilike(like_expression),
                 User.username.ilike(like_expression),
                 pull_request_alias.title.ilike(like_expression),
                 pull_request_alias.description.ilike(like_expression),
             ))
 
         order_map = {
             'name_raw': pull_request_alias.pull_request_id,
             'title': pull_request_alias.title,
             'updated_on_raw': pull_request_alias.updated_on,
             'target_repo': pull_request_alias.target_repo_id
         }
         if order_by and order_by in order_map:
             if order_dir == 'asc':
                 q = q.order_by(order_map[order_by].asc())
             else:
                 q = q.order_by(order_map[order_by].desc())
 
         return q
 
     def count_im_participating_in_for_review(self, user_id, statuses=None, query=''):
         q = self._prepare_participating_in_for_review_query(user_id, statuses=statuses, query=query)
         return q.count()
 
     def get_im_participating_in_for_review(
             self, user_id, statuses=None, query='', offset=0,
             length=None, order_by=None, order_dir='desc'):
         """
         Get all Pull requests that needs user approval or rejection
         """
 
         q = self._prepare_participating_in_for_review_query(
             user_id, statuses=statuses, query=query, order_by=order_by,
             order_dir=order_dir)
 
         if length:
             pull_requests = q.limit(length).offset(offset).all()
         else:
             pull_requests = q.all()
 
         return pull_requests
 
     def get_versions(self, pull_request):
         """
         returns version of pull request sorted by ID descending
         """
         return PullRequestVersion.query()\
             .filter(PullRequestVersion.pull_request == pull_request)\
             .order_by(PullRequestVersion.pull_request_version_id.asc())\
             .all()
 
     def get_pr_version(self, pull_request_id, version=None):
         at_version = None
 
         if version and version == 'latest':
             pull_request_ver = PullRequest.get(pull_request_id)
             pull_request_obj = pull_request_ver
             _org_pull_request_obj = pull_request_obj
             at_version = 'latest'
         elif version:
             pull_request_ver = PullRequestVersion.get_or_404(version)
             pull_request_obj = pull_request_ver
             _org_pull_request_obj = pull_request_ver.pull_request
             at_version = pull_request_ver.pull_request_version_id
         else:
             _org_pull_request_obj = pull_request_obj = PullRequest.get_or_404(
                 pull_request_id)
 
         pull_request_display_obj = PullRequest.get_pr_display_object(
             pull_request_obj, _org_pull_request_obj)
 
         return _org_pull_request_obj, pull_request_obj, \
                pull_request_display_obj, at_version
 
     def pr_commits_versions(self, versions):
         """
         Maps the pull-request commits into all known PR versions. This way we can obtain
         each pr version the commit was introduced in.
         """
         commit_versions = collections.defaultdict(list)
         num_versions = [x.pull_request_version_id for x in versions]
         for ver in versions:
             for commit_id in ver.revisions:
                 ver_idx = ChangesetComment.get_index_from_version(
                     ver.pull_request_version_id, num_versions=num_versions)
                 commit_versions[commit_id].append(ver_idx)
         return commit_versions
 
     def create(self, created_by, source_repo, source_ref, target_repo,
                target_ref, revisions, reviewers, observers, title, description=None,
                common_ancestor_id=None,
                description_renderer=None,
                reviewer_data=None, translator=None, auth_user=None):
         translator = translator or get_current_request().translate
 
         created_by_user = self._get_user(created_by)
         auth_user = auth_user or created_by_user.AuthUser()
         source_repo = self._get_repo(source_repo)
         target_repo = self._get_repo(target_repo)
 
         pull_request = PullRequest()
         pull_request.source_repo = source_repo
         pull_request.source_ref = source_ref
         pull_request.target_repo = target_repo
         pull_request.target_ref = target_ref
         pull_request.revisions = revisions
         pull_request.title = title
         pull_request.description = description
         pull_request.description_renderer = description_renderer
         pull_request.author = created_by_user
         pull_request.reviewer_data = reviewer_data
         pull_request.pull_request_state = pull_request.STATE_CREATING
         pull_request.common_ancestor_id = common_ancestor_id
 
         Session().add(pull_request)
         Session().flush()
 
         reviewer_ids = set()
         # members / reviewers
         for reviewer_object in reviewers:
             user_id, reasons, mandatory, role, rules = reviewer_object
             user = self._get_user(user_id)
 
             # skip duplicates
             if user.user_id in reviewer_ids:
                 continue
 
             reviewer_ids.add(user.user_id)
 
             reviewer = PullRequestReviewers()
             reviewer.user = user
             reviewer.pull_request = pull_request
             reviewer.reasons = reasons
             reviewer.mandatory = mandatory
             reviewer.role = role
 
             # NOTE(marcink): pick only first rule for now
             rule_id = list(rules)[0] if rules else None
             rule = RepoReviewRule.get(rule_id) if rule_id else None
             if rule:
                 review_group = rule.user_group_vote_rule(user_id)
                 # we check if this particular reviewer is member of a voting group
                 if review_group:
                     # NOTE(marcink):
                     # can be that user is member of more but we pick the first same,
                     # same as default reviewers algo
                     review_group = review_group[0]
 
                     rule_data = {
                         'rule_name':
                             rule.review_rule_name,
                         'rule_user_group_entry_id':
                             review_group.repo_review_rule_users_group_id,
                         'rule_user_group_name':
                             review_group.users_group.users_group_name,
                         'rule_user_group_members':
                             [x.user.username for x in review_group.users_group.members],
                         'rule_user_group_members_id':
                             [x.user.user_id for x in review_group.users_group.members],
                     }
                     # e.g {'vote_rule': -1, 'mandatory': True}
                     rule_data.update(review_group.rule_data())
 
                     reviewer.rule_data = rule_data
 
             Session().add(reviewer)
             Session().flush()
 
         for observer_object in observers:
             user_id, reasons, mandatory, role, rules = observer_object
             user = self._get_user(user_id)
 
             # skip duplicates from reviewers
             if user.user_id in reviewer_ids:
                 continue
 
             #reviewer_ids.add(user.user_id)
 
             observer = PullRequestReviewers()
             observer.user = user
             observer.pull_request = pull_request
             observer.reasons = reasons
             observer.mandatory = mandatory
             observer.role = role
 
             # NOTE(marcink): pick only first rule for now
             rule_id = list(rules)[0] if rules else None
             rule = RepoReviewRule.get(rule_id) if rule_id else None
             if rule:
                 # TODO(marcink): do we need this for observers ??
                 pass
 
             Session().add(observer)
             Session().flush()
 
         # Set approval status to "Under Review" for all commits which are
         # part of this pull request.
         ChangesetStatusModel().set_status(
             repo=target_repo,
             status=ChangesetStatus.STATUS_UNDER_REVIEW,
             user=created_by_user,
             pull_request=pull_request
         )
         # we commit early at this point. This has to do with a fact
         # that before queries do some row-locking. And because of that
         # we need to commit and finish transaction before below validate call
         # that for large repos could be long resulting in long row locks
         Session().commit()
 
         # prepare workspace, and run initial merge simulation. Set state during that
         # operation
         pull_request = PullRequest.get(pull_request.pull_request_id)
 
         # set as merging, for merge simulation, and if finished to created so we mark
         # simulation is working fine
         with pull_request.set_state(PullRequest.STATE_MERGING,
                                     final_state=PullRequest.STATE_CREATED) as state_obj:
             MergeCheck.validate(
                 pull_request, auth_user=auth_user, translator=translator)
 
         self.notify_reviewers(pull_request, reviewer_ids, created_by_user)
         self.trigger_pull_request_hook(pull_request, created_by_user, 'create')
 
         creation_data = pull_request.get_api_data(with_merge_state=False)
         self._log_audit_action(
             'repo.pull_request.create', {'data': creation_data},
             auth_user, pull_request)
 
         return pull_request
 
     def trigger_pull_request_hook(self, pull_request, user, action, data=None):
         pull_request = self.__get_pull_request(pull_request)
         target_scm = pull_request.target_repo.scm_instance()
         if action == 'create':
             trigger_hook = hooks_utils.trigger_create_pull_request_hook
         elif action == 'merge':
             trigger_hook = hooks_utils.trigger_merge_pull_request_hook
         elif action == 'close':
             trigger_hook = hooks_utils.trigger_close_pull_request_hook
         elif action == 'review_status_change':
             trigger_hook = hooks_utils.trigger_review_pull_request_hook
         elif action == 'update':
             trigger_hook = hooks_utils.trigger_update_pull_request_hook
         elif action == 'comment':
             trigger_hook = hooks_utils.trigger_comment_pull_request_hook
         elif action == 'comment_edit':
             trigger_hook = hooks_utils.trigger_comment_pull_request_edit_hook
         else:
             return
 
         log.debug('Handling pull_request %s trigger_pull_request_hook with action %s and hook: %s',
                   pull_request, action, trigger_hook)
         trigger_hook(
             username=user.username,
             repo_name=pull_request.target_repo.repo_name,
             repo_type=target_scm.alias,
             pull_request=pull_request,
             data=data)
 
     def _get_commit_ids(self, pull_request):
         """
         Return the commit ids of the merged pull request.
 
         This method is not dealing correctly yet with the lack of autoupdates
         nor with the implicit target updates.
         For example: if a commit in the source repo is already in the target it
         will be reported anyways.
         """
         merge_rev = pull_request.merge_rev
         if merge_rev is None:
             raise ValueError('This pull request was not merged yet')
 
         commit_ids = list(pull_request.revisions)
         if merge_rev not in commit_ids:
             commit_ids.append(merge_rev)
 
         return commit_ids
 
     def merge_repo(self, pull_request, user, extras):
+        repo_type = pull_request.source_repo.repo_type
         log.debug("Merging pull request %s", pull_request.pull_request_id)
-        extras['user_agent'] = 'internal-merge'
+        extras['user_agent'] = '{}/internal-merge'.format(repo_type)
         merge_state = self._merge_pull_request(pull_request, user, extras)
         if merge_state.executed:
             log.debug("Merge was successful, updating the pull request comments.")
             self._comment_and_close_pr(pull_request, user, merge_state)
 
             self._log_audit_action(
                 'repo.pull_request.merge',
                 {'merge_state': merge_state.__dict__},
                 user, pull_request)
 
         else:
             log.warn("Merge failed, not updating the pull request.")
         return merge_state
 
     def _merge_pull_request(self, pull_request, user, extras, merge_msg=None):
         target_vcs = pull_request.target_repo.scm_instance()
         source_vcs = pull_request.source_repo.scm_instance()
 
         message = safe_unicode(merge_msg or vcs_settings.MERGE_MESSAGE_TMPL).format(
             pr_id=pull_request.pull_request_id,
             pr_title=pull_request.title,
             pr_desc=pull_request.description,
             source_repo=source_vcs.name,
             source_ref_name=pull_request.source_ref_parts.name,
             target_repo=target_vcs.name,
             target_ref_name=pull_request.target_ref_parts.name,
         )
 
         workspace_id = self._workspace_id(pull_request)
         repo_id = pull_request.target_repo.repo_id
         use_rebase = self._use_rebase_for_merging(pull_request)
         close_branch = self._close_branch_before_merging(pull_request)
         user_name = self._user_name_for_merging(pull_request, user)
 
         target_ref = self._refresh_reference(
             pull_request.target_ref_parts, target_vcs)
 
         callback_daemon, extras = prepare_callback_daemon(
             extras, protocol=vcs_settings.HOOKS_PROTOCOL,
             host=vcs_settings.HOOKS_HOST,
             use_direct_calls=vcs_settings.HOOKS_DIRECT_CALLS)
 
         with callback_daemon:
             # TODO: johbo: Implement a clean way to run a config_override
             # for a single call.
             target_vcs.config.set(
                 'rhodecode', 'RC_SCM_DATA', json.dumps(extras))
 
             merge_state = target_vcs.merge(
                 repo_id, workspace_id, target_ref, source_vcs,
                 pull_request.source_ref_parts,
                 user_name=user_name, user_email=user.email,
                 message=message, use_rebase=use_rebase,
                 close_branch=close_branch)
         return merge_state
 
     def _comment_and_close_pr(self, pull_request, user, merge_state, close_msg=None):
         pull_request.merge_rev = merge_state.merge_ref.commit_id
         pull_request.updated_on = datetime.datetime.now()
         close_msg = close_msg or 'Pull request merged and closed'
 
         CommentsModel().create(
             text=safe_unicode(close_msg),
             repo=pull_request.target_repo.repo_id,
             user=user.user_id,
             pull_request=pull_request.pull_request_id,
             f_path=None,
             line_no=None,
             closing_pr=True
         )
 
         Session().add(pull_request)
         Session().flush()
         # TODO: paris: replace invalidation with less radical solution
         ScmModel().mark_for_invalidation(
             pull_request.target_repo.repo_name)
         self.trigger_pull_request_hook(pull_request, user, 'merge')
 
     def has_valid_update_type(self, pull_request):
         source_ref_type = pull_request.source_ref_parts.type
         return source_ref_type in self.REF_TYPES
 
     def get_flow_commits(self, pull_request):
 
         # source repo
         source_ref_name = pull_request.source_ref_parts.name
         source_ref_type = pull_request.source_ref_parts.type
         source_ref_id = pull_request.source_ref_parts.commit_id
         source_repo = pull_request.source_repo.scm_instance()
 
         try:
             if source_ref_type in self.REF_TYPES:
                 source_commit = source_repo.get_commit(
                     source_ref_name, reference_obj=pull_request.source_ref_parts)
             else:
                 source_commit = source_repo.get_commit(source_ref_id)
         except CommitDoesNotExistError:
             raise SourceRefMissing()
 
         # target repo
         target_ref_name = pull_request.target_ref_parts.name
         target_ref_type = pull_request.target_ref_parts.type
         target_ref_id = pull_request.target_ref_parts.commit_id
         target_repo = pull_request.target_repo.scm_instance()
 
         try:
             if target_ref_type in self.REF_TYPES:
                 target_commit = target_repo.get_commit(
                     target_ref_name, reference_obj=pull_request.target_ref_parts)
             else:
                 target_commit = target_repo.get_commit(target_ref_id)
         except CommitDoesNotExistError:
             raise TargetRefMissing()
 
         return source_commit, target_commit
 
     def update_commits(self, pull_request, updating_user):
         """
         Get the updated list of commits for the pull request
         and return the new pull request version and the list
         of commits processed by this update action
 
         updating_user is the user_object who triggered the update
         """
         pull_request = self.__get_pull_request(pull_request)
         source_ref_type = pull_request.source_ref_parts.type
         source_ref_name = pull_request.source_ref_parts.name
         source_ref_id = pull_request.source_ref_parts.commit_id
 
         target_ref_type = pull_request.target_ref_parts.type
         target_ref_name = pull_request.target_ref_parts.name
         target_ref_id = pull_request.target_ref_parts.commit_id
 
         if not self.has_valid_update_type(pull_request):
             log.debug("Skipping update of pull request %s due to ref type: %s",
                       pull_request, source_ref_type)
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.WRONG_REF_TYPE,
                 old=pull_request, new=None, common_ancestor_id=None, commit_changes=None,
                 source_changed=False, target_changed=False)
 
         try:
             source_commit, target_commit = self.get_flow_commits(pull_request)
         except SourceRefMissing:
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.MISSING_SOURCE_REF,
                 old=pull_request, new=None, common_ancestor_id=None, commit_changes=None,
                 source_changed=False, target_changed=False)
         except TargetRefMissing:
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.MISSING_TARGET_REF,
                 old=pull_request, new=None, common_ancestor_id=None, commit_changes=None,
                 source_changed=False, target_changed=False)
 
         source_changed = source_ref_id != source_commit.raw_id
         target_changed = target_ref_id != target_commit.raw_id
 
         if not (source_changed or target_changed):
             log.debug("Nothing changed in pull request %s", pull_request)
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.NO_CHANGE,
                 old=pull_request, new=None, common_ancestor_id=None, commit_changes=None,
                 source_changed=target_changed, target_changed=source_changed)
 
         change_in_found = 'target repo' if target_changed else 'source repo'
         log.debug('Updating pull request because of change in %s detected',
                   change_in_found)
 
         # Finally there is a need for an update, in case of source change
         # we create a new version, else just an update
         if source_changed:
             pull_request_version = self._create_version_from_snapshot(pull_request)
             self._link_comments_to_version(pull_request_version)
         else:
             try:
                 ver = pull_request.versions[-1]
             except IndexError:
                 ver = None
 
             pull_request.pull_request_version_id = \
                 ver.pull_request_version_id if ver else None
             pull_request_version = pull_request
 
         source_repo = pull_request.source_repo.scm_instance()
         target_repo = pull_request.target_repo.scm_instance()
 
         # re-compute commit ids
         old_commit_ids = pull_request.revisions
         pre_load = ["author", "date", "message", "branch"]
         commit_ranges = target_repo.compare(
             target_commit.raw_id, source_commit.raw_id, source_repo, merge=True,
             pre_load=pre_load)
 
         target_ref = target_commit.raw_id
         source_ref = source_commit.raw_id
         ancestor_commit_id = target_repo.get_common_ancestor(
             target_ref, source_ref, source_repo)
 
         if not ancestor_commit_id:
             raise ValueError(
                 'cannot calculate diff info without a common ancestor. '
                 'Make sure both repositories are related, and have a common forking commit.')
 
         pull_request.common_ancestor_id = ancestor_commit_id
 
         pull_request.source_ref = '%s:%s:%s' % (
             source_ref_type, source_ref_name, source_commit.raw_id)
         pull_request.target_ref = '%s:%s:%s' % (
             target_ref_type, target_ref_name, ancestor_commit_id)
 
         pull_request.revisions = [
             commit.raw_id for commit in reversed(commit_ranges)]
         pull_request.updated_on = datetime.datetime.now()
         Session().add(pull_request)
         new_commit_ids = pull_request.revisions
 
         old_diff_data, new_diff_data = self._generate_update_diffs(
             pull_request, pull_request_version)
 
         # calculate commit and file changes
         commit_changes = self._calculate_commit_id_changes(
             old_commit_ids, new_commit_ids)
         file_changes = self._calculate_file_changes(
             old_diff_data, new_diff_data)
 
         # set comments as outdated if DIFFS changed
         CommentsModel().outdate_comments(
             pull_request, old_diff_data=old_diff_data,
             new_diff_data=new_diff_data)
 
         valid_commit_changes = (commit_changes.added or commit_changes.removed)
         file_node_changes = (
             file_changes.added or file_changes.modified or file_changes.removed)
         pr_has_changes = valid_commit_changes or file_node_changes
 
         # Add an automatic comment to the pull request, in case
         # anything has changed
         if pr_has_changes:
             update_comment = CommentsModel().create(
                 text=self._render_update_message(ancestor_commit_id, commit_changes, file_changes),
                 repo=pull_request.target_repo,
                 user=pull_request.author,
                 pull_request=pull_request,
                 send_email=False, renderer=DEFAULT_COMMENTS_RENDERER)
 
             # Update status to "Under Review" for added commits
             for commit_id in commit_changes.added:
                 ChangesetStatusModel().set_status(
                     repo=pull_request.source_repo,
                     status=ChangesetStatus.STATUS_UNDER_REVIEW,
                     comment=update_comment,
                     user=pull_request.author,
                     pull_request=pull_request,
                     revision=commit_id)
 
         # initial commit
         Session().commit()
 
         if pr_has_changes:
             # send update email to users
             try:
                 self.notify_users(pull_request=pull_request, updating_user=updating_user,
                                   ancestor_commit_id=ancestor_commit_id,
                                   commit_changes=commit_changes,
                                   file_changes=file_changes)
                 Session().commit()
             except Exception:
                 log.exception('Failed to send email notification to users')
                 Session().rollback()
 
         log.debug(
             'Updated pull request %s, added_ids: %s, common_ids: %s, '
             'removed_ids: %s', pull_request.pull_request_id,
             commit_changes.added, commit_changes.common, commit_changes.removed)
         log.debug(
             'Updated pull request with the following file changes: %s',
             file_changes)
 
         log.info(
             "Updated pull request %s from commit %s to commit %s, "
             "stored new version %s of this pull request.",
             pull_request.pull_request_id, source_ref_id,
             pull_request.source_ref_parts.commit_id,
             pull_request_version.pull_request_version_id)
 
         self.trigger_pull_request_hook(pull_request, pull_request.author, 'update')
 
         return UpdateResponse(
             executed=True, reason=UpdateFailureReason.NONE,
             old=pull_request, new=pull_request_version,
             common_ancestor_id=ancestor_commit_id, commit_changes=commit_changes,
             source_changed=source_changed, target_changed=target_changed)
 
     def _create_version_from_snapshot(self, pull_request):
         version = PullRequestVersion()
         version.title = pull_request.title
         version.description = pull_request.description
         version.status = pull_request.status
         version.pull_request_state = pull_request.pull_request_state
         version.created_on = datetime.datetime.now()
         version.updated_on = pull_request.updated_on
         version.user_id = pull_request.user_id
         version.source_repo = pull_request.source_repo
         version.source_ref = pull_request.source_ref
         version.target_repo = pull_request.target_repo
         version.target_ref = pull_request.target_ref
 
         version._last_merge_source_rev = pull_request._last_merge_source_rev
         version._last_merge_target_rev = pull_request._last_merge_target_rev
         version.last_merge_status = pull_request.last_merge_status
         version.last_merge_metadata = pull_request.last_merge_metadata
         version.shadow_merge_ref = pull_request.shadow_merge_ref
         version.merge_rev = pull_request.merge_rev
         version.reviewer_data = pull_request.reviewer_data
 
         version.revisions = pull_request.revisions
         version.common_ancestor_id = pull_request.common_ancestor_id
         version.pull_request = pull_request
         Session().add(version)
         Session().flush()
 
         return version
 
     def _generate_update_diffs(self, pull_request, pull_request_version):
 
         diff_context = (
             self.DIFF_CONTEXT +
             CommentsModel.needed_extra_diff_context())
         hide_whitespace_changes = False
         source_repo = pull_request_version.source_repo
         source_ref_id = pull_request_version.source_ref_parts.commit_id
         target_ref_id = pull_request_version.target_ref_parts.commit_id
         old_diff = self._get_diff_from_pr_or_version(
             source_repo, source_ref_id, target_ref_id,
             hide_whitespace_changes=hide_whitespace_changes, diff_context=diff_context)
 
         source_repo = pull_request.source_repo
         source_ref_id = pull_request.source_ref_parts.commit_id
         target_ref_id = pull_request.target_ref_parts.commit_id
 
         new_diff = self._get_diff_from_pr_or_version(
             source_repo, source_ref_id, target_ref_id,
             hide_whitespace_changes=hide_whitespace_changes, diff_context=diff_context)
 
         old_diff_data = diffs.DiffProcessor(old_diff)
         old_diff_data.prepare()
         new_diff_data = diffs.DiffProcessor(new_diff)
         new_diff_data.prepare()
 
         return old_diff_data, new_diff_data
 
     def _link_comments_to_version(self, pull_request_version):
         """
         Link all unlinked comments of this pull request to the given version.
 
         :param pull_request_version: The `PullRequestVersion` to which
             the comments shall be linked.
 
         """
         pull_request = pull_request_version.pull_request
         comments = ChangesetComment.query()\
             .filter(
                 # TODO: johbo: Should we query for the repo at all here?
                 # Pending decision on how comments of PRs are to be related
                 # to either the source repo, the target repo or no repo at all.
                 ChangesetComment.repo_id == pull_request.target_repo.repo_id,
                 ChangesetComment.pull_request == pull_request,
                 ChangesetComment.pull_request_version == None)\
             .order_by(ChangesetComment.comment_id.asc())
 
         # TODO: johbo: Find out why this breaks if it is done in a bulk
         # operation.
         for comment in comments:
             comment.pull_request_version_id = (
                 pull_request_version.pull_request_version_id)
             Session().add(comment)
 
     def _calculate_commit_id_changes(self, old_ids, new_ids):
         added = [x for x in new_ids if x not in old_ids]
         common = [x for x in new_ids if x in old_ids]
         removed = [x for x in old_ids if x not in new_ids]
         total = new_ids
         return ChangeTuple(added, common, removed, total)
 
     def _calculate_file_changes(self, old_diff_data, new_diff_data):
 
         old_files = OrderedDict()
         for diff_data in old_diff_data.parsed_diff:
             old_files[diff_data['filename']] = md5_safe(diff_data['raw_diff'])
 
         added_files = []
         modified_files = []
         removed_files = []
         for diff_data in new_diff_data.parsed_diff:
             new_filename = diff_data['filename']
             new_hash = md5_safe(diff_data['raw_diff'])
 
             old_hash = old_files.get(new_filename)
             if not old_hash:
                 # file is not present in old diff, we have to figure out from parsed diff
                 # operation ADD/REMOVE
                 operations_dict = diff_data['stats']['ops']
                 if diffs.DEL_FILENODE in operations_dict:
                     removed_files.append(new_filename)
                 else:
                     added_files.append(new_filename)
             else:
                 if new_hash != old_hash:
                     modified_files.append(new_filename)
                 # now remove a file from old, since we have seen it already
                 del old_files[new_filename]
 
         # removed files is when there are present in old, but not in NEW,
         # since we remove old files that are present in new diff, left-overs
         # if any should be the removed files
         removed_files.extend(old_files.keys())
 
         return FileChangeTuple(added_files, modified_files, removed_files)
 
     def _render_update_message(self, ancestor_commit_id, changes, file_changes):
         """
         render the message using DEFAULT_COMMENTS_RENDERER (RST renderer),
         so it's always looking the same disregarding on which default
         renderer system is using.
 
         :param ancestor_commit_id: ancestor raw_id
         :param changes: changes named tuple
         :param file_changes: file changes named tuple
 
         """
         new_status = ChangesetStatus.get_status_lbl(
             ChangesetStatus.STATUS_UNDER_REVIEW)
 
         changed_files = (
             file_changes.added + file_changes.modified + file_changes.removed)
 
         params = {
             'under_review_label': new_status,
             'added_commits': changes.added,
             'removed_commits': changes.removed,
             'changed_files': changed_files,
             'added_files': file_changes.added,
             'modified_files': file_changes.modified,
             'removed_files': file_changes.removed,
             'ancestor_commit_id': ancestor_commit_id
         }
         renderer = RstTemplateRenderer()
         return renderer.render('pull_request_update.mako', **params)
 
     def edit(self, pull_request, title, description, description_renderer, user):
         pull_request = self.__get_pull_request(pull_request)
         old_data = pull_request.get_api_data(with_merge_state=False)
         if pull_request.is_closed():
             raise ValueError('This pull request is closed')
         if title:
             pull_request.title = title
         pull_request.description = description
         pull_request.updated_on = datetime.datetime.now()
         pull_request.description_renderer = description_renderer
         Session().add(pull_request)
         self._log_audit_action(
             'repo.pull_request.edit', {'old_data': old_data},
             user, pull_request)
 
     def update_reviewers(self, pull_request, reviewer_data, user):
         """
         Update the reviewers in the pull request
 
         :param pull_request: the pr to update
         :param reviewer_data: list of tuples
             [(user, ['reason1', 'reason2'], mandatory_flag, role, [rules])]
         :param user: current use who triggers this action
         """
 
         pull_request = self.__get_pull_request(pull_request)
         if pull_request.is_closed():
             raise ValueError('This pull request is closed')
 
         reviewers = {}
         for user_id, reasons, mandatory, role, rules in reviewer_data:
             if isinstance(user_id, (int, compat.string_types)):
                 user_id = self._get_user(user_id).user_id
             reviewers[user_id] = {
                 'reasons': reasons, 'mandatory': mandatory, 'role': role}
 
         reviewers_ids = set(reviewers.keys())
         current_reviewers = PullRequestReviewers.get_pull_request_reviewers(
             pull_request.pull_request_id, role=PullRequestReviewers.ROLE_REVIEWER)
 
         current_reviewers_ids = set([x.user.user_id for x in current_reviewers])
 
         ids_to_add = reviewers_ids.difference(current_reviewers_ids)
         ids_to_remove = current_reviewers_ids.difference(reviewers_ids)
 
         log.debug("Adding %s reviewers", ids_to_add)
         log.debug("Removing %s reviewers", ids_to_remove)
         changed = False
         added_audit_reviewers = []
         removed_audit_reviewers = []
 
         for uid in ids_to_add:
             changed = True
             _usr = self._get_user(uid)
             reviewer = PullRequestReviewers()
             reviewer.user = _usr
             reviewer.pull_request = pull_request
             reviewer.reasons = reviewers[uid]['reasons']
             # NOTE(marcink): mandatory shouldn't be changed now
             # reviewer.mandatory = reviewers[uid]['reasons']
             # NOTE(marcink): role should be hardcoded, so we won't edit it.
             reviewer.role = PullRequestReviewers.ROLE_REVIEWER
             Session().add(reviewer)
             added_audit_reviewers.append(reviewer.get_dict())
 
         for uid in ids_to_remove:
             changed = True
             # NOTE(marcink): we fetch "ALL" reviewers objects using .all().
             # This is an edge case that handles previous state of having the same reviewer twice.
             # this CAN happen due to the lack of DB checks
             reviewers = PullRequestReviewers.query()\
                 .filter(PullRequestReviewers.user_id == uid,
                         PullRequestReviewers.role == PullRequestReviewers.ROLE_REVIEWER,
                         PullRequestReviewers.pull_request == pull_request)\
                 .all()
 
             for obj in reviewers:
                 added_audit_reviewers.append(obj.get_dict())
                 Session().delete(obj)
 
         if changed:
             Session().expire_all()
             pull_request.updated_on = datetime.datetime.now()
             Session().add(pull_request)
 
         # finally store audit logs
         for user_data in added_audit_reviewers:
             self._log_audit_action(
                 'repo.pull_request.reviewer.add', {'data': user_data},
                 user, pull_request)
         for user_data in removed_audit_reviewers:
             self._log_audit_action(
                 'repo.pull_request.reviewer.delete', {'old_data': user_data},
                 user, pull_request)
 
         self.notify_reviewers(pull_request, ids_to_add, user)
         return ids_to_add, ids_to_remove
 
     def update_observers(self, pull_request, observer_data, user):
         """
         Update the observers in the pull request
 
         :param pull_request: the pr to update
         :param observer_data: list of tuples
             [(user, ['reason1', 'reason2'], mandatory_flag, role, [rules])]
         :param user: current use who triggers this action
         """
         pull_request = self.__get_pull_request(pull_request)
         if pull_request.is_closed():
             raise ValueError('This pull request is closed')
 
         observers = {}
         for user_id, reasons, mandatory, role, rules in observer_data:
             if isinstance(user_id, (int, compat.string_types)):
                 user_id = self._get_user(user_id).user_id
             observers[user_id] = {
                 'reasons': reasons, 'observers': mandatory, 'role': role}
 
         observers_ids = set(observers.keys())
         current_observers = PullRequestReviewers.get_pull_request_reviewers(
             pull_request.pull_request_id, role=PullRequestReviewers.ROLE_OBSERVER)
 
         current_observers_ids = set([x.user.user_id for x in current_observers])
 
         ids_to_add = observers_ids.difference(current_observers_ids)
         ids_to_remove = current_observers_ids.difference(observers_ids)
 
         log.debug("Adding %s observer", ids_to_add)
         log.debug("Removing %s observer", ids_to_remove)
         changed = False
         added_audit_observers = []
         removed_audit_observers = []
 
         for uid in ids_to_add:
             changed = True
             _usr = self._get_user(uid)
             observer = PullRequestReviewers()
             observer.user = _usr
             observer.pull_request = pull_request
             observer.reasons = observers[uid]['reasons']
             # NOTE(marcink): mandatory shouldn't be changed now
             # observer.mandatory = observer[uid]['reasons']
 
             # NOTE(marcink): role should be hardcoded, so we won't edit it.
             observer.role = PullRequestReviewers.ROLE_OBSERVER
             Session().add(observer)
             added_audit_observers.append(observer.get_dict())
 
         for uid in ids_to_remove:
             changed = True
             # NOTE(marcink): we fetch "ALL" reviewers objects using .all().
             # This is an edge case that handles previous state of having the same reviewer twice.
             # this CAN happen due to the lack of DB checks
             observers = PullRequestReviewers.query()\
                 .filter(PullRequestReviewers.user_id == uid,
                         PullRequestReviewers.role == PullRequestReviewers.ROLE_OBSERVER,
                         PullRequestReviewers.pull_request == pull_request)\
                 .all()
 
             for obj in observers:
                 added_audit_observers.append(obj.get_dict())
                 Session().delete(obj)
 
         if changed:
             Session().expire_all()
             pull_request.updated_on = datetime.datetime.now()
             Session().add(pull_request)
 
         # finally store audit logs
         for user_data in added_audit_observers:
             self._log_audit_action(
                 'repo.pull_request.observer.add', {'data': user_data},
                 user, pull_request)
         for user_data in removed_audit_observers:
             self._log_audit_action(
                 'repo.pull_request.observer.delete', {'old_data': user_data},
                 user, pull_request)
 
         self.notify_observers(pull_request, ids_to_add, user)
         return ids_to_add, ids_to_remove
 
     def get_url(self, pull_request, request=None, permalink=False):
         if not request:
             request = get_current_request()
 
         if permalink:
             return request.route_url(
                 'pull_requests_global',
                 pull_request_id=pull_request.pull_request_id,)
         else:
             return request.route_url('pullrequest_show',
                 repo_name=safe_str(pull_request.target_repo.repo_name),
                 pull_request_id=pull_request.pull_request_id,)
 
     def get_shadow_clone_url(self, pull_request, request=None):
         """
         Returns qualified url pointing to the shadow repository. If this pull
         request is closed there is no shadow repository and ``None`` will be
         returned.
         """
         if pull_request.is_closed():
             return None
         else:
             pr_url = urllib.unquote(self.get_url(pull_request, request=request))
             return safe_unicode('{pr_url}/repository'.format(pr_url=pr_url))
 
     def _notify_reviewers(self, pull_request, user_ids, role, user):
         # notification to reviewers/observers
         if not user_ids:
             return
 
         log.debug('Notify following %s users about pull-request %s', role, user_ids)
 
         pull_request_obj = pull_request
         # get the current participants of this pull request
         recipients = user_ids
         notification_type = EmailNotificationModel.TYPE_PULL_REQUEST
 
         pr_source_repo = pull_request_obj.source_repo
         pr_target_repo = pull_request_obj.target_repo
 
         pr_url = h.route_url('pullrequest_show',
                              repo_name=pr_target_repo.repo_name,
                              pull_request_id=pull_request_obj.pull_request_id,)
 
         # set some variables for email notification
         pr_target_repo_url = h.route_url(
             'repo_summary', repo_name=pr_target_repo.repo_name)
 
         pr_source_repo_url = h.route_url(
             'repo_summary', repo_name=pr_source_repo.repo_name)
 
         # pull request specifics
         pull_request_commits = [
             (x.raw_id, x.message)
             for x in map(pr_source_repo.get_commit, pull_request.revisions)]
 
         current_rhodecode_user = user
         kwargs = {
             'user': current_rhodecode_user,
             'pull_request_author': pull_request.author,
             'pull_request': pull_request_obj,
             'pull_request_commits': pull_request_commits,
 
             'pull_request_target_repo': pr_target_repo,
             'pull_request_target_repo_url': pr_target_repo_url,
 
             'pull_request_source_repo': pr_source_repo,
             'pull_request_source_repo_url': pr_source_repo_url,
 
             'pull_request_url': pr_url,
             'thread_ids': [pr_url],
             'user_role': role
         }
 
         # create notification objects, and emails
         NotificationModel().create(
             created_by=current_rhodecode_user,
             notification_subject='',  # Filled in based on the notification_type
             notification_body='',  # Filled in based on the notification_type
             notification_type=notification_type,
             recipients=recipients,
             email_kwargs=kwargs,
         )
 
     def notify_reviewers(self, pull_request, reviewers_ids, user):
         return self._notify_reviewers(pull_request, reviewers_ids,
                                       PullRequestReviewers.ROLE_REVIEWER, user)
 
     def notify_observers(self, pull_request, observers_ids, user):
         return self._notify_reviewers(pull_request, observers_ids,
                                       PullRequestReviewers.ROLE_OBSERVER, user)
 
     def notify_users(self, pull_request, updating_user, ancestor_commit_id,
                      commit_changes, file_changes):
 
         updating_user_id = updating_user.user_id
         reviewers = set([x.user.user_id for x in pull_request.get_pull_request_reviewers()])
         # NOTE(marcink): send notification to all other users except to
         # person who updated the PR
         recipients = reviewers.difference(set([updating_user_id]))
 
         log.debug('Notify following recipients about pull-request update %s', recipients)
 
         pull_request_obj = pull_request
 
         # send email about the update
         changed_files = (
                 file_changes.added + file_changes.modified + file_changes.removed)
 
         pr_source_repo = pull_request_obj.source_repo
         pr_target_repo = pull_request_obj.target_repo
 
         pr_url = h.route_url('pullrequest_show',
                              repo_name=pr_target_repo.repo_name,
                              pull_request_id=pull_request_obj.pull_request_id,)
 
         # set some variables for email notification
         pr_target_repo_url = h.route_url(
             'repo_summary', repo_name=pr_target_repo.repo_name)
 
         pr_source_repo_url = h.route_url(
             'repo_summary', repo_name=pr_source_repo.repo_name)
 
         email_kwargs = {
             'date': datetime.datetime.now(),
             'updating_user': updating_user,
 
             'pull_request': pull_request_obj,
 
             'pull_request_target_repo': pr_target_repo,
             'pull_request_target_repo_url': pr_target_repo_url,
 
             'pull_request_source_repo': pr_source_repo,
             'pull_request_source_repo_url': pr_source_repo_url,
 
             'pull_request_url': pr_url,
 
             'ancestor_commit_id': ancestor_commit_id,
             'added_commits': commit_changes.added,
             'removed_commits': commit_changes.removed,
             'changed_files': changed_files,
             'added_files': file_changes.added,
             'modified_files': file_changes.modified,
             'removed_files': file_changes.removed,
             'thread_ids': [pr_url],
         }
 
         # create notification objects, and emails
         NotificationModel().create(
             created_by=updating_user,
             notification_subject='',  # Filled in based on the notification_type
             notification_body='',  # Filled in based on the notification_type
             notification_type=EmailNotificationModel.TYPE_PULL_REQUEST_UPDATE,
             recipients=recipients,
             email_kwargs=email_kwargs,
         )
 
     def delete(self, pull_request, user=None):
         if not user:
             user = getattr(get_current_rhodecode_user(), 'username', None)
 
         pull_request = self.__get_pull_request(pull_request)
         old_data = pull_request.get_api_data(with_merge_state=False)
         self._cleanup_merge_workspace(pull_request)
         self._log_audit_action(
             'repo.pull_request.delete', {'old_data': old_data},
             user, pull_request)
         Session().delete(pull_request)
 
     def close_pull_request(self, pull_request, user):
         pull_request = self.__get_pull_request(pull_request)
         self._cleanup_merge_workspace(pull_request)
         pull_request.status = PullRequest.STATUS_CLOSED
         pull_request.updated_on = datetime.datetime.now()
         Session().add(pull_request)
         self.trigger_pull_request_hook(pull_request, pull_request.author, 'close')
 
         pr_data = pull_request.get_api_data(with_merge_state=False)
         self._log_audit_action(
             'repo.pull_request.close', {'data': pr_data}, user, pull_request)
 
     def close_pull_request_with_comment(
             self, pull_request, user, repo, message=None, auth_user=None):
 
         pull_request_review_status = pull_request.calculated_review_status()
 
         if pull_request_review_status == ChangesetStatus.STATUS_APPROVED:
             # approved only if we have voting consent
             status = ChangesetStatus.STATUS_APPROVED
         else:
             status = ChangesetStatus.STATUS_REJECTED
         status_lbl = ChangesetStatus.get_status_lbl(status)
 
         default_message = (
             'Closing with status change {transition_icon} {status}.'
         ).format(transition_icon='>', status=status_lbl)
         text = message or default_message
 
         # create a comment, and link it to new status
         comment = CommentsModel().create(
             text=text,
             repo=repo.repo_id,
             user=user.user_id,
             pull_request=pull_request.pull_request_id,
             status_change=status_lbl,
             status_change_type=status,
             closing_pr=True,
             auth_user=auth_user,
         )
 
         # calculate old status before we change it
         old_calculated_status = pull_request.calculated_review_status()
         ChangesetStatusModel().set_status(
             repo.repo_id,
             status,
             user.user_id,
             comment=comment,
             pull_request=pull_request.pull_request_id
         )
 
         Session().flush()
 
         self.trigger_pull_request_hook(pull_request, user, 'comment',
                                        data={'comment': comment})
 
         # we now calculate the status of pull request again, and based on that
         # calculation trigger status change. This might happen in cases
         # that non-reviewer admin closes a pr, which means his vote doesn't
         # change the status, while if he's a reviewer this might change it.
         calculated_status = pull_request.calculated_review_status()
         if old_calculated_status != calculated_status:
             self.trigger_pull_request_hook(pull_request, user, 'review_status_change',
                                            data={'status': calculated_status})
 
         # finally close the PR
         PullRequestModel().close_pull_request(pull_request.pull_request_id, user)
 
         return comment, status
 
     def merge_status(self, pull_request, translator=None, force_shadow_repo_refresh=False):
         _ = translator or get_current_request().translate
 
         if not self._is_merge_enabled(pull_request):
             return None, False, _('Server-side pull request merging is disabled.')
 
         if pull_request.is_closed():
             return None, False, _('This pull request is closed.')
 
         merge_possible, msg = self._check_repo_requirements(
             target=pull_request.target_repo, source=pull_request.source_repo,
             translator=_)
         if not merge_possible:
             return None, merge_possible, msg
 
         try:
             merge_response = self._try_merge(
                 pull_request, force_shadow_repo_refresh=force_shadow_repo_refresh)
             log.debug("Merge response: %s", merge_response)
             return merge_response, merge_response.possible, merge_response.merge_status_message
         except NotImplementedError:
             return None, False, _('Pull request merging is not supported.')
 
     def _check_repo_requirements(self, target, source, translator):
         """
         Check if `target` and `source` have compatible requirements.
 
         Currently this is just checking for largefiles.
         """
         _ = translator
         target_has_largefiles = self._has_largefiles(target)
         source_has_largefiles = self._has_largefiles(source)
         merge_possible = True
         message = u''
 
         if target_has_largefiles != source_has_largefiles:
             merge_possible = False
             if source_has_largefiles:
                 message = _(
                     'Target repository large files support is disabled.')
             else:
                 message = _(
                     'Source repository large files support is disabled.')
 
         return merge_possible, message
 
     def _has_largefiles(self, repo):
         largefiles_ui = VcsSettingsModel(repo=repo).get_ui_settings(
             'extensions', 'largefiles')
         return largefiles_ui and largefiles_ui[0].active
 
     def _try_merge(self, pull_request, force_shadow_repo_refresh=False):
         """
         Try to merge the pull request and return the merge status.
         """
         log.debug(
             "Trying out if the pull request %s can be merged. Force_refresh=%s",
             pull_request.pull_request_id, force_shadow_repo_refresh)
         target_vcs = pull_request.target_repo.scm_instance()
         # Refresh the target reference.
         try:
             target_ref = self._refresh_reference(
                 pull_request.target_ref_parts, target_vcs)
         except CommitDoesNotExistError:
             merge_state = MergeResponse(
                 False, False, None, MergeFailureReason.MISSING_TARGET_REF,
                 metadata={'target_ref': pull_request.target_ref_parts})
             return merge_state
 
         target_locked = pull_request.target_repo.locked
         if target_locked and target_locked[0]:
             locked_by = 'user:{}'.format(target_locked[0])
             log.debug("The target repository is locked by %s.", locked_by)
             merge_state = MergeResponse(
                 False, False, None, MergeFailureReason.TARGET_IS_LOCKED,
                 metadata={'locked_by': locked_by})
         elif force_shadow_repo_refresh or self._needs_merge_state_refresh(
                 pull_request, target_ref):
             log.debug("Refreshing the merge status of the repository.")
             merge_state = self._refresh_merge_state(
                 pull_request, target_vcs, target_ref)
         else:
             possible = pull_request.last_merge_status == MergeFailureReason.NONE
             metadata = {
                 'unresolved_files': '',
                 'target_ref': pull_request.target_ref_parts,
                 'source_ref': pull_request.source_ref_parts,
             }
             if pull_request.last_merge_metadata:
                 metadata.update(pull_request.last_merge_metadata_parsed)
 
             if not possible and target_ref.type == 'branch':
                 # NOTE(marcink): case for mercurial multiple heads on branch
                 heads = target_vcs._heads(target_ref.name)
                 if len(heads) != 1:
                     heads = '\n,'.join(target_vcs._heads(target_ref.name))
                     metadata.update({
                         'heads': heads
                     })
 
             merge_state = MergeResponse(
                 possible, False, None, pull_request.last_merge_status, metadata=metadata)
 
         return merge_state
 
     def _refresh_reference(self, reference, vcs_repository):
         if reference.type in self.UPDATABLE_REF_TYPES:
             name_or_id = reference.name
         else:
             name_or_id = reference.commit_id
 
         refreshed_commit = vcs_repository.get_commit(name_or_id)
         refreshed_reference = Reference(
             reference.type, reference.name, refreshed_commit.raw_id)
         return refreshed_reference
 
     def _needs_merge_state_refresh(self, pull_request, target_reference):
         return not(
             pull_request.revisions and
             pull_request.revisions[0] == pull_request._last_merge_source_rev and
             target_reference.commit_id == pull_request._last_merge_target_rev)
 
     def _refresh_merge_state(self, pull_request, target_vcs, target_reference):
         workspace_id = self._workspace_id(pull_request)
         source_vcs = pull_request.source_repo.scm_instance()
         repo_id = pull_request.target_repo.repo_id
         use_rebase = self._use_rebase_for_merging(pull_request)
         close_branch = self._close_branch_before_merging(pull_request)
         merge_state = target_vcs.merge(
             repo_id, workspace_id,
             target_reference, source_vcs, pull_request.source_ref_parts,
             dry_run=True, use_rebase=use_rebase,
             close_branch=close_branch)
 
         # Do not store the response if there was an unknown error.
         if merge_state.failure_reason != MergeFailureReason.UNKNOWN:
             pull_request._last_merge_source_rev = \
                 pull_request.source_ref_parts.commit_id
             pull_request._last_merge_target_rev = target_reference.commit_id
             pull_request.last_merge_status = merge_state.failure_reason
             pull_request.last_merge_metadata = merge_state.metadata
 
             pull_request.shadow_merge_ref = merge_state.merge_ref
             Session().add(pull_request)
             Session().commit()
 
         return merge_state
 
     def _workspace_id(self, pull_request):
         workspace_id = 'pr-%s' % pull_request.pull_request_id
         return workspace_id
 
     def generate_repo_data(self, repo, commit_id=None, branch=None,
                            bookmark=None, translator=None):
         from rhodecode.model.repo import RepoModel
 
         all_refs, selected_ref = \
             self._get_repo_pullrequest_sources(
                 repo.scm_instance(), commit_id=commit_id,
                 branch=branch, bookmark=bookmark, translator=translator)
 
         refs_select2 = []
         for element in all_refs:
             children = [{'id': x[0], 'text': x[1]} for x in element[0]]
             refs_select2.append({'text': element[1], 'children': children})
 
         return {
             'user': {
                 'user_id': repo.user.user_id,
                 'username': repo.user.username,
                 'firstname': repo.user.first_name,
                 'lastname': repo.user.last_name,
                 'gravatar_link': h.gravatar_url(repo.user.email, 14),
             },
             'name': repo.repo_name,
             'link': RepoModel().get_url(repo),
             'description': h.chop_at_smart(repo.description_safe, '\n'),
             'refs': {
                 'all_refs': all_refs,
                 'selected_ref': selected_ref,
                 'select2_refs': refs_select2
             }
         }
 
     def generate_pullrequest_title(self, source, source_ref, target):
         return u'{source}#{at_ref} to {target}'.format(
             source=source,
             at_ref=source_ref,
             target=target,
         )
 
     def _cleanup_merge_workspace(self, pull_request):
         # Merging related cleanup
         repo_id = pull_request.target_repo.repo_id
         target_scm = pull_request.target_repo.scm_instance()
         workspace_id = self._workspace_id(pull_request)
 
         try:
             target_scm.cleanup_merge_workspace(repo_id, workspace_id)
         except NotImplementedError:
             pass
 
     def _get_repo_pullrequest_sources(
             self, repo, commit_id=None, branch=None, bookmark=None,
             translator=None):
         """
         Return a structure with repo's interesting commits, suitable for
         the selectors in pullrequest controller
 
         :param commit_id: a commit that must be in the list somehow
             and selected by default
         :param branch: a branch that must be in the list and selected
             by default - even if closed
         :param bookmark: a bookmark that must be in the list and selected
         """
         _ = translator or get_current_request().translate
 
         commit_id = safe_str(commit_id) if commit_id else None
         branch = safe_unicode(branch) if branch else None
         bookmark = safe_unicode(bookmark) if bookmark else None
 
         selected = None
 
         # order matters: first source that has commit_id in it will be selected
         sources = []
         sources.append(('book', repo.bookmarks.items(), _('Bookmarks'), bookmark))
         sources.append(('branch', repo.branches.items(), _('Branches'), branch))
 
         if commit_id:
             ref_commit = (h.short_id(commit_id), commit_id)
             sources.append(('rev', [ref_commit], _('Commit IDs'), commit_id))
 
         sources.append(
             ('branch', repo.branches_closed.items(), _('Closed Branches'), branch),
         )
 
         groups = []
 
         for group_key, ref_list, group_name, match in sources:
             group_refs = []
             for ref_name, ref_id in ref_list:
                 ref_key = u'{}:{}:{}'.format(group_key, ref_name, ref_id)
                 group_refs.append((ref_key, ref_name))
 
                 if not selected:
                     if set([commit_id, match]) & set([ref_id, ref_name]):
                         selected = ref_key
 
             if group_refs:
                 groups.append((group_refs, group_name))
 
         if not selected:
             ref = commit_id or branch or bookmark
             if ref:
                 raise CommitDoesNotExistError(
                     u'No commit refs could be found matching: {}'.format(ref))
             elif repo.DEFAULT_BRANCH_NAME in repo.branches:
                 selected = u'branch:{}:{}'.format(
                     safe_unicode(repo.DEFAULT_BRANCH_NAME),
                     safe_unicode(repo.branches[repo.DEFAULT_BRANCH_NAME])
                 )
             elif repo.commit_ids:
                 # make the user select in this case
                 selected = None
             else:
                 raise EmptyRepositoryError()
         return groups, selected
 
     def get_diff(self, source_repo, source_ref_id, target_ref_id,
                  hide_whitespace_changes, diff_context):
 
         return self._get_diff_from_pr_or_version(
             source_repo, source_ref_id, target_ref_id,
             hide_whitespace_changes=hide_whitespace_changes, diff_context=diff_context)
 
     def _get_diff_from_pr_or_version(
             self, source_repo, source_ref_id, target_ref_id,
             hide_whitespace_changes, diff_context):
 
         target_commit = source_repo.get_commit(
             commit_id=safe_str(target_ref_id))
         source_commit = source_repo.get_commit(
             commit_id=safe_str(source_ref_id), maybe_unreachable=True)
         if isinstance(source_repo, Repository):
             vcs_repo = source_repo.scm_instance()
         else:
             vcs_repo = source_repo
 
         # TODO: johbo: In the context of an update, we cannot reach
         # the old commit anymore with our normal mechanisms. It needs
         # some sort of special support in the vcs layer to avoid this
         # workaround.
         if (source_commit.raw_id == vcs_repo.EMPTY_COMMIT_ID and
                 vcs_repo.alias == 'git'):
             source_commit.raw_id = safe_str(source_ref_id)
 
         log.debug('calculating diff between '
                   'source_ref:%s and target_ref:%s for repo `%s`',
                   target_ref_id, source_ref_id,
                   safe_unicode(vcs_repo.path))
 
         vcs_diff = vcs_repo.get_diff(
             commit1=target_commit, commit2=source_commit,
             ignore_whitespace=hide_whitespace_changes, context=diff_context)
         return vcs_diff
 
     def _is_merge_enabled(self, pull_request):
         return self._get_general_setting(
             pull_request, 'rhodecode_pr_merge_enabled')
 
     def _use_rebase_for_merging(self, pull_request):
         repo_type = pull_request.target_repo.repo_type
         if repo_type == 'hg':
             return self._get_general_setting(
                 pull_request, 'rhodecode_hg_use_rebase_for_merging')
         elif repo_type == 'git':
             return self._get_general_setting(
                 pull_request, 'rhodecode_git_use_rebase_for_merging')
 
         return False
 
     def _user_name_for_merging(self, pull_request, user):
         env_user_name_attr = os.environ.get('RC_MERGE_USER_NAME_ATTR', '')
         if env_user_name_attr and hasattr(user, env_user_name_attr):
             user_name_attr = env_user_name_attr
         else:
             user_name_attr = 'short_contact'
 
         user_name = getattr(user, user_name_attr)
         return user_name
 
     def _close_branch_before_merging(self, pull_request):
         repo_type = pull_request.target_repo.repo_type
         if repo_type == 'hg':
             return self._get_general_setting(
                 pull_request, 'rhodecode_hg_close_branch_before_merging')
         elif repo_type == 'git':
             return self._get_general_setting(
                 pull_request, 'rhodecode_git_close_branch_before_merging')
 
         return False
 
     def _get_general_setting(self, pull_request, settings_key, default=False):
         settings_model = VcsSettingsModel(repo=pull_request.target_repo)
         settings = settings_model.get_general_settings()
         return settings.get(settings_key, default)
 
     def _log_audit_action(self, action, action_data, user, pull_request):
         audit_logger.store(
             action=action,
             action_data=action_data,
             user=user,
             repo=pull_request.target_repo)
 
     def get_reviewer_functions(self):
         """
         Fetches functions for validation and fetching default reviewers.
         If available we use the EE package, else we fallback to CE
         package functions
         """
         try:
             from rc_reviewers.utils import get_default_reviewers_data
             from rc_reviewers.utils import validate_default_reviewers
             from rc_reviewers.utils import validate_observers
         except ImportError:
             from rhodecode.apps.repository.utils import get_default_reviewers_data
             from rhodecode.apps.repository.utils import validate_default_reviewers
             from rhodecode.apps.repository.utils import validate_observers
 
         return get_default_reviewers_data, validate_default_reviewers, validate_observers
 
 
 class MergeCheck(object):
     """
     Perform Merge Checks and returns a check object which stores information
     about merge errors, and merge conditions
     """
     TODO_CHECK = 'todo'
     PERM_CHECK = 'perm'
     REVIEW_CHECK = 'review'
     MERGE_CHECK = 'merge'
     WIP_CHECK = 'wip'
 
     def __init__(self):
         self.review_status = None
         self.merge_possible = None
         self.merge_msg = ''
         self.merge_response = None
         self.failed = None
         self.errors = []
         self.error_details = OrderedDict()
         self.source_commit = AttributeDict()
         self.target_commit = AttributeDict()
         self.reviewers_count = 0
         self.observers_count = 0
 
     def __repr__(self):
         return '<MergeCheck(possible:{}, failed:{}, errors:{})>'.format(
             self.merge_possible, self.failed, self.errors)
 
     def push_error(self, error_type, message, error_key, details):
         self.failed = True
         self.errors.append([error_type, message])
         self.error_details[error_key] = dict(
             details=details,
             error_type=error_type,
             message=message
         )
 
     @classmethod
     def validate(cls, pull_request, auth_user, translator, fail_early=False,
                  force_shadow_repo_refresh=False):
         _ = translator
         merge_check = cls()
 
         # title has WIP:
         if pull_request.work_in_progress:
             log.debug("MergeCheck: cannot merge, title has wip: marker.")
 
             msg = _('WIP marker in title prevents from accidental merge.')
             merge_check.push_error('error', msg, cls.WIP_CHECK, pull_request.title)
             if fail_early:
                 return merge_check
 
         # permissions to merge
         user_allowed_to_merge = PullRequestModel().check_user_merge(pull_request, auth_user)
         if not user_allowed_to_merge:
             log.debug("MergeCheck: cannot merge, approval is pending.")
 
             msg = _('User `{}` not allowed to perform merge.').format(auth_user.username)
             merge_check.push_error('error', msg, cls.PERM_CHECK, auth_user.username)
             if fail_early:
                 return merge_check
 
         # permission to merge into the target branch
         target_commit_id = pull_request.target_ref_parts.commit_id
         if pull_request.target_ref_parts.type == 'branch':
             branch_name = pull_request.target_ref_parts.name
         else:
             # for mercurial we can always figure out the branch from the commit
             # in case of bookmark
             target_commit = pull_request.target_repo.get_commit(target_commit_id)
             branch_name = target_commit.branch
 
         rule, branch_perm = auth_user.get_rule_and_branch_permission(
             pull_request.target_repo.repo_name, branch_name)
         if branch_perm and branch_perm == 'branch.none':
             msg = _('Target branch `{}` changes rejected by rule {}.').format(
                 branch_name, rule)
             merge_check.push_error('error', msg, cls.PERM_CHECK, auth_user.username)
             if fail_early:
                 return merge_check
 
         # review status, must be always present
         review_status = pull_request.calculated_review_status()
         merge_check.review_status = review_status
         merge_check.reviewers_count = pull_request.reviewers_count
         merge_check.observers_count = pull_request.observers_count
 
         status_approved = review_status == ChangesetStatus.STATUS_APPROVED
         if not status_approved and merge_check.reviewers_count:
             log.debug("MergeCheck: cannot merge, approval is pending.")
             msg = _('Pull request reviewer approval is pending.')
 
             merge_check.push_error('warning', msg, cls.REVIEW_CHECK, review_status)
 
             if fail_early:
                 return merge_check
 
         # left over TODOs
         todos = CommentsModel().get_pull_request_unresolved_todos(pull_request)
         if todos:
             log.debug("MergeCheck: cannot merge, {} "
                       "unresolved TODOs left.".format(len(todos)))
 
             if len(todos) == 1:
                 msg = _('Cannot merge, {} TODO still not resolved.').format(
                     len(todos))
             else:
                 msg = _('Cannot merge, {} TODOs still not resolved.').format(
                     len(todos))
 
             merge_check.push_error('warning', msg, cls.TODO_CHECK, todos)
 
             if fail_early:
                 return merge_check
 
         # merge possible, here is the filesystem simulation + shadow repo
         merge_response, merge_status, msg = PullRequestModel().merge_status(
             pull_request, translator=translator,
             force_shadow_repo_refresh=force_shadow_repo_refresh)
 
         merge_check.merge_possible = merge_status
         merge_check.merge_msg = msg
         merge_check.merge_response = merge_response
 
         source_ref_id = pull_request.source_ref_parts.commit_id
         target_ref_id = pull_request.target_ref_parts.commit_id
 
         try:
             source_commit, target_commit = PullRequestModel().get_flow_commits(pull_request)
             merge_check.source_commit.changed = source_ref_id != source_commit.raw_id
             merge_check.source_commit.ref_spec = pull_request.source_ref_parts
             merge_check.source_commit.current_raw_id = source_commit.raw_id
             merge_check.source_commit.previous_raw_id = source_ref_id
 
             merge_check.target_commit.changed = target_ref_id != target_commit.raw_id
             merge_check.target_commit.ref_spec = pull_request.target_ref_parts
             merge_check.target_commit.current_raw_id = target_commit.raw_id
             merge_check.target_commit.previous_raw_id = target_ref_id
         except (SourceRefMissing, TargetRefMissing):
             pass
 
         if not merge_status:
             log.debug("MergeCheck: cannot merge, pull request merge not possible.")
             merge_check.push_error('warning', msg, cls.MERGE_CHECK, None)
 
             if fail_early:
                 return merge_check
 
         log.debug('MergeCheck: is failed: %s', merge_check.failed)
         return merge_check
 
     @classmethod
     def get_merge_conditions(cls, pull_request, translator):
         _ = translator
         merge_details = {}
 
         model = PullRequestModel()
         use_rebase = model._use_rebase_for_merging(pull_request)
 
         if use_rebase:
             merge_details['merge_strategy'] = dict(
                 details={},
                 message=_('Merge strategy: rebase')
             )
         else:
             merge_details['merge_strategy'] = dict(
                 details={},
                 message=_('Merge strategy: explicit merge commit')
             )
 
         close_branch = model._close_branch_before_merging(pull_request)
         if close_branch:
             repo_type = pull_request.target_repo.repo_type
             close_msg = ''
             if repo_type == 'hg':
                 close_msg = _('Source branch will be closed before the merge.')
             elif repo_type == 'git':
                 close_msg = _('Source branch will be deleted after the merge.')
 
             merge_details['close_branch'] = dict(
                 details={},
                 message=close_msg
             )
 
         return merge_details
 
 
 ChangeTuple = collections.namedtuple(
     'ChangeTuple', ['added', 'common', 'removed', 'total'])
 
 FileChangeTuple = collections.namedtuple(
     'FileChangeTuple', ['added', 'modified', 'removed'])