pull-requests: expose unresolved files in merge response.

rhodecode/lib/vcs/backends/base.py

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

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

@@ -1,1004 +1,1017 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 GIT repository module
 """
 
 import logging
 import os
 import re
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from rhodecode.lib.compat import OrderedDict
 from rhodecode.lib.datelib import (
     utcdate_fromtimestamp, makedate, date_astimestamp)
 from rhodecode.lib.utils import safe_unicode, safe_str
 from rhodecode.lib.utils2 import CachedProperty
 from rhodecode.lib.vcs import connection, path as vcspath
 from rhodecode.lib.vcs.backends.base import (
     BaseRepository, CollectionGenerator, Config, MergeResponse,
     MergeFailureReason, Reference)
 from rhodecode.lib.vcs.backends.git.commit import GitCommit
 from rhodecode.lib.vcs.backends.git.diff import GitDiff
 from rhodecode.lib.vcs.backends.git.inmemory import GitInMemoryCommit
 from rhodecode.lib.vcs.exceptions import (
     CommitDoesNotExistError, EmptyRepositoryError,
-    RepositoryError, TagAlreadyExistError, TagDoesNotExistError, VCSError)
+    RepositoryError, TagAlreadyExistError, TagDoesNotExistError, VCSError, UnresolvedFilesInRepo)
 
 
 SHA_PATTERN = re.compile(r'^[[0-9a-fA-F]{12}|[0-9a-fA-F]{40}]$')
 
 log = logging.getLogger(__name__)
 
 
 class GitRepository(BaseRepository):
     """
     Git repository backend.
     """
     DEFAULT_BRANCH_NAME = 'master'
 
     contact = BaseRepository.DEFAULT_CONTACT
 
     def __init__(self, repo_path, config=None, create=False, src_url=None,
                  do_workspace_checkout=False, with_wire=None, bare=False):
 
         self.path = safe_str(os.path.abspath(repo_path))
         self.config = config if config else self.get_default_config()
         self.with_wire = with_wire or {"cache": False}  # default should not use cache
 
         self._init_repo(create, src_url, do_workspace_checkout, bare)
 
         # caches
         self._commit_ids = {}
 
     @LazyProperty
     def _remote(self):
         repo_id = self.path
         return connection.Git(self.path, repo_id, self.config, with_wire=self.with_wire)
 
     @LazyProperty
     def bare(self):
         return self._remote.bare()
 
     @LazyProperty
     def head(self):
         return self._remote.head()
 
     @CachedProperty
     def commit_ids(self):
         """
         Returns list of commit ids, in ascending order.  Being lazy
         attribute allows external tools to inject commit ids from cache.
         """
         commit_ids = self._get_all_commit_ids()
         self._rebuild_cache(commit_ids)
         return commit_ids
 
     def _rebuild_cache(self, commit_ids):
         self._commit_ids = dict((commit_id, index)
                                 for index, commit_id in enumerate(commit_ids))
 
     def run_git_command(self, cmd, **opts):
         """
         Runs given ``cmd`` as git command and returns tuple
         (stdout, stderr).
 
         :param cmd: git command to be executed
         :param opts: env options to pass into Subprocess command
         """
         if not isinstance(cmd, list):
             raise ValueError('cmd must be a list, got %s instead' % type(cmd))
 
         skip_stderr_log = opts.pop('skip_stderr_log', False)
         out, err = self._remote.run_git_command(cmd, **opts)
         if err and not skip_stderr_log:
             log.debug('Stderr output of git command "%s":\n%s', cmd, err)
         return out, err
 
     @staticmethod
     def check_url(url, config):
         """
         Function will check given url and try to verify if it's a valid
         link. Sometimes it may happened that git will issue basic
         auth request that can cause whole API to hang when used from python
         or other external calls.
 
         On failures it'll raise urllib2.HTTPError, exception is also thrown
         when the return code is non 200
         """
         # check first if it's not an url
         if os.path.isdir(url) or url.startswith('file:'):
             return True
 
         if '+' in url.split('://', 1)[0]:
             url = url.split('+', 1)[1]
 
         # Request the _remote to verify the url
         return connection.Git.check_url(url, config.serialize())
 
     @staticmethod
     def is_valid_repository(path):
         if os.path.isdir(os.path.join(path, '.git')):
             return True
         # check case of bare repository
         try:
             GitRepository(path)
             return True
         except VCSError:
             pass
         return False
 
     def _init_repo(self, create, src_url=None, do_workspace_checkout=False,
                    bare=False):
         if create and os.path.exists(self.path):
             raise RepositoryError(
                 "Cannot create repository at %s, location already exist"
                 % self.path)
 
         if bare and do_workspace_checkout:
             raise RepositoryError("Cannot update a bare repository")
         try:
 
             if src_url:
                 # check URL before any actions
                 GitRepository.check_url(src_url, self.config)
 
             if create:
                 os.makedirs(self.path, mode=0o755)
 
                 if bare:
                     self._remote.init_bare()
                 else:
                     self._remote.init()
 
                 if src_url and bare:
                     # bare repository only allows a fetch and checkout is not allowed
                     self.fetch(src_url, commit_ids=None)
                 elif src_url:
                     self.pull(src_url, commit_ids=None,
                               update_after=do_workspace_checkout)
 
             else:
                 if not self._remote.assert_correct_path():
                     raise RepositoryError(
                         'Path "%s" does not contain a Git repository' %
                         (self.path,))
 
         # TODO: johbo: check if we have to translate the OSError here
         except OSError as err:
             raise RepositoryError(err)
 
     def _get_all_commit_ids(self):
         return self._remote.get_all_commit_ids()
 
     def _get_commit_ids(self, filters=None):
         # we must check if this repo is not empty, since later command
         # fails if it is. And it's cheaper to ask than throw the subprocess
         # errors
 
         head = self._remote.head(show_exc=False)
 
         if not head:
             return []
 
         rev_filter = ['--branches', '--tags']
         extra_filter = []
 
         if filters:
             if filters.get('since'):
                 extra_filter.append('--since=%s' % (filters['since']))
             if filters.get('until'):
                 extra_filter.append('--until=%s' % (filters['until']))
             if filters.get('branch_name'):
                 rev_filter = []
                 extra_filter.append(filters['branch_name'])
         rev_filter.extend(extra_filter)
 
             # if filters.get('start') or filters.get('end'):
             #     # skip is offset, max-count is limit
             #     if filters.get('start'):
             #         extra_filter += ' --skip=%s' % filters['start']
             #     if filters.get('end'):
             #         extra_filter += ' --max-count=%s' % (filters['end'] - (filters['start'] or 0))
 
         cmd = ['rev-list', '--reverse', '--date-order'] + rev_filter
         try:
             output, __ = self.run_git_command(cmd)
         except RepositoryError:
             # Can be raised for empty repositories
             return []
         return output.splitlines()
 
     def _lookup_commit(self, commit_id_or_idx, translate_tag=True):
         def is_null(value):
             return len(value) == commit_id_or_idx.count('0')
 
         if commit_id_or_idx in (None, '', 'tip', 'HEAD', 'head', -1):
             return self.commit_ids[-1]
         commit_missing_err = "Commit {} does not exist for `{}`".format(
             *map(safe_str, [commit_id_or_idx, self.name]))
 
         is_bstr = isinstance(commit_id_or_idx, (str, unicode))
         if ((is_bstr and commit_id_or_idx.isdigit() and len(commit_id_or_idx) < 12)
             or isinstance(commit_id_or_idx, int) or is_null(commit_id_or_idx)):
             try:
                 commit_id_or_idx = self.commit_ids[int(commit_id_or_idx)]
             except Exception:
                 raise CommitDoesNotExistError(commit_missing_err)
 
         elif is_bstr:
             # Need to call remote to translate id for tagging scenario
             try:
                 remote_data = self._remote.get_object(commit_id_or_idx)
                 commit_id_or_idx = remote_data["commit_id"]
             except (CommitDoesNotExistError,):
                 raise CommitDoesNotExistError(commit_missing_err)
 
         # Ensure we return full id
         if not SHA_PATTERN.match(str(commit_id_or_idx)):
             raise CommitDoesNotExistError(
                 "Given commit id %s not recognized" % commit_id_or_idx)
         return commit_id_or_idx
 
     def get_hook_location(self):
         """
         returns absolute path to location where hooks are stored
         """
         loc = os.path.join(self.path, 'hooks')
         if not self.bare:
             loc = os.path.join(self.path, '.git', 'hooks')
         return loc
 
     @LazyProperty
     def last_change(self):
         """
         Returns last change made on this repository as
         `datetime.datetime` object.
         """
         try:
             return self.get_commit().date
         except RepositoryError:
             tzoffset = makedate()[1]
             return utcdate_fromtimestamp(self._get_fs_mtime(), tzoffset)
 
     def _get_fs_mtime(self):
         idx_loc = '' if self.bare else '.git'
         # fallback to filesystem
         in_path = os.path.join(self.path, idx_loc, "index")
         he_path = os.path.join(self.path, idx_loc, "HEAD")
         if os.path.exists(in_path):
             return os.stat(in_path).st_mtime
         else:
             return os.stat(he_path).st_mtime
 
     @LazyProperty
     def description(self):
         description = self._remote.get_description()
         return safe_unicode(description or self.DEFAULT_DESCRIPTION)
 
     def _get_refs_entries(self, prefix='', reverse=False, strip_prefix=True):
         if self.is_empty():
             return OrderedDict()
 
         result = []
         for ref, sha in self._refs.iteritems():
             if ref.startswith(prefix):
                 ref_name = ref
                 if strip_prefix:
                     ref_name = ref[len(prefix):]
                 result.append((safe_unicode(ref_name), sha))
 
         def get_name(entry):
             return entry[0]
 
         return OrderedDict(sorted(result, key=get_name, reverse=reverse))
 
     def _get_branches(self):
         return self._get_refs_entries(prefix='refs/heads/', strip_prefix=True)
 
     @CachedProperty
     def branches(self):
         return self._get_branches()
 
     @CachedProperty
     def branches_closed(self):
         return {}
 
     @CachedProperty
     def bookmarks(self):
         return {}
 
     @CachedProperty
     def branches_all(self):
         all_branches = {}
         all_branches.update(self.branches)
         all_branches.update(self.branches_closed)
         return all_branches
 
     @CachedProperty
     def tags(self):
         return self._get_tags()
 
     def _get_tags(self):
         return self._get_refs_entries(prefix='refs/tags/', strip_prefix=True, reverse=True)
 
     def tag(self, name, user, commit_id=None, message=None, date=None,
             **kwargs):
         # TODO: fix this method to apply annotated tags correct with message
         """
         Creates and returns a tag for the given ``commit_id``.
 
         :param name: name for new tag
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param commit_id: commit id for which new tag would be created
         :param message: message of the tag's commit
         :param date: date of tag's commit
 
         :raises TagAlreadyExistError: if tag with same name already exists
         """
         if name in self.tags:
             raise TagAlreadyExistError("Tag %s already exists" % name)
         commit = self.get_commit(commit_id=commit_id)
         message = message or "Added tag %s for commit %s" % (name, commit.raw_id)
 
         self._remote.set_refs('refs/tags/%s' % name, commit.raw_id)
 
         self._invalidate_prop_cache('tags')
         self._invalidate_prop_cache('_refs')
 
         return commit
 
     def remove_tag(self, name, user, message=None, date=None):
         """
         Removes tag with the given ``name``.
 
         :param name: name of the tag to be removed
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param message: message of the tag's removal commit
         :param date: date of tag's removal commit
 
         :raises TagDoesNotExistError: if tag with given name does not exists
         """
         if name not in self.tags:
             raise TagDoesNotExistError("Tag %s does not exist" % name)
 
         self._remote.tag_remove(name)
         self._invalidate_prop_cache('tags')
         self._invalidate_prop_cache('_refs')
 
     def _get_refs(self):
         return self._remote.get_refs()
 
     @CachedProperty
     def _refs(self):
         return self._get_refs()
 
     @property
     def _ref_tree(self):
         node = tree = {}
         for ref, sha in self._refs.iteritems():
             path = ref.split('/')
             for bit in path[:-1]:
                 node = node.setdefault(bit, {})
             node[path[-1]] = sha
             node = tree
         return tree
 
     def get_remote_ref(self, ref_name):
         ref_key = 'refs/remotes/origin/{}'.format(safe_str(ref_name))
         try:
             return self._refs[ref_key]
         except Exception:
             return
 
     def get_commit(self, commit_id=None, commit_idx=None, pre_load=None, translate_tag=True):
         """
         Returns `GitCommit` object representing commit from git repository
         at the given `commit_id` or head (most recent commit) if None given.
         """
         if self.is_empty():
             raise EmptyRepositoryError("There are no commits yet")
 
         if commit_id is not None:
             self._validate_commit_id(commit_id)
             try:
                 # we have cached idx, use it without contacting the remote
                 idx = self._commit_ids[commit_id]
                 return GitCommit(self, commit_id, idx, pre_load=pre_load)
             except KeyError:
                 pass
 
         elif commit_idx is not None:
             self._validate_commit_idx(commit_idx)
             try:
                 _commit_id = self.commit_ids[commit_idx]
                 if commit_idx < 0:
                     commit_idx = self.commit_ids.index(_commit_id)
                 return GitCommit(self, _commit_id, commit_idx, pre_load=pre_load)
             except IndexError:
                 commit_id = commit_idx
         else:
             commit_id = "tip"
 
         if translate_tag:
             commit_id = self._lookup_commit(commit_id)
 
         try:
             idx = self._commit_ids[commit_id]
         except KeyError:
             idx = -1
 
         return GitCommit(self, commit_id, idx, pre_load=pre_load)
 
     def get_commits(
             self, start_id=None, end_id=None, start_date=None, end_date=None,
             branch_name=None, show_hidden=False, pre_load=None, translate_tags=True):
         """
         Returns generator of `GitCommit` objects from start to end (both
         are inclusive), in ascending date order.
 
         :param start_id: None, str(commit_id)
         :param end_id: None, str(commit_id)
         :param start_date: if specified, commits with commit date less than
           ``start_date`` would be filtered out from returned set
         :param end_date: if specified, commits with commit date greater than
           ``end_date`` would be filtered out from returned set
         :param branch_name: if specified, commits not reachable from given
           branch would be filtered out from returned set
         :param show_hidden: Show hidden commits such as obsolete or hidden from
             Mercurial evolve
         :raise BranchDoesNotExistError: If given `branch_name` does not
             exist.
         :raise CommitDoesNotExistError: If commits for given `start` or
           `end` could not be found.
 
         """
         if self.is_empty():
             raise EmptyRepositoryError("There are no commits yet")
 
         self._validate_branch_name(branch_name)
 
         if start_id is not None:
             self._validate_commit_id(start_id)
         if end_id is not None:
             self._validate_commit_id(end_id)
 
         start_raw_id = self._lookup_commit(start_id)
         start_pos = self._commit_ids[start_raw_id] if start_id else None
         end_raw_id = self._lookup_commit(end_id)
         end_pos = max(0, self._commit_ids[end_raw_id]) if end_id else None
 
         if None not in [start_id, end_id] and start_pos > end_pos:
             raise RepositoryError(
                 "Start commit '%s' cannot be after end commit '%s'" %
                 (start_id, end_id))
 
         if end_pos is not None:
             end_pos += 1
 
         filter_ = []
         if branch_name:
             filter_.append({'branch_name': branch_name})
         if start_date and not end_date:
             filter_.append({'since': start_date})
         if end_date and not start_date:
             filter_.append({'until': end_date})
         if start_date and end_date:
             filter_.append({'since': start_date})
             filter_.append({'until': end_date})
 
         # if start_pos or end_pos:
         #     filter_.append({'start': start_pos})
         #     filter_.append({'end': end_pos})
 
         if filter_:
             revfilters = {
                 'branch_name': branch_name,
                 'since': start_date.strftime('%m/%d/%y %H:%M:%S') if start_date else None,
                 'until': end_date.strftime('%m/%d/%y %H:%M:%S') if end_date else None,
                 'start': start_pos,
                 'end': end_pos,
             }
             commit_ids = self._get_commit_ids(filters=revfilters)
 
         else:
             commit_ids = self.commit_ids
 
         if start_pos or end_pos:
             commit_ids = commit_ids[start_pos: end_pos]
 
         return CollectionGenerator(self, commit_ids, pre_load=pre_load,
                                    translate_tag=translate_tags)
 
     def get_diff(
             self, commit1, commit2, path='', ignore_whitespace=False,
             context=3, path1=None):
         """
         Returns (git like) *diff*, as plain text. Shows changes introduced by
         ``commit2`` since ``commit1``.
 
         :param commit1: Entry point from which diff is shown. Can be
           ``self.EMPTY_COMMIT`` - in this case, patch showing all
           the changes since empty state of the repository until ``commit2``
         :param commit2: Until which commits changes should be shown.
         :param ignore_whitespace: If set to ``True``, would not show whitespace
           changes. Defaults to ``False``.
         :param context: How many lines before/after changed lines should be
           shown. Defaults to ``3``.
         """
         self._validate_diff_commits(commit1, commit2)
         if path1 is not None and path1 != path:
             raise ValueError("Diff of two different paths not supported.")
 
         if path:
             file_filter = path
         else:
             file_filter = None
 
         diff = self._remote.diff(
             commit1.raw_id, commit2.raw_id, file_filter=file_filter,
             opt_ignorews=ignore_whitespace,
             context=context)
         return GitDiff(diff)
 
     def strip(self, commit_id, branch_name):
         commit = self.get_commit(commit_id=commit_id)
         if commit.merge:
             raise Exception('Cannot reset to merge commit')
 
         # parent is going to be the new head now
         commit = commit.parents[0]
         self._remote.set_refs('refs/heads/%s' % branch_name, commit.raw_id)
 
         # clear cached properties
         self._invalidate_prop_cache('commit_ids')
         self._invalidate_prop_cache('_refs')
         self._invalidate_prop_cache('branches')
 
         return len(self.commit_ids)
 
     def get_common_ancestor(self, commit_id1, commit_id2, repo2):
         if commit_id1 == commit_id2:
             return commit_id1
 
         if self != repo2:
             commits = self._remote.get_missing_revs(
                 commit_id1, commit_id2, repo2.path)
             if commits:
                 commit = repo2.get_commit(commits[-1])
                 if commit.parents:
                     ancestor_id = commit.parents[0].raw_id
                 else:
                     ancestor_id = None
             else:
                 # no commits from other repo, ancestor_id is the commit_id2
                 ancestor_id = commit_id2
         else:
             output, __ = self.run_git_command(
                 ['merge-base', commit_id1, commit_id2])
             ancestor_id = re.findall(r'[0-9a-fA-F]{40}', output)[0]
 
         return ancestor_id
 
     def compare(self, commit_id1, commit_id2, repo2, merge, pre_load=None):
         repo1 = self
         ancestor_id = None
 
         if commit_id1 == commit_id2:
             commits = []
         elif repo1 != repo2:
             missing_ids = self._remote.get_missing_revs(commit_id1, commit_id2,
                                                         repo2.path)
             commits = [
                 repo2.get_commit(commit_id=commit_id, pre_load=pre_load)
                 for commit_id in reversed(missing_ids)]
         else:
             output, __ = repo1.run_git_command(
                 ['log', '--reverse', '--pretty=format: %H', '-s',
                  '%s..%s' % (commit_id1, commit_id2)])
             commits = [
                 repo1.get_commit(commit_id=commit_id, pre_load=pre_load)
                 for commit_id in re.findall(r'[0-9a-fA-F]{40}', output)]
 
         return commits
 
     @LazyProperty
     def in_memory_commit(self):
         """
         Returns ``GitInMemoryCommit`` object for this repository.
         """
         return GitInMemoryCommit(self)
 
     def pull(self, url, commit_ids=None, update_after=False):
         """
         Pull changes from external location. Pull is different in GIT
         that fetch since it's doing a checkout
 
         :param commit_ids: Optional. Can be set to a list of commit ids
            which shall be pulled from the other repository.
         """
         refs = None
         if commit_ids is not None:
             remote_refs = self._remote.get_remote_refs(url)
             refs = [ref for ref in remote_refs if remote_refs[ref] in commit_ids]
         self._remote.pull(url, refs=refs, update_after=update_after)
         self._remote.invalidate_vcs_cache()
 
     def fetch(self, url, commit_ids=None):
         """
         Fetch all git objects from external location.
         """
         self._remote.sync_fetch(url, refs=commit_ids)
         self._remote.invalidate_vcs_cache()
 
     def push(self, url):
         refs = None
         self._remote.sync_push(url, refs=refs)
 
     def set_refs(self, ref_name, commit_id):
         self._remote.set_refs(ref_name, commit_id)
         self._invalidate_prop_cache('_refs')
 
     def remove_ref(self, ref_name):
         self._remote.remove_ref(ref_name)
         self._invalidate_prop_cache('_refs')
 
     def _update_server_info(self):
         """
         runs gits update-server-info command in this repo instance
         """
         self._remote.update_server_info()
 
     def _current_branch(self):
         """
         Return the name of the current branch.
 
         It only works for non bare repositories (i.e. repositories with a
         working copy)
         """
         if self.bare:
             raise RepositoryError('Bare git repos do not have active branches')
 
         if self.is_empty():
             return None
 
         stdout, _ = self.run_git_command(['rev-parse', '--abbrev-ref', 'HEAD'])
         return stdout.strip()
 
     def _checkout(self, branch_name, create=False, force=False):
         """
         Checkout a branch in the working directory.
 
         It tries to create the branch if create is True, failing if the branch
         already exists.
 
         It only works for non bare repositories (i.e. repositories with a
         working copy)
         """
         if self.bare:
             raise RepositoryError('Cannot checkout branches in a bare git repo')
 
         cmd = ['checkout']
         if force:
             cmd.append('-f')
         if create:
             cmd.append('-b')
         cmd.append(branch_name)
         self.run_git_command(cmd, fail_on_stderr=False)
 
     def _create_branch(self, branch_name, commit_id):
         """
         creates a branch in a GIT repo
         """
         self._remote.create_branch(branch_name, commit_id)
 
     def _identify(self):
         """
         Return the current state of the working directory.
         """
         if self.bare:
             raise RepositoryError('Bare git repos do not have active branches')
 
         if self.is_empty():
             return None
 
         stdout, _ = self.run_git_command(['rev-parse', 'HEAD'])
         return stdout.strip()
 
     def _local_clone(self, clone_path, branch_name, source_branch=None):
         """
         Create a local clone of the current repo.
         """
         # N.B.(skreft): the --branch option is required as otherwise the shallow
         # clone will only fetch the active branch.
         cmd = ['clone', '--branch', branch_name,
                self.path, os.path.abspath(clone_path)]
 
         self.run_git_command(cmd, fail_on_stderr=False)
 
         # if we get the different source branch, make sure we also fetch it for
         # merge conditions
         if source_branch and source_branch != branch_name:
             # check if the ref exists.
             shadow_repo = GitRepository(os.path.abspath(clone_path))
             if shadow_repo.get_remote_ref(source_branch):
                 cmd = ['fetch', self.path, source_branch]
                 self.run_git_command(cmd, fail_on_stderr=False)
 
     def _local_fetch(self, repository_path, branch_name, use_origin=False):
         """
         Fetch a branch from a local repository.
         """
         repository_path = os.path.abspath(repository_path)
         if repository_path == self.path:
             raise ValueError('Cannot fetch from the same repository')
 
         if use_origin:
             branch_name = '+{branch}:refs/heads/{branch}'.format(
                 branch=branch_name)
 
         cmd = ['fetch', '--no-tags', '--update-head-ok',
                repository_path, branch_name]
         self.run_git_command(cmd, fail_on_stderr=False)
 
     def _local_reset(self, branch_name):
         branch_name = '{}'.format(branch_name)
         cmd = ['reset', '--hard', branch_name, '--']
         self.run_git_command(cmd, fail_on_stderr=False)
 
     def _last_fetch_heads(self):
         """
         Return the last fetched heads that need merging.
 
         The algorithm is defined at
         https://github.com/git/git/blob/v2.1.3/git-pull.sh#L283
         """
         if not self.bare:
             fetch_heads_path = os.path.join(self.path, '.git', 'FETCH_HEAD')
         else:
             fetch_heads_path = os.path.join(self.path, 'FETCH_HEAD')
 
         heads = []
         with open(fetch_heads_path) as f:
             for line in f:
                 if '    not-for-merge   ' in line:
                     continue
                 line = re.sub('\t.*', '', line, flags=re.DOTALL)
                 heads.append(line)
 
         return heads
 
     def get_shadow_instance(self, shadow_repository_path, enable_hooks=False, cache=False):
         return GitRepository(shadow_repository_path, with_wire={"cache": cache})
 
     def _local_pull(self, repository_path, branch_name, ff_only=True):
         """
         Pull a branch from a local repository.
         """
         if self.bare:
             raise RepositoryError('Cannot pull into a bare git repository')
         # N.B.(skreft): The --ff-only option is to make sure this is a
         # fast-forward (i.e., we are only pulling new changes and there are no
         # conflicts with our current branch)
         # Additionally, that option needs to go before --no-tags, otherwise git
         # pull complains about it being an unknown flag.
         cmd = ['pull']
         if ff_only:
             cmd.append('--ff-only')
         cmd.extend(['--no-tags', repository_path, branch_name])
         self.run_git_command(cmd, fail_on_stderr=False)
 
     def _local_merge(self, merge_message, user_name, user_email, heads):
         """
         Merge the given head into the checked out branch.
 
         It will force a merge commit.
 
         Currently it raises an error if the repo is empty, as it is not possible
         to create a merge commit in an empty repo.
 
         :param merge_message: The message to use for the merge commit.
         :param heads: the heads to merge.
         """
         if self.bare:
             raise RepositoryError('Cannot merge into a bare git repository')
 
         if not heads:
             return
 
         if self.is_empty():
-            # TODO(skreft): do somehting more robust in this case.
+            # TODO(skreft): do something more robust in this case.
             raise RepositoryError(
                 'Do not know how to merge into empty repositories yet')
+        unresolved = None
 
         # N.B.(skreft): the --no-ff option is used to enforce the creation of a
         # commit message. We also specify the user who is doing the merge.
         cmd = ['-c', 'user.name="%s"' % safe_str(user_name),
                '-c', 'user.email=%s' % safe_str(user_email),
                'merge', '--no-ff', '-m', safe_str(merge_message)]
         cmd.extend(heads)
         try:
             output = self.run_git_command(cmd, fail_on_stderr=False)
         except RepositoryError:
+            files = self.run_git_command(['diff', '--name-only', '--diff-filter', 'U'],
+                                         fail_on_stderr=False)[0].splitlines()
+            # NOTE(marcink): we add U notation for consistent with HG backend output
+            unresolved = ['U {}'.format(f) for f in files]
+
             # Cleanup any merge leftovers
             self.run_git_command(['merge', '--abort'], fail_on_stderr=False)
+
+            if unresolved:
+                raise UnresolvedFilesInRepo(unresolved)
+            else:
             raise
 
     def _local_push(
             self, source_branch, repository_path, target_branch,
             enable_hooks=False, rc_scm_data=None):
         """
         Push the source_branch to the given repository and target_branch.
 
         Currently it if the target_branch is not master and the target repo is
         empty, the push will work, but then GitRepository won't be able to find
         the pushed branch or the commits. As the HEAD will be corrupted (i.e.,
         pointing to master, which does not exist).
 
         It does not run the hooks in the target repo.
         """
         # TODO(skreft): deal with the case in which the target repo is empty,
         # and the target_branch is not master.
         target_repo = GitRepository(repository_path)
         if (not target_repo.bare and
                 target_repo._current_branch() == target_branch):
             # Git prevents pushing to the checked out branch, so simulate it by
             # pulling into the target repository.
             target_repo._local_pull(self.path, source_branch)
         else:
             cmd = ['push', os.path.abspath(repository_path),
                    '%s:%s' % (source_branch, target_branch)]
             gitenv = {}
             if rc_scm_data:
                 gitenv.update({'RC_SCM_DATA': rc_scm_data})
 
             if not enable_hooks:
                 gitenv['RC_SKIP_HOOKS'] = '1'
             self.run_git_command(cmd, fail_on_stderr=False, extra_env=gitenv)
 
     def _get_new_pr_branch(self, source_branch, target_branch):
         prefix = 'pr_%s-%s_' % (source_branch, target_branch)
         pr_branches = []
         for branch in self.branches:
             if branch.startswith(prefix):
                 pr_branches.append(int(branch[len(prefix):]))
 
         if not pr_branches:
             branch_id = 0
         else:
             branch_id = max(pr_branches) + 1
 
         return '%s%d' % (prefix, branch_id)
 
     def _maybe_prepare_merge_workspace(
             self, repo_id, workspace_id, target_ref, source_ref):
         shadow_repository_path = self._get_shadow_repository_path(
             self.path, repo_id, workspace_id)
         if not os.path.exists(shadow_repository_path):
             self._local_clone(
                 shadow_repository_path, target_ref.name, source_ref.name)
             log.debug('Prepared %s shadow repository in %s',
                       self.alias, shadow_repository_path)
 
         return shadow_repository_path
 
     def _merge_repo(self, repo_id, workspace_id, target_ref,
                     source_repo, source_ref, merge_message,
                     merger_name, merger_email, dry_run=False,
                     use_rebase=False, close_branch=False):
 
         log.debug('Executing merge_repo with %s strategy, dry_run mode:%s',
                   'rebase' if use_rebase else 'merge', dry_run)
         if target_ref.commit_id != self.branches[target_ref.name]:
             log.warning('Target ref %s commit mismatch %s vs %s', target_ref,
                         target_ref.commit_id, self.branches[target_ref.name])
             return MergeResponse(
                 False, False, None, MergeFailureReason.TARGET_IS_NOT_HEAD,
                 metadata={'target_ref': target_ref})
 
         shadow_repository_path = self._maybe_prepare_merge_workspace(
             repo_id, workspace_id, target_ref, source_ref)
         shadow_repo = self.get_shadow_instance(shadow_repository_path)
 
         # checkout source, if it's different. Otherwise we could not
         # fetch proper commits for merge testing
         if source_ref.name != target_ref.name:
             if shadow_repo.get_remote_ref(source_ref.name):
                 shadow_repo._checkout(source_ref.name, force=True)
 
         # checkout target, and fetch changes
         shadow_repo._checkout(target_ref.name, force=True)
 
         # fetch/reset pull the target, in case it is changed
         # this handles even force changes
         shadow_repo._local_fetch(self.path, target_ref.name, use_origin=True)
         shadow_repo._local_reset(target_ref.name)
 
         # Need to reload repo to invalidate the cache, or otherwise we cannot
         # retrieve the last target commit.
         shadow_repo = self.get_shadow_instance(shadow_repository_path)
         if target_ref.commit_id != shadow_repo.branches[target_ref.name]:
             log.warning('Shadow Target ref %s commit mismatch %s vs %s',
                         target_ref, target_ref.commit_id,
                         shadow_repo.branches[target_ref.name])
             return MergeResponse(
                 False, False, None, MergeFailureReason.TARGET_IS_NOT_HEAD,
                 metadata={'target_ref': target_ref})
 
         # calculate new branch
         pr_branch = shadow_repo._get_new_pr_branch(
             source_ref.name, target_ref.name)
         log.debug('using pull-request merge branch: `%s`', pr_branch)
         # checkout to temp branch, and fetch changes
         shadow_repo._checkout(pr_branch, create=True)
         try:
             shadow_repo._local_fetch(source_repo.path, source_ref.name)
         except RepositoryError:
             log.exception('Failure when doing local fetch on '
                           'shadow repo: %s', shadow_repo)
             return MergeResponse(
                 False, False, None, MergeFailureReason.MISSING_SOURCE_REF,
                 metadata={'source_ref': source_ref})
 
         merge_ref = None
         merge_failure_reason = MergeFailureReason.NONE
         metadata = {}
         try:
             shadow_repo._local_merge(merge_message, merger_name, merger_email,
                                      [source_ref.commit_id])
             merge_possible = True
 
             # Need to invalidate the cache, or otherwise we
             # cannot retrieve the merge commit.
             shadow_repo = shadow_repo.get_shadow_instance(shadow_repository_path)
             merge_commit_id = shadow_repo.branches[pr_branch]
 
             # Set a reference pointing to the merge commit. This reference may
             # be used to easily identify the last successful merge commit in
             # the shadow repository.
             shadow_repo.set_refs('refs/heads/pr-merge', merge_commit_id)
             merge_ref = Reference('branch', 'pr-merge', merge_commit_id)
-        except RepositoryError:
+        except RepositoryError as e:
             log.exception('Failure when doing local merge on git shadow repo')
+            if isinstance(e, UnresolvedFilesInRepo):
+                metadata['unresolved_files'] = 'file: ' + (', file: '.join(e.args[0]))
+
             merge_possible = False
             merge_failure_reason = MergeFailureReason.MERGE_FAILED
 
         if merge_possible and not dry_run:
             try:
                 shadow_repo._local_push(
                     pr_branch, self.path, target_ref.name, enable_hooks=True,
                     rc_scm_data=self.config.get('rhodecode', 'RC_SCM_DATA'))
                 merge_succeeded = True
             except RepositoryError:
                 log.exception(
                     'Failure when doing local push from the shadow '
                     'repository to the target repository at %s.', self.path)
                 merge_succeeded = False
                 merge_failure_reason = MergeFailureReason.PUSH_FAILED
                 metadata['target'] = 'git shadow repo'
                 metadata['merge_commit'] = pr_branch
         else:
             merge_succeeded = False
 
         return MergeResponse(
             merge_possible, merge_succeeded, merge_ref, merge_failure_reason,
             metadata=metadata)

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

@@ -1,952 +1,972 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 HG repository module
 """
 import os
 import logging
 import binascii
 import urllib
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from rhodecode.lib.compat import OrderedDict
 from rhodecode.lib.datelib import (
     date_to_timestamp_plus_offset, utcdate_fromtimestamp, makedate)
 from rhodecode.lib.utils import safe_unicode, safe_str
 from rhodecode.lib.utils2 import CachedProperty
 from rhodecode.lib.vcs import connection, exceptions
 from rhodecode.lib.vcs.backends.base import (
     BaseRepository, CollectionGenerator, Config, MergeResponse,
     MergeFailureReason, Reference, BasePathPermissionChecker)
 from rhodecode.lib.vcs.backends.hg.commit import MercurialCommit
 from rhodecode.lib.vcs.backends.hg.diff import MercurialDiff
 from rhodecode.lib.vcs.backends.hg.inmemory import MercurialInMemoryCommit
 from rhodecode.lib.vcs.exceptions import (
     EmptyRepositoryError, RepositoryError, TagAlreadyExistError,
-    TagDoesNotExistError, CommitDoesNotExistError, SubrepoMergeError)
+    TagDoesNotExistError, CommitDoesNotExistError, SubrepoMergeError, UnresolvedFilesInRepo)
 from rhodecode.lib.vcs.compat import configparser
 
 hexlify = binascii.hexlify
 nullid = "\0" * 20
 
 log = logging.getLogger(__name__)
 
 
 class MercurialRepository(BaseRepository):
     """
     Mercurial repository backend
     """
     DEFAULT_BRANCH_NAME = 'default'
 
     def __init__(self, repo_path, config=None, create=False, src_url=None,
                  do_workspace_checkout=False, with_wire=None, bare=False):
         """
         Raises RepositoryError if repository could not be find at the given
         ``repo_path``.
 
         :param repo_path: local path of the repository
         :param config: config object containing the repo configuration
         :param create=False: if set to True, would try to create repository if
            it does not exist rather than raising exception
         :param src_url=None: would try to clone repository from given location
         :param do_workspace_checkout=False: sets update of working copy after
            making a clone
         :param bare: not used, compatible with other VCS
         """
 
         self.path = safe_str(os.path.abspath(repo_path))
         # mercurial since 4.4.X requires certain configuration to be present
         # because sometimes we init the repos with config we need to meet
         # special requirements
         self.config = config if config else self.get_default_config(
             default=[('extensions', 'largefiles', '1')])
         self.with_wire = with_wire or {"cache": False}  # default should not use cache
 
         self._init_repo(create, src_url, do_workspace_checkout)
 
         # caches
         self._commit_ids = {}
 
     @LazyProperty
     def _remote(self):
         repo_id = self.path
         return connection.Hg(self.path, repo_id, self.config, with_wire=self.with_wire)
 
     @CachedProperty
     def commit_ids(self):
         """
         Returns list of commit ids, in ascending order.  Being lazy
         attribute allows external tools to inject shas from cache.
         """
         commit_ids = self._get_all_commit_ids()
         self._rebuild_cache(commit_ids)
         return commit_ids
 
     def _rebuild_cache(self, commit_ids):
         self._commit_ids = dict((commit_id, index)
                                 for index, commit_id in enumerate(commit_ids))
 
     @CachedProperty
     def branches(self):
         return self._get_branches()
 
     @CachedProperty
     def branches_closed(self):
         return self._get_branches(active=False, closed=True)
 
     @CachedProperty
     def branches_all(self):
         all_branches = {}
         all_branches.update(self.branches)
         all_branches.update(self.branches_closed)
         return all_branches
 
     def _get_branches(self, active=True, closed=False):
         """
         Gets branches for this repository
         Returns only not closed active branches by default
 
         :param active: return also active branches
         :param closed: return also closed branches
 
         """
         if self.is_empty():
             return {}
 
         def get_name(ctx):
             return ctx[0]
 
         _branches = [(safe_unicode(n), hexlify(h),) for n, h in
                      self._remote.branches(active, closed).items()]
 
         return OrderedDict(sorted(_branches, key=get_name, reverse=False))
 
     @CachedProperty
     def tags(self):
         """
         Gets tags for this repository
         """
         return self._get_tags()
 
     def _get_tags(self):
         if self.is_empty():
             return {}
 
         def get_name(ctx):
             return ctx[0]
 
         _tags = [(safe_unicode(n), hexlify(h),) for n, h in
                  self._remote.tags().items()]
 
         return OrderedDict(sorted(_tags, key=get_name, reverse=True))
 
     def tag(self, name, user, commit_id=None, message=None, date=None, **kwargs):
         """
         Creates and returns a tag for the given ``commit_id``.
 
         :param name: name for new tag
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param commit_id: commit id for which new tag would be created
         :param message: message of the tag's commit
         :param date: date of tag's commit
 
         :raises TagAlreadyExistError: if tag with same name already exists
         """
         if name in self.tags:
             raise TagAlreadyExistError("Tag %s already exists" % name)
 
         commit = self.get_commit(commit_id=commit_id)
         local = kwargs.setdefault('local', False)
 
         if message is None:
             message = "Added tag %s for commit %s" % (name, commit.short_id)
 
         date, tz = date_to_timestamp_plus_offset(date)
 
         self._remote.tag(name, commit.raw_id, message, local, user, date, tz)
         self._remote.invalidate_vcs_cache()
 
         # Reinitialize tags
         self._invalidate_prop_cache('tags')
         tag_id = self.tags[name]
 
         return self.get_commit(commit_id=tag_id)
 
     def remove_tag(self, name, user, message=None, date=None):
         """
         Removes tag with the given `name`.
 
         :param name: name of the tag to be removed
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param message: message of the tag's removal commit
         :param date: date of tag's removal commit
 
         :raises TagDoesNotExistError: if tag with given name does not exists
         """
         if name not in self.tags:
             raise TagDoesNotExistError("Tag %s does not exist" % name)
 
         if message is None:
             message = "Removed tag %s" % name
         local = False
 
         date, tz = date_to_timestamp_plus_offset(date)
 
         self._remote.tag(name, nullid, message, local, user, date, tz)
         self._remote.invalidate_vcs_cache()
         self._invalidate_prop_cache('tags')
 
     @LazyProperty
     def bookmarks(self):
         """
         Gets bookmarks for this repository
         """
         return self._get_bookmarks()
 
     def _get_bookmarks(self):
         if self.is_empty():
             return {}
 
         def get_name(ctx):
             return ctx[0]
 
         _bookmarks = [
             (safe_unicode(n), hexlify(h)) for n, h in
             self._remote.bookmarks().items()]
 
         return OrderedDict(sorted(_bookmarks, key=get_name))
 
     def _get_all_commit_ids(self):
         return self._remote.get_all_commit_ids('visible')
 
     def get_diff(
             self, commit1, commit2, path='', ignore_whitespace=False,
             context=3, path1=None):
         """
         Returns (git like) *diff*, as plain text. Shows changes introduced by
         `commit2` since `commit1`.
 
         :param commit1: Entry point from which diff is shown. Can be
           ``self.EMPTY_COMMIT`` - in this case, patch showing all
           the changes since empty state of the repository until `commit2`
         :param commit2: Until which commit changes should be shown.
         :param ignore_whitespace: If set to ``True``, would not show whitespace
           changes. Defaults to ``False``.
         :param context: How many lines before/after changed lines should be
           shown. Defaults to ``3``.
         """
         self._validate_diff_commits(commit1, commit2)
         if path1 is not None and path1 != path:
             raise ValueError("Diff of two different paths not supported.")
 
         if path:
             file_filter = [self.path, path]
         else:
             file_filter = None
 
         diff = self._remote.diff(
             commit1.raw_id, commit2.raw_id, file_filter=file_filter,
             opt_git=True, opt_ignorews=ignore_whitespace,
             context=context)
         return MercurialDiff(diff)
 
     def strip(self, commit_id, branch=None):
         self._remote.strip(commit_id, update=False, backup="none")
 
         self._remote.invalidate_vcs_cache()
         # clear cache
         self._invalidate_prop_cache('commit_ids')
 
         return len(self.commit_ids)
 
     def verify(self):
         verify = self._remote.verify()
 
         self._remote.invalidate_vcs_cache()
         return verify
 
     def hg_update_cache(self):
         update_cache = self._remote.hg_update_cache()
 
         self._remote.invalidate_vcs_cache()
         return update_cache
 
     def get_common_ancestor(self, commit_id1, commit_id2, repo2):
         if commit_id1 == commit_id2:
             return commit_id1
 
         ancestors = self._remote.revs_from_revspec(
             "ancestor(id(%s), id(%s))", commit_id1, commit_id2,
             other_path=repo2.path)
         return repo2[ancestors[0]].raw_id if ancestors else None
 
     def compare(self, commit_id1, commit_id2, repo2, merge, pre_load=None):
         if commit_id1 == commit_id2:
             commits = []
         else:
             if merge:
                 indexes = self._remote.revs_from_revspec(
                     "ancestors(id(%s)) - ancestors(id(%s)) - id(%s)",
                     commit_id2, commit_id1, commit_id1, other_path=repo2.path)
             else:
                 indexes = self._remote.revs_from_revspec(
                     "id(%s)..id(%s) - id(%s)", commit_id1, commit_id2,
                     commit_id1, other_path=repo2.path)
 
             commits = [repo2.get_commit(commit_idx=idx, pre_load=pre_load)
                        for idx in indexes]
 
         return commits
 
     @staticmethod
     def check_url(url, config):
         """
         Function will check given url and try to verify if it's a valid
         link. Sometimes it may happened that mercurial will issue basic
         auth request that can cause whole API to hang when used from python
         or other external calls.
 
         On failures it'll raise urllib2.HTTPError, exception is also thrown
         when the return code is non 200
         """
         # check first if it's not an local url
         if os.path.isdir(url) or url.startswith('file:'):
             return True
 
         # Request the _remote to verify the url
         return connection.Hg.check_url(url, config.serialize())
 
     @staticmethod
     def is_valid_repository(path):
         return os.path.isdir(os.path.join(path, '.hg'))
 
     def _init_repo(self, create, src_url=None, do_workspace_checkout=False):
         """
         Function will check for mercurial repository in given path. If there
         is no repository in that path it will raise an exception unless
         `create` parameter is set to True - in that case repository would
         be created.
 
         If `src_url` is given, would try to clone repository from the
         location at given clone_point. Additionally it'll make update to
         working copy accordingly to `do_workspace_checkout` flag.
         """
         if create and os.path.exists(self.path):
             raise RepositoryError(
                 "Cannot create repository at %s, location already exist"
                 % self.path)
 
         if src_url:
             url = str(self._get_url(src_url))
             MercurialRepository.check_url(url, self.config)
 
             self._remote.clone(url, self.path, do_workspace_checkout)
 
             # Don't try to create if we've already cloned repo
             create = False
 
         if create:
             os.makedirs(self.path, mode=0o755)
         self._remote.localrepository(create)
 
     @LazyProperty
     def in_memory_commit(self):
         return MercurialInMemoryCommit(self)
 
     @LazyProperty
     def description(self):
         description = self._remote.get_config_value(
             'web', 'description', untrusted=True)
         return safe_unicode(description or self.DEFAULT_DESCRIPTION)
 
     @LazyProperty
     def contact(self):
         contact = (
             self._remote.get_config_value("web", "contact") or
             self._remote.get_config_value("ui", "username"))
         return safe_unicode(contact or self.DEFAULT_CONTACT)
 
     @LazyProperty
     def last_change(self):
         """
         Returns last change made on this repository as
         `datetime.datetime` object.
         """
         try:
             return self.get_commit().date
         except RepositoryError:
             tzoffset = makedate()[1]
             return utcdate_fromtimestamp(self._get_fs_mtime(), tzoffset)
 
     def _get_fs_mtime(self):
         # fallback to filesystem
         cl_path = os.path.join(self.path, '.hg', "00changelog.i")
         st_path = os.path.join(self.path, '.hg', "store")
         if os.path.exists(cl_path):
             return os.stat(cl_path).st_mtime
         else:
             return os.stat(st_path).st_mtime
 
     def _get_url(self, url):
         """
         Returns normalized url. If schema is not given, would fall
         to filesystem
         (``file:///``) schema.
         """
         url = url.encode('utf8')
         if url != 'default' and '://' not in url:
             url = "file:" + urllib.pathname2url(url)
         return url
 
     def get_hook_location(self):
         """
         returns absolute path to location where hooks are stored
         """
         return os.path.join(self.path, '.hg', '.hgrc')
 
     def get_commit(self, commit_id=None, commit_idx=None, pre_load=None, translate_tag=None):
         """
         Returns ``MercurialCommit`` object representing repository's
         commit at the given `commit_id` or `commit_idx`.
         """
         if self.is_empty():
             raise EmptyRepositoryError("There are no commits yet")
 
         if commit_id is not None:
             self._validate_commit_id(commit_id)
             try:
                 # we have cached idx, use it without contacting the remote
                 idx = self._commit_ids[commit_id]
                 return MercurialCommit(self, commit_id, idx, pre_load=pre_load)
             except KeyError:
                 pass
 
         elif commit_idx is not None:
             self._validate_commit_idx(commit_idx)
             try:
                 _commit_id = self.commit_ids[commit_idx]
                 if commit_idx < 0:
                     commit_idx = self.commit_ids.index(_commit_id)
 
                 return MercurialCommit(self, _commit_id, commit_idx, pre_load=pre_load)
             except IndexError:
                 commit_id = commit_idx
         else:
             commit_id = "tip"
 
         if isinstance(commit_id, unicode):
             commit_id = safe_str(commit_id)
 
         try:
             raw_id, idx = self._remote.lookup(commit_id, both=True)
         except CommitDoesNotExistError:
             msg = "Commit {} does not exist for `{}`".format(
                 *map(safe_str, [commit_id, self.name]))
             raise CommitDoesNotExistError(msg)
 
         return MercurialCommit(self, raw_id, idx, pre_load=pre_load)
 
     def get_commits(
             self, start_id=None, end_id=None, start_date=None, end_date=None,
             branch_name=None, show_hidden=False, pre_load=None, translate_tags=None):
         """
         Returns generator of ``MercurialCommit`` objects from start to end
         (both are inclusive)
 
         :param start_id: None, str(commit_id)
         :param end_id: None, str(commit_id)
         :param start_date: if specified, commits with commit date less than
           ``start_date`` would be filtered out from returned set
         :param end_date: if specified, commits with commit date greater than
           ``end_date`` would be filtered out from returned set
         :param branch_name: if specified, commits not reachable from given
           branch would be filtered out from returned set
         :param show_hidden: Show hidden commits such as obsolete or hidden from
             Mercurial evolve
         :raise BranchDoesNotExistError: If given ``branch_name`` does not
             exist.
         :raise CommitDoesNotExistError: If commit for given ``start`` or
           ``end`` could not be found.
         """
         # actually we should check now if it's not an empty repo
         if self.is_empty():
             raise EmptyRepositoryError("There are no commits yet")
         self._validate_branch_name(branch_name)
 
         branch_ancestors = False
         if start_id is not None:
             self._validate_commit_id(start_id)
             c_start = self.get_commit(commit_id=start_id)
             start_pos = self._commit_ids[c_start.raw_id]
         else:
             start_pos = None
 
         if end_id is not None:
             self._validate_commit_id(end_id)
             c_end = self.get_commit(commit_id=end_id)
             end_pos = max(0, self._commit_ids[c_end.raw_id])
         else:
             end_pos = None
 
         if None not in [start_id, end_id] and start_pos > end_pos:
             raise RepositoryError(
                 "Start commit '%s' cannot be after end commit '%s'" %
                 (start_id, end_id))
 
         if end_pos is not None:
             end_pos += 1
 
         commit_filter = []
 
         if branch_name and not branch_ancestors:
             commit_filter.append('branch("%s")' % (branch_name,))
         elif branch_name and branch_ancestors:
             commit_filter.append('ancestors(branch("%s"))' % (branch_name,))
 
         if start_date and not end_date:
             commit_filter.append('date(">%s")' % (start_date,))
         if end_date and not start_date:
             commit_filter.append('date("<%s")' % (end_date,))
         if start_date and end_date:
             commit_filter.append(
                 'date(">%s") and date("<%s")' % (start_date, end_date))
 
         if not show_hidden:
             commit_filter.append('not obsolete()')
             commit_filter.append('not hidden()')
 
         # TODO: johbo: Figure out a simpler way for this solution
         collection_generator = CollectionGenerator
         if commit_filter:
             commit_filter = ' and '.join(map(safe_str, commit_filter))
             revisions = self._remote.rev_range([commit_filter])
             collection_generator = MercurialIndexBasedCollectionGenerator
         else:
             revisions = self.commit_ids
 
         if start_pos or end_pos:
             revisions = revisions[start_pos:end_pos]
 
         return collection_generator(self, revisions, pre_load=pre_load)
 
     def pull(self, url, commit_ids=None):
         """
         Pull changes from external location.
 
         :param commit_ids: Optional. Can be set to a list of commit ids
            which shall be pulled from the other repository.
         """
         url = self._get_url(url)
         self._remote.pull(url, commit_ids=commit_ids)
         self._remote.invalidate_vcs_cache()
 
     def fetch(self, url, commit_ids=None):
         """
         Backward compatibility with GIT fetch==pull
         """
         return self.pull(url, commit_ids=commit_ids)
 
     def push(self, url):
         url = self._get_url(url)
         self._remote.sync_push(url)
 
     def _local_clone(self, clone_path):
         """
         Create a local clone of the current repo.
         """
         self._remote.clone(self.path, clone_path, update_after_clone=True,
                            hooks=False)
 
     def _update(self, revision, clean=False):
         """
         Update the working copy to the specified revision.
         """
         log.debug('Doing checkout to commit: `%s` for %s', revision, self)
         self._remote.update(revision, clean=clean)
 
     def _identify(self):
         """
         Return the current state of the working directory.
         """
         return self._remote.identify().strip().rstrip('+')
 
     def _heads(self, branch=None):
         """
         Return the commit ids of the repository heads.
         """
         return self._remote.heads(branch=branch).strip().split(' ')
 
     def _ancestor(self, revision1, revision2):
         """
         Return the common ancestor of the two revisions.
         """
         return self._remote.ancestor(revision1, revision2)
 
     def _local_push(
             self, revision, repository_path, push_branches=False,
             enable_hooks=False):
         """
         Push the given revision to the specified repository.
 
         :param push_branches: allow to create branches in the target repo.
         """
         self._remote.push(
             [revision], repository_path, hooks=enable_hooks,
             push_branches=push_branches)
 
     def _local_merge(self, target_ref, merge_message, user_name, user_email,
                      source_ref, use_rebase=False, dry_run=False):
         """
         Merge the given source_revision into the checked out revision.
 
         Returns the commit id of the merge and a boolean indicating if the
         commit needs to be pushed.
         """
         self._update(target_ref.commit_id, clean=True)
 
         ancestor = self._ancestor(target_ref.commit_id, source_ref.commit_id)
         is_the_same_branch = self._is_the_same_branch(target_ref, source_ref)
 
         if ancestor == source_ref.commit_id:
             # Nothing to do, the changes were already integrated
             return target_ref.commit_id, False
 
         elif ancestor == target_ref.commit_id and is_the_same_branch:
             # In this case we should force a commit message
             return source_ref.commit_id, True
 
+        unresolved = None
         if use_rebase:
             try:
                 bookmark_name = 'rcbook%s%s' % (source_ref.commit_id,
                                                 target_ref.commit_id)
                 self.bookmark(bookmark_name, revision=source_ref.commit_id)
                 self._remote.rebase(
                     source=source_ref.commit_id, dest=target_ref.commit_id)
                 self._remote.invalidate_vcs_cache()
                 self._update(bookmark_name, clean=True)
                 return self._identify(), True
-            except RepositoryError:
+            except RepositoryError as e:
                 # The rebase-abort may raise another exception which 'hides'
                 # the original one, therefore we log it here.
                 log.exception('Error while rebasing shadow repo during merge.')
+                if 'unresolved conflicts' in e.message:
+                    unresolved = self._remote.get_unresolved_files()
+                    log.debug('unresolved files: %s', unresolved)
 
                 # Cleanup any rebase leftovers
                 self._remote.invalidate_vcs_cache()
                 self._remote.rebase(abort=True)
                 self._remote.invalidate_vcs_cache()
                 self._remote.update(clean=True)
+                if unresolved:
+                    raise UnresolvedFilesInRepo(unresolved)
+                else:
                 raise
         else:
             try:
                 self._remote.merge(source_ref.commit_id)
                 self._remote.invalidate_vcs_cache()
                 self._remote.commit(
                     message=safe_str(merge_message),
                     username=safe_str('%s <%s>' % (user_name, user_email)))
                 self._remote.invalidate_vcs_cache()
                 return self._identify(), True
-            except RepositoryError:
+            except RepositoryError as e:
+                # The merge-abort may raise another exception which 'hides'
+                # the original one, therefore we log it here.
+                log.exception('Error while merging shadow repo during merge.')
+                if 'unresolved merge conflicts' in e.message:
+                    unresolved = self._remote.get_unresolved_files()
+                    log.debug('unresolved files: %s', unresolved)
+
                 # Cleanup any merge leftovers
                 self._remote.update(clean=True)
+                if unresolved:
+                    raise UnresolvedFilesInRepo(unresolved)
+                else:
                 raise
 
     def _local_close(self, target_ref, user_name, user_email,
                      source_ref, close_message=''):
         """
         Close the branch of the given source_revision
 
         Returns the commit id of the close and a boolean indicating if the
         commit needs to be pushed.
         """
         self._update(source_ref.commit_id)
         message = close_message or "Closing branch: `{}`".format(source_ref.name)
         try:
             self._remote.commit(
                 message=safe_str(message),
                 username=safe_str('%s <%s>' % (user_name, user_email)),
                 close_branch=True)
             self._remote.invalidate_vcs_cache()
             return self._identify(), True
         except RepositoryError:
             # Cleanup any commit leftovers
             self._remote.update(clean=True)
             raise
 
     def _is_the_same_branch(self, target_ref, source_ref):
         return (
             self._get_branch_name(target_ref) ==
             self._get_branch_name(source_ref))
 
     def _get_branch_name(self, ref):
         if ref.type == 'branch':
             return ref.name
         return self._remote.ctx_branch(ref.commit_id)
 
     def _maybe_prepare_merge_workspace(
             self, repo_id, workspace_id, unused_target_ref, unused_source_ref):
         shadow_repository_path = self._get_shadow_repository_path(
             self.path, repo_id, workspace_id)
         if not os.path.exists(shadow_repository_path):
             self._local_clone(shadow_repository_path)
             log.debug(
                 'Prepared shadow repository in %s', shadow_repository_path)
 
         return shadow_repository_path
 
     def _merge_repo(self, repo_id, workspace_id, target_ref,
                     source_repo, source_ref, merge_message,
                     merger_name, merger_email, dry_run=False,
                     use_rebase=False, close_branch=False):
 
         log.debug('Executing merge_repo with %s strategy, dry_run mode:%s',
                   'rebase' if use_rebase else 'merge', dry_run)
         if target_ref.commit_id not in self._heads():
             return MergeResponse(
                 False, False, None, MergeFailureReason.TARGET_IS_NOT_HEAD,
                 metadata={'target_ref': target_ref})
 
         try:
             if target_ref.type == 'branch' and len(self._heads(target_ref.name)) != 1:
                 heads = '\n,'.join(self._heads(target_ref.name))
                 metadata = {
                     'target_ref': target_ref,
                     'source_ref': source_ref,
                     'heads': heads
                 }
                 return MergeResponse(
                     False, False, None,
                     MergeFailureReason.HG_TARGET_HAS_MULTIPLE_HEADS,
                     metadata=metadata)
         except CommitDoesNotExistError:
             log.exception('Failure when looking up branch heads on hg target')
             return MergeResponse(
                 False, False, None, MergeFailureReason.MISSING_TARGET_REF,
                 metadata={'target_ref': target_ref})
 
         shadow_repository_path = self._maybe_prepare_merge_workspace(
             repo_id, workspace_id, target_ref, source_ref)
         shadow_repo = self.get_shadow_instance(shadow_repository_path)
 
         log.debug('Pulling in target reference %s', target_ref)
         self._validate_pull_reference(target_ref)
         shadow_repo._local_pull(self.path, target_ref)
 
         try:
             log.debug('Pulling in source reference %s', source_ref)
             source_repo._validate_pull_reference(source_ref)
             shadow_repo._local_pull(source_repo.path, source_ref)
         except CommitDoesNotExistError:
             log.exception('Failure when doing local pull on hg shadow repo')
             return MergeResponse(
                 False, False, None, MergeFailureReason.MISSING_SOURCE_REF,
                 metadata={'source_ref': source_ref})
 
         merge_ref = None
         merge_commit_id = None
         close_commit_id = None
         merge_failure_reason = MergeFailureReason.NONE
         metadata = {}
 
         # enforce that close branch should be used only in case we source from
         # an actual Branch
         close_branch = close_branch and source_ref.type == 'branch'
 
         # don't allow to close branch if source and target are the same
         close_branch = close_branch and source_ref.name != target_ref.name
 
         needs_push_on_close = False
         if close_branch and not use_rebase and not dry_run:
             try:
                 close_commit_id, needs_push_on_close = shadow_repo._local_close(
                     target_ref, merger_name, merger_email, source_ref)
                 merge_possible = True
             except RepositoryError:
                 log.exception('Failure when doing close branch on '
                               'shadow repo: %s', shadow_repo)
                 merge_possible = False
                 merge_failure_reason = MergeFailureReason.MERGE_FAILED
         else:
             merge_possible = True
 
         needs_push = False
         if merge_possible:
             try:
                 merge_commit_id, needs_push = shadow_repo._local_merge(
                     target_ref, merge_message, merger_name, merger_email,
                     source_ref, use_rebase=use_rebase, dry_run=dry_run)
                 merge_possible = True
 
                 # read the state of the close action, if it
                 # maybe required a push
                 needs_push = needs_push or needs_push_on_close
 
                 # Set a bookmark pointing to the merge commit. This bookmark
                 # may be used to easily identify the last successful merge
                 # commit in the shadow repository.
                 shadow_repo.bookmark('pr-merge', revision=merge_commit_id)
                 merge_ref = Reference('book', 'pr-merge', merge_commit_id)
             except SubrepoMergeError:
                 log.exception(
                     'Subrepo merge error during local merge on hg shadow repo.')
                 merge_possible = False
                 merge_failure_reason = MergeFailureReason.SUBREPO_MERGE_FAILED
                 needs_push = False
-            except RepositoryError:
+            except RepositoryError as e:
                 log.exception('Failure when doing local merge on hg shadow repo')
+                if isinstance(e, UnresolvedFilesInRepo):
+                    metadata['unresolved_files'] = 'file: ' + (', file: '.join(e.args[0]))
+
                 merge_possible = False
                 merge_failure_reason = MergeFailureReason.MERGE_FAILED
                 needs_push = False
 
         if merge_possible and not dry_run:
             if needs_push:
                 # In case the target is a bookmark, update it, so after pushing
                 # the bookmarks is also updated in the target.
                 if target_ref.type == 'book':
                     shadow_repo.bookmark(
                         target_ref.name, revision=merge_commit_id)
                 try:
                     shadow_repo_with_hooks = self.get_shadow_instance(
                         shadow_repository_path,
                         enable_hooks=True)
                     # This is the actual merge action, we push from shadow
                     # into origin.
                     # Note: the push_branches option will push any new branch
                     # defined in the source repository to the target. This may
                     # be dangerous as branches are permanent in Mercurial.
                     # This feature was requested in issue #441.
                     shadow_repo_with_hooks._local_push(
                         merge_commit_id, self.path, push_branches=True,
                         enable_hooks=True)
 
                     # maybe we also need to push the close_commit_id
                     if close_commit_id:
                         shadow_repo_with_hooks._local_push(
                             close_commit_id, self.path, push_branches=True,
                             enable_hooks=True)
                     merge_succeeded = True
                 except RepositoryError:
                     log.exception(
                         'Failure when doing local push from the shadow '
                         'repository to the target repository at %s.', self.path)
                     merge_succeeded = False
                     merge_failure_reason = MergeFailureReason.PUSH_FAILED
                     metadata['target'] = 'hg shadow repo'
                     metadata['merge_commit'] = merge_commit_id
             else:
                 merge_succeeded = True
         else:
             merge_succeeded = False
 
         return MergeResponse(
             merge_possible, merge_succeeded, merge_ref, merge_failure_reason,
             metadata=metadata)
 
     def get_shadow_instance(self, shadow_repository_path, enable_hooks=False, cache=False):
         config = self.config.copy()
         if not enable_hooks:
             config.clear_section('hooks')
         return MercurialRepository(shadow_repository_path, config, with_wire={"cache": cache})
 
     def _validate_pull_reference(self, reference):
         if not (reference.name in self.bookmarks or
                 reference.name in self.branches or
                 self.get_commit(reference.commit_id)):
             raise CommitDoesNotExistError(
                 'Unknown branch, bookmark or commit id')
 
     def _local_pull(self, repository_path, reference):
         """
         Fetch a branch, bookmark or commit from a local repository.
         """
         repository_path = os.path.abspath(repository_path)
         if repository_path == self.path:
             raise ValueError('Cannot pull from the same repository')
 
         reference_type_to_option_name = {
             'book': 'bookmark',
             'branch': 'branch',
         }
         option_name = reference_type_to_option_name.get(
             reference.type, 'revision')
 
         if option_name == 'revision':
             ref = reference.commit_id
         else:
             ref = reference.name
 
         options = {option_name: [ref]}
         self._remote.pull_cmd(repository_path, hooks=False, **options)
         self._remote.invalidate_vcs_cache()
 
     def bookmark(self, bookmark, revision=None):
         if isinstance(bookmark, unicode):
             bookmark = safe_str(bookmark)
         self._remote.bookmark(bookmark, revision=revision)
         self._remote.invalidate_vcs_cache()
 
     def get_path_permissions(self, username):
         hgacl_file = os.path.join(self.path, '.hg/hgacl')
 
         def read_patterns(suffix):
             svalue = None
             for section, option in [
                 ('narrowacl', username + suffix),
                 ('narrowacl', 'default' + suffix),
                 ('narrowhgacl', username + suffix),
                 ('narrowhgacl', 'default' + suffix)
             ]:
                 try:
                     svalue = hgacl.get(section, option)
                     break  # stop at the first value we find
                 except configparser.NoOptionError:
                     pass
             if not svalue:
                 return None
             result = ['/']
             for pattern in svalue.split():
                 result.append(pattern)
                 if '*' not in pattern and '?' not in pattern:
                     result.append(pattern + '/*')
             return result
 
         if os.path.exists(hgacl_file):
             try:
                 hgacl = configparser.RawConfigParser()
                 hgacl.read(hgacl_file)
 
                 includes = read_patterns('.includes')
                 excludes = read_patterns('.excludes')
                 return BasePathPermissionChecker.create_from_patterns(
                     includes, excludes)
             except BaseException as e:
                 msg = 'Cannot read ACL settings from {} on {}: {}'.format(
                     hgacl_file, self.name, e)
                 raise exceptions.RepositoryRequirementError(msg)
         else:
             return None
 
 
 class MercurialIndexBasedCollectionGenerator(CollectionGenerator):
 
     def _commit_factory(self, commit_id):
         return self.repo.get_commit(
             commit_idx=commit_id, pre_load=self.pre_load)

rhodecode/lib/vcs/exceptions.py

@@ -1,226 +1,230 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 Custom vcs exceptions module.
 """
 import logging
 import functools
 import urllib2
 import rhodecode
 from pyramid import compat
 
 log = logging.getLogger(__name__)
 
 
 class VCSCommunicationError(Exception):
     pass
 
 
 class HttpVCSCommunicationError(VCSCommunicationError):
     pass
 
 
 class VCSError(Exception):
     pass
 
 
 class RepositoryError(VCSError):
     pass
 
 
 class RepositoryRequirementError(RepositoryError):
     pass
 
 
+class UnresolvedFilesInRepo(RepositoryError):
+    pass
+
+
 class VCSBackendNotSupportedError(VCSError):
     """
     Exception raised when VCSServer does not support requested backend
     """
 
 
 class EmptyRepositoryError(RepositoryError):
     pass
 
 
 class TagAlreadyExistError(RepositoryError):
     pass
 
 
 class TagDoesNotExistError(RepositoryError):
     pass
 
 
 class BranchAlreadyExistError(RepositoryError):
     pass
 
 
 class BranchDoesNotExistError(RepositoryError):
     pass
 
 
 class CommitError(RepositoryError):
     """
     Exceptions related to an existing commit
     """
 
 
 class CommitDoesNotExistError(CommitError):
     pass
 
 
 class CommittingError(RepositoryError):
     """
     Exceptions happening while creating a new commit
     """
 
 
 class NothingChangedError(CommittingError):
     pass
 
 
 class NodeError(VCSError):
     pass
 
 
 class RemovedFileNodeError(NodeError):
     pass
 
 
 class NodeAlreadyExistsError(CommittingError):
     pass
 
 
 class NodeAlreadyChangedError(CommittingError):
     pass
 
 
 class NodeDoesNotExistError(CommittingError):
     pass
 
 
 class NodeNotChangedError(CommittingError):
     pass
 
 
 class NodeAlreadyAddedError(CommittingError):
     pass
 
 
 class NodeAlreadyRemovedError(CommittingError):
     pass
 
 
 class SubrepoMergeError(RepositoryError):
     """
     This happens if we try to merge a repository which contains subrepos and
     the subrepos cannot be merged. The subrepos are not merged itself but
     their references in the root repo are merged.
     """
 
 
 class ImproperArchiveTypeError(VCSError):
     pass
 
 
 class CommandError(VCSError):
     pass
 
 
 class UnhandledException(VCSError):
     """
     Signals that something unexpected went wrong.
 
     This usually means we have a programming error on the side of the VCSServer
     and should inspect the logfile of the VCSServer to find more details.
     """
 
 
 _EXCEPTION_MAP = {
     'abort': RepositoryError,
     'archive': ImproperArchiveTypeError,
     'error': RepositoryError,
     'lookup': CommitDoesNotExistError,
     'repo_locked': RepositoryError,
     'requirement': RepositoryRequirementError,
     'unhandled': UnhandledException,
     # TODO: johbo: Define our own exception for this and stop abusing
     # urllib's exception class.
     'url_error': urllib2.URLError,
     'subrepo_merge_error': SubrepoMergeError,
 }
 
 
 def map_vcs_exceptions(func):
     """
     Utility to decorate functions so that plain exceptions are translated.
 
     The translation is based on `exc_map` which maps a `str` indicating
     the error type into an exception class representing this error inside
     of the vcs layer.
     """
 
     @functools.wraps(func)
     def wrapper(*args, **kwargs):
         try:
             return func(*args, **kwargs)
         except Exception as e:
             from rhodecode.lib.utils2 import str2bool
             debug = str2bool(rhodecode.CONFIG.get('debug'))
 
             # The error middleware adds information if it finds
             # __traceback_info__ in a frame object. This way the remote
             # traceback information is made available in error reports.
             remote_tb = getattr(e, '_vcs_server_traceback', None)
             org_remote_tb = getattr(e, '_vcs_server_org_exc_tb', '')
             __traceback_info__ = None
             if remote_tb:
                 if isinstance(remote_tb, compat.string_types):
                     remote_tb = [remote_tb]
                 __traceback_info__ = (
                     'Found VCSServer remote traceback information:\n'
                     '{}\n'
                     '+++ BEG SOURCE EXCEPTION +++\n\n'
                     '{}\n'
                     '+++ END SOURCE EXCEPTION +++\n'
                     ''.format('\n'.join(remote_tb), org_remote_tb)
                 )
 
                 # Avoid that remote_tb also appears in the frame
                 del remote_tb
 
             # Special vcs errors had an attribute "_vcs_kind" which is used
             # to translate them to the proper exception class in the vcs
             # client layer.
             kind = getattr(e, '_vcs_kind', None)
 
             if kind:
                 if any(e.args):
                     args = e.args
                 else:
                     args = [__traceback_info__ or 'unhandledException']
                 if debug or __traceback_info__ and kind not in ['unhandled', 'lookup']:
                     # for other than unhandled errors also log the traceback
                     # can be useful for debugging
                     log.error(__traceback_info__)
                 raise _EXCEPTION_MAP[kind](*args)
             else:
                 raise
     return wrapper

rhodecode/model/pull_request.py

@@ -1,1759 +1,1760 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2012-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 
 """
 pull request model for RhodeCode
 """
 
 
 import json
 import logging
 import datetime
 import urllib
 import collections
 
 from pyramid import compat
 from pyramid.threadlocal import get_current_request
 
 from rhodecode import events
 from rhodecode.translation import lazy_ugettext
 from rhodecode.lib import helpers as h, hooks_utils, diffs
 from rhodecode.lib import audit_logger
 from rhodecode.lib.compat import OrderedDict
 from rhodecode.lib.hooks_daemon import prepare_callback_daemon
 from rhodecode.lib.markup_renderer import (
     DEFAULT_COMMENTS_RENDERER, RstTemplateRenderer)
 from rhodecode.lib.utils2 import safe_unicode, safe_str, md5_safe
 from rhodecode.lib.vcs.backends.base import (
     Reference, MergeResponse, MergeFailureReason, UpdateFailureReason)
 from rhodecode.lib.vcs.conf import settings as vcs_settings
 from rhodecode.lib.vcs.exceptions import (
     CommitDoesNotExistError, EmptyRepositoryError)
 from rhodecode.model import BaseModel
 from rhodecode.model.changeset_status import ChangesetStatusModel
 from rhodecode.model.comment import CommentsModel
 from rhodecode.model.db import (
     or_, String, cast, PullRequest, PullRequestReviewers, ChangesetStatus,
     PullRequestVersion, ChangesetComment, Repository, RepoReviewRule)
 from rhodecode.model.meta import Session
 from rhodecode.model.notification import NotificationModel, \
     EmailNotificationModel
 from rhodecode.model.scm import ScmModel
 from rhodecode.model.settings import VcsSettingsModel
 
 
 log = logging.getLogger(__name__)
 
 
 # Data structure to hold the response data when updating commits during a pull
 # request update.
 UpdateResponse = collections.namedtuple('UpdateResponse', [
     'executed', 'reason', 'new', 'old', 'changes',
     'source_changed', 'target_changed'])
 
 
 class PullRequestModel(BaseModel):
 
     cls = PullRequest
 
     DIFF_CONTEXT = diffs.DEFAULT_CONTEXT
 
     UPDATE_STATUS_MESSAGES = {
         UpdateFailureReason.NONE: lazy_ugettext(
             'Pull request update successful.'),
         UpdateFailureReason.UNKNOWN: lazy_ugettext(
             'Pull request update failed because of an unknown error.'),
         UpdateFailureReason.NO_CHANGE: lazy_ugettext(
             'No update needed because the source and target have not changed.'),
         UpdateFailureReason.WRONG_REF_TYPE: lazy_ugettext(
             'Pull request cannot be updated because the reference type is '
             'not supported for an update. Only Branch, Tag or Bookmark is allowed.'),
         UpdateFailureReason.MISSING_TARGET_REF: lazy_ugettext(
             'This pull request cannot be updated because the target '
             'reference is missing.'),
         UpdateFailureReason.MISSING_SOURCE_REF: lazy_ugettext(
             'This pull request cannot be updated because the source '
             'reference is missing.'),
     }
     REF_TYPES = ['bookmark', 'book', 'tag', 'branch']
     UPDATABLE_REF_TYPES = ['bookmark', 'book', 'branch']
 
     def __get_pull_request(self, pull_request):
         return self._get_instance((
             PullRequest, PullRequestVersion), pull_request)
 
     def _check_perms(self, perms, pull_request, user, api=False):
         if not api:
             return h.HasRepoPermissionAny(*perms)(
                 user=user, repo_name=pull_request.target_repo.repo_name)
         else:
             return h.HasRepoPermissionAnyApi(*perms)(
                 user=user, repo_name=pull_request.target_repo.repo_name)
 
     def check_user_read(self, pull_request, user, api=False):
         _perms = ('repository.admin', 'repository.write', 'repository.read',)
         return self._check_perms(_perms, pull_request, user, api)
 
     def check_user_merge(self, pull_request, user, api=False):
         _perms = ('repository.admin', 'repository.write', 'hg.admin',)
         return self._check_perms(_perms, pull_request, user, api)
 
     def check_user_update(self, pull_request, user, api=False):
         owner = user.user_id == pull_request.user_id
         return self.check_user_merge(pull_request, user, api) or owner
 
     def check_user_delete(self, pull_request, user):
         owner = user.user_id == pull_request.user_id
         _perms = ('repository.admin',)
         return self._check_perms(_perms, pull_request, user) or owner
 
     def check_user_change_status(self, pull_request, user, api=False):
         reviewer = user.user_id in [x.user_id for x in
                                     pull_request.reviewers]
         return self.check_user_update(pull_request, user, api) or reviewer
 
     def check_user_comment(self, pull_request, user):
         owner = user.user_id == pull_request.user_id
         return self.check_user_read(pull_request, user) or owner
 
     def get(self, pull_request):
         return self.__get_pull_request(pull_request)
 
     def _prepare_get_all_query(self, repo_name, 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.filter(or_(
                 cast(PullRequest.pull_request_id, String).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)
 
         if order_by:
             order_map = {
                 'name_raw': PullRequest.pull_request_id,
                 'id': PullRequest.pull_request_id,
                 'title': PullRequest.title,
                 'updated_on_raw': PullRequest.updated_on,
                 'target_repo': PullRequest.target_repo_id
             }
             if order_dir == 'asc':
                 q = q.order_by(order_map[order_by].asc())
             else:
                 q = q.order_by(order_map[order_by].desc())
 
         return q
 
     def count_all(self, repo_name, 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, source=False, statuses=None,
                               opened_by=None):
         """
         Count the number of pull requests for a specific repository that are
         awaiting review.
 
         :param repo_name: target or source repo
         :param 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
         """
         pull_requests = self.get_awaiting_review(
             repo_name, search_q=search_q, source=source, statuses=statuses, opened_by=opened_by)
 
         return len(pull_requests)
 
     def get_awaiting_review(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 that are awaiting
         review.
 
         :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
         """
         pull_requests = self.get_all(
             repo_name, search_q=search_q, source=source, statuses=statuses,
             opened_by=opened_by, order_by=order_by, order_dir=order_dir)
 
         _filtered_pull_requests = []
         for pr in pull_requests:
             status = pr.calculated_review_status()
             if status in [ChangesetStatus.STATUS_NOT_REVIEWED,
                           ChangesetStatus.STATUS_UNDER_REVIEW]:
                 _filtered_pull_requests.append(pr)
         if length:
             return _filtered_pull_requests[offset:offset+length]
         else:
             return _filtered_pull_requests
 
     def count_awaiting_my_review(self, repo_name, search_q=None, source=False, statuses=None,
                                  opened_by=None, user_id=None):
         """
         Count the number of pull requests for a specific repository that are
         awaiting review from a specific user.
 
         :param repo_name: target or source repo
         :param 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 user_id: reviewer user of the pull request
         :returns: int number of pull requests
         """
         pull_requests = self.get_awaiting_my_review(
             repo_name, search_q=search_q, source=source, statuses=statuses,
             opened_by=opened_by, user_id=user_id)
 
         return len(pull_requests)
 
     def get_awaiting_my_review(self, repo_name, search_q=None, source=False, statuses=None,
                                opened_by=None, user_id=None, offset=0,
                                length=None, order_by=None, order_dir='desc'):
         """
         Get all pull requests for a specific repository that are awaiting
         review from a specific user.
 
         :param repo_name: target or source repo
         :param 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 user_id: reviewer user of the pull request
         :param offset: pagination offset
         :param length: length of returned list
         :param order_by: order of the returned list
         :param order_dir: 'asc' or 'desc' ordering direction
         :returns: list of pull requests
         """
         pull_requests = self.get_all(
             repo_name, search_q=search_q, source=source, statuses=statuses,
             opened_by=opened_by, order_by=order_by, order_dir=order_dir)
 
         _my = PullRequestModel().get_not_reviewed(user_id)
         my_participation = []
         for pr in pull_requests:
             if pr in _my:
                 my_participation.append(pr)
         _filtered_pull_requests = my_participation
         if length:
             return _filtered_pull_requests[offset:offset+length]
         else:
             return _filtered_pull_requests
 
     def get_not_reviewed(self, user_id):
         return [
             x.pull_request for x in PullRequestReviewers.query().filter(
                 PullRequestReviewers.user_id == user_id).all()
         ]
 
     def _prepare_participating_query(self, user_id=None, statuses=None,
                                      order_by=None, order_dir='desc'):
         q = PullRequest.query()
         if user_id:
             reviewers_subquery = Session().query(
                 PullRequestReviewers.pull_request_id).filter(
                 PullRequestReviewers.user_id == user_id).subquery()
             user_filter = or_(
                 PullRequest.user_id == user_id,
                 PullRequest.pull_request_id.in_(reviewers_subquery)
             )
             q = PullRequest.query().filter(user_filter)
 
         # closed,opened
         if statuses:
             q = q.filter(PullRequest.status.in_(statuses))
 
         if order_by:
             order_map = {
                 'name_raw': PullRequest.pull_request_id,
                 'title': PullRequest.title,
                 'updated_on_raw': PullRequest.updated_on,
                 'target_repo': PullRequest.target_repo_id
             }
             if order_dir == 'asc':
                 q = q.order_by(order_map[order_by].asc())
             else:
                 q = q.order_by(order_map[order_by].desc())
 
         return q
 
     def count_im_participating_in(self, user_id=None, statuses=None):
         q = self._prepare_participating_query(user_id, statuses=statuses)
         return q.count()
 
     def get_im_participating_in(
             self, user_id=None, statuses=None, offset=0,
             length=None, order_by=None, order_dir='desc'):
         """
         Get all Pull requests that i'm participating in, or i have opened
         """
 
         q = self._prepare_participating_query(
             user_id, statuses=statuses, order_by=order_by,
             order_dir=order_dir)
 
         if length:
             pull_requests = q.limit(length).offset(offset).all()
         else:
             pull_requests = q.all()
 
         return pull_requests
 
     def get_versions(self, pull_request):
         """
         returns version of pull request sorted by ID descending
         """
         return PullRequestVersion.query()\
             .filter(PullRequestVersion.pull_request == pull_request)\
             .order_by(PullRequestVersion.pull_request_version_id.asc())\
             .all()
 
     def get_pr_version(self, pull_request_id, version=None):
         at_version = None
 
         if version and version == 'latest':
             pull_request_ver = PullRequest.get(pull_request_id)
             pull_request_obj = pull_request_ver
             _org_pull_request_obj = pull_request_obj
             at_version = 'latest'
         elif version:
             pull_request_ver = PullRequestVersion.get_or_404(version)
             pull_request_obj = pull_request_ver
             _org_pull_request_obj = pull_request_ver.pull_request
             at_version = pull_request_ver.pull_request_version_id
         else:
             _org_pull_request_obj = pull_request_obj = PullRequest.get_or_404(
                 pull_request_id)
 
         pull_request_display_obj = PullRequest.get_pr_display_object(
             pull_request_obj, _org_pull_request_obj)
 
         return _org_pull_request_obj, pull_request_obj, \
                pull_request_display_obj, at_version
 
     def create(self, created_by, source_repo, source_ref, target_repo,
                target_ref, revisions, reviewers, title, description=None,
                description_renderer=None,
                reviewer_data=None, translator=None, auth_user=None):
         translator = translator or get_current_request().translate
 
         created_by_user = self._get_user(created_by)
         auth_user = auth_user or created_by_user.AuthUser()
         source_repo = self._get_repo(source_repo)
         target_repo = self._get_repo(target_repo)
 
         pull_request = PullRequest()
         pull_request.source_repo = source_repo
         pull_request.source_ref = source_ref
         pull_request.target_repo = target_repo
         pull_request.target_ref = target_ref
         pull_request.revisions = revisions
         pull_request.title = title
         pull_request.description = description
         pull_request.description_renderer = description_renderer
         pull_request.author = created_by_user
         pull_request.reviewer_data = reviewer_data
         pull_request.pull_request_state = pull_request.STATE_CREATING
         Session().add(pull_request)
         Session().flush()
 
         reviewer_ids = set()
         # members / reviewers
         for reviewer_object in reviewers:
             user_id, reasons, mandatory, rules = reviewer_object
             user = self._get_user(user_id)
 
             # skip duplicates
             if user.user_id in reviewer_ids:
                 continue
 
             reviewer_ids.add(user.user_id)
 
             reviewer = PullRequestReviewers()
             reviewer.user = user
             reviewer.pull_request = pull_request
             reviewer.reasons = reasons
             reviewer.mandatory = mandatory
 
             # NOTE(marcink): pick only first rule for now
             rule_id = list(rules)[0] if rules else None
             rule = RepoReviewRule.get(rule_id) if rule_id else None
             if rule:
                 review_group = rule.user_group_vote_rule(user_id)
                 # we check if this particular reviewer is member of a voting group
                 if review_group:
                     # NOTE(marcink):
                     # can be that user is member of more but we pick the first same,
                     # same as default reviewers algo
                     review_group = review_group[0]
 
                     rule_data = {
                         'rule_name':
                             rule.review_rule_name,
                         'rule_user_group_entry_id':
                             review_group.repo_review_rule_users_group_id,
                         'rule_user_group_name':
                             review_group.users_group.users_group_name,
                         'rule_user_group_members':
                             [x.user.username for x in review_group.users_group.members],
                         'rule_user_group_members_id':
                             [x.user.user_id for x in review_group.users_group.members],
                     }
                     # e.g {'vote_rule': -1, 'mandatory': True}
                     rule_data.update(review_group.rule_data())
 
                     reviewer.rule_data = rule_data
 
             Session().add(reviewer)
             Session().flush()
 
         # Set approval status to "Under Review" for all commits which are
         # part of this pull request.
         ChangesetStatusModel().set_status(
             repo=target_repo,
             status=ChangesetStatus.STATUS_UNDER_REVIEW,
             user=created_by_user,
             pull_request=pull_request
         )
         # we commit early at this point. This has to do with a fact
         # that before queries do some row-locking. And because of that
         # we need to commit and finish transaction before below validate call
         # that for large repos could be long resulting in long row locks
         Session().commit()
 
         # prepare workspace, and run initial merge simulation. Set state during that
         # operation
         pull_request = PullRequest.get(pull_request.pull_request_id)
 
         # set as merging, for 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)
         self.trigger_pull_request_hook(
             pull_request, created_by_user, 'create')
 
         creation_data = pull_request.get_api_data(with_merge_state=False)
         self._log_audit_action(
             'repo.pull_request.create', {'data': creation_data},
             auth_user, pull_request)
 
         return pull_request
 
     def trigger_pull_request_hook(self, pull_request, user, action, data=None):
         pull_request = self.__get_pull_request(pull_request)
         target_scm = pull_request.target_repo.scm_instance()
         if action == 'create':
             trigger_hook = hooks_utils.trigger_log_create_pull_request_hook
         elif action == 'merge':
             trigger_hook = hooks_utils.trigger_log_merge_pull_request_hook
         elif action == 'close':
             trigger_hook = hooks_utils.trigger_log_close_pull_request_hook
         elif action == 'review_status_change':
             trigger_hook = hooks_utils.trigger_log_review_pull_request_hook
         elif action == 'update':
             trigger_hook = hooks_utils.trigger_log_update_pull_request_hook
         elif action == 'comment':
             # dummy hook ! for comment. We want this function to handle all cases
             def trigger_hook(*args, **kwargs):
                 pass
             comment = data['comment']
             events.trigger(events.PullRequestCommentEvent(pull_request, comment))
         else:
             return
 
         trigger_hook(
             username=user.username,
             repo_name=pull_request.target_repo.repo_name,
             repo_alias=target_scm.alias,
             pull_request=pull_request,
             data=data)
 
     def _get_commit_ids(self, pull_request):
         """
         Return the commit ids of the merged pull request.
 
         This method is not dealing correctly yet with the lack of autoupdates
         nor with the implicit target updates.
         For example: if a commit in the source repo is already in the target it
         will be reported anyways.
         """
         merge_rev = pull_request.merge_rev
         if merge_rev is None:
             raise ValueError('This pull request was not merged yet')
 
         commit_ids = list(pull_request.revisions)
         if merge_rev not in commit_ids:
             commit_ids.append(merge_rev)
 
         return commit_ids
 
     def merge_repo(self, pull_request, user, extras):
         log.debug("Merging pull request %s", pull_request.pull_request_id)
         extras['user_agent'] = 'internal-merge'
         merge_state = self._merge_pull_request(pull_request, user, extras)
         if merge_state.executed:
             log.debug("Merge was successful, updating the pull request comments.")
             self._comment_and_close_pr(pull_request, user, merge_state)
 
             self._log_audit_action(
                 'repo.pull_request.merge',
                 {'merge_state': merge_state.__dict__},
                 user, pull_request)
 
         else:
             log.warn("Merge failed, not updating the pull request.")
         return merge_state
 
     def _merge_pull_request(self, pull_request, user, extras, merge_msg=None):
         target_vcs = pull_request.target_repo.scm_instance()
         source_vcs = pull_request.source_repo.scm_instance()
 
         message = safe_unicode(merge_msg or vcs_settings.MERGE_MESSAGE_TMPL).format(
             pr_id=pull_request.pull_request_id,
             pr_title=pull_request.title,
             source_repo=source_vcs.name,
             source_ref_name=pull_request.source_ref_parts.name,
             target_repo=target_vcs.name,
             target_ref_name=pull_request.target_ref_parts.name,
         )
 
         workspace_id = self._workspace_id(pull_request)
         repo_id = pull_request.target_repo.repo_id
         use_rebase = self._use_rebase_for_merging(pull_request)
         close_branch = self._close_branch_before_merging(pull_request)
 
         target_ref = self._refresh_reference(
             pull_request.target_ref_parts, target_vcs)
 
         callback_daemon, extras = prepare_callback_daemon(
             extras, protocol=vcs_settings.HOOKS_PROTOCOL,
             host=vcs_settings.HOOKS_HOST,
             use_direct_calls=vcs_settings.HOOKS_DIRECT_CALLS)
 
         with callback_daemon:
             # TODO: johbo: Implement a clean way to run a config_override
             # for a single call.
             target_vcs.config.set(
                 'rhodecode', 'RC_SCM_DATA', json.dumps(extras))
 
             user_name = user.short_contact
             merge_state = target_vcs.merge(
                 repo_id, workspace_id, target_ref, source_vcs,
                 pull_request.source_ref_parts,
                 user_name=user_name, user_email=user.email,
                 message=message, use_rebase=use_rebase,
                 close_branch=close_branch)
         return merge_state
 
     def _comment_and_close_pr(self, pull_request, user, merge_state, close_msg=None):
         pull_request.merge_rev = merge_state.merge_ref.commit_id
         pull_request.updated_on = datetime.datetime.now()
         close_msg = close_msg or 'Pull request merged and closed'
 
         CommentsModel().create(
             text=safe_unicode(close_msg),
             repo=pull_request.target_repo.repo_id,
             user=user.user_id,
             pull_request=pull_request.pull_request_id,
             f_path=None,
             line_no=None,
             closing_pr=True
         )
 
         Session().add(pull_request)
         Session().flush()
         # TODO: paris: replace invalidation with less radical solution
         ScmModel().mark_for_invalidation(
             pull_request.target_repo.repo_name)
         self.trigger_pull_request_hook(pull_request, user, 'merge')
 
     def has_valid_update_type(self, pull_request):
         source_ref_type = pull_request.source_ref_parts.type
         return source_ref_type in self.REF_TYPES
 
     def update_commits(self, pull_request):
         """
         Get the updated list of commits for the pull request
         and return the new pull request version and the list
         of commits processed by this update action
         """
         pull_request = self.__get_pull_request(pull_request)
         source_ref_type = pull_request.source_ref_parts.type
         source_ref_name = pull_request.source_ref_parts.name
         source_ref_id = pull_request.source_ref_parts.commit_id
 
         target_ref_type = pull_request.target_ref_parts.type
         target_ref_name = pull_request.target_ref_parts.name
         target_ref_id = pull_request.target_ref_parts.commit_id
 
         if not self.has_valid_update_type(pull_request):
             log.debug("Skipping update of pull request %s due to ref type: %s",
                       pull_request, source_ref_type)
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.WRONG_REF_TYPE,
                 old=pull_request, new=None, changes=None,
                 source_changed=False, target_changed=False)
 
         # source repo
         source_repo = pull_request.source_repo.scm_instance()
 
         try:
             source_commit = source_repo.get_commit(commit_id=source_ref_name)
         except CommitDoesNotExistError:
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.MISSING_SOURCE_REF,
                 old=pull_request, new=None, changes=None,
                 source_changed=False, target_changed=False)
 
         source_changed = source_ref_id != source_commit.raw_id
 
         # target repo
         target_repo = pull_request.target_repo.scm_instance()
 
         try:
             target_commit = target_repo.get_commit(commit_id=target_ref_name)
         except CommitDoesNotExistError:
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.MISSING_TARGET_REF,
                 old=pull_request, new=None, changes=None,
                 source_changed=False, target_changed=False)
         target_changed = target_ref_id != target_commit.raw_id
 
         if not (source_changed or target_changed):
             log.debug("Nothing changed in pull request %s", pull_request)
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.NO_CHANGE,
                 old=pull_request, new=None, changes=None,
                 source_changed=target_changed, target_changed=source_changed)
 
         change_in_found = 'target repo' if target_changed else 'source repo'
         log.debug('Updating pull request because of change in %s detected',
                   change_in_found)
 
         # Finally there is a need for an update, in case of source change
         # we create a new version, else just an update
         if source_changed:
             pull_request_version = self._create_version_from_snapshot(pull_request)
             self._link_comments_to_version(pull_request_version)
         else:
             try:
                 ver = pull_request.versions[-1]
             except IndexError:
                 ver = None
 
             pull_request.pull_request_version_id = \
                 ver.pull_request_version_id if ver else None
             pull_request_version = pull_request
 
         try:
             if target_ref_type in self.REF_TYPES:
                 target_commit = target_repo.get_commit(target_ref_name)
             else:
                 target_commit = target_repo.get_commit(target_ref_id)
         except CommitDoesNotExistError:
             return UpdateResponse(
                 executed=False,
                 reason=UpdateFailureReason.MISSING_TARGET_REF,
                 old=pull_request, new=None, changes=None,
                 source_changed=source_changed, target_changed=target_changed)
 
         # re-compute commit ids
         old_commit_ids = pull_request.revisions
         pre_load = ["author", "date", "message", "branch"]
         commit_ranges = target_repo.compare(
             target_commit.raw_id, source_commit.raw_id, source_repo, merge=True,
             pre_load=pre_load)
 
         ancestor = source_repo.get_common_ancestor(
             source_commit.raw_id, target_commit.raw_id, target_repo)
 
         pull_request.source_ref = '%s:%s:%s' % (
             source_ref_type, source_ref_name, source_commit.raw_id)
         pull_request.target_ref = '%s:%s:%s' % (
             target_ref_type, target_ref_name, ancestor)
 
         pull_request.revisions = [
             commit.raw_id for commit in reversed(commit_ranges)]
         pull_request.updated_on = datetime.datetime.now()
         Session().add(pull_request)
         new_commit_ids = pull_request.revisions
 
         old_diff_data, new_diff_data = self._generate_update_diffs(
             pull_request, pull_request_version)
 
         # calculate commit and file changes
         changes = self._calculate_commit_id_changes(
             old_commit_ids, new_commit_ids)
         file_changes = self._calculate_file_changes(
             old_diff_data, new_diff_data)
 
         # set comments as outdated if DIFFS changed
         CommentsModel().outdate_comments(
             pull_request, old_diff_data=old_diff_data,
             new_diff_data=new_diff_data)
 
         commit_changes = (changes.added or changes.removed)
         file_node_changes = (
             file_changes.added or file_changes.modified or file_changes.removed)
         pr_has_changes = commit_changes or file_node_changes
 
         # Add an automatic comment to the pull request, in case
         # anything has changed
         if pr_has_changes:
             update_comment = CommentsModel().create(
                 text=self._render_update_message(changes, file_changes),
                 repo=pull_request.target_repo,
                 user=pull_request.author,
                 pull_request=pull_request,
                 send_email=False, renderer=DEFAULT_COMMENTS_RENDERER)
 
             # Update status to "Under Review" for added commits
             for commit_id in changes.added:
                 ChangesetStatusModel().set_status(
                     repo=pull_request.source_repo,
                     status=ChangesetStatus.STATUS_UNDER_REVIEW,
                     comment=update_comment,
                     user=pull_request.author,
                     pull_request=pull_request,
                     revision=commit_id)
 
         log.debug(
             'Updated pull request %s, added_ids: %s, common_ids: %s, '
             'removed_ids: %s', pull_request.pull_request_id,
             changes.added, changes.common, changes.removed)
         log.debug(
             'Updated pull request with the following file changes: %s',
             file_changes)
 
         log.info(
             "Updated pull request %s from commit %s to commit %s, "
             "stored new version %s of this pull request.",
             pull_request.pull_request_id, source_ref_id,
             pull_request.source_ref_parts.commit_id,
             pull_request_version.pull_request_version_id)
         Session().commit()
         self.trigger_pull_request_hook(pull_request, pull_request.author, 'update')
 
         return UpdateResponse(
             executed=True, reason=UpdateFailureReason.NONE,
             old=pull_request, new=pull_request_version, changes=changes,
             source_changed=source_changed, target_changed=target_changed)
 
     def _create_version_from_snapshot(self, pull_request):
         version = PullRequestVersion()
         version.title = pull_request.title
         version.description = pull_request.description
         version.status = pull_request.status
         version.pull_request_state = pull_request.pull_request_state
         version.created_on = datetime.datetime.now()
         version.updated_on = pull_request.updated_on
         version.user_id = pull_request.user_id
         version.source_repo = pull_request.source_repo
         version.source_ref = pull_request.source_ref
         version.target_repo = pull_request.target_repo
         version.target_ref = pull_request.target_ref
 
         version._last_merge_source_rev = pull_request._last_merge_source_rev
         version._last_merge_target_rev = pull_request._last_merge_target_rev
         version.last_merge_status = pull_request.last_merge_status
         version.shadow_merge_ref = pull_request.shadow_merge_ref
         version.merge_rev = pull_request.merge_rev
         version.reviewer_data = pull_request.reviewer_data
 
         version.revisions = pull_request.revisions
         version.pull_request = pull_request
         Session().add(version)
         Session().flush()
 
         return version
 
     def _generate_update_diffs(self, pull_request, pull_request_version):
 
         diff_context = (
             self.DIFF_CONTEXT +
             CommentsModel.needed_extra_diff_context())
         hide_whitespace_changes = False
         source_repo = pull_request_version.source_repo
         source_ref_id = pull_request_version.source_ref_parts.commit_id
         target_ref_id = pull_request_version.target_ref_parts.commit_id
         old_diff = self._get_diff_from_pr_or_version(
             source_repo, source_ref_id, target_ref_id,
             hide_whitespace_changes=hide_whitespace_changes, diff_context=diff_context)
 
         source_repo = pull_request.source_repo
         source_ref_id = pull_request.source_ref_parts.commit_id
         target_ref_id = pull_request.target_ref_parts.commit_id
 
         new_diff = self._get_diff_from_pr_or_version(
             source_repo, source_ref_id, target_ref_id,
             hide_whitespace_changes=hide_whitespace_changes, diff_context=diff_context)
 
         old_diff_data = diffs.DiffProcessor(old_diff)
         old_diff_data.prepare()
         new_diff_data = diffs.DiffProcessor(new_diff)
         new_diff_data.prepare()
 
         return old_diff_data, new_diff_data
 
     def _link_comments_to_version(self, pull_request_version):
         """
         Link all unlinked comments of this pull request to the given version.
 
         :param pull_request_version: The `PullRequestVersion` to which
             the comments shall be linked.
 
         """
         pull_request = pull_request_version.pull_request
         comments = ChangesetComment.query()\
             .filter(
                 # TODO: johbo: Should we query for the repo at all here?
                 # Pending decision on how comments of PRs are to be related
                 # to either the source repo, the target repo or no repo at all.
                 ChangesetComment.repo_id == pull_request.target_repo.repo_id,
                 ChangesetComment.pull_request == pull_request,
                 ChangesetComment.pull_request_version == None)\
             .order_by(ChangesetComment.comment_id.asc())
 
         # TODO: johbo: Find out why this breaks if it is done in a bulk
         # operation.
         for comment in comments:
             comment.pull_request_version_id = (
                 pull_request_version.pull_request_version_id)
             Session().add(comment)
 
     def _calculate_commit_id_changes(self, old_ids, new_ids):
         added = [x for x in new_ids if x not in old_ids]
         common = [x for x in new_ids if x in old_ids]
         removed = [x for x in old_ids if x not in new_ids]
         total = new_ids
         return ChangeTuple(added, common, removed, total)
 
     def _calculate_file_changes(self, old_diff_data, new_diff_data):
 
         old_files = OrderedDict()
         for diff_data in old_diff_data.parsed_diff:
             old_files[diff_data['filename']] = md5_safe(diff_data['raw_diff'])
 
         added_files = []
         modified_files = []
         removed_files = []
         for diff_data in new_diff_data.parsed_diff:
             new_filename = diff_data['filename']
             new_hash = md5_safe(diff_data['raw_diff'])
 
             old_hash = old_files.get(new_filename)
             if not old_hash:
                 # file is not present in old diff, means it's added
                 added_files.append(new_filename)
             else:
                 if new_hash != old_hash:
                     modified_files.append(new_filename)
                 # now remove a file from old, since we have seen it already
                 del old_files[new_filename]
 
         # removed files is when there are present in old, but not in NEW,
         # since we remove old files that are present in new diff, left-overs
         # if any should be the removed files
         removed_files.extend(old_files.keys())
 
         return FileChangeTuple(added_files, modified_files, removed_files)
 
     def _render_update_message(self, changes, file_changes):
         """
         render the message using DEFAULT_COMMENTS_RENDERER (RST renderer),
         so it's always looking the same disregarding on which default
         renderer system is using.
 
         :param changes: changes named tuple
         :param file_changes: file changes named tuple
 
         """
         new_status = ChangesetStatus.get_status_lbl(
             ChangesetStatus.STATUS_UNDER_REVIEW)
 
         changed_files = (
             file_changes.added + file_changes.modified + file_changes.removed)
 
         params = {
             'under_review_label': new_status,
             'added_commits': changes.added,
             'removed_commits': changes.removed,
             'changed_files': changed_files,
             'added_files': file_changes.added,
             'modified_files': file_changes.modified,
             'removed_files': file_changes.removed,
         }
         renderer = RstTemplateRenderer()
         return renderer.render('pull_request_update.mako', **params)
 
     def edit(self, pull_request, title, description, description_renderer, user):
         pull_request = self.__get_pull_request(pull_request)
         old_data = pull_request.get_api_data(with_merge_state=False)
         if pull_request.is_closed():
             raise ValueError('This pull request is closed')
         if title:
             pull_request.title = title
         pull_request.description = description
         pull_request.updated_on = datetime.datetime.now()
         pull_request.description_renderer = description_renderer
         Session().add(pull_request)
         self._log_audit_action(
             'repo.pull_request.edit', {'old_data': old_data},
             user, pull_request)
 
     def update_reviewers(self, pull_request, reviewer_data, user):
         """
         Update the reviewers in the pull request
 
         :param pull_request: the pr to update
         :param reviewer_data: list of tuples
             [(user, ['reason1', 'reason2'], mandatory_flag, [rules])]
         """
         pull_request = self.__get_pull_request(pull_request)
         if pull_request.is_closed():
             raise ValueError('This pull request is closed')
 
         reviewers = {}
         for user_id, reasons, mandatory, rules in reviewer_data:
             if isinstance(user_id, (int, compat.string_types)):
                 user_id = self._get_user(user_id).user_id
             reviewers[user_id] = {
                 'reasons': reasons, 'mandatory': mandatory}
 
         reviewers_ids = set(reviewers.keys())
         current_reviewers = PullRequestReviewers.query()\
             .filter(PullRequestReviewers.pull_request ==
                     pull_request).all()
         current_reviewers_ids = set([x.user.user_id for x in current_reviewers])
 
         ids_to_add = reviewers_ids.difference(current_reviewers_ids)
         ids_to_remove = current_reviewers_ids.difference(reviewers_ids)
 
         log.debug("Adding %s reviewers", ids_to_add)
         log.debug("Removing %s reviewers", ids_to_remove)
         changed = False
         added_audit_reviewers = []
         removed_audit_reviewers = []
 
         for uid in ids_to_add:
             changed = True
             _usr = self._get_user(uid)
             reviewer = PullRequestReviewers()
             reviewer.user = _usr
             reviewer.pull_request = pull_request
             reviewer.reasons = reviewers[uid]['reasons']
             # NOTE(marcink): mandatory shouldn't be changed now
             # reviewer.mandatory = reviewers[uid]['reasons']
             Session().add(reviewer)
             added_audit_reviewers.append(reviewer.get_dict())
 
         for uid in ids_to_remove:
             changed = True
             # NOTE(marcink): we fetch "ALL" reviewers using .all(). This is an edge case
             # that prevents and fixes cases that we added the same reviewer twice.
             # this CAN happen due to the lack of DB checks
             reviewers = PullRequestReviewers.query()\
                 .filter(PullRequestReviewers.user_id == uid,
                         PullRequestReviewers.pull_request == pull_request)\
                 .all()
 
             for obj in reviewers:
                 added_audit_reviewers.append(obj.get_dict())
                 Session().delete(obj)
 
         if changed:
             Session().expire_all()
             pull_request.updated_on = datetime.datetime.now()
             Session().add(pull_request)
 
         # finally store audit logs
         for user_data in added_audit_reviewers:
             self._log_audit_action(
                 'repo.pull_request.reviewer.add', {'data': user_data},
                 user, pull_request)
         for user_data in removed_audit_reviewers:
             self._log_audit_action(
                 'repo.pull_request.reviewer.delete', {'old_data': user_data},
                 user, pull_request)
 
         self.notify_reviewers(pull_request, ids_to_add)
         return ids_to_add, ids_to_remove
 
     def get_url(self, pull_request, request=None, permalink=False):
         if not request:
             request = get_current_request()
 
         if permalink:
             return request.route_url(
                 'pull_requests_global',
                 pull_request_id=pull_request.pull_request_id,)
         else:
             return request.route_url('pullrequest_show',
                 repo_name=safe_str(pull_request.target_repo.repo_name),
                 pull_request_id=pull_request.pull_request_id,)
 
     def get_shadow_clone_url(self, pull_request, request=None):
         """
         Returns qualified url pointing to the shadow repository. If this pull
         request is closed there is no shadow repository and ``None`` will be
         returned.
         """
         if pull_request.is_closed():
             return None
         else:
             pr_url = urllib.unquote(self.get_url(pull_request, request=request))
             return safe_unicode('{pr_url}/repository'.format(pr_url=pr_url))
 
     def notify_reviewers(self, pull_request, reviewers_ids):
         # notification to reviewers
         if not reviewers_ids:
             return
 
         log.debug('Notify following reviewers about pull-request %s', reviewers_ids)
 
         pull_request_obj = pull_request
         # get the current participants of this pull request
         recipients = reviewers_ids
         notification_type = EmailNotificationModel.TYPE_PULL_REQUEST
 
         pr_source_repo = pull_request_obj.source_repo
         pr_target_repo = pull_request_obj.target_repo
 
         pr_url = h.route_url('pullrequest_show',
                              repo_name=pr_target_repo.repo_name,
                              pull_request_id=pull_request_obj.pull_request_id,)
 
         # set some variables for email notification
         pr_target_repo_url = h.route_url(
             'repo_summary', repo_name=pr_target_repo.repo_name)
 
         pr_source_repo_url = h.route_url(
             'repo_summary', repo_name=pr_source_repo.repo_name)
 
         # pull request specifics
         pull_request_commits = [
             (x.raw_id, x.message)
             for x in map(pr_source_repo.get_commit, pull_request.revisions)]
 
         kwargs = {
             'user': pull_request.author,
             'pull_request': pull_request_obj,
             'pull_request_commits': pull_request_commits,
 
             'pull_request_target_repo': pr_target_repo,
             'pull_request_target_repo_url': pr_target_repo_url,
 
             'pull_request_source_repo': pr_source_repo,
             'pull_request_source_repo_url': pr_source_repo_url,
 
             'pull_request_url': pr_url,
         }
 
         # pre-generate the subject for notification itself
         (subject,
          _h, _e,  # we don't care about those
          body_plaintext) = EmailNotificationModel().render_email(
             notification_type, **kwargs)
 
         # create notification objects, and emails
         NotificationModel().create(
             created_by=pull_request.author,
             notification_subject=subject,
             notification_body=body_plaintext,
             notification_type=notification_type,
             recipients=recipients,
             email_kwargs=kwargs,
         )
 
     def delete(self, pull_request, user):
         pull_request = self.__get_pull_request(pull_request)
         old_data = pull_request.get_api_data(with_merge_state=False)
         self._cleanup_merge_workspace(pull_request)
         self._log_audit_action(
             'repo.pull_request.delete', {'old_data': old_data},
             user, pull_request)
         Session().delete(pull_request)
 
     def close_pull_request(self, pull_request, user):
         pull_request = self.__get_pull_request(pull_request)
         self._cleanup_merge_workspace(pull_request)
         pull_request.status = PullRequest.STATUS_CLOSED
         pull_request.updated_on = datetime.datetime.now()
         Session().add(pull_request)
         self.trigger_pull_request_hook(
             pull_request, pull_request.author, 'close')
 
         pr_data = pull_request.get_api_data(with_merge_state=False)
         self._log_audit_action(
             'repo.pull_request.close', {'data': pr_data}, user, pull_request)
 
     def close_pull_request_with_comment(
             self, pull_request, user, repo, message=None, auth_user=None):
 
         pull_request_review_status = pull_request.calculated_review_status()
 
         if pull_request_review_status == ChangesetStatus.STATUS_APPROVED:
             # approved only if we have voting consent
             status = ChangesetStatus.STATUS_APPROVED
         else:
             status = ChangesetStatus.STATUS_REJECTED
         status_lbl = ChangesetStatus.get_status_lbl(status)
 
         default_message = (
             'Closing with status change {transition_icon} {status}.'
         ).format(transition_icon='>', status=status_lbl)
         text = message or default_message
 
         # create a comment, and link it to new status
         comment = CommentsModel().create(
             text=text,
             repo=repo.repo_id,
             user=user.user_id,
             pull_request=pull_request.pull_request_id,
             status_change=status_lbl,
             status_change_type=status,
             closing_pr=True,
             auth_user=auth_user,
         )
 
         # calculate old status before we change it
         old_calculated_status = pull_request.calculated_review_status()
         ChangesetStatusModel().set_status(
             repo.repo_id,
             status,
             user.user_id,
             comment=comment,
             pull_request=pull_request.pull_request_id
         )
 
         Session().flush()
         events.trigger(events.PullRequestCommentEvent(pull_request, comment))
         # we now calculate the status of pull request again, and based on that
         # calculation trigger status change. This might happen in cases
         # that non-reviewer admin closes a pr, which means his vote doesn't
         # change the status, while if he's a reviewer this might change it.
         calculated_status = pull_request.calculated_review_status()
         if old_calculated_status != calculated_status:
             self.trigger_pull_request_hook(
                 pull_request, user, 'review_status_change',
                 data={'status': calculated_status})
 
         # finally close the PR
         PullRequestModel().close_pull_request(
             pull_request.pull_request_id, user)
 
         return comment, status
 
     def merge_status(self, pull_request, translator=None,
                      force_shadow_repo_refresh=False):
         _ = translator or get_current_request().translate
 
         if not self._is_merge_enabled(pull_request):
             return False, _('Server-side pull request merging is disabled.')
         if pull_request.is_closed():
             return False, _('This pull request is closed.')
         merge_possible, msg = self._check_repo_requirements(
             target=pull_request.target_repo, source=pull_request.source_repo,
             translator=_)
         if not merge_possible:
             return merge_possible, msg
 
         try:
             resp = self._try_merge(
                 pull_request,
                 force_shadow_repo_refresh=force_shadow_repo_refresh)
             log.debug("Merge response: %s", resp)
             status = resp.possible, resp.merge_status_message
         except NotImplementedError:
             status = False, _('Pull request merging is not supported.')
 
         return status
 
     def _check_repo_requirements(self, target, source, translator):
         """
         Check if `target` and `source` have compatible requirements.
 
         Currently this is just checking for largefiles.
         """
         _ = translator
         target_has_largefiles = self._has_largefiles(target)
         source_has_largefiles = self._has_largefiles(source)
         merge_possible = True
         message = u''
 
         if target_has_largefiles != source_has_largefiles:
             merge_possible = False
             if source_has_largefiles:
                 message = _(
                     'Target repository large files support is disabled.')
             else:
                 message = _(
                     'Source repository large files support is disabled.')
 
         return merge_possible, message
 
     def _has_largefiles(self, repo):
         largefiles_ui = VcsSettingsModel(repo=repo).get_ui_settings(
             'extensions', 'largefiles')
         return largefiles_ui and largefiles_ui[0].active
 
     def _try_merge(self, pull_request, force_shadow_repo_refresh=False):
         """
         Try to merge the pull request and return the merge status.
         """
         log.debug(
             "Trying out if the pull request %s can be merged. Force_refresh=%s",
             pull_request.pull_request_id, force_shadow_repo_refresh)
         target_vcs = pull_request.target_repo.scm_instance()
         # Refresh the target reference.
         try:
             target_ref = self._refresh_reference(
                 pull_request.target_ref_parts, target_vcs)
         except CommitDoesNotExistError:
             merge_state = MergeResponse(
                 False, False, None, MergeFailureReason.MISSING_TARGET_REF,
                 metadata={'target_ref': pull_request.target_ref_parts})
             return merge_state
 
         target_locked = pull_request.target_repo.locked
         if target_locked and target_locked[0]:
             locked_by = 'user:{}'.format(target_locked[0])
             log.debug("The target repository is locked by %s.", locked_by)
             merge_state = MergeResponse(
                 False, False, None, MergeFailureReason.TARGET_IS_LOCKED,
                 metadata={'locked_by': locked_by})
         elif force_shadow_repo_refresh or self._needs_merge_state_refresh(
                 pull_request, target_ref):
             log.debug("Refreshing the merge status of the repository.")
             merge_state = self._refresh_merge_state(
                 pull_request, target_vcs, target_ref)
         else:
             possible = pull_request.last_merge_status == MergeFailureReason.NONE
             metadata = {
+                'unresolved_files': '',
                 'target_ref': pull_request.target_ref_parts,
                 'source_ref': pull_request.source_ref_parts,
             }
             if not possible and target_ref.type == 'branch':
                 # NOTE(marcink): case for mercurial multiple heads on branch
                 heads = target_vcs._heads(target_ref.name)
                 if len(heads) != 1:
                     heads = '\n,'.join(target_vcs._heads(target_ref.name))
                     metadata.update({
                         'heads': heads
                     })
             merge_state = MergeResponse(
                 possible, False, None, pull_request.last_merge_status, metadata=metadata)
 
         return merge_state
 
     def _refresh_reference(self, reference, vcs_repository):
         if reference.type in self.UPDATABLE_REF_TYPES:
             name_or_id = reference.name
         else:
             name_or_id = reference.commit_id
 
         refreshed_commit = vcs_repository.get_commit(name_or_id)
         refreshed_reference = Reference(
             reference.type, reference.name, refreshed_commit.raw_id)
         return refreshed_reference
 
     def _needs_merge_state_refresh(self, pull_request, target_reference):
         return not(
             pull_request.revisions and
             pull_request.revisions[0] == pull_request._last_merge_source_rev and
             target_reference.commit_id == pull_request._last_merge_target_rev)
 
     def _refresh_merge_state(self, pull_request, target_vcs, target_reference):
         workspace_id = self._workspace_id(pull_request)
         source_vcs = pull_request.source_repo.scm_instance()
         repo_id = pull_request.target_repo.repo_id
         use_rebase = self._use_rebase_for_merging(pull_request)
         close_branch = self._close_branch_before_merging(pull_request)
         merge_state = target_vcs.merge(
             repo_id, workspace_id,
             target_reference, source_vcs, pull_request.source_ref_parts,
             dry_run=True, use_rebase=use_rebase,
             close_branch=close_branch)
 
         # Do not store the response if there was an unknown error.
         if merge_state.failure_reason != MergeFailureReason.UNKNOWN:
             pull_request._last_merge_source_rev = \
                 pull_request.source_ref_parts.commit_id
             pull_request._last_merge_target_rev = target_reference.commit_id
             pull_request.last_merge_status = merge_state.failure_reason
             pull_request.shadow_merge_ref = merge_state.merge_ref
             Session().add(pull_request)
             Session().commit()
 
         return merge_state
 
     def _workspace_id(self, pull_request):
         workspace_id = 'pr-%s' % pull_request.pull_request_id
         return workspace_id
 
     def generate_repo_data(self, repo, commit_id=None, branch=None,
                            bookmark=None, translator=None):
         from rhodecode.model.repo import RepoModel
 
         all_refs, selected_ref = \
             self._get_repo_pullrequest_sources(
                 repo.scm_instance(), commit_id=commit_id,
                 branch=branch, bookmark=bookmark, translator=translator)
 
         refs_select2 = []
         for element in all_refs:
             children = [{'id': x[0], 'text': x[1]} for x in element[0]]
             refs_select2.append({'text': element[1], 'children': children})
 
         return {
             'user': {
                 'user_id': repo.user.user_id,
                 'username': repo.user.username,
                 'firstname': repo.user.first_name,
                 'lastname': repo.user.last_name,
                 'gravatar_link': h.gravatar_url(repo.user.email, 14),
             },
             'name': repo.repo_name,
             'link': RepoModel().get_url(repo),
             'description': h.chop_at_smart(repo.description_safe, '\n'),
             'refs': {
                 'all_refs': all_refs,
                 'selected_ref': selected_ref,
                 'select2_refs': refs_select2
             }
         }
 
     def generate_pullrequest_title(self, source, source_ref, target):
         return u'{source}#{at_ref} to {target}'.format(
             source=source,
             at_ref=source_ref,
             target=target,
         )
 
     def _cleanup_merge_workspace(self, pull_request):
         # Merging related cleanup
         repo_id = pull_request.target_repo.repo_id
         target_scm = pull_request.target_repo.scm_instance()
         workspace_id = self._workspace_id(pull_request)
 
         try:
             target_scm.cleanup_merge_workspace(repo_id, workspace_id)
         except NotImplementedError:
             pass
 
     def _get_repo_pullrequest_sources(
             self, repo, commit_id=None, branch=None, bookmark=None,
             translator=None):
         """
         Return a structure with repo's interesting commits, suitable for
         the selectors in pullrequest controller
 
         :param commit_id: a commit that must be in the list somehow
             and selected by default
         :param branch: a branch that must be in the list and selected
             by default - even if closed
         :param bookmark: a bookmark that must be in the list and selected
         """
         _ = translator or get_current_request().translate
 
         commit_id = safe_str(commit_id) if commit_id else None
         branch = safe_unicode(branch) if branch else None
         bookmark = safe_unicode(bookmark) if bookmark else None
 
         selected = None
 
         # order matters: first source that has commit_id in it will be selected
         sources = []
         sources.append(('book', repo.bookmarks.items(), _('Bookmarks'), bookmark))
         sources.append(('branch', repo.branches.items(), _('Branches'), branch))
 
         if commit_id:
             ref_commit = (h.short_id(commit_id), commit_id)
             sources.append(('rev', [ref_commit], _('Commit IDs'), commit_id))
 
         sources.append(
             ('branch', repo.branches_closed.items(), _('Closed Branches'), branch),
         )
 
         groups = []
 
         for group_key, ref_list, group_name, match in sources:
             group_refs = []
             for ref_name, ref_id in ref_list:
                 ref_key = u'{}:{}:{}'.format(group_key, ref_name, ref_id)
                 group_refs.append((ref_key, ref_name))
 
                 if not selected:
                     if set([commit_id, match]) & set([ref_id, ref_name]):
                         selected = ref_key
 
             if group_refs:
                 groups.append((group_refs, group_name))
 
         if not selected:
             ref = commit_id or branch or bookmark
             if ref:
                 raise CommitDoesNotExistError(
                     u'No commit refs could be found matching: {}'.format(ref))
             elif repo.DEFAULT_BRANCH_NAME in repo.branches:
                 selected = u'branch:{}:{}'.format(
                     safe_unicode(repo.DEFAULT_BRANCH_NAME),
                     safe_unicode(repo.branches[repo.DEFAULT_BRANCH_NAME])
                 )
             elif repo.commit_ids:
                 # make the user select in this case
                 selected = None
             else:
                 raise EmptyRepositoryError()
         return groups, selected
 
     def get_diff(self, source_repo, source_ref_id, target_ref_id,
                  hide_whitespace_changes, diff_context):
 
         return self._get_diff_from_pr_or_version(
             source_repo, source_ref_id, target_ref_id,
             hide_whitespace_changes=hide_whitespace_changes, diff_context=diff_context)
 
     def _get_diff_from_pr_or_version(
             self, source_repo, source_ref_id, target_ref_id,
             hide_whitespace_changes, diff_context):
 
         target_commit = source_repo.get_commit(
             commit_id=safe_str(target_ref_id))
         source_commit = source_repo.get_commit(
             commit_id=safe_str(source_ref_id))
         if isinstance(source_repo, Repository):
             vcs_repo = source_repo.scm_instance()
         else:
             vcs_repo = source_repo
 
         # TODO: johbo: In the context of an update, we cannot reach
         # the old commit anymore with our normal mechanisms. It needs
         # some sort of special support in the vcs layer to avoid this
         # workaround.
         if (source_commit.raw_id == vcs_repo.EMPTY_COMMIT_ID and
                 vcs_repo.alias == 'git'):
             source_commit.raw_id = safe_str(source_ref_id)
 
         log.debug('calculating diff between '
                   'source_ref:%s and target_ref:%s for repo `%s`',
                   target_ref_id, source_ref_id,
                   safe_unicode(vcs_repo.path))
 
         vcs_diff = vcs_repo.get_diff(
             commit1=target_commit, commit2=source_commit,
             ignore_whitespace=hide_whitespace_changes, context=diff_context)
         return vcs_diff
 
     def _is_merge_enabled(self, pull_request):
         return self._get_general_setting(
             pull_request, 'rhodecode_pr_merge_enabled')
 
     def _use_rebase_for_merging(self, pull_request):
         repo_type = pull_request.target_repo.repo_type
         if repo_type == 'hg':
             return self._get_general_setting(
                 pull_request, 'rhodecode_hg_use_rebase_for_merging')
         elif repo_type == 'git':
             return self._get_general_setting(
                 pull_request, 'rhodecode_git_use_rebase_for_merging')
 
         return False
 
     def _close_branch_before_merging(self, pull_request):
         repo_type = pull_request.target_repo.repo_type
         if repo_type == 'hg':
             return self._get_general_setting(
                 pull_request, 'rhodecode_hg_close_branch_before_merging')
         elif repo_type == 'git':
             return self._get_general_setting(
                 pull_request, 'rhodecode_git_close_branch_before_merging')
 
         return False
 
     def _get_general_setting(self, pull_request, settings_key, default=False):
         settings_model = VcsSettingsModel(repo=pull_request.target_repo)
         settings = settings_model.get_general_settings()
         return settings.get(settings_key, default)
 
     def _log_audit_action(self, action, action_data, user, pull_request):
         audit_logger.store(
             action=action,
             action_data=action_data,
             user=user,
             repo=pull_request.target_repo)
 
     def get_reviewer_functions(self):
         """
         Fetches functions for validation and fetching default reviewers.
         If available we use the EE package, else we fallback to CE
         package functions
         """
         try:
             from rc_reviewers.utils import get_default_reviewers_data
             from rc_reviewers.utils import validate_default_reviewers
         except ImportError:
             from rhodecode.apps.repository.utils import get_default_reviewers_data
             from rhodecode.apps.repository.utils import validate_default_reviewers
 
         return get_default_reviewers_data, validate_default_reviewers
 
 
 class MergeCheck(object):
     """
     Perform Merge Checks and returns a check object which stores information
     about merge errors, and merge conditions
     """
     TODO_CHECK = 'todo'
     PERM_CHECK = 'perm'
     REVIEW_CHECK = 'review'
     MERGE_CHECK = 'merge'
 
     def __init__(self):
         self.review_status = None
         self.merge_possible = None
         self.merge_msg = ''
         self.failed = None
         self.errors = []
         self.error_details = OrderedDict()
 
     def push_error(self, error_type, message, error_key, details):
         self.failed = True
         self.errors.append([error_type, message])
         self.error_details[error_key] = dict(
             details=details,
             error_type=error_type,
             message=message
         )
 
     @classmethod
     def validate(cls, pull_request, auth_user, translator, fail_early=False,
                  force_shadow_repo_refresh=False):
         _ = translator
         merge_check = cls()
 
         # permissions to merge
         user_allowed_to_merge = PullRequestModel().check_user_merge(
             pull_request, auth_user)
         if not user_allowed_to_merge:
             log.debug("MergeCheck: cannot merge, approval is pending.")
 
             msg = _('User `{}` not allowed to perform merge.').format(auth_user.username)
             merge_check.push_error('error', msg, cls.PERM_CHECK, auth_user.username)
             if fail_early:
                 return merge_check
 
         # permission to merge into the target branch
         target_commit_id = pull_request.target_ref_parts.commit_id
         if pull_request.target_ref_parts.type == 'branch':
             branch_name = pull_request.target_ref_parts.name
         else:
             # for mercurial we can always figure out the branch from the commit
             # in case of bookmark
             target_commit = pull_request.target_repo.get_commit(target_commit_id)
             branch_name = target_commit.branch
 
         rule, branch_perm = auth_user.get_rule_and_branch_permission(
             pull_request.target_repo.repo_name, branch_name)
         if branch_perm and branch_perm == 'branch.none':
             msg = _('Target branch `{}` changes rejected by rule {}.').format(
                 branch_name, rule)
             merge_check.push_error('error', msg, cls.PERM_CHECK, auth_user.username)
             if fail_early:
                 return merge_check
 
         # review status, must be always present
         review_status = pull_request.calculated_review_status()
         merge_check.review_status = review_status
 
         status_approved = review_status == ChangesetStatus.STATUS_APPROVED
         if not status_approved:
             log.debug("MergeCheck: cannot merge, approval is pending.")
 
             msg = _('Pull request reviewer approval is pending.')
 
             merge_check.push_error('warning', msg, cls.REVIEW_CHECK, review_status)
 
             if fail_early:
                 return merge_check
 
         # left over TODOs
         todos = CommentsModel().get_pull_request_unresolved_todos(pull_request)
         if todos:
             log.debug("MergeCheck: cannot merge, {} "
                       "unresolved TODOs left.".format(len(todos)))
 
             if len(todos) == 1:
                 msg = _('Cannot merge, {} TODO still not resolved.').format(
                     len(todos))
             else:
                 msg = _('Cannot merge, {} TODOs still not resolved.').format(
                     len(todos))
 
             merge_check.push_error('warning', msg, cls.TODO_CHECK, todos)
 
             if fail_early:
                 return merge_check
 
         # merge possible, here is the filesystem simulation + shadow repo
         merge_status, msg = PullRequestModel().merge_status(
             pull_request, translator=translator,
             force_shadow_repo_refresh=force_shadow_repo_refresh)
         merge_check.merge_possible = merge_status
         merge_check.merge_msg = msg
         if not merge_status:
             log.debug("MergeCheck: cannot merge, pull request merge not possible.")
             merge_check.push_error('warning', msg, cls.MERGE_CHECK, None)
 
             if fail_early:
                 return merge_check
 
         log.debug('MergeCheck: is failed: %s', merge_check.failed)
         return merge_check
 
     @classmethod
     def get_merge_conditions(cls, pull_request, translator):
         _ = translator
         merge_details = {}
 
         model = PullRequestModel()
         use_rebase = model._use_rebase_for_merging(pull_request)
 
         if use_rebase:
             merge_details['merge_strategy'] = dict(
                 details={},
                 message=_('Merge strategy: rebase')
             )
         else:
             merge_details['merge_strategy'] = dict(
                 details={},
                 message=_('Merge strategy: explicit merge commit')
             )
 
         close_branch = model._close_branch_before_merging(pull_request)
         if close_branch:
             repo_type = pull_request.target_repo.repo_type
             close_msg = ''
             if repo_type == 'hg':
                 close_msg = _('Source branch will be closed after merge.')
             elif repo_type == 'git':
                 close_msg = _('Source branch will be deleted after merge.')
 
             merge_details['close_branch'] = dict(
                 details={},
                 message=close_msg
             )
 
         return merge_details
 
 
 ChangeTuple = collections.namedtuple(
     'ChangeTuple', ['added', 'common', 'removed', 'total'])
 
 FileChangeTuple = collections.namedtuple(
     'FileChangeTuple', ['added', 'modified', 'removed'])

rhodecode/tests/models/test_pullrequest.py

@@ -1,964 +1,966 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import mock
 import pytest
 import textwrap
 
 import rhodecode
 from rhodecode.lib.utils2 import safe_unicode
 from rhodecode.lib.vcs.backends import get_backend
 from rhodecode.lib.vcs.backends.base import (
     MergeResponse, MergeFailureReason, Reference)
 from rhodecode.lib.vcs.exceptions import RepositoryError
 from rhodecode.lib.vcs.nodes import FileNode
 from rhodecode.model.comment import CommentsModel
 from rhodecode.model.db import PullRequest, Session
 from rhodecode.model.pull_request import PullRequestModel
 from rhodecode.model.user import UserModel
 from rhodecode.tests import TEST_USER_ADMIN_LOGIN
 
 
 pytestmark = [
     pytest.mark.backends("git", "hg"),
 ]
 
 
 @pytest.mark.usefixtures('config_stub')
 class TestPullRequestModel(object):
 
     @pytest.fixture()
     def pull_request(self, request, backend, pr_util):
         """
         A pull request combined with multiples patches.
         """
         BackendClass = get_backend(backend.alias)
         merge_resp = MergeResponse(
             False, False, None, MergeFailureReason.UNKNOWN,
             metadata={'exception': 'MockError'})
         self.merge_patcher = mock.patch.object(
             BackendClass, 'merge', return_value=merge_resp)
         self.workspace_remove_patcher = mock.patch.object(
             BackendClass, 'cleanup_merge_workspace')
 
         self.workspace_remove_mock = self.workspace_remove_patcher.start()
         self.merge_mock = self.merge_patcher.start()
         self.comment_patcher = mock.patch(
             'rhodecode.model.changeset_status.ChangesetStatusModel.set_status')
         self.comment_patcher.start()
         self.notification_patcher = mock.patch(
             'rhodecode.model.notification.NotificationModel.create')
         self.notification_patcher.start()
         self.helper_patcher = mock.patch(
             'rhodecode.lib.helpers.route_path')
         self.helper_patcher.start()
 
         self.hook_patcher = mock.patch.object(PullRequestModel,
                                               'trigger_pull_request_hook')
         self.hook_mock = self.hook_patcher.start()
 
         self.invalidation_patcher = mock.patch(
             'rhodecode.model.pull_request.ScmModel.mark_for_invalidation')
         self.invalidation_mock = self.invalidation_patcher.start()
 
         self.pull_request = pr_util.create_pull_request(
             mergeable=True, name_suffix=u'ąć')
         self.source_commit = self.pull_request.source_ref_parts.commit_id
         self.target_commit = self.pull_request.target_ref_parts.commit_id
         self.workspace_id = 'pr-%s' % self.pull_request.pull_request_id
         self.repo_id = self.pull_request.target_repo.repo_id
 
         @request.addfinalizer
         def cleanup_pull_request():
             calls = [mock.call(
                 self.pull_request, self.pull_request.author, 'create')]
             self.hook_mock.assert_has_calls(calls)
 
             self.workspace_remove_patcher.stop()
             self.merge_patcher.stop()
             self.comment_patcher.stop()
             self.notification_patcher.stop()
             self.helper_patcher.stop()
             self.hook_patcher.stop()
             self.invalidation_patcher.stop()
 
         return self.pull_request
 
     def test_get_all(self, pull_request):
         prs = PullRequestModel().get_all(pull_request.target_repo)
         assert isinstance(prs, list)
         assert len(prs) == 1
 
     def test_count_all(self, pull_request):
         pr_count = PullRequestModel().count_all(pull_request.target_repo)
         assert pr_count == 1
 
     def test_get_awaiting_review(self, pull_request):
         prs = PullRequestModel().get_awaiting_review(pull_request.target_repo)
         assert isinstance(prs, list)
         assert len(prs) == 1
 
     def test_count_awaiting_review(self, pull_request):
         pr_count = PullRequestModel().count_awaiting_review(
             pull_request.target_repo)
         assert pr_count == 1
 
     def test_get_awaiting_my_review(self, pull_request):
         PullRequestModel().update_reviewers(
             pull_request, [(pull_request.author, ['author'], False, [])],
             pull_request.author)
         Session().commit()
 
         prs = PullRequestModel().get_awaiting_my_review(
             pull_request.target_repo, user_id=pull_request.author.user_id)
         assert isinstance(prs, list)
         assert len(prs) == 1
 
     def test_count_awaiting_my_review(self, pull_request):
         PullRequestModel().update_reviewers(
             pull_request, [(pull_request.author, ['author'], False, [])],
             pull_request.author)
         Session().commit()
 
         pr_count = PullRequestModel().count_awaiting_my_review(
             pull_request.target_repo, user_id=pull_request.author.user_id)
         assert pr_count == 1
 
     def test_delete_calls_cleanup_merge(self, pull_request):
         repo_id = pull_request.target_repo.repo_id
         PullRequestModel().delete(pull_request, pull_request.author)
         Session().commit()
 
         self.workspace_remove_mock.assert_called_once_with(
             repo_id, self.workspace_id)
 
     def test_close_calls_cleanup_and_hook(self, pull_request):
         PullRequestModel().close_pull_request(
             pull_request, pull_request.author)
         Session().commit()
 
         repo_id = pull_request.target_repo.repo_id
 
         self.workspace_remove_mock.assert_called_once_with(
             repo_id, self.workspace_id)
         self.hook_mock.assert_called_with(
             self.pull_request, self.pull_request.author, 'close')
 
     def test_merge_status(self, pull_request):
         self.merge_mock.return_value = MergeResponse(
             True, False, None, MergeFailureReason.NONE)
 
         assert pull_request._last_merge_source_rev is None
         assert pull_request._last_merge_target_rev is None
         assert pull_request.last_merge_status is None
 
         status, msg = PullRequestModel().merge_status(pull_request)
         assert status is True
         assert msg == 'This pull request can be automatically merged.'
         self.merge_mock.assert_called_with(
             self.repo_id, self.workspace_id,
             pull_request.target_ref_parts,
             pull_request.source_repo.scm_instance(),
             pull_request.source_ref_parts, dry_run=True,
             use_rebase=False, close_branch=False)
 
         assert pull_request._last_merge_source_rev == self.source_commit
         assert pull_request._last_merge_target_rev == self.target_commit
         assert pull_request.last_merge_status is MergeFailureReason.NONE
 
         self.merge_mock.reset_mock()
         status, msg = PullRequestModel().merge_status(pull_request)
         assert status is True
         assert msg == 'This pull request can be automatically merged.'
         assert self.merge_mock.called is False
 
     def test_merge_status_known_failure(self, pull_request):
         self.merge_mock.return_value = MergeResponse(
-            False, False, None, MergeFailureReason.MERGE_FAILED)
+            False, False, None, MergeFailureReason.MERGE_FAILED,
+            metadata={'unresolved_files': 'file1'})
 
         assert pull_request._last_merge_source_rev is None
         assert pull_request._last_merge_target_rev is None
         assert pull_request.last_merge_status is None
 
         status, msg = PullRequestModel().merge_status(pull_request)
         assert status is False
-        assert msg == 'This pull request cannot be merged because of merge conflicts.'
+        assert msg == 'This pull request cannot be merged because of merge conflicts. file1'
         self.merge_mock.assert_called_with(
             self.repo_id, self.workspace_id,
             pull_request.target_ref_parts,
             pull_request.source_repo.scm_instance(),
             pull_request.source_ref_parts, dry_run=True,
             use_rebase=False, close_branch=False)
 
         assert pull_request._last_merge_source_rev == self.source_commit
         assert pull_request._last_merge_target_rev == self.target_commit
-        assert (
-            pull_request.last_merge_status is MergeFailureReason.MERGE_FAILED)
+        assert pull_request.last_merge_status is MergeFailureReason.MERGE_FAILED
 
         self.merge_mock.reset_mock()
         status, msg = PullRequestModel().merge_status(pull_request)
         assert status is False
         assert msg == 'This pull request cannot be merged because of merge conflicts.'
         assert self.merge_mock.called is False
 
     def test_merge_status_unknown_failure(self, pull_request):
         self.merge_mock.return_value = MergeResponse(
             False, False, None, MergeFailureReason.UNKNOWN,
             metadata={'exception': 'MockError'})
 
         assert pull_request._last_merge_source_rev is None
         assert pull_request._last_merge_target_rev is None
         assert pull_request.last_merge_status is None
 
         status, msg = PullRequestModel().merge_status(pull_request)
         assert status is False
         assert msg == (
             'This pull request cannot be merged because of an unhandled exception. '
             'MockError')
         self.merge_mock.assert_called_with(
             self.repo_id, self.workspace_id,
             pull_request.target_ref_parts,
             pull_request.source_repo.scm_instance(),
             pull_request.source_ref_parts, dry_run=True,
             use_rebase=False, close_branch=False)
 
         assert pull_request._last_merge_source_rev is None
         assert pull_request._last_merge_target_rev is None
         assert pull_request.last_merge_status is None
 
         self.merge_mock.reset_mock()
         status, msg = PullRequestModel().merge_status(pull_request)
         assert status is False
         assert msg == (
             'This pull request cannot be merged because of an unhandled exception. '
             'MockError')
         assert self.merge_mock.called is True
 
     def test_merge_status_when_target_is_locked(self, pull_request):
         pull_request.target_repo.locked = [1, u'12345.50', 'lock_web']
         status, msg = PullRequestModel().merge_status(pull_request)
         assert status is False
         assert msg == (
             'This pull request cannot be merged because the target repository '
             'is locked by user:1.')
 
     def test_merge_status_requirements_check_target(self, pull_request):
 
         def has_largefiles(self, repo):
             return repo == pull_request.source_repo
 
         patcher = mock.patch.object(PullRequestModel, '_has_largefiles', has_largefiles)
         with patcher:
             status, msg = PullRequestModel().merge_status(pull_request)
 
         assert status is False
         assert msg == 'Target repository large files support is disabled.'
 
     def test_merge_status_requirements_check_source(self, pull_request):
 
         def has_largefiles(self, repo):
             return repo == pull_request.target_repo
 
         patcher = mock.patch.object(PullRequestModel, '_has_largefiles', has_largefiles)
         with patcher:
             status, msg = PullRequestModel().merge_status(pull_request)
 
         assert status is False
         assert msg == 'Source repository large files support is disabled.'
 
     def test_merge(self, pull_request, merge_extras):
         user = UserModel().get_by_username(TEST_USER_ADMIN_LOGIN)
         merge_ref = Reference(
             'type', 'name', '6126b7bfcc82ad2d3deaee22af926b082ce54cc6')
         self.merge_mock.return_value = MergeResponse(
             True, True, merge_ref, MergeFailureReason.NONE)
 
         merge_extras['repository'] = pull_request.target_repo.repo_name
         PullRequestModel().merge_repo(
             pull_request, pull_request.author, extras=merge_extras)
         Session().commit()
 
         message = (
             u'Merge pull request !{pr_id} from {source_repo} {source_ref_name}'
             u'\n\n {pr_title}'.format(
                 pr_id=pull_request.pull_request_id,
                 source_repo=safe_unicode(
                     pull_request.source_repo.scm_instance().name),
                 source_ref_name=pull_request.source_ref_parts.name,
                 pr_title=safe_unicode(pull_request.title)
             )
         )
         self.merge_mock.assert_called_with(
             self.repo_id, self.workspace_id,
             pull_request.target_ref_parts,
             pull_request.source_repo.scm_instance(),
             pull_request.source_ref_parts,
             user_name=user.short_contact, user_email=user.email, message=message,
             use_rebase=False, close_branch=False
         )
         self.invalidation_mock.assert_called_once_with(
             pull_request.target_repo.repo_name)
 
         self.hook_mock.assert_called_with(
             self.pull_request, self.pull_request.author, 'merge')
 
         pull_request = PullRequest.get(pull_request.pull_request_id)
         assert pull_request.merge_rev == '6126b7bfcc82ad2d3deaee22af926b082ce54cc6'
 
     def test_merge_with_status_lock(self, pull_request, merge_extras):
         user = UserModel().get_by_username(TEST_USER_ADMIN_LOGIN)
         merge_ref = Reference(
             'type', 'name', '6126b7bfcc82ad2d3deaee22af926b082ce54cc6')
         self.merge_mock.return_value = MergeResponse(
             True, True, merge_ref, MergeFailureReason.NONE)
 
         merge_extras['repository'] = pull_request.target_repo.repo_name
 
         with pull_request.set_state(PullRequest.STATE_UPDATING):
             assert pull_request.pull_request_state == PullRequest.STATE_UPDATING
             PullRequestModel().merge_repo(
                 pull_request, pull_request.author, extras=merge_extras)
             Session().commit()
 
         assert pull_request.pull_request_state == PullRequest.STATE_CREATED
 
         message = (
             u'Merge pull request !{pr_id} from {source_repo} {source_ref_name}'
             u'\n\n {pr_title}'.format(
                 pr_id=pull_request.pull_request_id,
                 source_repo=safe_unicode(
                     pull_request.source_repo.scm_instance().name),
                 source_ref_name=pull_request.source_ref_parts.name,
                 pr_title=safe_unicode(pull_request.title)
             )
         )
         self.merge_mock.assert_called_with(
             self.repo_id, self.workspace_id,
             pull_request.target_ref_parts,
             pull_request.source_repo.scm_instance(),
             pull_request.source_ref_parts,
             user_name=user.short_contact, user_email=user.email, message=message,
             use_rebase=False, close_branch=False
         )
         self.invalidation_mock.assert_called_once_with(
             pull_request.target_repo.repo_name)
 
         self.hook_mock.assert_called_with(
             self.pull_request, self.pull_request.author, 'merge')
 
         pull_request = PullRequest.get(pull_request.pull_request_id)
         assert pull_request.merge_rev == '6126b7bfcc82ad2d3deaee22af926b082ce54cc6'
 
     def test_merge_failed(self, pull_request, merge_extras):
         user = UserModel().get_by_username(TEST_USER_ADMIN_LOGIN)
         merge_ref = Reference(
             'type', 'name', '6126b7bfcc82ad2d3deaee22af926b082ce54cc6')
         self.merge_mock.return_value = MergeResponse(
             False, False, merge_ref, MergeFailureReason.MERGE_FAILED)
 
         merge_extras['repository'] = pull_request.target_repo.repo_name
         PullRequestModel().merge_repo(
             pull_request, pull_request.author, extras=merge_extras)
         Session().commit()
 
         message = (
             u'Merge pull request !{pr_id} from {source_repo} {source_ref_name}'
             u'\n\n {pr_title}'.format(
                 pr_id=pull_request.pull_request_id,
                 source_repo=safe_unicode(
                     pull_request.source_repo.scm_instance().name),
                 source_ref_name=pull_request.source_ref_parts.name,
                 pr_title=safe_unicode(pull_request.title)
             )
         )
         self.merge_mock.assert_called_with(
             self.repo_id, self.workspace_id,
             pull_request.target_ref_parts,
             pull_request.source_repo.scm_instance(),
             pull_request.source_ref_parts,
             user_name=user.short_contact, user_email=user.email, message=message,
             use_rebase=False, close_branch=False
         )
 
         pull_request = PullRequest.get(pull_request.pull_request_id)
         assert self.invalidation_mock.called is False
         assert pull_request.merge_rev is None
 
     def test_get_commit_ids(self, pull_request):
         # The PR has been not merget yet, so expect an exception
         with pytest.raises(ValueError):
             PullRequestModel()._get_commit_ids(pull_request)
 
         # Merge revision is in the revisions list
         pull_request.merge_rev = pull_request.revisions[0]
         commit_ids = PullRequestModel()._get_commit_ids(pull_request)
         assert commit_ids == pull_request.revisions
 
         # Merge revision is not in the revisions list
         pull_request.merge_rev = 'f000' * 10
         commit_ids = PullRequestModel()._get_commit_ids(pull_request)
         assert commit_ids == pull_request.revisions + [pull_request.merge_rev]
 
     def test_get_diff_from_pr_version(self, pull_request):
         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
         diff = PullRequestModel()._get_diff_from_pr_or_version(
             source_repo, source_ref_id, target_ref_id,
             hide_whitespace_changes=False, diff_context=6)
         assert 'file_1' in diff.raw
 
     def test_generate_title_returns_unicode(self):
         title = PullRequestModel().generate_pullrequest_title(
             source='source-dummy',
             source_ref='source-ref-dummy',
             target='target-dummy',
         )
         assert type(title) == unicode
 
 
 @pytest.mark.usefixtures('config_stub')
 class TestIntegrationMerge(object):
     @pytest.mark.parametrize('extra_config', (
         {'vcs.hooks.protocol': 'http', 'vcs.hooks.direct_calls': False},
     ))
     def test_merge_triggers_push_hooks(
             self, pr_util, user_admin, capture_rcextensions, merge_extras,
             extra_config):
 
         pull_request = pr_util.create_pull_request(
             approved=True, mergeable=True)
         # TODO: johbo: Needed for sqlite, try to find an automatic way for it
         merge_extras['repository'] = pull_request.target_repo.repo_name
         Session().commit()
 
         with mock.patch.dict(rhodecode.CONFIG, extra_config, clear=False):
             merge_state = PullRequestModel().merge_repo(
                 pull_request, user_admin, extras=merge_extras)
             Session().commit()
 
         assert merge_state.executed
         assert '_pre_push_hook' in capture_rcextensions
         assert '_push_hook' in capture_rcextensions
 
     def test_merge_can_be_rejected_by_pre_push_hook(
             self, pr_util, user_admin, capture_rcextensions, merge_extras):
         pull_request = pr_util.create_pull_request(
             approved=True, mergeable=True)
         # TODO: johbo: Needed for sqlite, try to find an automatic way for it
         merge_extras['repository'] = pull_request.target_repo.repo_name
         Session().commit()
 
         with mock.patch('rhodecode.EXTENSIONS.PRE_PUSH_HOOK') as pre_pull:
             pre_pull.side_effect = RepositoryError("Disallow push!")
             merge_status = PullRequestModel().merge_repo(
                 pull_request, user_admin, extras=merge_extras)
             Session().commit()
 
         assert not merge_status.executed
         assert 'pre_push' not in capture_rcextensions
         assert 'post_push' not in capture_rcextensions
 
     def test_merge_fails_if_target_is_locked(
             self, pr_util, user_regular, merge_extras):
         pull_request = pr_util.create_pull_request(
             approved=True, mergeable=True)
         locked_by = [user_regular.user_id + 1, 12345.50, 'lock_web']
         pull_request.target_repo.locked = locked_by
         # TODO: johbo: Check if this can work based on the database, currently
         # all data is pre-computed, that's why just updating the DB is not
         # enough.
         merge_extras['locked_by'] = locked_by
         merge_extras['repository'] = pull_request.target_repo.repo_name
         # TODO: johbo: Needed for sqlite, try to find an automatic way for it
         Session().commit()
         merge_status = PullRequestModel().merge_repo(
             pull_request, user_regular, extras=merge_extras)
         Session().commit()
 
         assert not merge_status.executed
 
 
 @pytest.mark.parametrize('use_outdated, inlines_count, outdated_count', [
     (False, 1, 0),
     (True, 0, 1),
 ])
 def test_outdated_comments(
         pr_util, use_outdated, inlines_count, outdated_count, config_stub):
     pull_request = pr_util.create_pull_request()
     pr_util.create_inline_comment(file_path='not_in_updated_diff')
 
     with outdated_comments_patcher(use_outdated) as outdated_comment_mock:
         pr_util.add_one_commit()
         assert_inline_comments(
             pull_request, visible=inlines_count, outdated=outdated_count)
     outdated_comment_mock.assert_called_with(pull_request)
 
 
 @pytest.mark.parametrize('mr_type, expected_msg', [
     (MergeFailureReason.NONE,
      'This pull request can be automatically merged.'),
     (MergeFailureReason.UNKNOWN,
      'This pull request cannot be merged because of an unhandled exception. CRASH'),
     (MergeFailureReason.MERGE_FAILED,
-     'This pull request cannot be merged because of merge conflicts.'),
+     'This pull request cannot be merged because of merge conflicts. CONFLICT_FILE'),
     (MergeFailureReason.PUSH_FAILED,
      'This pull request could not be merged because push to target:`some-repo@merge_commit` failed.'),
     (MergeFailureReason.TARGET_IS_NOT_HEAD,
      'This pull request cannot be merged because the target `ref_name` is not a head.'),
     (MergeFailureReason.HG_SOURCE_HAS_MORE_BRANCHES,
      'This pull request cannot be merged because the source contains more branches than the target.'),
     (MergeFailureReason.HG_TARGET_HAS_MULTIPLE_HEADS,
      'This pull request cannot be merged because the target `ref_name` has multiple heads: `a,b,c`.'),
     (MergeFailureReason.TARGET_IS_LOCKED,
      'This pull request cannot be merged because the target repository is locked by user:123.'),
     (MergeFailureReason.MISSING_TARGET_REF,
      'This pull request cannot be merged because the target reference `ref_name` is missing.'),
     (MergeFailureReason.MISSING_SOURCE_REF,
      'This pull request cannot be merged because the source reference `ref_name` is missing.'),
     (MergeFailureReason.SUBREPO_MERGE_FAILED,
      'This pull request cannot be merged because of conflicts related to sub repositories.'),
 
 ])
 def test_merge_response_message(mr_type, expected_msg):
     merge_ref = Reference('type', 'ref_name', '6126b7bfcc82ad2d3deaee22af926b082ce54cc6')
     metadata = {
+        'unresolved_files': 'CONFLICT_FILE',
         'exception': "CRASH",
         'target': 'some-repo',
         'merge_commit': 'merge_commit',
         'target_ref': merge_ref,
         'source_ref': merge_ref,
         'heads': ','.join(['a', 'b', 'c']),
-        'locked_by': 'user:123'}
+        'locked_by': 'user:123'
+    }
 
     merge_response = MergeResponse(True, True, merge_ref, mr_type, metadata=metadata)
     assert merge_response.merge_status_message == expected_msg
 
 
 @pytest.fixture()
 def merge_extras(user_regular):
     """
     Context for the vcs operation when running a merge.
     """
     extras = {
         'ip': '127.0.0.1',
         'username': user_regular.username,
         'user_id': user_regular.user_id,
         'action': 'push',
         'repository': 'fake_target_repo_name',
         'scm': 'git',
         'config': 'fake_config_ini_path',
         'repo_store': '',
         'make_lock': None,
         'locked_by': [None, None, None],
         'server_url': 'http://test.example.com:5000',
         'hooks': ['push', 'pull'],
         'is_shadow_repo': False,
     }
     return extras
 
 
 @pytest.mark.usefixtures('config_stub')
 class TestUpdateCommentHandling(object):
 
     @pytest.fixture(autouse=True, scope='class')
     def enable_outdated_comments(self, request, baseapp):
         config_patch = mock.patch.dict(
             'rhodecode.CONFIG', {'rhodecode_use_outdated_comments': True})
         config_patch.start()
 
         @request.addfinalizer
         def cleanup():
             config_patch.stop()
 
     def test_comment_stays_unflagged_on_unchanged_diff(self, pr_util):
         commits = [
             {'message': 'a'},
             {'message': 'b', 'added': [FileNode('file_b', 'test_content\n')]},
             {'message': 'c', 'added': [FileNode('file_c', 'test_content\n')]},
         ]
         pull_request = pr_util.create_pull_request(
             commits=commits, target_head='a', source_head='b', revisions=['b'])
         pr_util.create_inline_comment(file_path='file_b')
         pr_util.add_one_commit(head='c')
 
         assert_inline_comments(pull_request, visible=1, outdated=0)
 
     def test_comment_stays_unflagged_on_change_above(self, pr_util):
         original_content = ''.join(
             ['line {}\n'.format(x) for x in range(1, 11)])
         updated_content = 'new_line_at_top\n' + original_content
         commits = [
             {'message': 'a'},
             {'message': 'b', 'added': [FileNode('file_b', original_content)]},
             {'message': 'c', 'changed': [FileNode('file_b', updated_content)]},
         ]
         pull_request = pr_util.create_pull_request(
             commits=commits, target_head='a', source_head='b', revisions=['b'])
 
         with outdated_comments_patcher():
             comment = pr_util.create_inline_comment(
                 line_no=u'n8', file_path='file_b')
             pr_util.add_one_commit(head='c')
 
         assert_inline_comments(pull_request, visible=1, outdated=0)
         assert comment.line_no == u'n9'
 
     def test_comment_stays_unflagged_on_change_below(self, pr_util):
         original_content = ''.join(['line {}\n'.format(x) for x in range(10)])
         updated_content = original_content + 'new_line_at_end\n'
         commits = [
             {'message': 'a'},
             {'message': 'b', 'added': [FileNode('file_b', original_content)]},
             {'message': 'c', 'changed': [FileNode('file_b', updated_content)]},
         ]
         pull_request = pr_util.create_pull_request(
             commits=commits, target_head='a', source_head='b', revisions=['b'])
         pr_util.create_inline_comment(file_path='file_b')
         pr_util.add_one_commit(head='c')
 
         assert_inline_comments(pull_request, visible=1, outdated=0)
 
     @pytest.mark.parametrize('line_no', ['n4', 'o4', 'n10', 'o9'])
     def test_comment_flagged_on_change_around_context(self, pr_util, line_no):
         base_lines = ['line {}\n'.format(x) for x in range(1, 13)]
         change_lines = list(base_lines)
         change_lines.insert(6, 'line 6a added\n')
 
         # Changes on the last line of sight
         update_lines = list(change_lines)
         update_lines[0] = 'line 1 changed\n'
         update_lines[-1] = 'line 12 changed\n'
 
         def file_b(lines):
             return FileNode('file_b', ''.join(lines))
 
         commits = [
             {'message': 'a', 'added': [file_b(base_lines)]},
             {'message': 'b', 'changed': [file_b(change_lines)]},
             {'message': 'c', 'changed': [file_b(update_lines)]},
         ]
 
         pull_request = pr_util.create_pull_request(
             commits=commits, target_head='a', source_head='b', revisions=['b'])
         pr_util.create_inline_comment(line_no=line_no, file_path='file_b')
 
         with outdated_comments_patcher():
             pr_util.add_one_commit(head='c')
             assert_inline_comments(pull_request, visible=0, outdated=1)
 
     @pytest.mark.parametrize("change, content", [
         ('changed', 'changed\n'),
         ('removed', ''),
     ], ids=['changed', 'removed'])
     def test_comment_flagged_on_change(self, pr_util, change, content):
         commits = [
             {'message': 'a'},
             {'message': 'b', 'added': [FileNode('file_b', 'test_content\n')]},
             {'message': 'c', change: [FileNode('file_b', content)]},
         ]
         pull_request = pr_util.create_pull_request(
             commits=commits, target_head='a', source_head='b', revisions=['b'])
         pr_util.create_inline_comment(file_path='file_b')
 
         with outdated_comments_patcher():
             pr_util.add_one_commit(head='c')
             assert_inline_comments(pull_request, visible=0, outdated=1)
 
 
 @pytest.mark.usefixtures('config_stub')
 class TestUpdateChangedFiles(object):
 
     def test_no_changes_on_unchanged_diff(self, pr_util):
         commits = [
             {'message': 'a'},
             {'message': 'b',
              'added': [FileNode('file_b', 'test_content b\n')]},
             {'message': 'c',
              'added': [FileNode('file_c', 'test_content c\n')]},
         ]
         # open a PR from a to b, adding file_b
         pull_request = pr_util.create_pull_request(
             commits=commits, target_head='a', source_head='b', revisions=['b'],
             name_suffix='per-file-review')
 
         # modify PR adding new file file_c
         pr_util.add_one_commit(head='c')
 
         assert_pr_file_changes(
             pull_request,
             added=['file_c'],
             modified=[],
             removed=[])
 
     def test_modify_and_undo_modification_diff(self, pr_util):
         commits = [
             {'message': 'a'},
             {'message': 'b',
              'added': [FileNode('file_b', 'test_content b\n')]},
             {'message': 'c',
              'changed': [FileNode('file_b', 'test_content b modified\n')]},
             {'message': 'd',
              'changed': [FileNode('file_b', 'test_content b\n')]},
         ]
         # open a PR from a to b, adding file_b
         pull_request = pr_util.create_pull_request(
             commits=commits, target_head='a', source_head='b', revisions=['b'],
             name_suffix='per-file-review')
 
         # modify PR modifying file file_b
         pr_util.add_one_commit(head='c')
 
         assert_pr_file_changes(
             pull_request,
             added=[],
             modified=['file_b'],
             removed=[])
 
         # move the head again to d, which rollbacks change,
         # meaning we should indicate no changes
         pr_util.add_one_commit(head='d')
 
         assert_pr_file_changes(
             pull_request,
             added=[],
             modified=[],
             removed=[])
 
     def test_updated_all_files_in_pr(self, pr_util):
         commits = [
             {'message': 'a'},
             {'message': 'b', 'added': [
                 FileNode('file_a', 'test_content a\n'),
                 FileNode('file_b', 'test_content b\n'),
                 FileNode('file_c', 'test_content c\n')]},
             {'message': 'c', 'changed': [
                 FileNode('file_a', 'test_content a changed\n'),
                 FileNode('file_b', 'test_content b changed\n'),
                 FileNode('file_c', 'test_content c changed\n')]},
         ]
         # open a PR from a to b, changing 3 files
         pull_request = pr_util.create_pull_request(
             commits=commits, target_head='a', source_head='b', revisions=['b'],
             name_suffix='per-file-review')
 
         pr_util.add_one_commit(head='c')
 
         assert_pr_file_changes(
             pull_request,
             added=[],
             modified=['file_a', 'file_b', 'file_c'],
             removed=[])
 
     def test_updated_and_removed_all_files_in_pr(self, pr_util):
         commits = [
             {'message': 'a'},
             {'message': 'b', 'added': [
                 FileNode('file_a', 'test_content a\n'),
                 FileNode('file_b', 'test_content b\n'),
                 FileNode('file_c', 'test_content c\n')]},
             {'message': 'c', 'removed': [
                 FileNode('file_a', 'test_content a changed\n'),
                 FileNode('file_b', 'test_content b changed\n'),
                 FileNode('file_c', 'test_content c changed\n')]},
         ]
         # open a PR from a to b, removing 3 files
         pull_request = pr_util.create_pull_request(
             commits=commits, target_head='a', source_head='b', revisions=['b'],
             name_suffix='per-file-review')
 
         pr_util.add_one_commit(head='c')
 
         assert_pr_file_changes(
             pull_request,
             added=[],
             modified=[],
             removed=['file_a', 'file_b', 'file_c'])
 
 
 def test_update_writes_snapshot_into_pull_request_version(pr_util, config_stub):
     model = PullRequestModel()
     pull_request = pr_util.create_pull_request()
     pr_util.update_source_repository()
 
     model.update_commits(pull_request)
 
     # Expect that it has a version entry now
     assert len(model.get_versions(pull_request)) == 1
 
 
 def test_update_skips_new_version_if_unchanged(pr_util, config_stub):
     pull_request = pr_util.create_pull_request()
     model = PullRequestModel()
     model.update_commits(pull_request)
 
     # Expect that it still has no versions
     assert len(model.get_versions(pull_request)) == 0
 
 
 def test_update_assigns_comments_to_the_new_version(pr_util, config_stub):
     model = PullRequestModel()
     pull_request = pr_util.create_pull_request()
     comment = pr_util.create_comment()
     pr_util.update_source_repository()
 
     model.update_commits(pull_request)
 
     # Expect that the comment is linked to the pr version now
     assert comment.pull_request_version == model.get_versions(pull_request)[0]
 
 
 def test_update_adds_a_comment_to_the_pull_request_about_the_change(pr_util, config_stub):
     model = PullRequestModel()
     pull_request = pr_util.create_pull_request()
     pr_util.update_source_repository()
     pr_util.update_source_repository()
 
     model.update_commits(pull_request)
 
     # Expect to find a new comment about the change
     expected_message = textwrap.dedent(
         """\
         Pull request updated. Auto status change to |under_review|
 
         .. role:: added
         .. role:: removed
         .. parsed-literal::
 
           Changed commits:
             * :added:`1 added`
             * :removed:`0 removed`
 
           Changed files:
             * `A file_2 <#a_c--92ed3b5f07b4>`_
 
         .. |under_review| replace:: *"Under Review"*"""
     )
     pull_request_comments = sorted(
         pull_request.comments, key=lambda c: c.modified_at)
     update_comment = pull_request_comments[-1]
     assert update_comment.text == expected_message
 
 
 def test_create_version_from_snapshot_updates_attributes(pr_util, config_stub):
     pull_request = pr_util.create_pull_request()
 
     # Avoiding default values
     pull_request.status = PullRequest.STATUS_CLOSED
     pull_request._last_merge_source_rev = "0" * 40
     pull_request._last_merge_target_rev = "1" * 40
     pull_request.last_merge_status = 1
     pull_request.merge_rev = "2" * 40
 
     # Remember automatic values
     created_on = pull_request.created_on
     updated_on = pull_request.updated_on
 
     # Create a new version of the pull request
     version = PullRequestModel()._create_version_from_snapshot(pull_request)
 
     # Check attributes
     assert version.title == pr_util.create_parameters['title']
     assert version.description == pr_util.create_parameters['description']
     assert version.status == PullRequest.STATUS_CLOSED
 
     # versions get updated created_on
     assert version.created_on != created_on
 
     assert version.updated_on == updated_on
     assert version.user_id == pull_request.user_id
     assert version.revisions == pr_util.create_parameters['revisions']
     assert version.source_repo == pr_util.source_repository
     assert version.source_ref == pr_util.create_parameters['source_ref']
     assert version.target_repo == pr_util.target_repository
     assert version.target_ref == pr_util.create_parameters['target_ref']
     assert version._last_merge_source_rev == pull_request._last_merge_source_rev
     assert version._last_merge_target_rev == pull_request._last_merge_target_rev
     assert version.last_merge_status == pull_request.last_merge_status
     assert version.merge_rev == pull_request.merge_rev
     assert version.pull_request == pull_request
 
 
 def test_link_comments_to_version_only_updates_unlinked_comments(pr_util, config_stub):
     version1 = pr_util.create_version_of_pull_request()
     comment_linked = pr_util.create_comment(linked_to=version1)
     comment_unlinked = pr_util.create_comment()
     version2 = pr_util.create_version_of_pull_request()
 
     PullRequestModel()._link_comments_to_version(version2)
     Session().commit()
 
     # Expect that only the new comment is linked to version2
     assert (
         comment_unlinked.pull_request_version_id ==
         version2.pull_request_version_id)
     assert (
         comment_linked.pull_request_version_id ==
         version1.pull_request_version_id)
     assert (
         comment_unlinked.pull_request_version_id !=
         comment_linked.pull_request_version_id)
 
 
 def test_calculate_commits():
     old_ids = [1, 2, 3]
     new_ids = [1, 3, 4, 5]
     change = PullRequestModel()._calculate_commit_id_changes(old_ids, new_ids)
     assert change.added == [4, 5]
     assert change.common == [1, 3]
     assert change.removed == [2]
     assert change.total == [1, 3, 4, 5]
 
 
 def assert_inline_comments(pull_request, visible=None, outdated=None):
     if visible is not None:
         inline_comments = CommentsModel().get_inline_comments(
             pull_request.target_repo.repo_id, pull_request=pull_request)
         inline_cnt = CommentsModel().get_inline_comments_count(
             inline_comments)
         assert inline_cnt == visible
     if outdated is not None:
         outdated_comments = CommentsModel().get_outdated_comments(
             pull_request.target_repo.repo_id, pull_request)
         assert len(outdated_comments) == outdated
 
 
 def assert_pr_file_changes(
         pull_request, added=None, modified=None, removed=None):
     pr_versions = PullRequestModel().get_versions(pull_request)
     # always use first version, ie original PR to calculate changes
     pull_request_version = pr_versions[0]
     old_diff_data, new_diff_data = PullRequestModel()._generate_update_diffs(
         pull_request, pull_request_version)
     file_changes = PullRequestModel()._calculate_file_changes(
         old_diff_data, new_diff_data)
 
     assert added == file_changes.added, \
         'expected added:%s vs value:%s' % (added, file_changes.added)
     assert modified == file_changes.modified, \
         'expected modified:%s vs value:%s' % (modified, file_changes.modified)
     assert removed == file_changes.removed, \
         'expected removed:%s vs value:%s' % (removed, file_changes.removed)
 
 
 def outdated_comments_patcher(use_outdated=True):
     return mock.patch.object(
         CommentsModel, 'use_outdated_comments',
         return_value=use_outdated)