archives: fixed bugs with serving archives from non-ascii repos, and also deliver archives at much bigger reading blocks for faster downloads

rhodecode/apps/repository/views/repo_files.py

@@ -1,1581 +1,1582 b''
 # Copyright (C) 2011-2023 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 itertools
 import logging
 import os
 import collections
 import urllib.request
 import urllib.parse
 import urllib.error
 import pathlib
 
 from pyramid.httpexceptions import HTTPNotFound, HTTPBadRequest, HTTPFound
 
 from pyramid.renderers import render
 from pyramid.response import Response
 
 import rhodecode
 from rhodecode.apps._base import RepoAppView
 
 
 from rhodecode.lib import diffs, helpers as h, rc_cache
 from rhodecode.lib import audit_logger
 from rhodecode.lib.hash_utils import sha1_safe
 from rhodecode.lib.rc_cache.archive_cache import get_archival_cache_store, get_archival_config, ReentrantLock
-from rhodecode.lib.str_utils import safe_bytes
+from rhodecode.lib.str_utils import safe_bytes, convert_special_chars
 from rhodecode.lib.view_utils import parse_path_ref
 from rhodecode.lib.exceptions import NonRelativePathError
 from rhodecode.lib.codeblocks import (
     filenode_as_lines_tokens, filenode_as_annotated_lines_tokens)
 from rhodecode.lib.utils2 import convert_line_endings, detect_mode
 from rhodecode.lib.type_utils import str2bool
 from rhodecode.lib.str_utils import safe_str, safe_int
 from rhodecode.lib.auth import (
     LoginRequired, HasRepoPermissionAnyDecorator, CSRFRequired)
 from rhodecode.lib.vcs import path as vcspath
 from rhodecode.lib.vcs.backends.base import EmptyCommit
 from rhodecode.lib.vcs.conf import settings
 from rhodecode.lib.vcs.nodes import FileNode
 from rhodecode.lib.vcs.exceptions import (
     RepositoryError, CommitDoesNotExistError, EmptyRepositoryError,
     ImproperArchiveTypeError, VCSError, NodeAlreadyExistsError,
     NodeDoesNotExistError, CommitError, NodeError)
 
 from rhodecode.model.scm import ScmModel
 from rhodecode.model.db import Repository
 
 log = logging.getLogger(__name__)
 
 
-def get_archive_name(db_repo_name, commit_sha, ext, subrepos=False, path_sha='', with_hash=True):
+def get_archive_name(db_repo_id, db_repo_name, commit_sha, ext, subrepos=False, path_sha='', with_hash=True):
     # original backward compat name of archive
-    clean_name = safe_str(db_repo_name.replace('/', '_'))
+    clean_name = safe_str(convert_special_chars(db_repo_name).replace('/', '_'))
 
-    # e.g vcsserver-sub-1-abcfdef-archive-all.zip
-    # vcsserver-sub-0-abcfdef-COMMIT_SHA-PATH_SHA.zip
-
+    # e.g vcsserver-id-abcd-sub-1-abcfdef-archive-all.zip
+    # vcsserver-id-abcd-sub-0-abcfdef-COMMIT_SHA-PATH_SHA.zip
+    id_sha = sha1_safe(str(db_repo_id))[:4]
     sub_repo = 'sub-1' if subrepos else 'sub-0'
     commit = commit_sha if with_hash else 'archive'
     path_marker = (path_sha if with_hash else '') or 'all'
-    archive_name = f'{clean_name}-{sub_repo}-{commit}-{path_marker}{ext}'
+    archive_name = f'{clean_name}-id-{id_sha}-{sub_repo}-{commit}-{path_marker}{ext}'
 
     return archive_name
 
 
 def get_path_sha(at_path):
     return safe_str(sha1_safe(at_path)[:8])
 
 
 def _get_archive_spec(fname):
     log.debug('Detecting archive spec for: `%s`', fname)
 
     fileformat = None
     ext = None
     content_type = None
     for a_type, content_type, extension in settings.ARCHIVE_SPECS:
 
         if fname.endswith(extension):
             fileformat = a_type
             log.debug('archive is of type: %s', fileformat)
             ext = extension
             break
 
     if not fileformat:
         raise ValueError()
 
     # left over part of whole fname is the commit
     commit_id = fname[:-len(ext)]
 
     return commit_id, ext, fileformat, content_type
 
 
 class RepoFilesView(RepoAppView):
 
     @staticmethod
     def adjust_file_path_for_svn(f_path, repo):
         """
         Computes the relative path of `f_path`.
 
         This is mainly based on prefix matching of the recognized tags and
         branches in the underlying repository.
         """
         tags_and_branches = itertools.chain(
             repo.branches.keys(),
             repo.tags.keys())
         tags_and_branches = sorted(tags_and_branches, key=len, reverse=True)
 
         for name in tags_and_branches:
             if f_path.startswith(f'{name}/'):
                 f_path = vcspath.relpath(f_path, name)
                 break
         return f_path
 
     def load_default_context(self):
         c = self._get_local_tmpl_context(include_app_defaults=True)
         c.rhodecode_repo = self.rhodecode_vcs_repo
         c.enable_downloads = self.db_repo.enable_downloads
         return c
 
     def _ensure_not_locked(self, commit_id='tip'):
         _ = self.request.translate
 
         repo = self.db_repo
         if repo.enable_locking and repo.locked[0]:
             h.flash(_('This repository has been locked by %s on %s')
                     % (h.person_by_id(repo.locked[0]),
                     h.format_date(h.time_to_datetime(repo.locked[1]))),
                     'warning')
             files_url = h.route_path(
                 'repo_files:default_path',
                 repo_name=self.db_repo_name, commit_id=commit_id)
             raise HTTPFound(files_url)
 
     def forbid_non_head(self, is_head, f_path, commit_id='tip', json_mode=False):
         _ = self.request.translate
 
         if not is_head:
             message = _('Cannot modify file. '
                         'Given commit `{}` is not head of a branch.').format(commit_id)
             h.flash(message, category='warning')
 
             if json_mode:
                 return message
 
             files_url = h.route_path(
                 'repo_files', repo_name=self.db_repo_name, commit_id=commit_id,
                 f_path=f_path)
             raise HTTPFound(files_url)
 
     def check_branch_permission(self, branch_name, commit_id='tip', json_mode=False):
         _ = self.request.translate
 
         rule, branch_perm = self._rhodecode_user.get_rule_and_branch_permission(
             self.db_repo_name, branch_name)
         if branch_perm and branch_perm not in ['branch.push', 'branch.push_force']:
             message = _('Branch `{}` changes forbidden by rule {}.').format(
                 h.escape(branch_name), h.escape(rule))
             h.flash(message, 'warning')
 
             if json_mode:
                 return message
 
             files_url = h.route_path(
                 'repo_files:default_path', repo_name=self.db_repo_name, commit_id=commit_id)
 
             raise HTTPFound(files_url)
 
     def _get_commit_and_path(self):
         default_commit_id = self.db_repo.landing_ref_name
         default_f_path = '/'
 
         commit_id = self.request.matchdict.get(
             'commit_id', default_commit_id)
         f_path = self._get_f_path(self.request.matchdict, default_f_path)
         return commit_id, f_path
 
     def _get_default_encoding(self, c):
         enc_list = getattr(c, 'default_encodings', [])
         return enc_list[0] if enc_list else 'UTF-8'
 
     def _get_commit_or_redirect(self, commit_id, redirect_after=True):
         """
         This is a safe way to get commit. If an error occurs it redirects to
         tip with proper message
 
         :param commit_id: id of commit to fetch
         :param redirect_after: toggle redirection
         """
         _ = self.request.translate
 
         try:
             return self.rhodecode_vcs_repo.get_commit(commit_id)
         except EmptyRepositoryError:
             if not redirect_after:
                 return None
 
             add_new = upload_new = ""
             if h.HasRepoPermissionAny(
                     'repository.write', 'repository.admin')(self.db_repo_name):
                 _url = h.route_path(
                     'repo_files_add_file',
                     repo_name=self.db_repo_name, commit_id=0, f_path='')
                 add_new = h.link_to(
                     _('add a new file'), _url, class_="alert-link")
 
                 _url_upld = h.route_path(
                     'repo_files_upload_file',
                     repo_name=self.db_repo_name, commit_id=0, f_path='')
                 upload_new = h.link_to(
                     _('upload a new file'), _url_upld, class_="alert-link")
 
             h.flash(h.literal(
                 _('There are no files yet. Click here to %s or %s.') % (add_new, upload_new)), category='warning')
             raise HTTPFound(
                 h.route_path('repo_summary', repo_name=self.db_repo_name))
 
         except (CommitDoesNotExistError, LookupError) as e:
             msg = _('No such commit exists for this repository. Commit: {}').format(commit_id)
             h.flash(msg, category='error')
             raise HTTPNotFound()
         except RepositoryError as e:
             h.flash(h.escape(safe_str(e)), category='error')
             raise HTTPNotFound()
 
     def _get_filenode_or_redirect(self, commit_obj, path, pre_load=None):
         """
         Returns file_node, if error occurs or given path is directory,
         it'll redirect to top level path
         """
         _ = self.request.translate
 
         try:
             file_node = commit_obj.get_node(path, pre_load=pre_load)
             if file_node.is_dir():
                 raise RepositoryError('The given path is a directory')
         except CommitDoesNotExistError:
             log.exception('No such commit exists for this repository')
             h.flash(_('No such commit exists for this repository'), category='error')
             raise HTTPNotFound()
         except RepositoryError as e:
             log.warning('Repository error while fetching filenode `%s`. Err:%s', path, e)
             h.flash(h.escape(safe_str(e)), category='error')
             raise HTTPNotFound()
 
         return file_node
 
     def _is_valid_head(self, commit_id, repo, landing_ref):
         branch_name = sha_commit_id = ''
         is_head = False
         log.debug('Checking if commit_id `%s` is a head for %s.', commit_id, repo)
 
         for _branch_name, branch_commit_id in repo.branches.items():
             # simple case we pass in branch name, it's a HEAD
             if commit_id == _branch_name:
                 is_head = True
                 branch_name = _branch_name
                 sha_commit_id = branch_commit_id
                 break
             # case when we pass in full sha commit_id, which is a head
             elif commit_id == branch_commit_id:
                 is_head = True
                 branch_name = _branch_name
                 sha_commit_id = branch_commit_id
                 break
 
         if h.is_svn(repo) and not repo.is_empty():
             # Note: Subversion only has one head.
             if commit_id == repo.get_commit(commit_idx=-1).raw_id:
                 is_head = True
                 return branch_name, sha_commit_id, is_head
 
         # checked branches, means we only need to try to get the branch/commit_sha
         if repo.is_empty():
             is_head = True
             branch_name = landing_ref
             sha_commit_id = EmptyCommit().raw_id
         else:
             commit = repo.get_commit(commit_id=commit_id)
             if commit:
                 branch_name = commit.branch
                 sha_commit_id = commit.raw_id
 
         return branch_name, sha_commit_id, is_head
 
     def _get_tree_at_commit(self, c, commit_id, f_path, full_load=False, at_rev=None):
 
         repo_id = self.db_repo.repo_id
         force_recache = self.get_recache_flag()
 
         cache_seconds = safe_int(
             rhodecode.CONFIG.get('rc_cache.cache_repo.expiration_time'))
         cache_on = not force_recache and cache_seconds > 0
         log.debug(
             'Computing FILE TREE for repo_id %s commit_id `%s` and path `%s`'
             'with caching: %s[TTL: %ss]' % (
                 repo_id, commit_id, f_path, cache_on, cache_seconds or 0))
 
         cache_namespace_uid = f'repo.{rc_cache.FILE_TREE_CACHE_VER}.{repo_id}'
         region = rc_cache.get_or_create_region('cache_repo', cache_namespace_uid)
 
         @region.conditional_cache_on_arguments(namespace=cache_namespace_uid, condition=cache_on)
         def compute_file_tree(_name_hash, _repo_id, _commit_id, _f_path, _full_load, _at_rev):
             log.debug('Generating cached file tree at for repo_id: %s, %s, %s',
                       _repo_id, _commit_id, _f_path)
 
             c.full_load = _full_load
             return render(
                 'rhodecode:templates/files/files_browser_tree.mako',
                 self._get_template_context(c), self.request, _at_rev)
 
         return compute_file_tree(
             self.db_repo.repo_name_hash, self.db_repo.repo_id, commit_id, f_path, full_load, at_rev)
 
     def create_pure_path(self, *parts):
         # Split paths and sanitize them, removing any ../ etc
         sanitized_path = [
             x for x in pathlib.PurePath(*parts).parts
             if x not in ['.', '..']]
 
         pure_path = pathlib.PurePath(*sanitized_path)
         return pure_path
 
     def _is_lf_enabled(self, target_repo):
         lf_enabled = False
 
         lf_key_for_vcs_map = {
             'hg': 'extensions_largefiles',
             'git': 'vcs_git_lfs_enabled'
         }
 
         lf_key_for_vcs = lf_key_for_vcs_map.get(target_repo.repo_type)
 
         if lf_key_for_vcs:
             lf_enabled = self._get_repo_setting(target_repo, lf_key_for_vcs)
 
         return lf_enabled
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     def repo_archivefile(self):
         # archive cache config
         from rhodecode import CONFIG
         _ = self.request.translate
         self.load_default_context()
         default_at_path = '/'
         fname = self.request.matchdict['fname']
         subrepos = self.request.GET.get('subrepos') == 'true'
         with_hash = str2bool(self.request.GET.get('with_hash', '1'))
         at_path = self.request.GET.get('at_path') or default_at_path
 
         if not self.db_repo.enable_downloads:
             return Response(_('Downloads disabled'))
 
         try:
             commit_id, ext, fileformat, content_type = \
                 _get_archive_spec(fname)
         except ValueError:
             return Response(_('Unknown archive type for: `{}`').format(
                 h.escape(fname)))
 
         try:
             commit = self.rhodecode_vcs_repo.get_commit(commit_id)
         except CommitDoesNotExistError:
             return Response(_('Unknown commit_id {}').format(
                 h.escape(commit_id)))
         except EmptyRepositoryError:
             return Response(_('Empty repository'))
 
         # we used a ref, or a shorter version, lets redirect client ot use explicit hash
         if commit_id != commit.raw_id:
             fname=f'{commit.raw_id}{ext}'
             raise HTTPFound(self.request.current_route_path(fname=fname))
 
         try:
             at_path = commit.get_node(at_path).path or default_at_path
         except Exception:
             return Response(_('No node at path {} for this repository').format(h.escape(at_path)))
 
         path_sha = get_path_sha(at_path)
 
         # used for cache etc, consistent unique archive name
         archive_name_key = get_archive_name(
-            self.db_repo_name, commit_sha=commit.short_id, ext=ext, subrepos=subrepos,
+            self.db_repo.repo_id, self.db_repo_name, commit_sha=commit.short_id, ext=ext, subrepos=subrepos,
             path_sha=path_sha, with_hash=True)
 
         if not with_hash:
             path_sha = ''
 
         # what end client gets served
         response_archive_name = get_archive_name(
-            self.db_repo_name, commit_sha=commit.short_id, ext=ext, subrepos=subrepos,
+            self.db_repo.repo_id, self.db_repo_name, commit_sha=commit.short_id, ext=ext, subrepos=subrepos,
             path_sha=path_sha, with_hash=with_hash)
 
         # remove extension from our archive directory name
         archive_dir_name = response_archive_name[:-len(ext)]
 
         archive_cache_disable = self.request.GET.get('no_cache')
 
         d_cache = get_archival_cache_store(config=CONFIG)
         # NOTE: we get the config to pass to a call to lazy-init the SAME type of cache on vcsserver
         d_cache_conf = get_archival_config(config=CONFIG)
 
         reentrant_lock_key = archive_name_key + '.lock'
         with ReentrantLock(d_cache, reentrant_lock_key):
             # This is also a cache key
             use_cached_archive = False
             if archive_name_key in d_cache and not archive_cache_disable:
                 reader, tag = d_cache.get(archive_name_key, read=True, tag=True, retry=True)
                 use_cached_archive = True
                 log.debug('Found cached archive as key=%s tag=%s, serving archive from cache reader=%s',
                           archive_name_key, tag, reader.name)
             else:
                 reader = None
                 log.debug('Archive with key=%s is not yet cached, creating one now...', archive_name_key)
 
             # generate new archive, as previous was not found in the cache
             if not reader:
 
                 try:
                     commit.archive_repo(archive_name_key, archive_dir_name=archive_dir_name,
                                         kind=fileformat, subrepos=subrepos,
                                         archive_at_path=at_path, cache_config=d_cache_conf)
                 except ImproperArchiveTypeError:
                     return _('Unknown archive type')
 
             reader, tag = d_cache.get(archive_name_key, read=True, tag=True, retry=True)
 
         if not reader:
             raise ValueError('archive cache reader is empty, failed to fetch file from distributed archive cache')
 
-        def archive_iterator(_reader):
+        def archive_iterator(_reader, block_size: int = 4096*512):
+            # 4096 * 64 = 64KB
             while 1:
-                data = _reader.read(1024)
+                data = _reader.read(block_size)
                 if not data:
                     break
                 yield data
 
         response = Response(app_iter=archive_iterator(reader))
         response.content_disposition = f'attachment; filename={response_archive_name}'
         response.content_type = str(content_type)
 
         try:
             return response
         finally:
             # store download action
             audit_logger.store_web(
                 'repo.archive.download', action_data={
                     'user_agent': self.request.user_agent,
                     'archive_name': archive_name_key,
                     'archive_spec': fname,
                     'archive_cached': use_cached_archive},
                 user=self._rhodecode_user,
                 repo=self.db_repo,
                 commit=True
             )
 
     def _get_file_node(self, commit_id, f_path):
         if commit_id not in ['', None, 'None', '0' * 12, '0' * 40]:
             commit = self.rhodecode_vcs_repo.get_commit(commit_id=commit_id)
             try:
                 node = commit.get_node(f_path)
                 if node.is_dir():
                     raise NodeError(f'{node} path is a {type(node)} not a file')
             except NodeDoesNotExistError:
                 commit = EmptyCommit(
                     commit_id=commit_id,
                     idx=commit.idx,
                     repo=commit.repository,
                     alias=commit.repository.alias,
                     message=commit.message,
                     author=commit.author,
                     date=commit.date)
                 node = FileNode(safe_bytes(f_path), b'', commit=commit)
         else:
             commit = EmptyCommit(
                 repo=self.rhodecode_vcs_repo,
                 alias=self.rhodecode_vcs_repo.alias)
             node = FileNode(safe_bytes(f_path), b'', commit=commit)
         return node
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     def repo_files_diff(self):
         c = self.load_default_context()
         f_path = self._get_f_path(self.request.matchdict)
         diff1 = self.request.GET.get('diff1', '')
         diff2 = self.request.GET.get('diff2', '')
 
         path1, diff1 = parse_path_ref(diff1, default_path=f_path)
 
         ignore_whitespace = str2bool(self.request.GET.get('ignorews'))
         line_context = self.request.GET.get('context', 3)
 
         if not any((diff1, diff2)):
             h.flash(
                 'Need query parameter "diff1" or "diff2" to generate a diff.',
                 category='error')
             raise HTTPBadRequest()
 
         c.action = self.request.GET.get('diff')
         if c.action not in ['download', 'raw']:
             compare_url = h.route_path(
                 'repo_compare',
                 repo_name=self.db_repo_name,
                 source_ref_type='rev',
                 source_ref=diff1,
                 target_repo=self.db_repo_name,
                 target_ref_type='rev',
                 target_ref=diff2,
                 _query=dict(f_path=f_path))
             # redirect to new view if we render diff
             raise HTTPFound(compare_url)
 
         try:
             node1 = self._get_file_node(diff1, path1)
             node2 = self._get_file_node(diff2, f_path)
         except (RepositoryError, NodeError):
             log.exception("Exception while trying to get node from repository")
             raise HTTPFound(
                 h.route_path('repo_files', repo_name=self.db_repo_name,
                              commit_id='tip', f_path=f_path))
 
         if all(isinstance(node.commit, EmptyCommit)
                for node in (node1, node2)):
             raise HTTPNotFound()
 
         c.commit_1 = node1.commit
         c.commit_2 = node2.commit
 
         if c.action == 'download':
             _diff = diffs.get_gitdiff(node1, node2,
                                       ignore_whitespace=ignore_whitespace,
                                       context=line_context)
             # NOTE: this was using diff_format='gitdiff'
             diff = diffs.DiffProcessor(_diff, diff_format='newdiff')
 
             response = Response(self.path_filter.get_raw_patch(diff))
             response.content_type = 'text/plain'
             response.content_disposition = (
                 f'attachment; filename={f_path}_{diff1}_vs_{diff2}.diff'
             )
             charset = self._get_default_encoding(c)
             if charset:
                 response.charset = charset
             return response
 
         elif c.action == 'raw':
             _diff = diffs.get_gitdiff(node1, node2,
                                       ignore_whitespace=ignore_whitespace,
                                       context=line_context)
             # NOTE: this was using diff_format='gitdiff'
             diff = diffs.DiffProcessor(_diff, diff_format='newdiff')
 
             response = Response(self.path_filter.get_raw_patch(diff))
             response.content_type = 'text/plain'
             charset = self._get_default_encoding(c)
             if charset:
                 response.charset = charset
             return response
 
         # in case we ever end up here
         raise HTTPNotFound()
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     def repo_files_diff_2way_redirect(self):
         """
         Kept only to make OLD links work
         """
         f_path = self._get_f_path_unchecked(self.request.matchdict)
         diff1 = self.request.GET.get('diff1', '')
         diff2 = self.request.GET.get('diff2', '')
 
         if not any((diff1, diff2)):
             h.flash(
                 'Need query parameter "diff1" or "diff2" to generate a diff.',
                 category='error')
             raise HTTPBadRequest()
 
         compare_url = h.route_path(
             'repo_compare',
             repo_name=self.db_repo_name,
             source_ref_type='rev',
             source_ref=diff1,
             target_ref_type='rev',
             target_ref=diff2,
             _query=dict(f_path=f_path, diffmode='sideside',
                         target_repo=self.db_repo_name,))
         raise HTTPFound(compare_url)
 
     @LoginRequired()
     def repo_files_default_commit_redirect(self):
         """
         Special page that redirects to the landing page of files based on the default
         commit for repository
         """
         c = self.load_default_context()
         ref_name = c.rhodecode_db_repo.landing_ref_name
         landing_url = h.repo_files_by_ref_url(
             c.rhodecode_db_repo.repo_name,
             c.rhodecode_db_repo.repo_type,
             f_path='',
             ref_name=ref_name,
             commit_id='tip',
             query=dict(at=ref_name)
         )
 
         raise HTTPFound(landing_url)
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     def repo_files(self):
         c = self.load_default_context()
 
         view_name = getattr(self.request.matched_route, 'name', None)
 
         c.annotate = view_name == 'repo_files:annotated'
         # default is false, but .rst/.md files later are auto rendered, we can
         # overwrite auto rendering by setting this GET flag
         c.renderer = view_name == 'repo_files:rendered' or not self.request.GET.get('no-render', False)
 
         commit_id, f_path = self._get_commit_and_path()
 
         c.commit = self._get_commit_or_redirect(commit_id)
         c.branch = self.request.GET.get('branch', None)
         c.f_path = f_path
         at_rev = self.request.GET.get('at')
 
         # prev link
         try:
             prev_commit = c.commit.prev(c.branch)
             c.prev_commit = prev_commit
             c.url_prev = h.route_path(
                 'repo_files', repo_name=self.db_repo_name,
                 commit_id=prev_commit.raw_id, f_path=f_path)
             if c.branch:
                 c.url_prev += '?branch=%s' % c.branch
         except (CommitDoesNotExistError, VCSError):
             c.url_prev = '#'
             c.prev_commit = EmptyCommit()
 
         # next link
         try:
             next_commit = c.commit.next(c.branch)
             c.next_commit = next_commit
             c.url_next = h.route_path(
                 'repo_files', repo_name=self.db_repo_name,
                 commit_id=next_commit.raw_id, f_path=f_path)
             if c.branch:
                 c.url_next += '?branch=%s' % c.branch
         except (CommitDoesNotExistError, VCSError):
             c.url_next = '#'
             c.next_commit = EmptyCommit()
 
         # files or dirs
         try:
             c.file = c.commit.get_node(f_path, pre_load=['is_binary', 'size', 'data'])
 
             c.file_author = True
             c.file_tree = ''
 
             # load file content
             if c.file.is_file():
                 c.lf_node = {}
 
                 has_lf_enabled = self._is_lf_enabled(self.db_repo)
                 if has_lf_enabled:
                     c.lf_node = c.file.get_largefile_node()
 
                 c.file_source_page = 'true'
                 c.file_last_commit = c.file.last_commit
 
                 c.file_size_too_big = c.file.size > c.visual.cut_off_limit_file
 
                 if not (c.file_size_too_big or c.file.is_binary):
                     if c.annotate:  # annotation has precedence over renderer
                         c.annotated_lines = filenode_as_annotated_lines_tokens(
                             c.file
                         )
                     else:
                         c.renderer = (
                             c.renderer and h.renderer_from_filename(c.file.path)
                         )
                         if not c.renderer:
                             c.lines = filenode_as_lines_tokens(c.file)
 
                 _branch_name, _sha_commit_id, is_head = \
                     self._is_valid_head(commit_id, self.rhodecode_vcs_repo,
                                         landing_ref=self.db_repo.landing_ref_name)
                 c.on_branch_head = is_head
 
                 branch = c.commit.branch if (
                     c.commit.branch and '/' not in c.commit.branch) else None
                 c.branch_or_raw_id = branch or c.commit.raw_id
                 c.branch_name = c.commit.branch or h.short_id(c.commit.raw_id)
 
                 author = c.file_last_commit.author
                 c.authors = [[
                     h.email(author),
                     h.person(author, 'username_or_name_or_email'),
                     1
                 ]]
 
             else:  # load tree content at path
                 c.file_source_page = 'false'
                 c.authors = []
                 # this loads a simple tree without metadata to speed things up
                 # later via ajax we call repo_nodetree_full and fetch whole
                 c.file_tree = self._get_tree_at_commit(c, c.commit.raw_id, f_path, at_rev=at_rev)
 
                 c.readme_data, c.readme_file = \
                     self._get_readme_data(self.db_repo, c.visual.default_renderer,
                                           c.commit.raw_id, f_path)
 
         except RepositoryError as e:
             h.flash(h.escape(safe_str(e)), category='error')
             raise HTTPNotFound()
 
         if self.request.environ.get('HTTP_X_PJAX'):
             html = render('rhodecode:templates/files/files_pjax.mako',
                           self._get_template_context(c), self.request)
         else:
             html = render('rhodecode:templates/files/files.mako',
                           self._get_template_context(c), self.request)
         return Response(html)
 
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     def repo_files_annotated_previous(self):
         self.load_default_context()
 
         commit_id, f_path = self._get_commit_and_path()
         commit = self._get_commit_or_redirect(commit_id)
         prev_commit_id = commit.raw_id
         line_anchor = self.request.GET.get('line_anchor')
         is_file = False
         try:
             _file = commit.get_node(f_path)
             is_file = _file.is_file()
         except (NodeDoesNotExistError, CommitDoesNotExistError, VCSError):
             pass
 
         if is_file:
             history = commit.get_path_history(f_path)
             prev_commit_id = history[1].raw_id \
                 if len(history) > 1 else prev_commit_id
         prev_url = h.route_path(
             'repo_files:annotated', repo_name=self.db_repo_name,
             commit_id=prev_commit_id, f_path=f_path,
             _anchor=f'L{line_anchor}')
 
         raise HTTPFound(prev_url)
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     def repo_nodetree_full(self):
         """
         Returns rendered html of file tree that contains commit date,
         author, commit_id for the specified combination of
         repo, commit_id and file path
         """
         c = self.load_default_context()
 
         commit_id, f_path = self._get_commit_and_path()
         commit = self._get_commit_or_redirect(commit_id)
         try:
             dir_node = commit.get_node(f_path)
         except RepositoryError as e:
             return Response(f'error: {h.escape(safe_str(e))}')
 
         if dir_node.is_file():
             return Response('')
 
         c.file = dir_node
         c.commit = commit
         at_rev = self.request.GET.get('at')
 
         html = self._get_tree_at_commit(
             c, commit.raw_id, dir_node.path, full_load=True, at_rev=at_rev)
 
         return Response(html)
 
     def _get_attachement_headers(self, f_path):
         f_name = safe_str(f_path.split(Repository.NAME_SEP)[-1])
         safe_path = f_name.replace('"', '\\"')
         encoded_path = urllib.parse.quote(f_name)
 
         return "attachment; " \
                "filename=\"{}\"; " \
                "filename*=UTF-8\'\'{}".format(safe_path, encoded_path)
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     def repo_file_raw(self):
         """
         Action for show as raw, some mimetypes are "rendered",
         those include images, icons.
         """
         c = self.load_default_context()
 
         commit_id, f_path = self._get_commit_and_path()
         commit = self._get_commit_or_redirect(commit_id)
         file_node = self._get_filenode_or_redirect(commit, f_path)
 
         raw_mimetype_mapping = {
             # map original mimetype to a mimetype used for "show as raw"
             # you can also provide a content-disposition to override the
             # default "attachment" disposition.
             # orig_type: (new_type, new_dispo)
 
             # show images inline:
             # Do not re-add SVG: it is unsafe and permits XSS attacks. One can
             # for example render an SVG with javascript inside or even render
             # HTML.
             'image/x-icon': ('image/x-icon', 'inline'),
             'image/png': ('image/png', 'inline'),
             'image/gif': ('image/gif', 'inline'),
             'image/jpeg': ('image/jpeg', 'inline'),
             'application/pdf': ('application/pdf', 'inline'),
         }
 
         mimetype = file_node.mimetype
         try:
             mimetype, disposition = raw_mimetype_mapping[mimetype]
         except KeyError:
             # we don't know anything special about this, handle it safely
             if file_node.is_binary:
                 # do same as download raw for binary files
                 mimetype, disposition = 'application/octet-stream', 'attachment'
             else:
                 # do not just use the original mimetype, but force text/plain,
                 # otherwise it would serve text/html and that might be unsafe.
                 # Note: underlying vcs library fakes text/plain mimetype if the
                 # mimetype can not be determined and it thinks it is not
                 # binary.This might lead to erroneous text display in some
                 # cases, but helps in other cases, like with text files
                 # without extension.
                 mimetype, disposition = 'text/plain', 'inline'
 
         if disposition == 'attachment':
             disposition = self._get_attachement_headers(f_path)
 
         stream_content = file_node.stream_bytes()
 
         response = Response(app_iter=stream_content)
         response.content_disposition = disposition
         response.content_type = mimetype
 
         charset = self._get_default_encoding(c)
         if charset:
             response.charset = charset
 
         return response
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     def repo_file_download(self):
         c = self.load_default_context()
 
         commit_id, f_path = self._get_commit_and_path()
         commit = self._get_commit_or_redirect(commit_id)
         file_node = self._get_filenode_or_redirect(commit, f_path)
 
         if self.request.GET.get('lf'):
             # only if lf get flag is passed, we download this file
             # as LFS/Largefile
             lf_node = file_node.get_largefile_node()
             if lf_node:
                 # overwrite our pointer with the REAL large-file
                 file_node = lf_node
 
         disposition = self._get_attachement_headers(f_path)
 
         stream_content = file_node.stream_bytes()
 
         response = Response(app_iter=stream_content)
         response.content_disposition = disposition
         response.content_type = file_node.mimetype
 
         charset = self._get_default_encoding(c)
         if charset:
             response.charset = charset
 
         return response
 
     def _get_nodelist_at_commit(self, repo_name, repo_id, commit_id, f_path):
 
         cache_seconds = safe_int(
             rhodecode.CONFIG.get('rc_cache.cache_repo.expiration_time'))
         cache_on = cache_seconds > 0
         log.debug(
             'Computing FILE SEARCH for repo_id %s commit_id `%s` and path `%s`'
             'with caching: %s[TTL: %ss]' % (
                 repo_id, commit_id, f_path, cache_on, cache_seconds or 0))
 
         cache_namespace_uid = f'repo.{repo_id}'
         region = rc_cache.get_or_create_region('cache_repo', cache_namespace_uid)
 
         @region.conditional_cache_on_arguments(namespace=cache_namespace_uid, condition=cache_on)
         def compute_file_search(_name_hash, _repo_id, _commit_id, _f_path):
             log.debug('Generating cached nodelist for repo_id:%s, %s, %s',
                       _repo_id, commit_id, f_path)
             try:
                 _d, _f = ScmModel().get_quick_filter_nodes(repo_name, _commit_id, _f_path)
             except (RepositoryError, CommitDoesNotExistError, Exception) as e:
                 log.exception(safe_str(e))
                 h.flash(h.escape(safe_str(e)), category='error')
                 raise HTTPFound(h.route_path(
                     'repo_files', repo_name=self.db_repo_name,
                     commit_id='tip', f_path='/'))
 
             return _d + _f
 
         result = compute_file_search(self.db_repo.repo_name_hash, self.db_repo.repo_id,
                                      commit_id, f_path)
         return filter(lambda n: self.path_filter.path_access_allowed(n['name']), result)
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     def repo_nodelist(self):
         self.load_default_context()
 
         commit_id, f_path = self._get_commit_and_path()
         commit = self._get_commit_or_redirect(commit_id)
 
         metadata = self._get_nodelist_at_commit(
             self.db_repo_name, self.db_repo.repo_id, commit.raw_id, f_path)
         return {'nodes': [x for x in metadata]}
 
     def _create_references(self, branches_or_tags, symbolic_reference, f_path, ref_type):
         items = []
         for name, commit_id in branches_or_tags.items():
             sym_ref = symbolic_reference(commit_id, name, f_path, ref_type)
             items.append((sym_ref, name, ref_type))
         return items
 
     def _symbolic_reference(self, commit_id, name, f_path, ref_type):
         return commit_id
 
     def _symbolic_reference_svn(self, commit_id, name, f_path, ref_type):
         return commit_id
 
         # NOTE(dan): old code we used in "diff" mode compare
         new_f_path = vcspath.join(name, f_path)
         return f'{new_f_path}@{commit_id}'
 
     def _get_node_history(self, commit_obj, f_path, commits=None):
         """
         get commit history for given node
 
         :param commit_obj: commit to calculate history
         :param f_path: path for node to calculate history for
         :param commits: if passed don't calculate history and take
             commits defined in this list
         """
         _ = self.request.translate
 
         # calculate history based on tip
         tip = self.rhodecode_vcs_repo.get_commit()
         if commits is None:
             pre_load = ["author", "branch"]
             try:
                 commits = tip.get_path_history(f_path, pre_load=pre_load)
             except (NodeDoesNotExistError, CommitError):
                 # this node is not present at tip!
                 commits = commit_obj.get_path_history(f_path, pre_load=pre_load)
 
         history = []
         commits_group = ([], _("Changesets"))
         for commit in commits:
             branch = ' (%s)' % commit.branch if commit.branch else ''
             n_desc = f'r{commit.idx}:{commit.short_id}{branch}'
             commits_group[0].append((commit.raw_id, n_desc, 'sha'))
         history.append(commits_group)
 
         symbolic_reference = self._symbolic_reference
 
         if self.rhodecode_vcs_repo.alias == 'svn':
             adjusted_f_path = RepoFilesView.adjust_file_path_for_svn(
                 f_path, self.rhodecode_vcs_repo)
             if adjusted_f_path != f_path:
                 log.debug(
                     'Recognized svn tag or branch in file "%s", using svn '
                     'specific symbolic references', f_path)
                 f_path = adjusted_f_path
                 symbolic_reference = self._symbolic_reference_svn
 
         branches = self._create_references(
             self.rhodecode_vcs_repo.branches, symbolic_reference, f_path, 'branch')
         branches_group = (branches, _("Branches"))
 
         tags = self._create_references(
             self.rhodecode_vcs_repo.tags, symbolic_reference, f_path, 'tag')
         tags_group = (tags, _("Tags"))
 
         history.append(branches_group)
         history.append(tags_group)
 
         return history, commits
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     def repo_file_history(self):
         self.load_default_context()
 
         commit_id, f_path = self._get_commit_and_path()
         commit = self._get_commit_or_redirect(commit_id)
         file_node = self._get_filenode_or_redirect(commit, f_path)
 
         if file_node.is_file():
             file_history, _hist = self._get_node_history(commit, f_path)
 
             res = []
             for section_items, section in file_history:
                 items = []
                 for obj_id, obj_text, obj_type in section_items:
                     at_rev = ''
                     if obj_type in ['branch', 'bookmark', 'tag']:
                         at_rev = obj_text
                     entry = {
                         'id': obj_id,
                         'text': obj_text,
                         'type': obj_type,
                         'at_rev': at_rev
                     }
 
                     items.append(entry)
 
                 res.append({
                     'text': section,
                     'children': items
                 })
 
             data = {
                 'more': False,
                 'results': res
             }
             return data
 
         log.warning('Cannot fetch history for directory')
         raise HTTPBadRequest()
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     def repo_file_authors(self):
         c = self.load_default_context()
 
         commit_id, f_path = self._get_commit_and_path()
         commit = self._get_commit_or_redirect(commit_id)
         file_node = self._get_filenode_or_redirect(commit, f_path)
 
         if not file_node.is_file():
             raise HTTPBadRequest()
 
         c.file_last_commit = file_node.last_commit
         if self.request.GET.get('annotate') == '1':
             # use _hist from annotation if annotation mode is on
             commit_ids = {x[1] for x in file_node.annotate}
             _hist = (
                 self.rhodecode_vcs_repo.get_commit(commit_id)
                 for commit_id in commit_ids)
         else:
             _f_history, _hist = self._get_node_history(commit, f_path)
         c.file_author = False
 
         unique = collections.OrderedDict()
         for commit in _hist:
             author = commit.author
             if author not in unique:
                 unique[commit.author] = [
                     h.email(author),
                     h.person(author, 'username_or_name_or_email'),
                     1  # counter
                 ]
 
             else:
                 # increase counter
                 unique[commit.author][2] += 1
 
         c.authors = [val for val in unique.values()]
 
         return self._get_template_context(c)
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator('repository.write', 'repository.admin')
     def repo_files_check_head(self):
         self.load_default_context()
 
         commit_id, f_path = self._get_commit_and_path()
         _branch_name, _sha_commit_id, is_head = \
             self._is_valid_head(commit_id, self.rhodecode_vcs_repo,
                                 landing_ref=self.db_repo.landing_ref_name)
 
         new_path = self.request.POST.get('path')
         operation = self.request.POST.get('operation')
         path_exist = ''
 
         if new_path and operation in ['create', 'upload']:
             new_f_path = os.path.join(f_path.lstrip('/'), new_path)
             try:
                 commit_obj = self.rhodecode_vcs_repo.get_commit(commit_id)
                 # NOTE(dan): construct whole path without leading /
                 file_node = commit_obj.get_node(new_f_path)
                 if file_node is not None:
                     path_exist = new_f_path
             except EmptyRepositoryError:
                 pass
             except Exception:
                 pass
 
         return {
             'branch': _branch_name,
             'sha': _sha_commit_id,
             'is_head': is_head,
             'path_exists': path_exist
         }
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator('repository.write', 'repository.admin')
     def repo_files_remove_file(self):
         _ = self.request.translate
         c = self.load_default_context()
         commit_id, f_path = self._get_commit_and_path()
 
         self._ensure_not_locked()
         _branch_name, _sha_commit_id, is_head = \
             self._is_valid_head(commit_id, self.rhodecode_vcs_repo,
                                 landing_ref=self.db_repo.landing_ref_name)
 
         self.forbid_non_head(is_head, f_path)
         self.check_branch_permission(_branch_name)
 
         c.commit = self._get_commit_or_redirect(commit_id)
         c.file = self._get_filenode_or_redirect(c.commit, f_path)
 
         c.default_message = _(
             'Deleted file {} via RhodeCode Enterprise').format(f_path)
         c.f_path = f_path
 
         return self._get_template_context(c)
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator('repository.write', 'repository.admin')
     @CSRFRequired()
     def repo_files_delete_file(self):
         _ = self.request.translate
 
         c = self.load_default_context()
         commit_id, f_path = self._get_commit_and_path()
 
         self._ensure_not_locked()
         _branch_name, _sha_commit_id, is_head = \
             self._is_valid_head(commit_id, self.rhodecode_vcs_repo,
                                 landing_ref=self.db_repo.landing_ref_name)
 
         self.forbid_non_head(is_head, f_path)
         self.check_branch_permission(_branch_name)
 
         c.commit = self._get_commit_or_redirect(commit_id)
         c.file = self._get_filenode_or_redirect(c.commit, f_path)
 
         c.default_message = _(
             'Deleted file {} via RhodeCode Enterprise').format(f_path)
         c.f_path = f_path
         node_path = f_path
         author = self._rhodecode_db_user.full_contact
         message = self.request.POST.get('message') or c.default_message
         try:
             nodes = {
                 safe_bytes(node_path): {
                     'content': b''
                 }
             }
             ScmModel().delete_nodes(
                 user=self._rhodecode_db_user.user_id, repo=self.db_repo,
                 message=message,
                 nodes=nodes,
                 parent_commit=c.commit,
                 author=author,
             )
 
             h.flash(
                 _('Successfully deleted file `{}`').format(
                     h.escape(f_path)), category='success')
         except Exception:
             log.exception('Error during commit operation')
             h.flash(_('Error occurred during commit'), category='error')
         raise HTTPFound(
             h.route_path('repo_commit', repo_name=self.db_repo_name,
                          commit_id='tip'))
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator('repository.write', 'repository.admin')
     def repo_files_edit_file(self):
         _ = self.request.translate
         c = self.load_default_context()
         commit_id, f_path = self._get_commit_and_path()
 
         self._ensure_not_locked()
         _branch_name, _sha_commit_id, is_head = \
             self._is_valid_head(commit_id, self.rhodecode_vcs_repo,
                                 landing_ref=self.db_repo.landing_ref_name)
 
         self.forbid_non_head(is_head, f_path, commit_id=commit_id)
         self.check_branch_permission(_branch_name, commit_id=commit_id)
 
         c.commit = self._get_commit_or_redirect(commit_id)
         c.file = self._get_filenode_or_redirect(c.commit, f_path)
 
         if c.file.is_binary:
             files_url = h.route_path(
                 'repo_files',
                 repo_name=self.db_repo_name,
                 commit_id=c.commit.raw_id, f_path=f_path)
             raise HTTPFound(files_url)
 
         c.default_message = _('Edited file {} via RhodeCode Enterprise').format(f_path)
         c.f_path = f_path
 
         return self._get_template_context(c)
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator('repository.write', 'repository.admin')
     @CSRFRequired()
     def repo_files_update_file(self):
         _ = self.request.translate
         c = self.load_default_context()
         commit_id, f_path = self._get_commit_and_path()
 
         self._ensure_not_locked()
 
         c.commit = self._get_commit_or_redirect(commit_id)
         c.file = self._get_filenode_or_redirect(c.commit, f_path)
 
         if c.file.is_binary:
             raise HTTPFound(h.route_path('repo_files', repo_name=self.db_repo_name,
                                          commit_id=c.commit.raw_id, f_path=f_path))
 
         _branch_name, _sha_commit_id, is_head = \
             self._is_valid_head(commit_id, self.rhodecode_vcs_repo,
                                 landing_ref=self.db_repo.landing_ref_name)
 
         self.forbid_non_head(is_head, f_path, commit_id=commit_id)
         self.check_branch_permission(_branch_name, commit_id=commit_id)
 
         c.default_message = _('Edited file {} via RhodeCode Enterprise').format(f_path)
         c.f_path = f_path
 
         old_content = c.file.str_content
         sl = old_content.splitlines(1)
         first_line = sl[0] if sl else ''
 
         r_post = self.request.POST
         # line endings:  0 - Unix, 1 - Mac, 2 - DOS
         line_ending_mode = detect_mode(first_line, 0)
         content = convert_line_endings(r_post.get('content', ''), line_ending_mode)
 
         message = r_post.get('message') or c.default_message
 
         org_node_path = c.file.str_path
         filename = r_post['filename']
 
         root_path = c.file.dir_path
         pure_path = self.create_pure_path(root_path, filename)
         node_path = pure_path.as_posix()
 
         default_redirect_url = h.route_path('repo_commit', repo_name=self.db_repo_name,
                                             commit_id=commit_id)
         if content == old_content and node_path == org_node_path:
             h.flash(_('No changes detected on {}').format(h.escape(org_node_path)),
                     category='warning')
             raise HTTPFound(default_redirect_url)
 
         try:
             mapping = {
                 c.file.bytes_path: {
                     'org_filename': org_node_path,
                     'filename': safe_bytes(node_path),
                     'content': safe_bytes(content),
                     'lexer': '',
                     'op': 'mod',
                     'mode': c.file.mode
                 }
             }
 
             commit = ScmModel().update_nodes(
                 user=self._rhodecode_db_user.user_id,
                 repo=self.db_repo,
                 message=message,
                 nodes=mapping,
                 parent_commit=c.commit,
             )
 
             h.flash(_('Successfully committed changes to file `{}`').format(
                     h.escape(f_path)), category='success')
             default_redirect_url = h.route_path(
                 'repo_commit', repo_name=self.db_repo_name, commit_id=commit.raw_id)
 
         except Exception:
             log.exception('Error occurred during commit')
             h.flash(_('Error occurred during commit'), category='error')
 
         raise HTTPFound(default_redirect_url)
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator('repository.write', 'repository.admin')
     def repo_files_add_file(self):
         _ = self.request.translate
         c = self.load_default_context()
         commit_id, f_path = self._get_commit_and_path()
 
         self._ensure_not_locked()
 
         c.commit = self._get_commit_or_redirect(commit_id, redirect_after=False)
         if c.commit is None:
             c.commit = EmptyCommit(alias=self.rhodecode_vcs_repo.alias)
 
         if self.rhodecode_vcs_repo.is_empty():
             # for empty repository we cannot check for current branch, we rely on
             # c.commit.branch instead
             _branch_name, _sha_commit_id, is_head = c.commit.branch, '', True
         else:
             _branch_name, _sha_commit_id, is_head = \
                 self._is_valid_head(commit_id, self.rhodecode_vcs_repo,
                                     landing_ref=self.db_repo.landing_ref_name)
 
         self.forbid_non_head(is_head, f_path, commit_id=commit_id)
         self.check_branch_permission(_branch_name, commit_id=commit_id)
 
         c.default_message = (_('Added file via RhodeCode Enterprise'))
         c.f_path = f_path.lstrip('/')  # ensure not relative path
 
         return self._get_template_context(c)
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator('repository.write', 'repository.admin')
     @CSRFRequired()
     def repo_files_create_file(self):
         _ = self.request.translate
         c = self.load_default_context()
         commit_id, f_path = self._get_commit_and_path()
 
         self._ensure_not_locked()
 
         c.commit = self._get_commit_or_redirect(commit_id, redirect_after=False)
         if c.commit is None:
             c.commit = EmptyCommit(alias=self.rhodecode_vcs_repo.alias)
 
         # calculate redirect URL
         if self.rhodecode_vcs_repo.is_empty():
             default_redirect_url = h.route_path(
                 'repo_summary', repo_name=self.db_repo_name)
         else:
             default_redirect_url = h.route_path(
                 'repo_commit', repo_name=self.db_repo_name, commit_id='tip')
 
         if self.rhodecode_vcs_repo.is_empty():
             # for empty repository we cannot check for current branch, we rely on
             # c.commit.branch instead
             _branch_name, _sha_commit_id, is_head = c.commit.branch, '', True
         else:
             _branch_name, _sha_commit_id, is_head = \
                 self._is_valid_head(commit_id, self.rhodecode_vcs_repo,
                                     landing_ref=self.db_repo.landing_ref_name)
 
         self.forbid_non_head(is_head, f_path, commit_id=commit_id)
         self.check_branch_permission(_branch_name, commit_id=commit_id)
 
         c.default_message = (_('Added file via RhodeCode Enterprise'))
         c.f_path = f_path
 
         r_post = self.request.POST
         message = r_post.get('message') or c.default_message
         filename = r_post.get('filename')
         unix_mode = 0
 
         if not filename:
             # If there's no commit, redirect to repo summary
             if type(c.commit) is EmptyCommit:
                 redirect_url = h.route_path(
                     'repo_summary', repo_name=self.db_repo_name)
             else:
                 redirect_url = default_redirect_url
             h.flash(_('No filename specified'), category='warning')
             raise HTTPFound(redirect_url)
 
         root_path = f_path
         pure_path = self.create_pure_path(root_path, filename)
         node_path = pure_path.as_posix().lstrip('/')
 
         author = self._rhodecode_db_user.full_contact
         content = convert_line_endings(r_post.get('content', ''), unix_mode)
         nodes = {
             safe_bytes(node_path): {
                 'content': safe_bytes(content)
             }
         }
 
         try:
 
             commit = ScmModel().create_nodes(
                 user=self._rhodecode_db_user.user_id,
                 repo=self.db_repo,
                 message=message,
                 nodes=nodes,
                 parent_commit=c.commit,
                 author=author,
             )
 
             h.flash(_('Successfully committed new file `{}`').format(
                     h.escape(node_path)), category='success')
 
             default_redirect_url = h.route_path(
                 'repo_commit', repo_name=self.db_repo_name, commit_id=commit.raw_id)
 
         except NonRelativePathError:
             log.exception('Non Relative path found')
             h.flash(_('The location specified must be a relative path and must not '
                       'contain .. in the path'), category='warning')
             raise HTTPFound(default_redirect_url)
         except (NodeError, NodeAlreadyExistsError) as e:
             h.flash(h.escape(safe_str(e)), category='error')
         except Exception:
             log.exception('Error occurred during commit')
             h.flash(_('Error occurred during commit'), category='error')
 
         raise HTTPFound(default_redirect_url)
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator('repository.write', 'repository.admin')
     @CSRFRequired()
     def repo_files_upload_file(self):
         _ = self.request.translate
         c = self.load_default_context()
         commit_id, f_path = self._get_commit_and_path()
 
         self._ensure_not_locked()
 
         c.commit = self._get_commit_or_redirect(commit_id, redirect_after=False)
         if c.commit is None:
             c.commit = EmptyCommit(alias=self.rhodecode_vcs_repo.alias)
 
         # calculate redirect URL
         if self.rhodecode_vcs_repo.is_empty():
             default_redirect_url = h.route_path(
                 'repo_summary', repo_name=self.db_repo_name)
         else:
             default_redirect_url = h.route_path(
                 'repo_commit', repo_name=self.db_repo_name, commit_id='tip')
 
         if self.rhodecode_vcs_repo.is_empty():
             # for empty repository we cannot check for current branch, we rely on
             # c.commit.branch instead
             _branch_name, _sha_commit_id, is_head = c.commit.branch, '', True
         else:
             _branch_name, _sha_commit_id, is_head = \
                 self._is_valid_head(commit_id, self.rhodecode_vcs_repo,
                                     landing_ref=self.db_repo.landing_ref_name)
 
         error = self.forbid_non_head(is_head, f_path, json_mode=True)
         if error:
             return {
                 'error': error,
                 'redirect_url': default_redirect_url
             }
         error = self.check_branch_permission(_branch_name, json_mode=True)
         if error:
             return {
                 'error': error,
                 'redirect_url': default_redirect_url
             }
 
         c.default_message = (_('Uploaded file via RhodeCode Enterprise'))
         c.f_path = f_path
 
         r_post = self.request.POST
 
         message = c.default_message
         user_message = r_post.getall('message')
         if isinstance(user_message, list) and user_message:
             # we take the first from duplicated results if it's not empty
             message = user_message[0] if user_message[0] else message
 
         nodes = {}
 
         for file_obj in r_post.getall('files_upload') or []:
             content = file_obj.file
             filename = file_obj.filename
 
             root_path = f_path
             pure_path = self.create_pure_path(root_path, filename)
             node_path = pure_path.as_posix().lstrip('/')
 
             nodes[safe_bytes(node_path)] = {
                 'content': content
             }
 
         if not nodes:
             error = 'missing files'
             return {
                 'error': error,
                 'redirect_url': default_redirect_url
             }
 
         author = self._rhodecode_db_user.full_contact
 
         try:
             commit = ScmModel().create_nodes(
                 user=self._rhodecode_db_user.user_id,
                 repo=self.db_repo,
                 message=message,
                 nodes=nodes,
                 parent_commit=c.commit,
                 author=author,
             )
             if len(nodes) == 1:
                 flash_message = _('Successfully committed {} new files').format(len(nodes))
             else:
                 flash_message = _('Successfully committed 1 new file')
 
             h.flash(flash_message, category='success')
 
             default_redirect_url = h.route_path(
                 'repo_commit', repo_name=self.db_repo_name, commit_id=commit.raw_id)
 
         except NonRelativePathError:
             log.exception('Non Relative path found')
             error = _('The location specified must be a relative path and must not '
                       'contain .. in the path')
             h.flash(error, category='warning')
 
             return {
                 'error': error,
                 'redirect_url': default_redirect_url
             }
         except (NodeError, NodeAlreadyExistsError) as e:
             error = h.escape(e)
             h.flash(error, category='error')
 
             return {
                 'error': error,
                 'redirect_url': default_redirect_url
             }
         except Exception:
             log.exception('Error occurred during commit')
             error = _('Error occurred during commit')
             h.flash(error, category='error')
             return {
                 'error': error,
                 'redirect_url': default_redirect_url
             }
 
         return {
             'error': None,
             'redirect_url': default_redirect_url
         }

rhodecode/lib/vcs/backends/base.py

@@ -1,1985 +1,1984 b''
 # Copyright (C) 2014-2023 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 dataclasses
 import warnings
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 
 import rhodecode
 from rhodecode.translation import lazy_ugettext
 from rhodecode.lib.utils2 import safe_str, CachedProperty
 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
 
 
 @dataclasses.dataclass
 class Reference:
     type: str
     name: str
     commit_id: str
 
     def __iter__(self):
         yield self.type
         yield self.name
         yield self.commit_id
 
     @property
     def branch(self):
         if self.type == 'branch':
             return self.name
 
     @property
     def bookmark(self):
         if self.type == 'book':
             return self.name
 
     @property
     def to_str(self):
         return reference_to_unicode(self)
 
     def asdict(self):
         return dict(
             type=self.type,
             name=self.name,
             commit_id=self.commit_id
         )
 
 
 def unicode_to_reference(raw: str):
     """
     Convert a unicode (or string) to a reference object.
     If unicode evaluates to False it returns None.
     """
     if raw:
         refs = raw.split(':')
         return Reference(*refs)
     else:
         return None
 
 
 def reference_to_unicode(ref: Reference):
     """
     Convert a reference object to unicode.
     If reference is None it returns None.
     """
     if ref:
         return ':'.join(ref)
     else:
         return None
 
 
 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(
             'This pull request can be automatically merged.'),
         MergeFailureReason.UNKNOWN: lazy_ugettext(
             'This pull request cannot be merged because of an unhandled exception. '
             '{exception}'),
         MergeFailureReason.MERGE_FAILED: lazy_ugettext(
             'This pull request cannot be merged because of merge conflicts. {unresolved_files}'),
         MergeFailureReason.PUSH_FAILED: lazy_ugettext(
             'This pull request could not be merged because push to '
             'target:`{target}@{merge_commit}` failed.'),
         MergeFailureReason.TARGET_IS_NOT_HEAD: lazy_ugettext(
             'This pull request cannot be merged because the target '
             '`{target_ref.name}` is not a head.'),
         MergeFailureReason.HG_SOURCE_HAS_MORE_BRANCHES: lazy_ugettext(
             'This pull request cannot be merged because the source contains '
             'more branches than the target.'),
         MergeFailureReason.HG_TARGET_HAS_MULTIPLE_HEADS: lazy_ugettext(
             'This pull request cannot be merged because the target `{target_ref.name}` '
             'has multiple heads: `{heads}`.'),
         MergeFailureReason.TARGET_IS_LOCKED: lazy_ugettext(
             'This pull request cannot be merged because the target repository is '
             'locked by {locked_by}.'),
 
         MergeFailureReason.MISSING_TARGET_REF: lazy_ugettext(
             'This pull request cannot be merged because the target '
             'reference `{target_ref.name}` is missing.'),
         MergeFailureReason.MISSING_SOURCE_REF: lazy_ugettext(
             'This pull request cannot be merged because the source '
             'reference `{source_ref.name}` is missing.'),
         MergeFailureReason.SUBREPO_MERGE_FAILED: lazy_ugettext(
             'This pull request cannot be merged because of conflicts related '
             'to sub repositories.'),
 
         # Deprecations
         MergeFailureReason._DEPRECATED_MISSING_COMMIT: lazy_ugettext(
             'This pull request cannot be merged because the target or the '
             '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 f'<MergeResponse:{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_str(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 TargetRefMissing(ValueError):
     pass
 
 
 class SourceRefMissing(ValueError):
     pass
 
 
 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 = "Unknown"
     DEFAULT_DESCRIPTION = "unknown"
     EMPTY_COMMIT_ID = '0' * 40
     COMMIT_ID_PAT = re.compile(r'[0-9a-fA-F]{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 f'<{self.__class__.__name__} at {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_str(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, maybe_unreachable=False, reference_obj=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),
             f'.__shadow_{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),
                 f'.__shadow_repo_{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, str):
             raise TypeError(f"commit_id must be a string value got {type(commit_id)} instead")
 
     def _validate_commit_idx(self, commit_idx):
         if not isinstance(commit_idx, int):
             raise TypeError(f"commit_idx must be a numeric value, got {type(commit_idx)}")
 
     def _validate_branch_name(self, branch_name):
         if branch_name and branch_name not in self.branches_all:
             msg = (f"Branch {branch_name} not found in {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, str):
             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, str):
             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()
 
     def vcsserver_invalidate_cache(self, delete=False):
         return self._remote.vcsserver_invalidate_cache(delete)
 
 
 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
 
     """
     repository = None
     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 = '{repo_name}-{short_id}'
     """
     This template is used to generate a default prefix for repository archives
     if no prefix has been specified.
     """
 
     def __repr__(self):
         return self.__str__()
 
     def __str__(self):
         return f'<{self.__class__.__name__} at {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_remote(self):
         return self._remote
 
     def serialize(self):
         return self.__json__()
 
     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: bytes):
         """
         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 node_md5_hash(self, path):
         """
         Returns md5 hash of a node data
         """
         raise NotImplementedError
 
     def get_file_content(self, path) -> bytes:
         """
         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, pre_load=None):
         """
         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_name_key, kind='tgz', subrepos=None,
                      archive_dir_name=None, write_metadata=False, mtime=None,
                      archive_at_path='/', cache_config=None):
         """
         Creates an archive containing the contents of the repository.
 
         :param archive_name_key: unique key under this archive should be generated
-        :param kind: one of following: ``"tbz2"``, ``"tgz"``, ``"zip"``.
+        :param kind: one of the following: ``"tbz2"``, ``"tgz"``, ``"zip"``.
         :param archive_dir_name: 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 '/')
         :param cache_config: config spec to send to vcsserver to configure the backend to store files
 
         :raise VCSError: If prefix has a problem.
         """
         cache_config = cache_config or {}
         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))
+                f'Archive kind ({kind}) not supported use one of {allowed_kinds}')
 
         archive_dir_name = self._validate_archive_prefix(archive_dir_name)
         mtime = mtime is not None or time.mktime(self.date.timetuple())
         commit_id = self.raw_id
 
         return self.repository._remote.archive_repo(
             archive_name_key, kind, mtime, archive_at_path,
             archive_dir_name, commit_id, cache_config)
 
     def _validate_archive_prefix(self, archive_dir_name):
         if archive_dir_name is None:
             archive_dir_name = self._ARCHIVE_PREFIX_TEMPLATE.format(
                 repo_name=safe_str(self.repository.name),
                 short_id=self.short_id)
         elif not isinstance(archive_dir_name, str):
             raise ValueError(f"archive_dir_name is not str object but: {type(archive_dir_name)}")
         elif archive_dir_name.startswith('/'):
             raise VCSError("Prefix cannot start with leading slash")
         elif archive_dir_name.strip() == '':
             raise VCSError("Prefix cannot be empty")
         elif not archive_dir_name.isascii():
             raise VCSError("Prefix cannot contain non ascii characters")
         return archive_dir_name
 
     @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 = range(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 = range(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
         (top_node, dirnodes, filenodes).
         """
         from rhodecode.lib.vcs.nodes import DirNode
 
         if isinstance(topurl, DirNode):
             top_node = topurl
         else:
             top_node = self.get_node(topurl)
 
         has_default_pre_load = False
         if isinstance(top_node, DirNode):
             # used to inject as we walk same defaults as given top_node
             default_pre_load = top_node.default_pre_load
             has_default_pre_load = True
 
         if not top_node.is_dir():
             return
         yield top_node, top_node.dirs, top_node.files
         for dir_node in top_node.dirs:
             if has_default_pre_load:
                 dir_node.default_pre_load = default_pre_load
             yield from self.walk(dir_node)
 
     def get_filenodes_generator(self):
         """
         Returns generator that yields *all* file nodes.
         """
         for topnode, dirs, files in self.walk():
             yield from files
 
     #
     # Utilities for sub classes to support consistent behavior
     #
 
     def no_node_at_path(self, path):
         return NodeDoesNotExistError(
             f"There is no file nor directory at the given path: "
             f"`{safe_str(path)}` at commit {self.short_id}")
 
     def _fix_path(self, path: str) -> str:
         """
         Paths are stored without trailing slash so we need to get rid off it if
         needed.
         """
         return safe_str(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().__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(
                         f"Node `{node.path}` already exists at {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().__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, pre_load=None):
         return self
 
     def get_file_content(self, path) -> bytes:
         return b''
 
     def get_file_content_streamed(self, path):
         yield self.get_file_content(path)
 
     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().__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(b'')
 
 
 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
         self.collection_size = collection_size
         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 __getitem__(self, key):
         """Return either a single element by index, or a sliced collection."""
 
         if isinstance(key, slice):
             commit_ids = self.commit_ids[key.start:key.stop]
 
         else:
             # single item
             commit_ids = self.commit_ids[key]
 
         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({} sections) at {}>'.format(
             len(self._values), hex(id(self)))
 
     def items(self, section):
         return self._values.get(section, {}).items()
 
     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: bytes = re.compile(br"")
 
     def __init__(self, raw_diff: bytes):
         if not isinstance(raw_diff, bytes):
             raise Exception(f'raw_diff must be bytes - got {type(raw_diff)}')
 
         self.raw = memoryview(raw_diff)
 
     def get_header_re(self):
         return self._header_re
 
     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 = (b'\n' + bytes(self.raw)).split(b'\ndiff --git')
 
         chunks = diff_parts[1:]
         total_chunks = len(chunks)
 
         def diff_iter(_chunks):
             for cur_chunk, chunk in enumerate(_chunks, start=1):
                 yield DiffChunk(chunk, self, cur_chunk == total_chunks)
         return diff_iter(chunks)
 
 
 class DiffChunk(object):
 
     def __init__(self, chunk: bytes, diff_obj: Diff, is_last_chunk: bool):
         self.diff_obj = diff_obj
 
         # 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 is_last_chunk:
             chunk += b'\n'
         header_re = self.diff_obj.get_header_re()
         match = header_re.match(chunk)
         self.header = match.groupdict()
         self.diff = chunk[match.end():]
         self.raw = chunk
 
     @property
     def header_as_str(self):
         if self.header:
             def safe_str_on_bytes(val):
                 if isinstance(val, bytes):
                     return safe_str(val)
                 return val
             return {safe_str(k): safe_str_on_bytes(v) for k, v in self.header.items()}
 
     def __repr__(self):
         return f'DiffChunk({self.header_as_str})'
 
 
 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 NotImplementedError()
 
     def has_access(self, path):
         raise NotImplementedError()
 
 
 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