file-nodes: added streaming remote attributes for vcsserver.

- this will enable doing a streaming raw content or raw downloads of large files without transfering them over to enterprise for pack & repack using msgpack
- msgpack has a limit of 4gb and generally pack+repack for 2gb is very slow

rhodecode/apps/repository/views/repo_files.py

@@ -1,1547 +1,1545 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2011-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import itertools
 import logging
 import os
 import shutil
 import tempfile
 import collections
 import urllib
 import pathlib2
 
 from pyramid.httpexceptions import HTTPNotFound, HTTPBadRequest, HTTPFound
 from pyramid.view import view_config
 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.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, safe_str, str2bool, safe_int, sha1, safe_unicode)
 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__)
 
 
 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.iterkeys(),
             repo.tags.iterkeys())
         tags_and_branches = sorted(tags_and_branches, key=len, reverse=True)
 
         for name in tags_and_branches:
             if f_path.startswith('{}/'.format(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(
                 branch_name, 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_rev[1]
         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
 
             _url = h.route_path(
                 'repo_files_add_file',
                 repo_name=self.db_repo_name, commit_id=0, f_path='')
 
             if h.HasRepoPermissionAny(
                     'repository.write', 'repository.admin')(self.db_repo_name):
                 add_new = h.link_to(
                     _('Click here to add a new file.'), _url, class_="alert-link")
             else:
                 add_new = ""
 
             h.flash(h.literal(
                 _('There are no files yet. %s') % add_new), category='warning')
             raise HTTPFound(
                 h.route_path('repo_summary', repo_name=self.db_repo_name))
 
         except (CommitDoesNotExistError, LookupError):
             msg = _('No such commit exists for this repository')
             h.flash(msg, category='error')
             raise HTTPNotFound()
         except RepositoryError as e:
             h.flash(safe_str(h.escape(e)), category='error')
             raise HTTPNotFound()
 
     def _get_filenode_or_redirect(self, commit_obj, path):
         """
         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)
             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(safe_str(h.escape(e)), category='error')
             raise HTTPNotFound()
 
         return file_node
 
     def _is_valid_head(self, commit_id, repo):
         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 not repo.is_empty():
             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):
 
         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 = 'cache_repo.{}'.format(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(ver, repo_id, commit_id, f_path, full_load):
             log.debug('Generating cached file tree at ver:%s for repo_id: %s, %s, %s',
                       ver, 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)
 
         return compute_file_tree('v1', self.db_repo.repo_id, commit_id, f_path, full_load)
 
     def _get_archive_spec(self, 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
 
     def create_pure_path(self, *parts):
         # Split paths and sanitize them, removing any ../ etc
         sanitized_path = [
             x for x in pathlib2.PurePath(*parts).parts
             if x not in ['.', '..']]
 
         pure_path = pathlib2.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')
     @view_config(
         route_name='repo_archivefile', request_method='GET',
         renderer=None)
     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'
         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 = \
                 self._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'))
 
         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(at_path))
 
         path_sha = sha1(at_path)[:8]
 
         # original backward compat name of archive
         clean_name = safe_str(self.db_repo_name.replace('/', '_'))
         short_sha = safe_str(commit.short_id)
 
         if at_path == default_at_path:
             archive_name = '{}-{}{}{}'.format(
                 clean_name,
                 '-sub' if subrepos else '',
                 short_sha,
                 ext)
         # custom path and new name
         else:
             archive_name = '{}-{}{}-{}{}'.format(
                 clean_name,
                 '-sub' if subrepos else '',
                 short_sha,
                 path_sha,
                 ext)
 
         use_cached_archive = False
         archive_cache_enabled = CONFIG.get(
             'archive_cache_dir') and not self.request.GET.get('no_cache')
         cached_archive_path = None
 
         if archive_cache_enabled:
             # check if we it's ok to write
             if not os.path.isdir(CONFIG['archive_cache_dir']):
                 os.makedirs(CONFIG['archive_cache_dir'])
             cached_archive_path = os.path.join(
                 CONFIG['archive_cache_dir'], archive_name)
             if os.path.isfile(cached_archive_path):
                 log.debug('Found cached archive in %s', cached_archive_path)
                 fd, archive = None, cached_archive_path
                 use_cached_archive = True
             else:
                 log.debug('Archive %s is not yet cached', archive_name)
 
         if not use_cached_archive:
             # generate new archive
             fd, archive = tempfile.mkstemp()
             log.debug('Creating new temp archive in %s', archive)
             try:
                 commit.archive_repo(archive, kind=fileformat, subrepos=subrepos,
                                     archive_at_path=at_path)
             except ImproperArchiveTypeError:
                 return _('Unknown archive type')
             if archive_cache_enabled:
                 # if we generated the archive and we have cache enabled
                 # let's use this for future
                 log.debug('Storing new archive in %s', cached_archive_path)
                 shutil.move(archive, cached_archive_path)
                 archive = cached_archive_path
 
         # store download action
         audit_logger.store_web(
             'repo.archive.download', action_data={
                 'user_agent': self.request.user_agent,
                 'archive_name': archive_name,
                 'archive_spec': fname,
                 'archive_cached': use_cached_archive},
             user=self._rhodecode_user,
             repo=self.db_repo,
             commit=True
         )
 
         def get_chunked_archive(archive_path):
             with open(archive_path, 'rb') as stream:
                 while True:
                     data = stream.read(16 * 1024)
                     if not data:
                         if fd:  # fd means we used temporary file
                             os.close(fd)
                         if not archive_cache_enabled:
                             log.debug('Destroying temp archive %s', archive_path)
                             os.remove(archive_path)
                         break
                     yield data
 
         response = Response(app_iter=get_chunked_archive(archive))
         response.content_disposition = str(
             'attachment; filename=%s' % archive_name)
         response.content_type = str(content_type)
 
         return response
 
     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('%s path is a %s not a file'
                                     % (node, type(node)))
             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(f_path, '', commit=commit)
         else:
             commit = EmptyCommit(
                 repo=self.rhodecode_vcs_repo,
                 alias=self.rhodecode_vcs_repo.alias)
             node = FileNode(f_path, '', commit=commit)
         return node
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @view_config(
         route_name='repo_files_diff', request_method='GET',
         renderer=None)
     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)
             diff = diffs.DiffProcessor(_diff, format='gitdiff')
 
             response = Response(self.path_filter.get_raw_patch(diff))
             response.content_type = 'text/plain'
             response.content_disposition = (
                 'attachment; filename=%s_%s_vs_%s.diff' % (f_path, diff1, diff2)
             )
             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)
             diff = diffs.DiffProcessor(_diff, format='gitdiff')
 
             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')
     @view_config(
         route_name='repo_files_diff_2way_redirect', request_method='GET',
         renderer=None)
     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()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @view_config(
         route_name='repo_files', request_method='GET',
         renderer=None)
     @view_config(
         route_name='repo_files:default_path', request_method='GET',
         renderer=None)
     @view_config(
         route_name='repo_files:default_commit', request_method='GET',
         renderer=None)
     @view_config(
         route_name='repo_files:rendered', request_method='GET',
         renderer=None)
     @view_config(
         route_name='repo_files:annotated', request_method='GET',
         renderer=None)
     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)
 
         # redirect to given commit_id from form if given
         get_commit_id = self.request.GET.get('at_rev', None)
         if get_commit_id:
             self._get_commit_or_redirect(get_commit_id)
 
         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
 
         # 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)
             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
                 if c.file.size < c.visual.cut_off_limit_diff:
                     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)
                 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)
 
         except RepositoryError as e:
             h.flash(safe_str(h.escape(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')
     @view_config(
         route_name='repo_files:annotated_previous', request_method='GET',
         renderer=None)
     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='L{}'.format(line_anchor))
 
         raise HTTPFound(prev_url)
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @view_config(
         route_name='repo_nodetree_full', request_method='GET',
         renderer=None, xhr=True)
     @view_config(
         route_name='repo_nodetree_full:default_path', request_method='GET',
         renderer=None, xhr=True)
     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('error: {}'.format(h.escape(safe_str(e))))
 
         if dir_node.is_file():
             return Response('')
 
         c.file = dir_node
         c.commit = commit
 
         html = self._get_tree_at_commit(
             c, commit.raw_id, dir_node.path, full_load=True)
 
         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.quote(f_name)
 
         return "attachment; " \
                "filename=\"{}\"; " \
                "filename*=UTF-8\'\'{}".format(safe_path, encoded_path)
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @view_config(
         route_name='repo_file_raw', request_method='GET',
         renderer=None)
     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)
 
-        def stream_node():
-            yield file_node.raw_bytes
+        stream_content = file_node.stream_bytes()
 
-        response = Response(app_iter=stream_node())
+        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')
     @view_config(
         route_name='repo_file_download', request_method='GET',
         renderer=None)
     @view_config(
         route_name='repo_file_download:legacy', request_method='GET',
         renderer=None)
     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)
 
-        def stream_node():
-            yield file_node.raw_bytes
+        stream_content = file_node.stream_bytes()
 
-        response = Response(app_iter=stream_node())
+        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 = 'cache_repo.{}'.format(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(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_nodes(
                     repo_name, commit_id, f_path, flat=False)
             except (RepositoryError, CommitDoesNotExistError, Exception) as e:
                 log.exception(safe_str(e))
                 h.flash(safe_str(h.escape(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_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')
     @view_config(
         route_name='repo_files_nodelist', request_method='GET',
         renderer='json_ext', xhr=True)
     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': 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):
         new_f_path = vcspath.join(name, f_path)
         return u'%s@%s' % (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 = 'r%s:%s%s' % (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')
     @view_config(
         route_name='repo_file_history', request_method='GET',
         renderer='json_ext')
     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 obj in file_history:
                 res.append({
                     'text': obj[1],
                     'children': [{'id': o[0], 'text': o[1], 'type': o[2]} for o in obj[0]]
                 })
 
             data = {
                 'more': False,
                 'results': res
             }
             return data
 
         log.warning('Cannot fetch history for directory')
         raise HTTPBadRequest()
 
     @LoginRequired()
     @HasRepoPermissionAnyDecorator(
         'repository.read', 'repository.write', 'repository.admin')
     @view_config(
         route_name='repo_file_authors', request_method='GET',
         renderer='rhodecode:templates/files/file_authors_box.mako')
     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 = set(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')
     @view_config(
         route_name='repo_files_remove_file', request_method='GET',
         renderer='rhodecode:templates/files/files_delete.mako')
     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)
 
         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()
     @view_config(
         route_name='repo_files_delete_file', request_method='POST',
         renderer=None)
     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)
 
         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 = {
                 node_path: {
                     'content': ''
                 }
             }
             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')
     @view_config(
         route_name='repo_files_edit_file', request_method='GET',
         renderer='rhodecode:templates/files/files_edit.mako')
     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)
 
         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()
     @view_config(
         route_name='repo_files_update_file', request_method='POST',
         renderer=None)
     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)
 
         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.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.unicode_path
         filename = r_post['filename']
 
         root_path = c.file.dir_path
         pure_path = self.create_pure_path(root_path, filename)
         node_path = safe_unicode(bytes(pure_path))
 
         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(org_node_path),
                     category='warning')
             raise HTTPFound(default_redirect_url)
 
         try:
             mapping = {
                 org_node_path: {
                     'org_filename': org_node_path,
                     'filename': node_path,
                     'content': 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')
     @view_config(
         route_name='repo_files_add_file', request_method='GET',
         renderer='rhodecode:templates/files/files_add.mako')
     @view_config(
         route_name='repo_files_upload_file', request_method='GET',
         renderer='rhodecode:templates/files/files_upload.mako')
     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)
 
         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()
     @view_config(
         route_name='repo_files_create_file', request_method='POST',
         renderer=None)
     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)
 
         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
         content = convert_line_endings(r_post.get('content', ''), unix_mode)
 
         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 = safe_unicode(bytes(pure_path).lstrip('/'))
 
         author = self._rhodecode_db_user.full_contact
         nodes = {
             node_path: {
                 'content': 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(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()
     @view_config(
         route_name='repo_files_upload_file', request_method='POST',
         renderer='json_ext')
     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)
 
         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 = safe_unicode(bytes(pure_path).lstrip('/'))
 
             nodes[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/__init__.py

@@ -1,188 +1,184 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 Various version Control System version lib (vcs) management abstraction layer
 for Python. Build with server client architecture.
 """
 import atexit
 import logging
 import urlparse
 from cStringIO import StringIO
 
 import rhodecode
 from rhodecode.lib.vcs.conf import settings
 from rhodecode.lib.vcs.backends import get_vcs_instance, get_backend
 from rhodecode.lib.vcs.exceptions import (
     VCSError, RepositoryError, CommitError, VCSCommunicationError)
 
 VERSION = (0, 5, 0, 'dev')
 
 __version__ = '.'.join((str(each) for each in VERSION[:4]))
 
 __all__ = [
     'get_version', 'get_vcs_instance', 'get_backend',
     'VCSError', 'RepositoryError', 'CommitError', 'VCSCommunicationError'
 ]
 
 log = logging.getLogger(__name__)
 
 # The pycurl library directly accesses C API functions and is not patched by
 # gevent. This will potentially lead to deadlocks due to incompatibility to
 # gevent. Therefore we check if gevent is active and import a gevent compatible
 # wrapper in that case.
 try:
     from gevent import monkey
     if monkey.is_module_patched('__builtin__'):
         import geventcurl as pycurl
         log.debug('Using gevent comapatible pycurl: %s', pycurl)
     else:
         import pycurl
 except ImportError:
     import pycurl
 
 
 def get_version():
     """
     Returns shorter version (digit parts only) as string.
     """
     return '.'.join((str(each) for each in VERSION[:3]))
 
 
 def connect_http(server_and_port):
     from rhodecode.lib.vcs import connection, client_http
     from rhodecode.lib.middleware.utils import scm_app
 
     session_factory = client_http.ThreadlocalSessionFactory()
 
-    connection.Git = client_http.RepoMaker(
+    connection.Git = client_http.RemoteVCSMaker(
         server_and_port, '/git', 'git', session_factory)
-    connection.Hg = client_http.RepoMaker(
+    connection.Hg = client_http.RemoteVCSMaker(
         server_and_port, '/hg', 'hg', session_factory)
-    connection.Svn = client_http.RepoMaker(
+    connection.Svn = client_http.RemoteVCSMaker(
         server_and_port, '/svn', 'svn', session_factory)
     connection.Service = client_http.ServiceConnection(
         server_and_port, '/_service', session_factory)
 
     scm_app.HG_REMOTE_WSGI = client_http.VcsHttpProxy(
         server_and_port, '/proxy/hg')
     scm_app.GIT_REMOTE_WSGI = client_http.VcsHttpProxy(
         server_and_port, '/proxy/git')
 
     @atexit.register
     def free_connection_resources():
         connection.Git = None
         connection.Hg = None
         connection.Svn = None
         connection.Service = None
 
 
 def connect_vcs(server_and_port, protocol):
     """
     Initializes the connection to the vcs server.
 
     :param server_and_port: str, e.g. "localhost:9900"
     :param protocol: str or "http"
     """
     if protocol == 'http':
         connect_http(server_and_port)
     else:
         raise Exception('Invalid vcs server protocol "{}"'.format(protocol))
 
 
-def create_vcsserver_proxy(server_and_port, protocol):
-    if protocol == 'http':
-        return _create_vcsserver_proxy_http(server_and_port)
-    else:
-        raise Exception('Invalid vcs server protocol "{}"'.format(protocol))
-
-
-def _create_vcsserver_proxy_http(server_and_port):
-    from rhodecode.lib.vcs import client_http
-
-    session = _create_http_rpc_session()
-    url = urlparse.urljoin('http://%s' % server_and_port, '/server')
-    return client_http.RemoteObject(url, session)
-
-
 class CurlSession(object):
     """
     Modeled so that it provides a subset of the requests interface.
 
     This has been created so that it does only provide a minimal API for our
     needs. The parts which it provides are based on the API of the library
     `requests` which allows us to easily benchmark against it.
 
     Please have a look at the class :class:`requests.Session` when you extend
     it.
     """
 
     def __init__(self):
         curl = pycurl.Curl()
         # TODO: johbo: I did test with 7.19 of libcurl. This version has
         # trouble with 100 - continue being set in the expect header. This
         # can lead to massive performance drops, switching it off here.
         curl.setopt(curl.HTTPHEADER, ["Expect:"])
         curl.setopt(curl.TCP_NODELAY, True)
         curl.setopt(curl.PROTOCOLS, curl.PROTO_HTTP)
         curl.setopt(curl.USERAGENT, 'RhodeCode HTTP {}'.format(rhodecode.__version__))
         self._curl = curl
 
     def post(self, url, data, allow_redirects=False):
         response_buffer = StringIO()
 
         curl = self._curl
         curl.setopt(curl.URL, url)
         curl.setopt(curl.POST, True)
         curl.setopt(curl.POSTFIELDS, data)
         curl.setopt(curl.FOLLOWLOCATION, allow_redirects)
         curl.setopt(curl.WRITEDATA, response_buffer)
         curl.perform()
 
         status_code = curl.getinfo(pycurl.HTTP_CODE)
 
         return CurlResponse(response_buffer, status_code)
 
 
 class CurlResponse(object):
     """
     The response of a request, modeled after the requests API.
 
     This class provides a subset of the response interface known from the
     library `requests`. It is intentionally kept similar, so that we can use
     `requests` as a drop in replacement for benchmarking purposes.
     """
 
     def __init__(self, response_buffer, status_code):
         self._response_buffer = response_buffer
         self._status_code = status_code
 
     @property
     def content(self):
-        return self._response_buffer.getvalue()
+        try:
+            return self._response_buffer.getvalue()
+        finally:
+            self._response_buffer.close()
 
     @property
     def status_code(self):
         return self._status_code
 
+    def iter_content(self, chunk_size):
+        self._response_buffer.seek(0)
+        while 1:
+            chunk = self._response_buffer.read(chunk_size)
+            if not chunk:
+                break
+            yield chunk
+
 
 def _create_http_rpc_session():
     session = CurlSession()
     return session

rhodecode/lib/vcs/backends/base.py

@@ -1,1881 +1,1890 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 Base module for all VCS systems
 """
 import os
 import re
 import time
 import shutil
 import datetime
 import fnmatch
 import itertools
 import logging
 import collections
 import warnings
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from pyramid import compat
 
 import rhodecode
 from rhodecode.translation import lazy_ugettext
 from rhodecode.lib.utils2 import safe_str, safe_unicode, CachedProperty
 from rhodecode.lib.vcs import connection
 from rhodecode.lib.vcs.utils import author_name, author_email
 from rhodecode.lib.vcs.conf import settings
 from rhodecode.lib.vcs.exceptions import (
     CommitError, EmptyRepositoryError, NodeAlreadyAddedError,
     NodeAlreadyChangedError, NodeAlreadyExistsError, NodeAlreadyRemovedError,
     NodeDoesNotExistError, NodeNotChangedError, VCSError,
     ImproperArchiveTypeError, BranchDoesNotExistError, CommitDoesNotExistError,
     RepositoryError)
 
 
 log = logging.getLogger(__name__)
 
 
 FILEMODE_DEFAULT = 0o100644
 FILEMODE_EXECUTABLE = 0o100755
 EMPTY_COMMIT_ID = '0' * 40
 
 Reference = collections.namedtuple('Reference', ('type', 'name', 'commit_id'))
 
 
 class MergeFailureReason(object):
     """
     Enumeration with all the reasons why the server side merge could fail.
 
     DO NOT change the number of the reasons, as they may be stored in the
     database.
 
     Changing the name of a reason is acceptable and encouraged to deprecate old
     reasons.
     """
 
     # Everything went well.
     NONE = 0
 
     # An unexpected exception was raised. Check the logs for more details.
     UNKNOWN = 1
 
     # The merge was not successful, there are conflicts.
     MERGE_FAILED = 2
 
     # The merge succeeded but we could not push it to the target repository.
     PUSH_FAILED = 3
 
     # The specified target is not a head in the target repository.
     TARGET_IS_NOT_HEAD = 4
 
     # The source repository contains more branches than the target. Pushing
     # the merge will create additional branches in the target.
     HG_SOURCE_HAS_MORE_BRANCHES = 5
 
     # The target reference has multiple heads. That does not allow to correctly
     # identify the target location. This could only happen for mercurial
     # branches.
     HG_TARGET_HAS_MULTIPLE_HEADS = 6
 
     # The target repository is locked
     TARGET_IS_LOCKED = 7
 
     # Deprecated, use MISSING_TARGET_REF or MISSING_SOURCE_REF instead.
     # A involved commit could not be found.
     _DEPRECATED_MISSING_COMMIT = 8
 
     # The target repo reference is missing.
     MISSING_TARGET_REF = 9
 
     # The source repo reference is missing.
     MISSING_SOURCE_REF = 10
 
     # The merge was not successful, there are conflicts related to sub
     # repositories.
     SUBREPO_MERGE_FAILED = 11
 
 
 class UpdateFailureReason(object):
     """
     Enumeration with all the reasons why the pull request update could fail.
 
     DO NOT change the number of the reasons, as they may be stored in the
     database.
 
     Changing the name of a reason is acceptable and encouraged to deprecate old
     reasons.
     """
 
     # Everything went well.
     NONE = 0
 
     # An unexpected exception was raised. Check the logs for more details.
     UNKNOWN = 1
 
     # The pull request is up to date.
     NO_CHANGE = 2
 
     # The pull request has a reference type that is not supported for update.
     WRONG_REF_TYPE = 3
 
     # Update failed because the target reference is missing.
     MISSING_TARGET_REF = 4
 
     # Update failed because the source reference is missing.
     MISSING_SOURCE_REF = 5
 
 
 class MergeResponse(object):
 
     # uses .format(**metadata) for variables
     MERGE_STATUS_MESSAGES = {
         MergeFailureReason.NONE: lazy_ugettext(
             u'This pull request can be automatically merged.'),
         MergeFailureReason.UNKNOWN: lazy_ugettext(
             u'This pull request cannot be merged because of an unhandled exception. '
             u'{exception}'),
         MergeFailureReason.MERGE_FAILED: lazy_ugettext(
             u'This pull request cannot be merged because of merge conflicts.'),
         MergeFailureReason.PUSH_FAILED: lazy_ugettext(
             u'This pull request could not be merged because push to '
             u'target:`{target}@{merge_commit}` failed.'),
         MergeFailureReason.TARGET_IS_NOT_HEAD: lazy_ugettext(
             u'This pull request cannot be merged because the target '
             u'`{target_ref.name}` is not a head.'),
         MergeFailureReason.HG_SOURCE_HAS_MORE_BRANCHES: lazy_ugettext(
             u'This pull request cannot be merged because the source contains '
             u'more branches than the target.'),
         MergeFailureReason.HG_TARGET_HAS_MULTIPLE_HEADS: lazy_ugettext(
             u'This pull request cannot be merged because the target `{target_ref.name}` '
             u'has multiple heads: `{heads}`.'),
         MergeFailureReason.TARGET_IS_LOCKED: lazy_ugettext(
             u'This pull request cannot be merged because the target repository is '
             u'locked by {locked_by}.'),
 
         MergeFailureReason.MISSING_TARGET_REF: lazy_ugettext(
             u'This pull request cannot be merged because the target '
             u'reference `{target_ref.name}` is missing.'),
         MergeFailureReason.MISSING_SOURCE_REF: lazy_ugettext(
             u'This pull request cannot be merged because the source '
             u'reference `{source_ref.name}` is missing.'),
         MergeFailureReason.SUBREPO_MERGE_FAILED: lazy_ugettext(
             u'This pull request cannot be merged because of conflicts related '
             u'to sub repositories.'),
 
         # Deprecations
         MergeFailureReason._DEPRECATED_MISSING_COMMIT: lazy_ugettext(
             u'This pull request cannot be merged because the target or the '
             u'source reference is missing.'),
 
     }
 
     def __init__(self, possible, executed, merge_ref, failure_reason, metadata=None):
         self.possible = possible
         self.executed = executed
         self.merge_ref = merge_ref
         self.failure_reason = failure_reason
         self.metadata = metadata or {}
 
     def __repr__(self):
         return '<MergeResponse:{} {}>'.format(self.label, self.failure_reason)
 
     def __eq__(self, other):
         same_instance = isinstance(other, self.__class__)
         return same_instance \
                and self.possible == other.possible \
                and self.executed == other.executed \
                and self.failure_reason == other.failure_reason
 
     @property
     def label(self):
         label_dict = dict((v, k) for k, v in MergeFailureReason.__dict__.items() if
                           not k.startswith('_'))
         return label_dict.get(self.failure_reason)
 
     @property
     def merge_status_message(self):
         """
         Return a human friendly error message for the given merge status code.
         """
         msg = safe_unicode(self.MERGE_STATUS_MESSAGES[self.failure_reason])
         try:
             return msg.format(**self.metadata)
         except Exception:
             log.exception('Failed to format %s message', self)
             return msg
 
     def asdict(self):
         data = {}
         for k in ['possible', 'executed', 'merge_ref', 'failure_reason',
                   'merge_status_message']:
             data[k] = getattr(self, k)
         return data
 
 
 class BaseRepository(object):
     """
     Base Repository for final backends
 
     .. attribute:: DEFAULT_BRANCH_NAME
 
        name of default branch (i.e. "trunk" for svn, "master" for git etc.
 
     .. attribute:: commit_ids
 
        list of all available commit ids, in ascending order
 
     .. attribute:: path
 
        absolute path to the repository
 
     .. attribute:: bookmarks
 
        Mapping from name to :term:`Commit ID` of the bookmark. Empty in case
        there are no bookmarks or the backend implementation does not support
        bookmarks.
 
     .. attribute:: tags
 
        Mapping from name to :term:`Commit ID` of the tag.
 
     """
 
     DEFAULT_BRANCH_NAME = None
     DEFAULT_CONTACT = u"Unknown"
     DEFAULT_DESCRIPTION = u"unknown"
     EMPTY_COMMIT_ID = '0' * 40
 
     path = None
 
     _is_empty = None
     _commit_ids = {}
 
     def __init__(self, repo_path, config=None, create=False, **kwargs):
         """
         Initializes repository. Raises RepositoryError if repository could
         not be find at the given ``repo_path`` or directory at ``repo_path``
         exists and ``create`` is set to True.
 
         :param repo_path: local path of the repository
         :param config: repository configuration
         :param create=False: if set to True, would try to create repository.
         :param src_url=None: if set, should be proper url from which repository
           would be cloned; requires ``create`` parameter to be set to True -
           raises RepositoryError if src_url is set and create evaluates to
           False
         """
         raise NotImplementedError
 
     def __repr__(self):
         return '<%s at %s>' % (self.__class__.__name__, self.path)
 
     def __len__(self):
         return self.count()
 
     def __eq__(self, other):
         same_instance = isinstance(other, self.__class__)
         return same_instance and other.path == self.path
 
     def __ne__(self, other):
         return not self.__eq__(other)
 
     def get_create_shadow_cache_pr_path(self, db_repo):
         path = db_repo.cached_diffs_dir
         if not os.path.exists(path):
             os.makedirs(path, 0o755)
         return path
 
     @classmethod
     def get_default_config(cls, default=None):
         config = Config()
         if default and isinstance(default, list):
             for section, key, val in default:
                 config.set(section, key, val)
         return config
 
     @LazyProperty
     def _remote(self):
         raise NotImplementedError
 
     def _heads(self, branch=None):
         return []
 
     @LazyProperty
     def EMPTY_COMMIT(self):
         return EmptyCommit(self.EMPTY_COMMIT_ID)
 
     @LazyProperty
     def alias(self):
         for k, v in settings.BACKENDS.items():
             if v.split('.')[-1] == str(self.__class__.__name__):
                 return k
 
     @LazyProperty
     def name(self):
         return safe_unicode(os.path.basename(self.path))
 
     @LazyProperty
     def description(self):
         raise NotImplementedError
 
     def refs(self):
         """
         returns a `dict` with branches, bookmarks, tags, and closed_branches
         for this repository
         """
         return dict(
             branches=self.branches,
             branches_closed=self.branches_closed,
             tags=self.tags,
             bookmarks=self.bookmarks
         )
 
     @LazyProperty
     def branches(self):
         """
         A `dict` which maps branch names to commit ids.
         """
         raise NotImplementedError
 
     @LazyProperty
     def branches_closed(self):
         """
         A `dict` which maps tags names to commit ids.
         """
         raise NotImplementedError
 
     @LazyProperty
     def bookmarks(self):
         """
         A `dict` which maps tags names to commit ids.
         """
         raise NotImplementedError
 
     @LazyProperty
     def tags(self):
         """
         A `dict` which maps tags names to commit ids.
         """
         raise NotImplementedError
 
     @LazyProperty
     def size(self):
         """
         Returns combined size in bytes for all repository files
         """
         tip = self.get_commit()
         return tip.size
 
     def size_at_commit(self, commit_id):
         commit = self.get_commit(commit_id)
         return commit.size
 
     def _check_for_empty(self):
         no_commits = len(self._commit_ids) == 0
         if no_commits:
             # check on remote to be sure
             return self._remote.is_empty()
         else:
             return False
 
     def is_empty(self):
         if rhodecode.is_test:
             return self._check_for_empty()
 
         if self._is_empty is None:
             # cache empty for production, but not tests
             self._is_empty = self._check_for_empty()
 
         return self._is_empty
 
     @staticmethod
     def check_url(url, config):
         """
         Function will check given url and try to verify if it's a valid
         link.
         """
         raise NotImplementedError
 
     @staticmethod
     def is_valid_repository(path):
         """
         Check if given `path` contains a valid repository of this backend
         """
         raise NotImplementedError
 
     # ==========================================================================
     # COMMITS
     # ==========================================================================
 
     @CachedProperty
     def commit_ids(self):
         raise NotImplementedError
 
     def append_commit_id(self, commit_id):
         if commit_id not in self.commit_ids:
             self._rebuild_cache(self.commit_ids + [commit_id])
 
         # clear cache
         self._invalidate_prop_cache('commit_ids')
         self._is_empty = False
 
     def get_commit(self, commit_id=None, commit_idx=None, pre_load=None, translate_tag=None):
         """
         Returns instance of `BaseCommit` class. If `commit_id` and `commit_idx`
         are both None, most recent commit is returned.
 
         :param pre_load: Optional. List of commit attributes to load.
 
         :raises ``EmptyRepositoryError``: if there are no commits
         """
         raise NotImplementedError
 
     def __iter__(self):
         for commit_id in self.commit_ids:
             yield self.get_commit(commit_id=commit_id)
 
     def get_commits(
             self, start_id=None, end_id=None, start_date=None, end_date=None,
             branch_name=None, show_hidden=False, pre_load=None, translate_tags=None):
         """
         Returns iterator of `BaseCommit` objects from start to end
         not inclusive. This should behave just like a list, ie. end is not
         inclusive.
 
         :param start_id: None or str, must be a valid commit id
         :param end_id: None or str, must be a valid commit id
         :param start_date:
         :param end_date:
         :param branch_name:
         :param show_hidden:
         :param pre_load:
         :param translate_tags:
         """
         raise NotImplementedError
 
     def __getitem__(self, key):
         """
         Allows index based access to the commit objects of this repository.
         """
         pre_load = ["author", "branch", "date", "message", "parents"]
         if isinstance(key, slice):
             return self._get_range(key, pre_load)
         return self.get_commit(commit_idx=key, pre_load=pre_load)
 
     def _get_range(self, slice_obj, pre_load):
         for commit_id in self.commit_ids.__getitem__(slice_obj):
             yield self.get_commit(commit_id=commit_id, pre_load=pre_load)
 
     def count(self):
         return len(self.commit_ids)
 
     def tag(self, name, user, commit_id=None, message=None, date=None, **opts):
         """
         Creates and returns a tag for the given ``commit_id``.
 
         :param name: name for new tag
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param commit_id: commit id for which new tag would be created
         :param message: message of the tag's commit
         :param date: date of tag's commit
 
         :raises TagAlreadyExistError: if tag with same name already exists
         """
         raise NotImplementedError
 
     def remove_tag(self, name, user, message=None, date=None):
         """
         Removes tag with the given ``name``.
 
         :param name: name of the tag to be removed
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param message: message of the tag's removal commit
         :param date: date of tag's removal commit
 
         :raises TagDoesNotExistError: if tag with given name does not exists
         """
         raise NotImplementedError
 
     def get_diff(
             self, commit1, commit2, path=None, ignore_whitespace=False,
             context=3, path1=None):
         """
         Returns (git like) *diff*, as plain text. Shows changes introduced by
         `commit2` since `commit1`.
 
         :param commit1: Entry point from which diff is shown. Can be
           ``self.EMPTY_COMMIT`` - in this case, patch showing all
           the changes since empty state of the repository until `commit2`
         :param commit2: Until which commit changes should be shown.
         :param path: Can be set to a path of a file to create a diff of that
           file. If `path1` is also set, this value is only associated to
           `commit2`.
         :param ignore_whitespace: If set to ``True``, would not show whitespace
           changes. Defaults to ``False``.
         :param context: How many lines before/after changed lines should be
           shown. Defaults to ``3``.
         :param path1: Can be set to a path to associate with `commit1`. This
           parameter works only for backends which support diff generation for
           different paths. Other backends will raise a `ValueError` if `path1`
           is set and has a different value than `path`.
         :param file_path: filter this diff by given path pattern
         """
         raise NotImplementedError
 
     def strip(self, commit_id, branch=None):
         """
         Strip given commit_id from the repository
         """
         raise NotImplementedError
 
     def get_common_ancestor(self, commit_id1, commit_id2, repo2):
         """
         Return a latest common ancestor commit if one exists for this repo
         `commit_id1` vs `commit_id2` from `repo2`.
 
         :param commit_id1: Commit it from this repository to use as a
            target for the comparison.
         :param commit_id2: Source commit id to use for comparison.
         :param repo2: Source repository to use for comparison.
         """
         raise NotImplementedError
 
     def compare(self, commit_id1, commit_id2, repo2, merge, pre_load=None):
         """
         Compare this repository's revision `commit_id1` with `commit_id2`.
 
         Returns a tuple(commits, ancestor) that would be merged from
         `commit_id2`. Doing a normal compare (``merge=False``), ``None``
         will be returned as ancestor.
 
         :param commit_id1: Commit it from this repository to use as a
            target for the comparison.
         :param commit_id2: Source commit id to use for comparison.
         :param repo2: Source repository to use for comparison.
         :param merge: If set to ``True`` will do a merge compare which also
            returns the common ancestor.
         :param pre_load: Optional. List of commit attributes to load.
         """
         raise NotImplementedError
 
     def merge(self, repo_id, workspace_id, target_ref, source_repo, source_ref,
               user_name='', user_email='', message='', dry_run=False,
               use_rebase=False, close_branch=False):
         """
         Merge the revisions specified in `source_ref` from `source_repo`
         onto the `target_ref` of this repository.
 
         `source_ref` and `target_ref` are named tupls with the following
         fields `type`, `name` and `commit_id`.
 
         Returns a MergeResponse named tuple with the following fields
         'possible', 'executed', 'source_commit', 'target_commit',
         'merge_commit'.
 
         :param repo_id: `repo_id` target repo id.
         :param workspace_id: `workspace_id` unique identifier.
         :param target_ref: `target_ref` points to the commit on top of which
             the `source_ref` should be merged.
         :param source_repo: The repository that contains the commits to be
             merged.
         :param source_ref: `source_ref` points to the topmost commit from
             the `source_repo` which should be merged.
         :param user_name: Merge commit `user_name`.
         :param user_email: Merge commit `user_email`.
         :param message: Merge commit `message`.
         :param dry_run: If `True` the merge will not take place.
         :param use_rebase: If `True` commits from the source will be rebased
             on top of the target instead of being merged.
         :param close_branch: If `True` branch will be close before merging it
         """
         if dry_run:
             message = message or settings.MERGE_DRY_RUN_MESSAGE
             user_email = user_email or settings.MERGE_DRY_RUN_EMAIL
             user_name = user_name or settings.MERGE_DRY_RUN_USER
         else:
             if not user_name:
                 raise ValueError('user_name cannot be empty')
             if not user_email:
                 raise ValueError('user_email cannot be empty')
             if not message:
                 raise ValueError('message cannot be empty')
 
         try:
             return self._merge_repo(
                 repo_id, workspace_id, target_ref, source_repo,
                 source_ref, message, user_name, user_email, dry_run=dry_run,
                 use_rebase=use_rebase, close_branch=close_branch)
         except RepositoryError as exc:
             log.exception('Unexpected failure when running merge, dry-run=%s', dry_run)
             return MergeResponse(
                 False, False, None, MergeFailureReason.UNKNOWN,
                 metadata={'exception': str(exc)})
 
     def _merge_repo(self, repo_id, workspace_id, target_ref,
                     source_repo, source_ref, merge_message,
                     merger_name, merger_email, dry_run=False,
                     use_rebase=False, close_branch=False):
         """Internal implementation of merge."""
         raise NotImplementedError
 
     def _maybe_prepare_merge_workspace(
             self, repo_id, workspace_id, target_ref, source_ref):
         """
         Create the merge workspace.
 
         :param workspace_id: `workspace_id` unique identifier.
         """
         raise NotImplementedError
 
     def _get_legacy_shadow_repository_path(self, workspace_id):
         """
         Legacy version that was used before. We still need it for
         backward compat
         """
         return os.path.join(
             os.path.dirname(self.path),
             '.__shadow_%s_%s' % (os.path.basename(self.path), workspace_id))
 
     def _get_shadow_repository_path(self, repo_id, workspace_id):
         # The name of the shadow repository must start with '.', so it is
         # skipped by 'rhodecode.lib.utils.get_filesystem_repos'.
         legacy_repository_path = self._get_legacy_shadow_repository_path(workspace_id)
         if os.path.exists(legacy_repository_path):
             return legacy_repository_path
         else:
             return os.path.join(
                 os.path.dirname(self.path),
                 '.__shadow_repo_%s_%s' % (repo_id, workspace_id))
 
     def cleanup_merge_workspace(self, repo_id, workspace_id):
         """
         Remove merge workspace.
 
         This function MUST not fail in case there is no workspace associated to
         the given `workspace_id`.
 
         :param workspace_id: `workspace_id` unique identifier.
         """
         shadow_repository_path = self._get_shadow_repository_path(repo_id, workspace_id)
         shadow_repository_path_del = '{}.{}.delete'.format(
             shadow_repository_path, time.time())
 
         # move the shadow repo, so it never conflicts with the one used.
         # we use this method because shutil.rmtree had some edge case problems
         # removing symlinked repositories
         if not os.path.isdir(shadow_repository_path):
             return
 
         shutil.move(shadow_repository_path, shadow_repository_path_del)
         try:
             shutil.rmtree(shadow_repository_path_del, ignore_errors=False)
         except Exception:
             log.exception('Failed to gracefully remove shadow repo under %s',
                           shadow_repository_path_del)
             shutil.rmtree(shadow_repository_path_del, ignore_errors=True)
 
     # ========== #
     # COMMIT API #
     # ========== #
 
     @LazyProperty
     def in_memory_commit(self):
         """
         Returns :class:`InMemoryCommit` object for this repository.
         """
         raise NotImplementedError
 
     # ======================== #
     # UTILITIES FOR SUBCLASSES #
     # ======================== #
 
     def _validate_diff_commits(self, commit1, commit2):
         """
         Validates that the given commits are related to this repository.
 
         Intended as a utility for sub classes to have a consistent validation
         of input parameters in methods like :meth:`get_diff`.
         """
         self._validate_commit(commit1)
         self._validate_commit(commit2)
         if (isinstance(commit1, EmptyCommit) and
                 isinstance(commit2, EmptyCommit)):
             raise ValueError("Cannot compare two empty commits")
 
     def _validate_commit(self, commit):
         if not isinstance(commit, BaseCommit):
             raise TypeError(
                 "%s is not of type BaseCommit" % repr(commit))
         if commit.repository != self and not isinstance(commit, EmptyCommit):
             raise ValueError(
                 "Commit %s must be a valid commit from this repository %s, "
                 "related to this repository instead %s." %
                 (commit, self, commit.repository))
 
     def _validate_commit_id(self, commit_id):
         if not isinstance(commit_id, compat.string_types):
             raise TypeError("commit_id must be a string value")
 
     def _validate_commit_idx(self, commit_idx):
         if not isinstance(commit_idx, (int, long)):
             raise TypeError("commit_idx must be a numeric value")
 
     def _validate_branch_name(self, branch_name):
         if branch_name and branch_name not in self.branches_all:
             msg = ("Branch %s not found in %s" % (branch_name, self))
             raise BranchDoesNotExistError(msg)
 
     #
     # Supporting deprecated API parts
     # TODO: johbo: consider to move this into a mixin
     #
 
     @property
     def EMPTY_CHANGESET(self):
         warnings.warn(
             "Use EMPTY_COMMIT or EMPTY_COMMIT_ID instead", DeprecationWarning)
         return self.EMPTY_COMMIT_ID
 
     @property
     def revisions(self):
         warnings.warn("Use commits attribute instead", DeprecationWarning)
         return self.commit_ids
 
     @revisions.setter
     def revisions(self, value):
         warnings.warn("Use commits attribute instead", DeprecationWarning)
         self.commit_ids = value
 
     def get_changeset(self, revision=None, pre_load=None):
         warnings.warn("Use get_commit instead", DeprecationWarning)
         commit_id = None
         commit_idx = None
         if isinstance(revision, compat.string_types):
             commit_id = revision
         else:
             commit_idx = revision
         return self.get_commit(
             commit_id=commit_id, commit_idx=commit_idx, pre_load=pre_load)
 
     def get_changesets(
             self, start=None, end=None, start_date=None, end_date=None,
             branch_name=None, pre_load=None):
         warnings.warn("Use get_commits instead", DeprecationWarning)
         start_id = self._revision_to_commit(start)
         end_id = self._revision_to_commit(end)
         return self.get_commits(
             start_id=start_id, end_id=end_id, start_date=start_date,
             end_date=end_date, branch_name=branch_name, pre_load=pre_load)
 
     def _revision_to_commit(self, revision):
         """
         Translates a revision to a commit_id
 
         Helps to support the old changeset based API which allows to use
         commit ids and commit indices interchangeable.
         """
         if revision is None:
             return revision
 
         if isinstance(revision, compat.string_types):
             commit_id = revision
         else:
             commit_id = self.commit_ids[revision]
         return commit_id
 
     @property
     def in_memory_changeset(self):
         warnings.warn("Use in_memory_commit instead", DeprecationWarning)
         return self.in_memory_commit
 
     def get_path_permissions(self, username):
         """
         Returns a path permission checker or None if not supported
 
         :param username: session user name
         :return: an instance of BasePathPermissionChecker or None
         """
         return None
 
     def install_hooks(self, force=False):
         return self._remote.install_hooks(force)
 
     def get_hooks_info(self):
         return self._remote.get_hooks_info()
 
 
 class BaseCommit(object):
     """
     Each backend should implement it's commit representation.
 
     **Attributes**
 
         ``repository``
             repository object within which commit exists
 
         ``id``
             The commit id, may be ``raw_id`` or i.e. for mercurial's tip
             just ``tip``.
 
         ``raw_id``
             raw commit representation (i.e. full 40 length sha for git
             backend)
 
         ``short_id``
             shortened (if apply) version of ``raw_id``; it would be simple
             shortcut for ``raw_id[:12]`` for git/mercurial backends or same
             as ``raw_id`` for subversion
 
         ``idx``
             commit index
 
         ``files``
             list of ``FileNode`` (``Node`` with NodeKind.FILE) objects
 
         ``dirs``
             list of ``DirNode`` (``Node`` with NodeKind.DIR) objects
 
         ``nodes``
             combined list of ``Node`` objects
 
         ``author``
             author of the commit, as unicode
 
         ``message``
             message of the commit, as unicode
 
         ``parents``
             list of parent commits
 
     """
 
     branch = None
     """
     Depending on the backend this should be set to the branch name of the
     commit. Backends not supporting branches on commits should leave this
     value as ``None``.
     """
 
     _ARCHIVE_PREFIX_TEMPLATE = b'{repo_name}-{short_id}'
     """
     This template is used to generate a default prefix for repository archives
     if no prefix has been specified.
     """
 
     def __str__(self):
         return '<%s at %s:%s>' % (
             self.__class__.__name__, self.idx, self.short_id)
 
     def __repr__(self):
         return self.__str__()
 
     def __unicode__(self):
         return u'%s:%s' % (self.idx, self.short_id)
 
     def __eq__(self, other):
         same_instance = isinstance(other, self.__class__)
         return same_instance and self.raw_id == other.raw_id
 
     def __json__(self):
         parents = []
         try:
             for parent in self.parents:
                 parents.append({'raw_id': parent.raw_id})
         except NotImplementedError:
             # empty commit doesn't have parents implemented
             pass
 
         return {
             'short_id': self.short_id,
             'raw_id': self.raw_id,
             'revision': self.idx,
             'message': self.message,
             'date': self.date,
             'author': self.author,
             'parents': parents,
             'branch': self.branch
         }
 
     def __getstate__(self):
         d = self.__dict__.copy()
         d.pop('_remote', None)
         d.pop('repository', None)
         return d
 
     def _get_refs(self):
         return {
             'branches': [self.branch] if self.branch else [],
             'bookmarks': getattr(self, 'bookmarks', []),
             'tags': self.tags
         }
 
     @LazyProperty
     def last(self):
         """
         ``True`` if this is last commit in repository, ``False``
         otherwise; trying to access this attribute while there is no
         commits would raise `EmptyRepositoryError`
         """
         if self.repository is None:
             raise CommitError("Cannot check if it's most recent commit")
         return self.raw_id == self.repository.commit_ids[-1]
 
     @LazyProperty
     def parents(self):
         """
         Returns list of parent commits.
         """
         raise NotImplementedError
 
     @LazyProperty
     def first_parent(self):
         """
         Returns list of parent commits.
         """
         return self.parents[0] if self.parents else EmptyCommit()
 
     @property
     def merge(self):
         """
         Returns boolean if commit is a merge.
         """
         return len(self.parents) > 1
 
     @LazyProperty
     def children(self):
         """
         Returns list of child commits.
         """
         raise NotImplementedError
 
     @LazyProperty
     def id(self):
         """
         Returns string identifying this commit.
         """
         raise NotImplementedError
 
     @LazyProperty
     def raw_id(self):
         """
         Returns raw string identifying this commit.
         """
         raise NotImplementedError
 
     @LazyProperty
     def short_id(self):
         """
         Returns shortened version of ``raw_id`` attribute, as string,
         identifying this commit, useful for presentation to users.
         """
         raise NotImplementedError
 
     @LazyProperty
     def idx(self):
         """
         Returns integer identifying this commit.
         """
         raise NotImplementedError
 
     @LazyProperty
     def committer(self):
         """
         Returns committer for this commit
         """
         raise NotImplementedError
 
     @LazyProperty
     def committer_name(self):
         """
         Returns committer name for this commit
         """
 
         return author_name(self.committer)
 
     @LazyProperty
     def committer_email(self):
         """
         Returns committer email address for this commit
         """
 
         return author_email(self.committer)
 
     @LazyProperty
     def author(self):
         """
         Returns author for this commit
         """
 
         raise NotImplementedError
 
     @LazyProperty
     def author_name(self):
         """
         Returns author name for this commit
         """
 
         return author_name(self.author)
 
     @LazyProperty
     def author_email(self):
         """
         Returns author email address for this commit
         """
 
         return author_email(self.author)
 
     def get_file_mode(self, path):
         """
         Returns stat mode of the file at `path`.
         """
         raise NotImplementedError
 
     def is_link(self, path):
         """
         Returns ``True`` if given `path` is a symlink
         """
         raise NotImplementedError
 
     def get_file_content(self, path):
         """
         Returns content of the file at the given `path`.
         """
         raise NotImplementedError
 
+    def get_file_content_streamed(self, path):
+        """
+        returns a streaming response from vcsserver with file content
+        """
+        raise NotImplementedError
+
     def get_file_size(self, path):
         """
         Returns size of the file at the given `path`.
         """
         raise NotImplementedError
 
     def get_path_commit(self, path, pre_load=None):
         """
         Returns last commit of the file at the given `path`.
 
         :param pre_load: Optional. List of commit attributes to load.
         """
         commits = self.get_path_history(path, limit=1, pre_load=pre_load)
         if not commits:
             raise RepositoryError(
                 'Failed to fetch history for path {}. '
                 'Please check if such path exists in your repository'.format(
                     path))
         return commits[0]
 
     def get_path_history(self, path, limit=None, pre_load=None):
         """
         Returns history of file as reversed list of :class:`BaseCommit`
         objects for which file at given `path` has been modified.
 
         :param limit: Optional. Allows to limit the size of the returned
            history. This is intended as a hint to the underlying backend, so
            that it can apply optimizations depending on the limit.
         :param pre_load: Optional. List of commit attributes to load.
         """
         raise NotImplementedError
 
     def get_file_annotate(self, path, pre_load=None):
         """
         Returns a generator of four element tuples with
         lineno, sha, commit lazy loader and line
 
         :param pre_load: Optional. List of commit attributes to load.
         """
         raise NotImplementedError
 
     def get_nodes(self, path):
         """
         Returns combined ``DirNode`` and ``FileNode`` objects list representing
         state of commit at the given ``path``.
 
         :raises ``CommitError``: if node at the given ``path`` is not
           instance of ``DirNode``
         """
         raise NotImplementedError
 
     def get_node(self, path):
         """
         Returns ``Node`` object from the given ``path``.
 
         :raises ``NodeDoesNotExistError``: if there is no node at the given
           ``path``
         """
         raise NotImplementedError
 
     def get_largefile_node(self, path):
         """
         Returns the path to largefile from Mercurial/Git-lfs storage.
         or None if it's not a largefile node
         """
         return None
 
     def archive_repo(self, archive_dest_path, kind='tgz', subrepos=None,
                      prefix=None, write_metadata=False, mtime=None, archive_at_path='/'):
         """
         Creates an archive containing the contents of the repository.
 
         :param archive_dest_path: path to the file which to create the archive.
         :param kind: one of following: ``"tbz2"``, ``"tgz"``, ``"zip"``.
         :param prefix: name of root directory in archive.
             Default is repository name and commit's short_id joined with dash:
             ``"{repo_name}-{short_id}"``.
         :param write_metadata: write a metadata file into archive.
         :param mtime: custom modification time for archive creation, defaults
             to time.time() if not given.
         :param archive_at_path: pack files at this path (default '/')
 
         :raise VCSError: If prefix has a problem.
         """
         allowed_kinds = [x[0] for x in settings.ARCHIVE_SPECS]
         if kind not in allowed_kinds:
             raise ImproperArchiveTypeError(
                 'Archive kind (%s) not supported use one of %s' %
                 (kind, allowed_kinds))
 
         prefix = self._validate_archive_prefix(prefix)
 
         mtime = mtime is not None or time.mktime(self.date.timetuple())
 
         file_info = []
         cur_rev = self.repository.get_commit(commit_id=self.raw_id)
         for _r, _d, files in cur_rev.walk(archive_at_path):
             for f in files:
                 f_path = os.path.join(prefix, f.path)
                 file_info.append(
                     (f_path, f.mode, f.is_link(), f.raw_bytes))
 
         if write_metadata:
             metadata = [
                 ('repo_name', self.repository.name),
                 ('commit_id', self.raw_id),
                 ('mtime', mtime),
                 ('branch', self.branch),
                 ('tags', ','.join(self.tags)),
             ]
             meta = ["%s:%s" % (f_name, value) for f_name, value in metadata]
             file_info.append(('.archival.txt', 0o644, False, '\n'.join(meta)))
 
         connection.Hg.archive_repo(archive_dest_path, mtime, file_info, kind)
 
     def _validate_archive_prefix(self, prefix):
         if prefix is None:
             prefix = self._ARCHIVE_PREFIX_TEMPLATE.format(
                 repo_name=safe_str(self.repository.name),
                 short_id=self.short_id)
         elif not isinstance(prefix, str):
             raise ValueError("prefix not a bytes object: %s" % repr(prefix))
         elif prefix.startswith('/'):
             raise VCSError("Prefix cannot start with leading slash")
         elif prefix.strip() == '':
             raise VCSError("Prefix cannot be empty")
         return prefix
 
     @LazyProperty
     def root(self):
         """
         Returns ``RootNode`` object for this commit.
         """
         return self.get_node('')
 
     def next(self, branch=None):
         """
         Returns next commit from current, if branch is gives it will return
         next commit belonging to this branch
 
         :param branch: show commits within the given named branch
         """
         indexes = xrange(self.idx + 1, self.repository.count())
         return self._find_next(indexes, branch)
 
     def prev(self, branch=None):
         """
         Returns previous commit from current, if branch is gives it will
         return previous commit belonging to this branch
 
         :param branch: show commit within the given named branch
         """
         indexes = xrange(self.idx - 1, -1, -1)
         return self._find_next(indexes, branch)
 
     def _find_next(self, indexes, branch=None):
         if branch and self.branch != branch:
             raise VCSError('Branch option used on commit not belonging '
                            'to that branch')
 
         for next_idx in indexes:
             commit = self.repository.get_commit(commit_idx=next_idx)
             if branch and branch != commit.branch:
                 continue
             return commit
         raise CommitDoesNotExistError
 
     def diff(self, ignore_whitespace=True, context=3):
         """
         Returns a `Diff` object representing the change made by this commit.
         """
         parent = self.first_parent
         diff = self.repository.get_diff(
             parent, self,
             ignore_whitespace=ignore_whitespace,
             context=context)
         return diff
 
     @LazyProperty
     def added(self):
         """
         Returns list of added ``FileNode`` objects.
         """
         raise NotImplementedError
 
     @LazyProperty
     def changed(self):
         """
         Returns list of modified ``FileNode`` objects.
         """
         raise NotImplementedError
 
     @LazyProperty
     def removed(self):
         """
         Returns list of removed ``FileNode`` objects.
         """
         raise NotImplementedError
 
     @LazyProperty
     def size(self):
         """
         Returns total number of bytes from contents of all filenodes.
         """
         return sum((node.size for node in self.get_filenodes_generator()))
 
     def walk(self, topurl=''):
         """
         Similar to os.walk method. Insted of filesystem it walks through
         commit starting at given ``topurl``.  Returns generator of tuples
         (topnode, dirnodes, filenodes).
         """
         topnode = self.get_node(topurl)
         if not topnode.is_dir():
             return
         yield (topnode, topnode.dirs, topnode.files)
         for dirnode in topnode.dirs:
             for tup in self.walk(dirnode.path):
                 yield tup
 
     def get_filenodes_generator(self):
         """
         Returns generator that yields *all* file nodes.
         """
         for topnode, dirs, files in self.walk():
             for node in files:
                 yield node
 
     #
     # Utilities for sub classes to support consistent behavior
     #
 
     def no_node_at_path(self, path):
         return NodeDoesNotExistError(
             u"There is no file nor directory at the given path: "
             u"`%s` at commit %s" % (safe_unicode(path), self.short_id))
 
     def _fix_path(self, path):
         """
         Paths are stored without trailing slash so we need to get rid off it if
         needed.
         """
         return path.rstrip('/')
 
     #
     # Deprecated API based on changesets
     #
 
     @property
     def revision(self):
         warnings.warn("Use idx instead", DeprecationWarning)
         return self.idx
 
     @revision.setter
     def revision(self, value):
         warnings.warn("Use idx instead", DeprecationWarning)
         self.idx = value
 
     def get_file_changeset(self, path):
         warnings.warn("Use get_path_commit instead", DeprecationWarning)
         return self.get_path_commit(path)
 
 
 class BaseChangesetClass(type):
 
     def __instancecheck__(self, instance):
         return isinstance(instance, BaseCommit)
 
 
 class BaseChangeset(BaseCommit):
 
     __metaclass__ = BaseChangesetClass
 
     def __new__(cls, *args, **kwargs):
         warnings.warn(
             "Use BaseCommit instead of BaseChangeset", DeprecationWarning)
         return super(BaseChangeset, cls).__new__(cls, *args, **kwargs)
 
 
 class BaseInMemoryCommit(object):
     """
     Represents differences between repository's state (most recent head) and
     changes made *in place*.
 
     **Attributes**
 
         ``repository``
             repository object for this in-memory-commit
 
         ``added``
             list of ``FileNode`` objects marked as *added*
 
         ``changed``
             list of ``FileNode`` objects marked as *changed*
 
         ``removed``
             list of ``FileNode`` or ``RemovedFileNode`` objects marked to be
             *removed*
 
         ``parents``
             list of :class:`BaseCommit` instances representing parents of
             in-memory commit. Should always be 2-element sequence.
 
     """
 
     def __init__(self, repository):
         self.repository = repository
         self.added = []
         self.changed = []
         self.removed = []
         self.parents = []
 
     def add(self, *filenodes):
         """
         Marks given ``FileNode`` objects as *to be committed*.
 
         :raises ``NodeAlreadyExistsError``: if node with same path exists at
            latest commit
         :raises ``NodeAlreadyAddedError``: if node with same path is already
            marked as *added*
         """
         # Check if not already marked as *added* first
         for node in filenodes:
             if node.path in (n.path for n in self.added):
                 raise NodeAlreadyAddedError(
                     "Such FileNode %s is already marked for addition"
                     % node.path)
         for node in filenodes:
             self.added.append(node)
 
     def change(self, *filenodes):
         """
         Marks given ``FileNode`` objects to be *changed* in next commit.
 
         :raises ``EmptyRepositoryError``: if there are no commits yet
         :raises ``NodeAlreadyExistsError``: if node with same path is already
            marked to be *changed*
         :raises ``NodeAlreadyRemovedError``: if node with same path is already
            marked to be *removed*
         :raises ``NodeDoesNotExistError``: if node doesn't exist in latest
            commit
         :raises ``NodeNotChangedError``: if node hasn't really be changed
         """
         for node in filenodes:
             if node.path in (n.path for n in self.removed):
                 raise NodeAlreadyRemovedError(
                     "Node at %s is already marked as removed" % node.path)
         try:
             self.repository.get_commit()
         except EmptyRepositoryError:
             raise EmptyRepositoryError(
                 "Nothing to change - try to *add* new nodes rather than "
                 "changing them")
         for node in filenodes:
             if node.path in (n.path for n in self.changed):
                 raise NodeAlreadyChangedError(
                     "Node at '%s' is already marked as changed" % node.path)
             self.changed.append(node)
 
     def remove(self, *filenodes):
         """
         Marks given ``FileNode`` (or ``RemovedFileNode``) objects to be
         *removed* in next commit.
 
         :raises ``NodeAlreadyRemovedError``: if node has been already marked to
            be *removed*
         :raises ``NodeAlreadyChangedError``: if node has been already marked to
            be *changed*
         """
         for node in filenodes:
             if node.path in (n.path for n in self.removed):
                 raise NodeAlreadyRemovedError(
                     "Node is already marked to for removal at %s" % node.path)
             if node.path in (n.path for n in self.changed):
                 raise NodeAlreadyChangedError(
                     "Node is already marked to be changed at %s" % node.path)
             # We only mark node as *removed* - real removal is done by
             # commit method
             self.removed.append(node)
 
     def reset(self):
         """
         Resets this instance to initial state (cleans ``added``, ``changed``
         and ``removed`` lists).
         """
         self.added = []
         self.changed = []
         self.removed = []
         self.parents = []
 
     def get_ipaths(self):
         """
         Returns generator of paths from nodes marked as added, changed or
         removed.
         """
         for node in itertools.chain(self.added, self.changed, self.removed):
             yield node.path
 
     def get_paths(self):
         """
         Returns list of paths from nodes marked as added, changed or removed.
         """
         return list(self.get_ipaths())
 
     def check_integrity(self, parents=None):
         """
         Checks in-memory commit's integrity. Also, sets parents if not
         already set.
 
         :raises CommitError: if any error occurs (i.e.
           ``NodeDoesNotExistError``).
         """
         if not self.parents:
             parents = parents or []
             if len(parents) == 0:
                 try:
                     parents = [self.repository.get_commit(), None]
                 except EmptyRepositoryError:
                     parents = [None, None]
             elif len(parents) == 1:
                 parents += [None]
             self.parents = parents
 
         # Local parents, only if not None
         parents = [p for p in self.parents if p]
 
         # Check nodes marked as added
         for p in parents:
             for node in self.added:
                 try:
                     p.get_node(node.path)
                 except NodeDoesNotExistError:
                     pass
                 else:
                     raise NodeAlreadyExistsError(
                         "Node `%s` already exists at %s" % (node.path, p))
 
         # Check nodes marked as changed
         missing = set(self.changed)
         not_changed = set(self.changed)
         if self.changed and not parents:
             raise NodeDoesNotExistError(str(self.changed[0].path))
         for p in parents:
             for node in self.changed:
                 try:
                     old = p.get_node(node.path)
                     missing.remove(node)
                     # if content actually changed, remove node from not_changed
                     if old.content != node.content:
                         not_changed.remove(node)
                 except NodeDoesNotExistError:
                     pass
         if self.changed and missing:
             raise NodeDoesNotExistError(
                 "Node `%s` marked as modified but missing in parents: %s"
                 % (node.path, parents))
 
         if self.changed and not_changed:
             raise NodeNotChangedError(
                 "Node `%s` wasn't actually changed (parents: %s)"
                 % (not_changed.pop().path, parents))
 
         # Check nodes marked as removed
         if self.removed and not parents:
             raise NodeDoesNotExistError(
                 "Cannot remove node at %s as there "
                 "were no parents specified" % self.removed[0].path)
         really_removed = set()
         for p in parents:
             for node in self.removed:
                 try:
                     p.get_node(node.path)
                     really_removed.add(node)
                 except CommitError:
                     pass
         not_removed = set(self.removed) - really_removed
         if not_removed:
             # TODO: johbo: This code branch does not seem to be covered
             raise NodeDoesNotExistError(
                 "Cannot remove node at %s from "
                 "following parents: %s" % (not_removed, parents))
 
     def commit(self, message, author, parents=None, branch=None, date=None, **kwargs):
         """
         Performs in-memory commit (doesn't check workdir in any way) and
         returns newly created :class:`BaseCommit`. Updates repository's
         attribute `commits`.
 
         .. note::
 
            While overriding this method each backend's should call
            ``self.check_integrity(parents)`` in the first place.
 
         :param message: message of the commit
         :param author: full username, i.e. "Joe Doe <[e-mail]>"
         :param parents: single parent or sequence of parents from which commit
            would be derived
         :param date: ``datetime.datetime`` instance. Defaults to
            ``datetime.datetime.now()``.
         :param branch: branch name, as string. If none given, default backend's
            branch would be used.
 
         :raises ``CommitError``: if any error occurs while committing
         """
         raise NotImplementedError
 
 
 class BaseInMemoryChangesetClass(type):
 
     def __instancecheck__(self, instance):
         return isinstance(instance, BaseInMemoryCommit)
 
 
 class BaseInMemoryChangeset(BaseInMemoryCommit):
 
     __metaclass__ = BaseInMemoryChangesetClass
 
     def __new__(cls, *args, **kwargs):
         warnings.warn(
             "Use BaseCommit instead of BaseInMemoryCommit", DeprecationWarning)
         return super(BaseInMemoryChangeset, cls).__new__(cls, *args, **kwargs)
 
 
 class EmptyCommit(BaseCommit):
     """
     An dummy empty commit. It's possible to pass hash when creating
     an EmptyCommit
     """
 
     def __init__(
             self, commit_id=EMPTY_COMMIT_ID, repo=None, alias=None, idx=-1,
             message='', author='', date=None):
         self._empty_commit_id = commit_id
         # TODO: johbo: Solve idx parameter, default value does not make
         # too much sense
         self.idx = idx
         self.message = message
         self.author = author
         self.date = date or datetime.datetime.fromtimestamp(0)
         self.repository = repo
         self.alias = alias
 
     @LazyProperty
     def raw_id(self):
         """
         Returns raw string identifying this commit, useful for web
         representation.
         """
 
         return self._empty_commit_id
 
     @LazyProperty
     def branch(self):
         if self.alias:
             from rhodecode.lib.vcs.backends import get_backend
             return get_backend(self.alias).DEFAULT_BRANCH_NAME
 
     @LazyProperty
     def short_id(self):
         return self.raw_id[:12]
 
     @LazyProperty
     def id(self):
         return self.raw_id
 
     def get_path_commit(self, path):
         return self
 
     def get_file_content(self, path):
         return u''
 
+    def get_file_content_streamed(self, path):
+        yield self.get_file_content()
+
     def get_file_size(self, path):
         return 0
 
 
 class EmptyChangesetClass(type):
 
     def __instancecheck__(self, instance):
         return isinstance(instance, EmptyCommit)
 
 
 class EmptyChangeset(EmptyCommit):
 
     __metaclass__ = EmptyChangesetClass
 
     def __new__(cls, *args, **kwargs):
         warnings.warn(
             "Use EmptyCommit instead of EmptyChangeset", DeprecationWarning)
         return super(EmptyCommit, cls).__new__(cls, *args, **kwargs)
 
     def __init__(self, cs=EMPTY_COMMIT_ID, repo=None, requested_revision=None,
                  alias=None, revision=-1, message='', author='', date=None):
         if requested_revision is not None:
             warnings.warn(
                 "Parameter requested_revision not supported anymore",
                 DeprecationWarning)
         super(EmptyChangeset, self).__init__(
             commit_id=cs, repo=repo, alias=alias, idx=revision,
             message=message, author=author, date=date)
 
         @property
         def revision(self):
             warnings.warn("Use idx instead", DeprecationWarning)
             return self.idx
 
         @revision.setter
         def revision(self, value):
             warnings.warn("Use idx instead", DeprecationWarning)
             self.idx = value
 
 
 class EmptyRepository(BaseRepository):
     def __init__(self, repo_path=None, config=None, create=False, **kwargs):
         pass
 
     def get_diff(self, *args, **kwargs):
         from rhodecode.lib.vcs.backends.git.diff import GitDiff
         return GitDiff('')
 
 
 class CollectionGenerator(object):
 
     def __init__(self, repo, commit_ids, collection_size=None, pre_load=None, translate_tag=None):
         self.repo = repo
         self.commit_ids = commit_ids
         # TODO: (oliver) this isn't currently hooked up
         self.collection_size = None
         self.pre_load = pre_load
         self.translate_tag = translate_tag
 
     def __len__(self):
         if self.collection_size is not None:
             return self.collection_size
         return self.commit_ids.__len__()
 
     def __iter__(self):
         for commit_id in self.commit_ids:
             # TODO: johbo: Mercurial passes in commit indices or commit ids
             yield self._commit_factory(commit_id)
 
     def _commit_factory(self, commit_id):
         """
         Allows backends to override the way commits are generated.
         """
         return self.repo.get_commit(
             commit_id=commit_id, pre_load=self.pre_load,
             translate_tag=self.translate_tag)
 
     def __getslice__(self, i, j):
         """
         Returns an iterator of sliced repository
         """
         commit_ids = self.commit_ids[i:j]
         return self.__class__(
             self.repo, commit_ids, pre_load=self.pre_load,
             translate_tag=self.translate_tag)
 
     def __repr__(self):
         return '<CollectionGenerator[len:%s]>' % (self.__len__())
 
 
 class Config(object):
     """
     Represents the configuration for a repository.
 
     The API is inspired by :class:`ConfigParser.ConfigParser` from the
     standard library. It implements only the needed subset.
     """
 
     def __init__(self):
         self._values = {}
 
     def copy(self):
         clone = Config()
         for section, values in self._values.items():
             clone._values[section] = values.copy()
         return clone
 
     def __repr__(self):
         return '<Config(%s sections) at %s>' % (
             len(self._values), hex(id(self)))
 
     def items(self, section):
         return self._values.get(section, {}).iteritems()
 
     def get(self, section, option):
         return self._values.get(section, {}).get(option)
 
     def set(self, section, option, value):
         section_values = self._values.setdefault(section, {})
         section_values[option] = value
 
     def clear_section(self, section):
         self._values[section] = {}
 
     def serialize(self):
         """
         Creates a list of three tuples (section, key, value) representing
         this config object.
         """
         items = []
         for section in self._values:
             for option, value in self._values[section].items():
                 items.append(
                     (safe_str(section), safe_str(option), safe_str(value)))
         return items
 
 
 class Diff(object):
     """
     Represents a diff result from a repository backend.
 
     Subclasses have to provide a backend specific value for
     :attr:`_header_re` and :attr:`_meta_re`.
     """
     _meta_re = None
     _header_re = None
 
     def __init__(self, raw_diff):
         self.raw = raw_diff
 
     def chunks(self):
         """
         split the diff in chunks of separate --git a/file b/file chunks
         to make diffs consistent we must prepend with \n, and make sure
         we can detect last chunk as this was also has special rule
         """
 
         diff_parts = ('\n' + self.raw).split('\ndiff --git')
         header = diff_parts[0]
 
         if self._meta_re:
             match = self._meta_re.match(header)
 
         chunks = diff_parts[1:]
         total_chunks = len(chunks)
 
         return (
             DiffChunk(chunk, self, cur_chunk == total_chunks)
             for cur_chunk, chunk in enumerate(chunks, start=1))
 
 
 class DiffChunk(object):
 
     def __init__(self, chunk, diff, last_chunk):
         self._diff = diff
 
         # since we split by \ndiff --git that part is lost from original diff
         # we need to re-apply it at the end, EXCEPT ! if it's last chunk
         if not last_chunk:
             chunk += '\n'
 
         match = self._diff._header_re.match(chunk)
         self.header = match.groupdict()
         self.diff = chunk[match.end():]
         self.raw = chunk
 
 
 class BasePathPermissionChecker(object):
 
     @staticmethod
     def create_from_patterns(includes, excludes):
         if includes and '*' in includes and not excludes:
             return AllPathPermissionChecker()
         elif excludes and '*' in excludes:
             return NonePathPermissionChecker()
         else:
             return PatternPathPermissionChecker(includes, excludes)
 
     @property
     def has_full_access(self):
         raise NotImplemented()
 
     def has_access(self, path):
         raise NotImplemented()
 
 
 class AllPathPermissionChecker(BasePathPermissionChecker):
 
     @property
     def has_full_access(self):
         return True
 
     def has_access(self, path):
         return True
 
 
 class NonePathPermissionChecker(BasePathPermissionChecker):
 
     @property
     def has_full_access(self):
         return False
 
     def has_access(self, path):
         return False
 
 
 class PatternPathPermissionChecker(BasePathPermissionChecker):
 
     def __init__(self, includes, excludes):
         self.includes = includes
         self.excludes = excludes
         self.includes_re = [] if not includes else [
             re.compile(fnmatch.translate(pattern)) for pattern in includes]
         self.excludes_re = [] if not excludes else [
             re.compile(fnmatch.translate(pattern)) for pattern in excludes]
 
     @property
     def has_full_access(self):
         return '*' in self.includes and not self.excludes
 
     def has_access(self, path):
         for regex in self.excludes_re:
             if regex.match(path):
                 return False
         for regex in self.includes_re:
             if regex.match(path):
                 return True
         return False

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

@@ -1,474 +1,479 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 GIT commit module
 """
 
 import re
 import stat
 from itertools import chain
 from StringIO import StringIO
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from rhodecode.lib.datelib import utcdate_fromtimestamp
 from rhodecode.lib.utils import safe_unicode, safe_str
 from rhodecode.lib.utils2 import safe_int
 from rhodecode.lib.vcs.conf import settings
 from rhodecode.lib.vcs.backends import base
 from rhodecode.lib.vcs.exceptions import CommitError, NodeDoesNotExistError
 from rhodecode.lib.vcs.nodes import (
     FileNode, DirNode, NodeKind, RootNode, SubModuleNode,
     ChangedFileNodesGenerator, AddedFileNodesGenerator,
     RemovedFileNodesGenerator, LargeFileNode)
 from rhodecode.lib.vcs.compat import configparser
 
 
 class GitCommit(base.BaseCommit):
     """
     Represents state of the repository at single commit id.
     """
 
     _filter_pre_load = [
         # done through a more complex tree walk on parents
         "affected_files",
         # done through subprocess not remote call
         "children",
         # done through a more complex tree walk on parents
         "status",
         # mercurial specific property not supported here
         "_file_paths",
         # mercurial specific property not supported here
         'obsolete',
         # mercurial specific property not supported here
         'phase',
         # mercurial specific property not supported here
         'hidden'
     ]
 
     def __init__(self, repository, raw_id, idx, pre_load=None):
         self.repository = repository
         self._remote = repository._remote
         # TODO: johbo: Tweak of raw_id should not be necessary
         self.raw_id = safe_str(raw_id)
         self.idx = idx
 
         self._set_bulk_properties(pre_load)
 
         # caches
         self._stat_modes = {}  # stat info for paths
         self._paths = {}  # path processed with parse_tree
         self.nodes = {}
         self._submodules = None
 
     def _set_bulk_properties(self, pre_load):
 
         if not pre_load:
             return
         pre_load = [entry for entry in pre_load
                     if entry not in self._filter_pre_load]
         if not pre_load:
             return
 
         result = self._remote.bulk_request(self.raw_id, pre_load)
         for attr, value in result.items():
             if attr in ["author", "message"]:
                 if value:
                     value = safe_unicode(value)
             elif attr == "date":
                 value = utcdate_fromtimestamp(*value)
             elif attr == "parents":
                 value = self._make_commits(value)
             elif attr == "branch":
                 value = value[0] if value else None
             self.__dict__[attr] = value
 
     @LazyProperty
     def _commit(self):
         return self._remote[self.raw_id]
 
     @LazyProperty
     def _tree_id(self):
         return self._remote[self._commit['tree']]['id']
 
     @LazyProperty
     def id(self):
         return self.raw_id
 
     @LazyProperty
     def short_id(self):
         return self.raw_id[:12]
 
     @LazyProperty
     def message(self):
         return safe_unicode(self._remote.message(self.id))
 
     @LazyProperty
     def committer(self):
         return safe_unicode(self._remote.author(self.id))
 
     @LazyProperty
     def author(self):
         return safe_unicode(self._remote.author(self.id))
 
     @LazyProperty
     def date(self):
         unix_ts, tz = self._remote.date(self.raw_id)
         return utcdate_fromtimestamp(unix_ts, tz)
 
     @LazyProperty
     def status(self):
         """
         Returns modified, added, removed, deleted files for current commit
         """
         return self.changed, self.added, self.removed
 
     @LazyProperty
     def tags(self):
         tags = [safe_unicode(name) for name,
                 commit_id in self.repository.tags.iteritems()
                 if commit_id == self.raw_id]
         return tags
 
     @LazyProperty
     def commit_branches(self):
         branches = []
         for name, commit_id in self.repository.branches.iteritems():
             if commit_id == self.raw_id:
                 branches.append(name)
         return branches
 
     @LazyProperty
     def branch(self):
         branches = self._remote.branch(self.raw_id)
 
         if branches:
             # actually commit can have multiple branches in git
             return safe_unicode(branches[0])
 
     def _get_tree_id_for_path(self, path):
         path = safe_str(path)
         if path in self._paths:
             return self._paths[path]
 
         tree_id = self._tree_id
 
         path = path.strip('/')
         if path == '':
             data = [tree_id, "tree"]
             self._paths[''] = data
             return data
 
         tree_id, tree_type, tree_mode = \
             self._remote.tree_and_type_for_path(self.raw_id, path)
         if tree_id is None:
             raise self.no_node_at_path(path)
 
         self._paths[path] = [tree_id, tree_type]
         self._stat_modes[path] = tree_mode
 
         if path not in self._paths:
             raise self.no_node_at_path(path)
 
         return self._paths[path]
 
     def _get_kind(self, path):
         tree_id, type_ = self._get_tree_id_for_path(path)
         if type_ == 'blob':
             return NodeKind.FILE
         elif type_ == 'tree':
             return NodeKind.DIR
         elif type_ == 'link':
             return NodeKind.SUBMODULE
         return None
 
     def _get_filectx(self, path):
         path = self._fix_path(path)
         if self._get_kind(path) != NodeKind.FILE:
             raise CommitError(
                 "File does not exist for commit %s at  '%s'" % (self.raw_id, path))
         return path
 
     def _get_file_nodes(self):
         return chain(*(t[2] for t in self.walk()))
 
     @LazyProperty
     def parents(self):
         """
         Returns list of parent commits.
         """
         parent_ids = self._remote.parents(self.id)
         return self._make_commits(parent_ids)
 
     @LazyProperty
     def children(self):
         """
         Returns list of child commits.
         """
 
         children = self._remote.children(self.raw_id)
         return self._make_commits(children)
 
     def _make_commits(self, commit_ids):
         def commit_maker(_commit_id):
             return self.repository.get_commit(commit_id=commit_id)
 
         return [commit_maker(commit_id) for commit_id in commit_ids]
 
     def get_file_mode(self, path):
         """
         Returns stat mode of the file at the given `path`.
         """
         path = safe_str(path)
         # ensure path is traversed
         self._get_tree_id_for_path(path)
         return self._stat_modes[path]
 
     def is_link(self, path):
         return stat.S_ISLNK(self.get_file_mode(path))
 
     def get_file_content(self, path):
         """
         Returns content of the file at given `path`.
         """
         tree_id, _ = self._get_tree_id_for_path(path)
         return self._remote.blob_as_pretty_string(tree_id)
 
+    def get_file_content_streamed(self, path):
+        tree_id, _ = self._get_tree_id_for_path(path)
+        stream_method = getattr(self._remote, 'stream:blob_as_pretty_string')
+        return stream_method(tree_id)
+
     def get_file_size(self, path):
         """
         Returns size of the file at given `path`.
         """
         tree_id, _ = self._get_tree_id_for_path(path)
         return self._remote.blob_raw_length(tree_id)
 
     def get_path_history(self, path, limit=None, pre_load=None):
         """
         Returns history of file as reversed list of `GitCommit` objects for
         which file at given `path` has been modified.
         """
 
         path = self._get_filectx(path)
         hist = self._remote.node_history(self.raw_id, path, limit)
         return [
             self.repository.get_commit(commit_id=commit_id, pre_load=pre_load)
             for commit_id in hist]
 
     def get_file_annotate(self, path, pre_load=None):
         """
         Returns a generator of four element tuples with
             lineno, commit_id, commit lazy loader and line
         """
 
         result = self._remote.node_annotate(self.raw_id, path)
 
         for ln_no, commit_id, content in result:
             yield (
                 ln_no, commit_id,
                 lambda: self.repository.get_commit(commit_id=commit_id, pre_load=pre_load),
                 content)
 
     def get_nodes(self, path):
 
         if self._get_kind(path) != NodeKind.DIR:
             raise CommitError(
                 "Directory does not exist for commit %s at '%s'" % (self.raw_id, path))
         path = self._fix_path(path)
 
         tree_id, _ = self._get_tree_id_for_path(path)
 
         dirnodes = []
         filenodes = []
 
         # extracted tree ID gives us our files...
         for name, stat_, id_, type_ in self._remote.tree_items(tree_id):
             if type_ == 'link':
                 url = self._get_submodule_url('/'.join((path, name)))
                 dirnodes.append(SubModuleNode(
                     name, url=url, commit=id_, alias=self.repository.alias))
                 continue
 
             if path != '':
                 obj_path = '/'.join((path, name))
             else:
                 obj_path = name
             if obj_path not in self._stat_modes:
                 self._stat_modes[obj_path] = stat_
 
             if type_ == 'tree':
                 dirnodes.append(DirNode(obj_path, commit=self))
             elif type_ == 'blob':
                 filenodes.append(FileNode(obj_path, commit=self, mode=stat_))
             else:
                 raise CommitError(
                     "Requested object should be Tree or Blob, is %s", type_)
 
         nodes = dirnodes + filenodes
         for node in nodes:
             if node.path not in self.nodes:
                 self.nodes[node.path] = node
         nodes.sort()
         return nodes
 
     def get_node(self, path, pre_load=None):
         if isinstance(path, unicode):
             path = path.encode('utf-8')
         path = self._fix_path(path)
         if path not in self.nodes:
             try:
                 tree_id, type_ = self._get_tree_id_for_path(path)
             except CommitError:
                 raise NodeDoesNotExistError(
                     "Cannot find one of parents' directories for a given "
                     "path: %s" % path)
 
             if type_ == 'link':
                 url = self._get_submodule_url(path)
                 node = SubModuleNode(path, url=url, commit=tree_id,
                                      alias=self.repository.alias)
             elif type_ == 'tree':
                 if path == '':
                     node = RootNode(commit=self)
                 else:
                     node = DirNode(path, commit=self)
             elif type_ == 'blob':
                 node = FileNode(path, commit=self, pre_load=pre_load)
                 self._stat_modes[path] = node.mode
             else:
                 raise self.no_node_at_path(path)
 
             # cache node
             self.nodes[path] = node
 
         return self.nodes[path]
 
     def get_largefile_node(self, path):
         tree_id, _ = self._get_tree_id_for_path(path)
         pointer_spec = self._remote.is_large_file(tree_id)
 
         if pointer_spec:
             # content of that file regular FileNode is the hash of largefile
             file_id = pointer_spec.get('oid_hash')
             if self._remote.in_largefiles_store(file_id):
                 lf_path = self._remote.store_path(file_id)
                 return LargeFileNode(lf_path, commit=self, org_path=path)
 
     @LazyProperty
     def affected_files(self):
         """
         Gets a fast accessible file changes for given commit
         """
         added, modified, deleted = self._changes_cache
         return list(added.union(modified).union(deleted))
 
     @LazyProperty
     def _changes_cache(self):
         added = set()
         modified = set()
         deleted = set()
         _r = self._remote
 
         parents = self.parents
         if not self.parents:
             parents = [base.EmptyCommit()]
         for parent in parents:
             if isinstance(parent, base.EmptyCommit):
                 oid = None
             else:
                 oid = parent.raw_id
             changes = _r.tree_changes(oid, self.raw_id)
             for (oldpath, newpath), (_, _), (_, _) in changes:
                 if newpath and oldpath:
                     modified.add(newpath)
                 elif newpath and not oldpath:
                     added.add(newpath)
                 elif not newpath and oldpath:
                     deleted.add(oldpath)
         return added, modified, deleted
 
     def _get_paths_for_status(self, status):
         """
         Returns sorted list of paths for given ``status``.
 
         :param status: one of: *added*, *modified* or *deleted*
         """
         added, modified, deleted = self._changes_cache
         return sorted({
             'added': list(added),
             'modified': list(modified),
             'deleted': list(deleted)}[status]
         )
 
     @LazyProperty
     def added(self):
         """
         Returns list of added ``FileNode`` objects.
         """
         if not self.parents:
             return list(self._get_file_nodes())
         return AddedFileNodesGenerator(
             [n for n in self._get_paths_for_status('added')], self)
 
     @LazyProperty
     def changed(self):
         """
         Returns list of modified ``FileNode`` objects.
         """
         if not self.parents:
             return []
         return ChangedFileNodesGenerator(
             [n for n in self._get_paths_for_status('modified')], self)
 
     @LazyProperty
     def removed(self):
         """
         Returns list of removed ``FileNode`` objects.
         """
         if not self.parents:
             return []
         return RemovedFileNodesGenerator(
             [n for n in self._get_paths_for_status('deleted')], self)
 
     def _get_submodule_url(self, submodule_path):
         git_modules_path = '.gitmodules'
 
         if self._submodules is None:
             self._submodules = {}
 
             try:
                 submodules_node = self.get_node(git_modules_path)
             except NodeDoesNotExistError:
                 return None
 
             content = submodules_node.content
 
             # ConfigParser fails if there are whitespaces
             content = '\n'.join(l.strip() for l in content.split('\n'))
 
             parser = configparser.ConfigParser()
             parser.readfp(StringIO(content))
 
             for section in parser.sections():
                 path = parser.get(section, 'path')
                 url = parser.get(section, 'url')
                 if path and url:
                     self._submodules[path.strip('/')] = url
 
         return self._submodules.get(submodule_path.strip('/'))

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

@@ -1,380 +1,385 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 HG commit module
 """
 
 import os
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from rhodecode.lib.datelib import utcdate_fromtimestamp
 from rhodecode.lib.utils import safe_str, safe_unicode
 from rhodecode.lib.vcs import path as vcspath
 from rhodecode.lib.vcs.backends import base
 from rhodecode.lib.vcs.backends.hg.diff import MercurialDiff
 from rhodecode.lib.vcs.exceptions import CommitError
 from rhodecode.lib.vcs.nodes import (
     AddedFileNodesGenerator, ChangedFileNodesGenerator, DirNode, FileNode,
     NodeKind, RemovedFileNodesGenerator, RootNode, SubModuleNode,
     LargeFileNode, LARGEFILE_PREFIX)
 from rhodecode.lib.vcs.utils.paths import get_dirs_for_path
 
 
 class MercurialCommit(base.BaseCommit):
     """
     Represents state of the repository at the single commit.
     """
 
     _filter_pre_load = [
         # git specific property not supported here
         "_commit",
     ]
 
     def __init__(self, repository, raw_id, idx, pre_load=None):
         raw_id = safe_str(raw_id)
 
         self.repository = repository
         self._remote = repository._remote
 
         self.raw_id = raw_id
         self.idx = idx
 
         self._set_bulk_properties(pre_load)
 
         # caches
         self.nodes = {}
 
     def _set_bulk_properties(self, pre_load):
         if not pre_load:
             return
         pre_load = [entry for entry in pre_load
                     if entry not in self._filter_pre_load]
         if not pre_load:
             return
 
         result = self._remote.bulk_request(self.raw_id, pre_load)
         for attr, value in result.items():
             if attr in ["author", "branch", "message"]:
                 value = safe_unicode(value)
             elif attr == "affected_files":
                 value = map(safe_unicode, value)
             elif attr == "date":
                 value = utcdate_fromtimestamp(*value)
             elif attr in ["children", "parents"]:
                 value = self._make_commits(value)
             elif attr in ["phase"]:
                 value = self._get_phase_text(value)
             self.__dict__[attr] = value
 
     @LazyProperty
     def tags(self):
         tags = [name for name, commit_id in self.repository.tags.iteritems()
                 if commit_id == self.raw_id]
         return tags
 
     @LazyProperty
     def branch(self):
         return safe_unicode(self._remote.ctx_branch(self.raw_id))
 
     @LazyProperty
     def bookmarks(self):
         bookmarks = [
             name for name, commit_id in self.repository.bookmarks.iteritems()
             if commit_id == self.raw_id]
         return bookmarks
 
     @LazyProperty
     def message(self):
         return safe_unicode(self._remote.ctx_description(self.raw_id))
 
     @LazyProperty
     def committer(self):
         return safe_unicode(self.author)
 
     @LazyProperty
     def author(self):
         return safe_unicode(self._remote.ctx_user(self.raw_id))
 
     @LazyProperty
     def date(self):
         return utcdate_fromtimestamp(*self._remote.ctx_date(self.raw_id))
 
     @LazyProperty
     def status(self):
         """
         Returns modified, added, removed, deleted files for current commit
         """
         return self._remote.ctx_status(self.raw_id)
 
     @LazyProperty
     def _file_paths(self):
         return self._remote.ctx_list(self.raw_id)
 
     @LazyProperty
     def _dir_paths(self):
         p = list(set(get_dirs_for_path(*self._file_paths)))
         p.insert(0, '')
         return p
 
     @LazyProperty
     def _paths(self):
         return self._dir_paths + self._file_paths
 
     @LazyProperty
     def id(self):
         if self.last:
             return u'tip'
         return self.short_id
 
     @LazyProperty
     def short_id(self):
         return self.raw_id[:12]
 
     def _make_commits(self, indexes, pre_load=None):
         return [self.repository.get_commit(commit_idx=idx, pre_load=pre_load)
                 for idx in indexes if idx >= 0]
 
     @LazyProperty
     def parents(self):
         """
         Returns list of parent commits.
         """
         parents = self._remote.ctx_parents(self.raw_id)
         return self._make_commits(parents)
 
     def _get_phase_text(self, phase_id):
         return {
                    0: 'public',
                    1: 'draft',
                    2: 'secret',
                }.get(phase_id) or ''
 
     @LazyProperty
     def phase(self):
         phase_id = self._remote.ctx_phase(self.raw_id)
         phase_text = self._get_phase_text(phase_id)
 
         return safe_unicode(phase_text)
 
     @LazyProperty
     def obsolete(self):
         obsolete = self._remote.ctx_obsolete(self.raw_id)
         return obsolete
 
     @LazyProperty
     def hidden(self):
         hidden = self._remote.ctx_hidden(self.raw_id)
         return hidden
 
     @LazyProperty
     def children(self):
         """
         Returns list of child commits.
         """
         children = self._remote.ctx_children(self.raw_id)
         return self._make_commits(children)
 
     def _fix_path(self, path):
         """
         Mercurial keeps filenodes as str so we need to encode from unicode
         to str.
         """
         return safe_str(super(MercurialCommit, self)._fix_path(path))
 
     def _get_kind(self, path):
         path = self._fix_path(path)
         if path in self._file_paths:
             return NodeKind.FILE
         elif path in self._dir_paths:
             return NodeKind.DIR
         else:
             raise CommitError(
                 "Node does not exist at the given path '%s'" % (path, ))
 
     def _get_filectx(self, path):
         path = self._fix_path(path)
         if self._get_kind(path) != NodeKind.FILE:
             raise CommitError(
                 "File does not exist for idx %s at '%s'" % (self.raw_id, path))
         return path
 
     def get_file_mode(self, path):
         """
         Returns stat mode of the file at the given ``path``.
         """
         path = self._get_filectx(path)
         if 'x' in self._remote.fctx_flags(self.raw_id, path):
             return base.FILEMODE_EXECUTABLE
         else:
             return base.FILEMODE_DEFAULT
 
     def is_link(self, path):
         path = self._get_filectx(path)
         return 'l' in self._remote.fctx_flags(self.raw_id, path)
 
     def get_file_content(self, path):
         """
         Returns content of the file at given ``path``.
         """
         path = self._get_filectx(path)
         return self._remote.fctx_node_data(self.raw_id, path)
 
+    def get_file_content_streamed(self, path):
+        path = self._get_filectx(path)
+        stream_method = getattr(self._remote, 'stream:fctx_node_data')
+        return stream_method(self.raw_id, path)
+
     def get_file_size(self, path):
         """
         Returns size of the file at given ``path``.
         """
         path = self._get_filectx(path)
         return self._remote.fctx_size(self.raw_id, path)
 
     def get_path_history(self, path, limit=None, pre_load=None):
         """
         Returns history of file as reversed list of `MercurialCommit` objects
         for which file at given ``path`` has been modified.
         """
         path = self._get_filectx(path)
         hist = self._remote.node_history(self.raw_id, path, limit)
         return [
             self.repository.get_commit(commit_id=commit_id, pre_load=pre_load)
             for commit_id in hist]
 
     def get_file_annotate(self, path, pre_load=None):
         """
         Returns a generator of four element tuples with
             lineno, commit_id, commit lazy loader and line
         """
         result = self._remote.fctx_annotate(self.raw_id, path)
 
         for ln_no, commit_id, content in result:
             yield (
                 ln_no, commit_id,
                 lambda: self.repository.get_commit(commit_id=commit_id, pre_load=pre_load),
                 content)
 
     def get_nodes(self, path):
         """
         Returns combined ``DirNode`` and ``FileNode`` objects list representing
         state of commit at the given ``path``. If node at the given ``path``
         is not instance of ``DirNode``, CommitError would be raised.
         """
 
         if self._get_kind(path) != NodeKind.DIR:
             raise CommitError(
                 "Directory does not exist for idx %s at '%s'" % (self.raw_id, path))
         path = self._fix_path(path)
 
         filenodes = [
             FileNode(f, commit=self) for f in self._file_paths
             if os.path.dirname(f) == path]
         # TODO: johbo: Check if this can be done in a more obvious way
         dirs = path == '' and '' or [
             d for d in self._dir_paths
             if d and vcspath.dirname(d) == path]
         dirnodes = [
             DirNode(d, commit=self) for d in dirs
             if os.path.dirname(d) == path]
 
         alias = self.repository.alias
         for k, vals in self._submodules.iteritems():
             if vcspath.dirname(k) == path:
                 loc = vals[0]
                 commit = vals[1]
                 dirnodes.append(SubModuleNode(k, url=loc, commit=commit, alias=alias))
 
         nodes = dirnodes + filenodes
         for node in nodes:
             if node.path not in self.nodes:
                 self.nodes[node.path] = node
         nodes.sort()
 
         return nodes
 
     def get_node(self, path, pre_load=None):
         """
         Returns `Node` object from the given `path`. If there is no node at
         the given `path`, `NodeDoesNotExistError` would be raised.
         """
         path = self._fix_path(path)
 
         if path not in self.nodes:
             if path in self._file_paths:
                 node = FileNode(path, commit=self, pre_load=pre_load)
             elif path in self._dir_paths:
                 if path == '':
                     node = RootNode(commit=self)
                 else:
                     node = DirNode(path, commit=self)
             else:
                 raise self.no_node_at_path(path)
 
             # cache node
             self.nodes[path] = node
         return self.nodes[path]
 
     def get_largefile_node(self, path):
         pointer_spec = self._remote.is_large_file(path)
         if pointer_spec:
             # content of that file regular FileNode is the hash of largefile
             file_id = self.get_file_content(path).strip()
 
             if self._remote.in_largefiles_store(file_id):
                 lf_path = self._remote.store_path(file_id)
                 return LargeFileNode(lf_path, commit=self, org_path=path)
             elif self._remote.in_user_cache(file_id):
                 lf_path = self._remote.store_path(file_id)
                 self._remote.link(file_id, path)
                 return LargeFileNode(lf_path, commit=self, org_path=path)
 
     @LazyProperty
     def _submodules(self):
         """
         Returns a dictionary with submodule information from substate file
         of hg repository.
         """
         return self._remote.ctx_substate(self.raw_id)
 
     @LazyProperty
     def affected_files(self):
         """
         Gets a fast accessible file changes for given commit
         """
         return self._remote.ctx_files(self.raw_id)
 
     @property
     def added(self):
         """
         Returns list of added ``FileNode`` objects.
         """
         return AddedFileNodesGenerator([n for n in self.status[1]], self)
 
     @property
     def changed(self):
         """
         Returns list of modified ``FileNode`` objects.
         """
         return ChangedFileNodesGenerator([n for n in self.status[0]], self)
 
     @property
     def removed(self):
         """
         Returns list of removed ``FileNode`` objects.
         """
         return RemovedFileNodesGenerator([n for n in self.status[2]], self)

rhodecode/lib/vcs/backends/svn/commit.py

@@ -1,236 +1,241 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 SVN commit module
 """
 
 
 import dateutil.parser
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from rhodecode.lib.utils import safe_str, safe_unicode
 from rhodecode.lib.vcs import nodes, path as vcspath
 from rhodecode.lib.vcs.backends import base
 from rhodecode.lib.vcs.exceptions import CommitError, NodeDoesNotExistError
 
 
 _SVN_PROP_TRUE = '*'
 
 
 class SubversionCommit(base.BaseCommit):
     """
     Subversion specific implementation of commits
 
     .. attribute:: branch
 
        The Subversion backend does not support to assign branches to
        specific commits. This attribute has always the value `None`.
 
     """
 
     def __init__(self, repository, commit_id):
         self.repository = repository
         self.idx = self.repository._get_commit_idx(commit_id)
         self._svn_rev = self.idx + 1
         self._remote = repository._remote
         # TODO: handling of raw_id should be a method on repository itself,
         # which knows how to translate commit index and commit id
         self.raw_id = commit_id
         self.short_id = commit_id
         self.id = 'r%s' % (commit_id, )
 
         # TODO: Implement the following placeholder attributes
         self.nodes = {}
         self.tags = []
 
     @property
     def author(self):
         return safe_unicode(self._properties.get('svn:author'))
 
     @property
     def date(self):
         return _date_from_svn_properties(self._properties)
 
     @property
     def message(self):
         return safe_unicode(self._properties.get('svn:log'))
 
     @LazyProperty
     def _properties(self):
         return self._remote.revision_properties(self._svn_rev)
 
     @LazyProperty
     def parents(self):
         parent_idx = self.idx - 1
         if parent_idx >= 0:
             parent = self.repository.get_commit(commit_idx=parent_idx)
             return [parent]
         return []
 
     @LazyProperty
     def children(self):
         child_idx = self.idx + 1
         if child_idx < len(self.repository.commit_ids):
             child = self.repository.get_commit(commit_idx=child_idx)
             return [child]
         return []
 
     def get_file_mode(self, path):
         # Note: Subversion flags files which are executable with a special
         # property `svn:executable` which is set to the value ``"*"``.
         if self._get_file_property(path, 'svn:executable') == _SVN_PROP_TRUE:
             return base.FILEMODE_EXECUTABLE
         else:
             return base.FILEMODE_DEFAULT
 
     def is_link(self, path):
         # Note: Subversion has a flag for special files, the content of the
         # file contains the type of that file.
         if self._get_file_property(path, 'svn:special') == _SVN_PROP_TRUE:
             return self.get_file_content(path).startswith('link')
         return False
 
     def _get_file_property(self, path, name):
         file_properties = self._remote.node_properties(
             safe_str(path), self._svn_rev)
         return file_properties.get(name)
 
     def get_file_content(self, path):
         path = self._fix_path(path)
         return self._remote.get_file_content(safe_str(path), self._svn_rev)
 
+    def get_file_content_streamed(self, path):
+        path = self._fix_path(path)
+        stream_method = getattr(self._remote, 'stream:get_file_content')
+        return stream_method(safe_str(path), self._svn_rev)
+
     def get_file_size(self, path):
         path = self._fix_path(path)
         return self._remote.get_file_size(safe_str(path), self._svn_rev)
 
     def get_path_history(self, path, limit=None, pre_load=None):
         path = safe_str(self._fix_path(path))
         history = self._remote.node_history(path, self._svn_rev, limit)
         return [
             self.repository.get_commit(commit_id=str(svn_rev))
             for svn_rev in history]
 
     def get_file_annotate(self, path, pre_load=None):
         result = self._remote.file_annotate(safe_str(path), self._svn_rev)
 
         for zero_based_line_no, svn_rev, content in result:
             commit_id = str(svn_rev)
             line_no = zero_based_line_no + 1
             yield (
                 line_no,
                 commit_id,
                 lambda: self.repository.get_commit(commit_id=commit_id),
                 content)
 
     def get_node(self, path, pre_load=None):
         path = self._fix_path(path)
         if path not in self.nodes:
 
             if path == '':
                 node = nodes.RootNode(commit=self)
             else:
                 node_type = self._remote.get_node_type(
                     safe_str(path), self._svn_rev)
                 if node_type == 'dir':
                     node = nodes.DirNode(path, commit=self)
                 elif node_type == 'file':
                     node = nodes.FileNode(path, commit=self, pre_load=pre_load)
                 else:
                     raise self.no_node_at_path(path)
 
             self.nodes[path] = node
         return self.nodes[path]
 
     def get_nodes(self, path):
         if self._get_kind(path) != nodes.NodeKind.DIR:
             raise CommitError(
                 "Directory does not exist for commit %s at "
                 " '%s'" % (self.raw_id, path))
         path = self._fix_path(path)
 
         path_nodes = []
         for name, kind in self._remote.get_nodes(
                 safe_str(path), revision=self._svn_rev):
             node_path = vcspath.join(path, name)
             if kind == 'dir':
                 node = nodes.DirNode(node_path, commit=self)
             elif kind == 'file':
                 node = nodes.FileNode(node_path, commit=self)
             else:
                 raise ValueError("Node kind %s not supported." % (kind, ))
             self.nodes[node_path] = node
             path_nodes.append(node)
 
         return path_nodes
 
     def _get_kind(self, path):
         path = self._fix_path(path)
         kind = self._remote.get_node_type(path, self._svn_rev)
         if kind == 'file':
             return nodes.NodeKind.FILE
         elif kind == 'dir':
             return nodes.NodeKind.DIR
         else:
             raise CommitError(
                 "Node does not exist at the given path '%s'" % (path, ))
 
     @LazyProperty
     def _changes_cache(self):
         return self._remote.revision_changes(self._svn_rev)
 
     @LazyProperty
     def affected_files(self):
         changed_files = set()
         for files in self._changes_cache.itervalues():
             changed_files.update(files)
         return list(changed_files)
 
     @LazyProperty
     def id(self):
         return self.raw_id
 
     @property
     def added(self):
         return nodes.AddedFileNodesGenerator(
             self._changes_cache['added'], self)
 
     @property
     def changed(self):
         return nodes.ChangedFileNodesGenerator(
             self._changes_cache['changed'], self)
 
     @property
     def removed(self):
         return nodes.RemovedFileNodesGenerator(
             self._changes_cache['removed'], self)
 
 
 def _date_from_svn_properties(properties):
     """
     Parses the date out of given svn properties.
 
     :return: :class:`datetime.datetime` instance. The object is naive.
     """
 
     aware_date = dateutil.parser.parse(properties.get('svn:date'))
     # final_date = aware_date.astimezone(dateutil.tz.tzlocal())
     final_date = aware_date
     return final_date.replace(tzinfo=None)

rhodecode/lib/vcs/client_http.py

@@ -1,318 +1,346 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2016-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 Client for the VCSServer implemented based on HTTP.
 """
 
 import copy
 import logging
 import threading
 import time
 import urllib2
 import urlparse
 import uuid
 import traceback
 
 import pycurl
 import msgpack
 import requests
 from requests.packages.urllib3.util.retry import Retry
 
 import rhodecode
 from rhodecode.lib.system_info import get_cert_path
 from rhodecode.lib.vcs import exceptions, CurlSession
 
 log = logging.getLogger(__name__)
 
 
 # TODO: mikhail: Keep it in sync with vcsserver's
 # HTTPApplication.ALLOWED_EXCEPTIONS
 EXCEPTIONS_MAP = {
     'KeyError': KeyError,
     'URLError': urllib2.URLError,
 }
 
 
-class RepoMaker(object):
-
-    def __init__(self, server_and_port, backend_endpoint, backend_type, session_factory):
-        self.url = urlparse.urljoin('http://%s' % server_and_port, backend_endpoint)
-        self._session_factory = session_factory
-        self.backend_type = backend_type
-
-    def __call__(self, path, repo_id, config, with_wire=None):
-        log.debug('%s RepoMaker call on %s', self.backend_type.upper(), path)
-        return RemoteRepo(path, repo_id, config, self.url, self._session_factory(),
-                          with_wire=with_wire)
-
-    def __getattr__(self, name):
-        def f(*args, **kwargs):
-            return self._call(name, *args, **kwargs)
-        return f
-
-    @exceptions.map_vcs_exceptions
-    def _call(self, name, *args, **kwargs):
-        payload = {
-            'id': str(uuid.uuid4()),
-            'method': name,
-            'backend': self.backend_type,
-            'params': {'args': args, 'kwargs': kwargs}
-        }
-        return _remote_call(
-            self.url, payload, EXCEPTIONS_MAP, self._session_factory())
-
-
-class ServiceConnection(object):
-    def __init__(self, server_and_port, backend_endpoint, session_factory):
-        self.url = urlparse.urljoin('http://%s' % server_and_port, backend_endpoint)
-        self._session_factory = session_factory
-
-    def __getattr__(self, name):
-        def f(*args, **kwargs):
-            return self._call(name, *args, **kwargs)
-
-        return f
-
-    @exceptions.map_vcs_exceptions
-    def _call(self, name, *args, **kwargs):
-        payload = {
-            'id': str(uuid.uuid4()),
-            'method': name,
-            'params': {'args': args, 'kwargs': kwargs}
-        }
-        return _remote_call(
-            self.url, payload, EXCEPTIONS_MAP, self._session_factory())
-
-
-class RemoteRepo(object):
-
-    def __init__(self, path, repo_id, config, url, session, with_wire=None):
-        self.url = url
-        self._session = session
-        with_wire = with_wire or {}
-
-        repo_state_uid = with_wire.get('repo_state_uid') or 'state'
-        self._wire = {
-            "path": path,  # repo path
-            "repo_id": repo_id,
-            "config": config,
-            "repo_state_uid": repo_state_uid,
-            "context": self._create_vcs_cache_context(path, repo_state_uid)
-        }
-
-        if with_wire:
-            self._wire.update(with_wire)
-
-        # NOTE(johbo): Trading complexity for performance. Avoiding the call to
-        # log.debug brings a few percent gain even if is is not active.
-        if log.isEnabledFor(logging.DEBUG):
-            self._call_with_logging = True
-
-        self.cert_dir = get_cert_path(rhodecode.CONFIG.get('__file__'))
-
-    def __getattr__(self, name):
-        def f(*args, **kwargs):
-            return self._call(name, *args, **kwargs)
-        return f
-
-    @exceptions.map_vcs_exceptions
-    def _call(self, name, *args, **kwargs):
-        # TODO: oliver: This is currently necessary pre-call since the
-        # config object is being changed for hooking scenarios
-        wire = copy.deepcopy(self._wire)
-        wire["config"] = wire["config"].serialize()
-        wire["config"].append(('vcs', 'ssl_dir', self.cert_dir))
-
-        payload = {
-            'id': str(uuid.uuid4()),
-            'method': name,
-            'params': {'wire': wire, 'args': args, 'kwargs': kwargs}
-        }
-
-        if self._call_with_logging:
-            start = time.time()
-            context_uid = wire.get('context')
-            log.debug('Calling %s@%s with args:%.10240r. wire_context: %s',
-                      self.url, name, args, context_uid)
-        result = _remote_call(self.url, payload, EXCEPTIONS_MAP, self._session)
-        if self._call_with_logging:
-            log.debug('Call %s@%s took: %.4fs. wire_context: %s',
-                      self.url, name, time.time()-start, context_uid)
-        return result
-
-    def __getitem__(self, key):
-        return self.revision(key)
-
-    def _create_vcs_cache_context(self, *args):
-        """
-        Creates a unique string which is passed to the VCSServer on every
-        remote call. It is used as cache key in the VCSServer.
-        """
-        hash_key = '-'.join(map(str, args))
-        return str(uuid.uuid5(uuid.NAMESPACE_URL, hash_key))
-
-    def invalidate_vcs_cache(self):
-        """
-        This invalidates the context which is sent to the VCSServer on every
-        call to a remote method. It forces the VCSServer to create a fresh
-        repository instance on the next call to a remote method.
-        """
-        self._wire['context'] = str(uuid.uuid4())
-
-
-class RemoteObject(object):
-
-    def __init__(self, url, session):
-        self._url = url
-        self._session = session
-
-        # johbo: Trading complexity for performance. Avoiding the call to
-        # log.debug brings a few percent gain even if is is not active.
-        if log.isEnabledFor(logging.DEBUG):
-            self._call = self._call_with_logging
-
-    def __getattr__(self, name):
-        def f(*args, **kwargs):
-            return self._call(name, *args, **kwargs)
-        return f
-
-    @exceptions.map_vcs_exceptions
-    def _call(self, name, *args, **kwargs):
-        payload = {
-            'id': str(uuid.uuid4()),
-            'method': name,
-            'params': {'args': args, 'kwargs': kwargs}
-        }
-        return _remote_call(self._url, payload, EXCEPTIONS_MAP, self._session)
-
-    def _call_with_logging(self, name, *args, **kwargs):
-        log.debug('Calling %s@%s', self._url, name)
-        return RemoteObject._call(self, name, *args, **kwargs)
-
-
 def _remote_call(url, payload, exceptions_map, session):
     try:
         response = session.post(url, data=msgpack.packb(payload))
     except pycurl.error as e:
         msg = '{}. \npycurl traceback: {}'.format(e, traceback.format_exc())
         raise exceptions.HttpVCSCommunicationError(msg)
     except Exception as e:
         message = getattr(e, 'message', '')
         if 'Failed to connect' in message:
             # gevent doesn't return proper pycurl errors
             raise exceptions.HttpVCSCommunicationError(e)
         else:
             raise
 
     if response.status_code >= 400:
         log.error('Call to %s returned non 200 HTTP code: %s',
                   url, response.status_code)
         raise exceptions.HttpVCSCommunicationError(repr(response.content))
 
     try:
         response = msgpack.unpackb(response.content)
     except Exception:
         log.exception('Failed to decode response %r', response.content)
         raise
 
     error = response.get('error')
     if error:
         type_ = error.get('type', 'Exception')
         exc = exceptions_map.get(type_, Exception)
         exc = exc(error.get('message'))
         try:
             exc._vcs_kind = error['_vcs_kind']
         except KeyError:
             pass
 
         try:
             exc._vcs_server_traceback = error['traceback']
             exc._vcs_server_org_exc_name = error['org_exc']
             exc._vcs_server_org_exc_tb = error['org_exc_tb']
         except KeyError:
             pass
 
         raise exc
     return response.get('result')
 
 
+def _streaming_remote_call(url, payload, exceptions_map, session, chunk_size):
+    try:
+        response = session.post(url, data=msgpack.packb(payload))
+    except pycurl.error as e:
+        msg = '{}. \npycurl traceback: {}'.format(e, traceback.format_exc())
+        raise exceptions.HttpVCSCommunicationError(msg)
+    except Exception as e:
+        message = getattr(e, 'message', '')
+        if 'Failed to connect' in message:
+            # gevent doesn't return proper pycurl errors
+            raise exceptions.HttpVCSCommunicationError(e)
+        else:
+            raise
+
+    if response.status_code >= 400:
+        log.error('Call to %s returned non 200 HTTP code: %s',
+                  url, response.status_code)
+        raise exceptions.HttpVCSCommunicationError(repr(response.content))
+
+    return response.iter_content(chunk_size=chunk_size)
+
+
+class ServiceConnection(object):
+    def __init__(self, server_and_port, backend_endpoint, session_factory):
+        self.url = urlparse.urljoin('http://%s' % server_and_port, backend_endpoint)
+        self._session_factory = session_factory
+
+    def __getattr__(self, name):
+        def f(*args, **kwargs):
+            return self._call(name, *args, **kwargs)
+
+        return f
+
+    @exceptions.map_vcs_exceptions
+    def _call(self, name, *args, **kwargs):
+        payload = {
+            'id': str(uuid.uuid4()),
+            'method': name,
+            'params': {'args': args, 'kwargs': kwargs}
+        }
+        return _remote_call(
+            self.url, payload, EXCEPTIONS_MAP, self._session_factory())
+
+
+class RemoteVCSMaker(object):
+
+    def __init__(self, server_and_port, backend_endpoint, backend_type, session_factory):
+        self.url = urlparse.urljoin('http://%s' % server_and_port, backend_endpoint)
+        self.stream_url = urlparse.urljoin('http://%s' % server_and_port, backend_endpoint+'/stream')
+
+        self._session_factory = session_factory
+        self.backend_type = backend_type
+
+    def __call__(self, path, repo_id, config, with_wire=None):
+        log.debug('%s RepoMaker call on %s', self.backend_type.upper(), path)
+        return RemoteRepo(path, repo_id, config, self, with_wire=with_wire)
+
+    def __getattr__(self, name):
+        def remote_attr(*args, **kwargs):
+            return self._call(name, *args, **kwargs)
+        return remote_attr
+
+    @exceptions.map_vcs_exceptions
+    def _call(self, func_name, *args, **kwargs):
+        payload = {
+            'id': str(uuid.uuid4()),
+            'method': func_name,
+            'backend': self.backend_type,
+            'params': {'args': args, 'kwargs': kwargs}
+        }
+        url = self.url
+        return _remote_call(url, payload, EXCEPTIONS_MAP, self._session_factory())
+
+
+class RemoteRepo(object):
+    CHUNK_SIZE = 16384
+
+    def __init__(self, path, repo_id, config, remote_maker, with_wire=None):
+        self.url = remote_maker.url
+        self.stream_url = remote_maker.stream_url
+        self._session = remote_maker._session_factory()
+
+        with_wire = with_wire or {}
+
+        repo_state_uid = with_wire.get('repo_state_uid') or 'state'
+        self._wire = {
+            "path": path,  # repo path
+            "repo_id": repo_id,
+            "config": config,
+            "repo_state_uid": repo_state_uid,
+            "context": self._create_vcs_cache_context(path, repo_state_uid)
+        }
+
+        if with_wire:
+            self._wire.update(with_wire)
+
+        # NOTE(johbo): Trading complexity for performance. Avoiding the call to
+        # log.debug brings a few percent gain even if is is not active.
+        if log.isEnabledFor(logging.DEBUG):
+            self._call_with_logging = True
+
+        self.cert_dir = get_cert_path(rhodecode.CONFIG.get('__file__'))
+
+    def __getattr__(self, name):
+
+        if name.startswith('stream:'):
+            def repo_remote_attr(*args, **kwargs):
+                return self._call_stream(name, *args, **kwargs)
+        else:
+            def repo_remote_attr(*args, **kwargs):
+                return self._call(name, *args, **kwargs)
+
+        return repo_remote_attr
+
+    def _base_call(self, name, *args, **kwargs):
+        # TODO: oliver: This is currently necessary pre-call since the
+        # config object is being changed for hooking scenarios
+        wire = copy.deepcopy(self._wire)
+        wire["config"] = wire["config"].serialize()
+        wire["config"].append(('vcs', 'ssl_dir', self.cert_dir))
+
+        payload = {
+            'id': str(uuid.uuid4()),
+            'method': name,
+            'params': {'wire': wire, 'args': args, 'kwargs': kwargs}
+        }
+
+        context_uid = wire.get('context')
+        return context_uid, payload
+
+    @exceptions.map_vcs_exceptions
+    def _call(self, name, *args, **kwargs):
+        context_uid, payload = self._base_call(name, *args, **kwargs)
+        url = self.url
+
+        start = time.time()
+        if self._call_with_logging:
+            log.debug('Calling %s@%s with args:%.10240r. wire_context: %s',
+                      url, name, args, context_uid)
+
+        result = _remote_call(url, payload, EXCEPTIONS_MAP, self._session)
+        if self._call_with_logging:
+            log.debug('Call %s@%s took: %.4fs. wire_context: %s',
+                      url, name, time.time()-start, context_uid)
+        return result
+
+    @exceptions.map_vcs_exceptions
+    def _call_stream(self, name, *args, **kwargs):
+        context_uid, payload = self._base_call(name, *args, **kwargs)
+        payload['chunk_size'] = self.CHUNK_SIZE
+        url = self.stream_url
+
+        start = time.time()
+        if self._call_with_logging:
+            log.debug('Calling %s@%s with args:%.10240r. wire_context: %s',
+                      url, name, args, context_uid)
+
+        result = _streaming_remote_call(url, payload, EXCEPTIONS_MAP, self._session,
+                                        self.CHUNK_SIZE)
+
+        if self._call_with_logging:
+            log.debug('Call %s@%s took: %.4fs. wire_context: %s',
+                      url, name, time.time()-start, context_uid)
+        return result
+
+    def __getitem__(self, key):
+        return self.revision(key)
+
+    def _create_vcs_cache_context(self, *args):
+        """
+        Creates a unique string which is passed to the VCSServer on every
+        remote call. It is used as cache key in the VCSServer.
+        """
+        hash_key = '-'.join(map(str, args))
+        return str(uuid.uuid5(uuid.NAMESPACE_URL, hash_key))
+
+    def invalidate_vcs_cache(self):
+        """
+        This invalidates the context which is sent to the VCSServer on every
+        call to a remote method. It forces the VCSServer to create a fresh
+        repository instance on the next call to a remote method.
+        """
+        self._wire['context'] = str(uuid.uuid4())
+
+
 class VcsHttpProxy(object):
 
     CHUNK_SIZE = 16384
 
     def __init__(self, server_and_port, backend_endpoint):
         retries = Retry(total=5, connect=None, read=None, redirect=None)
 
         adapter = requests.adapters.HTTPAdapter(max_retries=retries)
         self.base_url = urlparse.urljoin('http://%s' % server_and_port, backend_endpoint)
         self.session = requests.Session()
         self.session.mount('http://', adapter)
 
     def handle(self, environment, input_data, *args, **kwargs):
         data = {
             'environment': environment,
             'input_data': input_data,
             'args': args,
             'kwargs': kwargs
         }
         result = self.session.post(
             self.base_url, msgpack.packb(data), stream=True)
         return self._get_result(result)
 
     def _deserialize_and_raise(self, error):
         exception = Exception(error['message'])
         try:
             exception._vcs_kind = error['_vcs_kind']
         except KeyError:
             pass
         raise exception
 
     def _iterate(self, result):
         unpacker = msgpack.Unpacker()
         for line in result.iter_content(chunk_size=self.CHUNK_SIZE):
             unpacker.feed(line)
             for chunk in unpacker:
                 yield chunk
 
     def _get_result(self, result):
         iterator = self._iterate(result)
         error = iterator.next()
         if error:
             self._deserialize_and_raise(error)
 
         status = iterator.next()
         headers = iterator.next()
 
         return iterator, status, headers
 
 
 class ThreadlocalSessionFactory(object):
     """
     Creates one CurlSession per thread on demand.
     """
 
     def __init__(self):
         self._thread_local = threading.local()
 
     def __call__(self):
         if not hasattr(self._thread_local, 'curl_session'):
             self._thread_local.curl_session = CurlSession()
         return self._thread_local.curl_session

rhodecode/lib/vcs/nodes.py

@@ -1,850 +1,870 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 """
 Module holding everything related to vcs nodes, with vcs2 architecture.
 """
 
 import os
 import stat
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
+import rhodecode
 from rhodecode.config.conf import LANGUAGES_EXTENSIONS_MAP
 from rhodecode.lib.utils import safe_unicode, safe_str
 from rhodecode.lib.utils2 import md5
 from rhodecode.lib.vcs import path as vcspath
 from rhodecode.lib.vcs.backends.base import EmptyCommit, FILEMODE_DEFAULT
 from rhodecode.lib.vcs.conf.mtypes import get_mimetypes_db
 from rhodecode.lib.vcs.exceptions import NodeError, RemovedFileNodeError
 
 LARGEFILE_PREFIX = '.hglf'
 
 
 class NodeKind:
     SUBMODULE = -1
     DIR = 1
     FILE = 2
     LARGEFILE = 3
 
 
 class NodeState:
     ADDED = u'added'
     CHANGED = u'changed'
     NOT_CHANGED = u'not changed'
     REMOVED = u'removed'
 
 
 class NodeGeneratorBase(object):
     """
     Base class for removed added and changed filenodes, it's a lazy generator
     class that will create filenodes only on iteration or call
 
     The len method doesn't need to create filenodes at all
     """
 
     def __init__(self, current_paths, cs):
         self.cs = cs
         self.current_paths = current_paths
 
     def __call__(self):
         return [n for n in self]
 
     def __getslice__(self, i, j):
         for p in self.current_paths[i:j]:
             yield self.cs.get_node(p)
 
     def __len__(self):
         return len(self.current_paths)
 
     def __iter__(self):
         for p in self.current_paths:
             yield self.cs.get_node(p)
 
 
 class AddedFileNodesGenerator(NodeGeneratorBase):
     """
     Class holding added files for current commit
     """
 
 
 class ChangedFileNodesGenerator(NodeGeneratorBase):
     """
     Class holding changed files for current commit
     """
 
 
 class RemovedFileNodesGenerator(NodeGeneratorBase):
     """
     Class holding removed files for current commit
     """
     def __iter__(self):
         for p in self.current_paths:
             yield RemovedFileNode(path=p)
 
     def __getslice__(self, i, j):
         for p in self.current_paths[i:j]:
             yield RemovedFileNode(path=p)
 
 
 class Node(object):
     """
     Simplest class representing file or directory on repository.  SCM backends
     should use ``FileNode`` and ``DirNode`` subclasses rather than ``Node``
     directly.
 
     Node's ``path`` cannot start with slash as we operate on *relative* paths
     only. Moreover, every single node is identified by the ``path`` attribute,
     so it cannot end with slash, too. Otherwise, path could lead to mistakes.
     """
     RTLO_MARKER = u"\u202E"  # RTLO marker allows swapping text, and certain
                              # security attacks could be used with this
     commit = None
 
     def __init__(self, path, kind):
         self._validate_path(path)  # can throw exception if path is invalid
         self.path = safe_str(path.rstrip('/'))  # we store paths as str
         if path == '' and kind != NodeKind.DIR:
             raise NodeError("Only DirNode and its subclasses may be "
                             "initialized with empty path")
         self.kind = kind
 
         if self.is_root() and not self.is_dir():
             raise NodeError("Root node cannot be FILE kind")
 
     def _validate_path(self, path):
         if path.startswith('/'):
             raise NodeError(
                 "Cannot initialize Node objects with slash at "
                 "the beginning as only relative paths are supported. "
                 "Got %s" % (path,))
 
     @LazyProperty
     def parent(self):
         parent_path = self.get_parent_path()
         if parent_path:
             if self.commit:
                 return self.commit.get_node(parent_path)
             return DirNode(parent_path)
         return None
 
     @LazyProperty
     def unicode_path(self):
         return safe_unicode(self.path)
 
     @LazyProperty
     def has_rtlo(self):
         """Detects if a path has right-to-left-override marker"""
         return self.RTLO_MARKER in self.unicode_path
 
     @LazyProperty
     def unicode_path_safe(self):
         """
         Special SAFE representation of path without the right-to-left-override.
         This should be only used for "showing" the file, cannot be used for any
         urls etc.
         """
         return safe_unicode(self.path).replace(self.RTLO_MARKER, '')
 
     @LazyProperty
     def dir_path(self):
         """
         Returns name of the directory from full path of this vcs node. Empty
         string is returned if there's no directory in the path
         """
         _parts = self.path.rstrip('/').rsplit('/', 1)
         if len(_parts) == 2:
             return safe_unicode(_parts[0])
         return u''
 
     @LazyProperty
     def name(self):
         """
         Returns name of the node so if its path
         then only last part is returned.
         """
         return safe_unicode(self.path.rstrip('/').split('/')[-1])
 
     @property
     def kind(self):
         return self._kind
 
     @kind.setter
     def kind(self, kind):
         if hasattr(self, '_kind'):
             raise NodeError("Cannot change node's kind")
         else:
             self._kind = kind
             # Post setter check (path's trailing slash)
             if self.path.endswith('/'):
                 raise NodeError("Node's path cannot end with slash")
 
     def __cmp__(self, other):
         """
         Comparator using name of the node, needed for quick list sorting.
         """
 
         kind_cmp = cmp(self.kind, other.kind)
         if kind_cmp:
             if isinstance(self, SubModuleNode):
                 # we make submodules equal to dirnode for "sorting" purposes
                 return NodeKind.DIR
             return kind_cmp
         return cmp(self.name, other.name)
 
     def __eq__(self, other):
         for attr in ['name', 'path', 'kind']:
             if getattr(self, attr) != getattr(other, attr):
                 return False
         if self.is_file():
             if self.content != other.content:
                 return False
         else:
             # For DirNode's check without entering each dir
             self_nodes_paths = list(sorted(n.path for n in self.nodes))
             other_nodes_paths = list(sorted(n.path for n in self.nodes))
             if self_nodes_paths != other_nodes_paths:
                 return False
         return True
 
     def __ne__(self, other):
         return not self.__eq__(other)
 
     def __repr__(self):
         return '<%s %r>' % (self.__class__.__name__, self.path)
 
     def __str__(self):
         return self.__repr__()
 
     def __unicode__(self):
         return self.name
 
     def get_parent_path(self):
         """
         Returns node's parent path or empty string if node is root.
         """
         if self.is_root():
             return ''
         return vcspath.dirname(self.path.rstrip('/')) + '/'
 
     def is_file(self):
         """
         Returns ``True`` if node's kind is ``NodeKind.FILE``, ``False``
         otherwise.
         """
         return self.kind == NodeKind.FILE
 
     def is_dir(self):
         """
         Returns ``True`` if node's kind is ``NodeKind.DIR``, ``False``
         otherwise.
         """
         return self.kind == NodeKind.DIR
 
     def is_root(self):
         """
         Returns ``True`` if node is a root node and ``False`` otherwise.
         """
         return self.kind == NodeKind.DIR and self.path == ''
 
     def is_submodule(self):
         """
         Returns ``True`` if node's kind is ``NodeKind.SUBMODULE``, ``False``
         otherwise.
         """
         return self.kind == NodeKind.SUBMODULE
 
     def is_largefile(self):
         """
         Returns ``True`` if node's kind is ``NodeKind.LARGEFILE``, ``False``
         otherwise
         """
         return self.kind == NodeKind.LARGEFILE
 
     def is_link(self):
         if self.commit:
             return self.commit.is_link(self.path)
         return False
 
     @LazyProperty
     def added(self):
         return self.state is NodeState.ADDED
 
     @LazyProperty
     def changed(self):
         return self.state is NodeState.CHANGED
 
     @LazyProperty
     def not_changed(self):
         return self.state is NodeState.NOT_CHANGED
 
     @LazyProperty
     def removed(self):
         return self.state is NodeState.REMOVED
 
 
 class FileNode(Node):
     """
     Class representing file nodes.
 
     :attribute: path: path to the node, relative to repository's root
     :attribute: content: if given arbitrary sets content of the file
     :attribute: commit: if given, first time content is accessed, callback
     :attribute: mode: stat mode for a node. Default is `FILEMODE_DEFAULT`.
     """
     _filter_pre_load = []
 
     def __init__(self, path, content=None, commit=None, mode=None, pre_load=None):
         """
         Only one of ``content`` and ``commit`` may be given. Passing both
         would raise ``NodeError`` exception.
 
         :param path: relative path to the node
         :param content: content may be passed to constructor
         :param commit: if given, will use it to lazily fetch content
         :param mode: ST_MODE (i.e. 0100644)
         """
         if content and commit:
             raise NodeError("Cannot use both content and commit")
         super(FileNode, self).__init__(path, kind=NodeKind.FILE)
         self.commit = commit
         self._content = content
         self._mode = mode or FILEMODE_DEFAULT
 
         self._set_bulk_properties(pre_load)
 
     def _set_bulk_properties(self, pre_load):
         if not pre_load:
             return
         pre_load = [entry for entry in pre_load
                     if entry not in self._filter_pre_load]
         if not pre_load:
             return
 
         for attr_name in pre_load:
             result = getattr(self, attr_name)
             if callable(result):
                 result = result()
             self.__dict__[attr_name] = result
 
     @LazyProperty
     def mode(self):
         """
         Returns lazily mode of the FileNode. If `commit` is not set, would
         use value given at initialization or `FILEMODE_DEFAULT` (default).
         """
         if self.commit:
             mode = self.commit.get_file_mode(self.path)
         else:
             mode = self._mode
         return mode
 
     @LazyProperty
     def raw_bytes(self):
         """
         Returns lazily the raw bytes of the FileNode.
         """
         if self.commit:
             if self._content is None:
                 self._content = self.commit.get_file_content(self.path)
             content = self._content
         else:
             content = self._content
         return content
 
+    def stream_bytes(self):
+        """
+        Returns an iterator that will stream the content of the file directly from
+        vcsserver without loading it to memory.
+        """
+        if self.commit:
+            return self.commit.get_file_content_streamed(self.path)
+        raise NodeError(
+            "Cannot retrieve message of the file without related "
+            "commit attribute")
+
     @LazyProperty
     def md5(self):
         """
         Returns md5 of the file node.
         """
         return md5(self.raw_bytes)
 
     def metadata_uncached(self):
         """
         Returns md5, binary flag of the file node, without any cache usage.
         """
 
         content = self.content_uncached()
 
         is_binary = content and '\0' in content
         size = 0
         if content:
             size = len(content)
 
         return is_binary, md5(content), size, content
 
     def content_uncached(self):
         """
         Returns lazily content of the FileNode. If possible, would try to
         decode content from UTF-8.
         """
         if self.commit:
             content = self.commit.get_file_content(self.path)
         else:
             content = self._content
         return content
 
     @LazyProperty
     def content(self):
         """
         Returns lazily content of the FileNode. If possible, would try to
         decode content from UTF-8.
         """
         content = self.raw_bytes
 
         if self.is_binary:
             return content
         return safe_unicode(content)
 
     @LazyProperty
     def size(self):
         if self.commit:
             return self.commit.get_file_size(self.path)
         raise NodeError(
             "Cannot retrieve size of the file without related "
             "commit attribute")
 
     @LazyProperty
     def message(self):
         if self.commit:
             return self.last_commit.message
         raise NodeError(
             "Cannot retrieve message of the file without related "
             "commit attribute")
 
     @LazyProperty
     def last_commit(self):
         if self.commit:
             pre_load = ["author", "date", "message", "parents"]
             return self.commit.get_path_commit(self.path, pre_load=pre_load)
         raise NodeError(
             "Cannot retrieve last commit of the file without "
             "related commit attribute")
 
     def get_mimetype(self):
         """
         Mimetype is calculated based on the file's content. If ``_mimetype``
         attribute is available, it will be returned (backends which store
         mimetypes or can easily recognize them, should set this private
         attribute to indicate that type should *NOT* be calculated).
         """
 
         if hasattr(self, '_mimetype'):
             if (isinstance(self._mimetype, (tuple, list,)) and
                     len(self._mimetype) == 2):
                 return self._mimetype
             else:
                 raise NodeError('given _mimetype attribute must be an 2 '
                                 'element list or tuple')
 
         db = get_mimetypes_db()
         mtype, encoding = db.guess_type(self.name)
 
         if mtype is None:
             if self.is_binary:
                 mtype = 'application/octet-stream'
                 encoding = None
             else:
                 mtype = 'text/plain'
                 encoding = None
 
                 # try with pygments
                 try:
                     from pygments.lexers import get_lexer_for_filename
                     mt = get_lexer_for_filename(self.name).mimetypes
                 except Exception:
                     mt = None
 
                 if mt:
                     mtype = mt[0]
 
         return mtype, encoding
 
     @LazyProperty
     def mimetype(self):
         """
         Wrapper around full mimetype info. It returns only type of fetched
         mimetype without the encoding part. use get_mimetype function to fetch
         full set of (type,encoding)
         """
         return self.get_mimetype()[0]
 
     @LazyProperty
     def mimetype_main(self):
         return self.mimetype.split('/')[0]
 
     @classmethod
     def get_lexer(cls, filename, content=None):
         from pygments import lexers
 
         extension = filename.split('.')[-1]
         lexer = None
 
         try:
             lexer = lexers.guess_lexer_for_filename(
                 filename, content, stripnl=False)
         except lexers.ClassNotFound:
             lexer = None
 
         # try our EXTENSION_MAP
         if not lexer:
             try:
                 lexer_class = LANGUAGES_EXTENSIONS_MAP.get(extension)
                 if lexer_class:
                     lexer = lexers.get_lexer_by_name(lexer_class[0])
             except lexers.ClassNotFound:
                 lexer = None
 
         if not lexer:
             lexer = lexers.TextLexer(stripnl=False)
 
         return lexer
 
     @LazyProperty
     def lexer(self):
         """
         Returns pygment's lexer class. Would try to guess lexer taking file's
         content, name and mimetype.
         """
         return self.get_lexer(self.name, self.content)
 
     @LazyProperty
     def lexer_alias(self):
         """
         Returns first alias of the lexer guessed for this file.
         """
         return self.lexer.aliases[0]
 
     @LazyProperty
     def history(self):
         """
         Returns a list of commit for this file in which the file was changed
         """
         if self.commit is None:
             raise NodeError('Unable to get commit for this FileNode')
         return self.commit.get_path_history(self.path)
 
     @LazyProperty
     def annotate(self):
         """
         Returns a list of three element tuples with lineno, commit and line
         """
         if self.commit is None:
             raise NodeError('Unable to get commit for this FileNode')
         pre_load = ["author", "date", "message", "parents"]
         return self.commit.get_file_annotate(self.path, pre_load=pre_load)
 
     @LazyProperty
     def state(self):
         if not self.commit:
             raise NodeError(
                 "Cannot check state of the node if it's not "
                 "linked with commit")
         elif self.path in (node.path for node in self.commit.added):
             return NodeState.ADDED
         elif self.path in (node.path for node in self.commit.changed):
             return NodeState.CHANGED
         else:
             return NodeState.NOT_CHANGED
 
     @LazyProperty
     def is_binary(self):
         """
         Returns True if file has binary content.
         """
         _bin = self.raw_bytes and '\0' in self.raw_bytes
         return _bin
 
     @LazyProperty
     def extension(self):
         """Returns filenode extension"""
         return self.name.split('.')[-1]
 
     @property
     def is_executable(self):
         """
         Returns ``True`` if file has executable flag turned on.
         """
         return bool(self.mode & stat.S_IXUSR)
 
     def get_largefile_node(self):
         """
         Try to return a Mercurial FileNode from this node. It does internal
         checks inside largefile store, if that file exist there it will
         create special instance of LargeFileNode which can get content from
         LF store.
         """
         if self.commit:
             return self.commit.get_largefile_node(self.path)
 
     def lines(self, count_empty=False):
         all_lines, empty_lines = 0, 0
 
         if not self.is_binary:
             content = self.content
             if count_empty:
                 all_lines = 0
                 empty_lines = 0
                 for line in content.splitlines(True):
                     if line == '\n':
                         empty_lines += 1
                     all_lines += 1
 
                 return all_lines, all_lines - empty_lines
             else:
                 # fast method
                 empty_lines = all_lines = content.count('\n')
                 if all_lines == 0 and content:
                     # one-line without a newline
                     empty_lines = all_lines = 1
 
         return all_lines, empty_lines
 
     def __repr__(self):
         return '<%s %r @ %s>' % (self.__class__.__name__, self.path,
                                  getattr(self.commit, 'short_id', ''))
 
 
 class RemovedFileNode(FileNode):
     """
     Dummy FileNode class - trying to access any public attribute except path,
     name, kind or state (or methods/attributes checking those two) would raise
     RemovedFileNodeError.
     """
     ALLOWED_ATTRIBUTES = [
         'name', 'path', 'state', 'is_root', 'is_file', 'is_dir', 'kind',
         'added', 'changed', 'not_changed', 'removed'
     ]
 
     def __init__(self, path):
         """
         :param path: relative path to the node
         """
         super(RemovedFileNode, self).__init__(path=path)
 
     def __getattribute__(self, attr):
         if attr.startswith('_') or attr in RemovedFileNode.ALLOWED_ATTRIBUTES:
             return super(RemovedFileNode, self).__getattribute__(attr)
         raise RemovedFileNodeError(
             "Cannot access attribute %s on RemovedFileNode" % attr)
 
     @LazyProperty
     def state(self):
         return NodeState.REMOVED
 
 
 class DirNode(Node):
     """
     DirNode stores list of files and directories within this node.
     Nodes may be used standalone but within repository context they
     lazily fetch data within same repositorty's commit.
     """
 
     def __init__(self, path, nodes=(), commit=None):
         """
         Only one of ``nodes`` and ``commit`` may be given. Passing both
         would raise ``NodeError`` exception.
 
         :param path: relative path to the node
         :param nodes: content may be passed to constructor
         :param commit: if given, will use it to lazily fetch content
         """
         if nodes and commit:
             raise NodeError("Cannot use both nodes and commit")
         super(DirNode, self).__init__(path, NodeKind.DIR)
         self.commit = commit
         self._nodes = nodes
 
     @LazyProperty
     def content(self):
         raise NodeError(
             "%s represents a dir and has no `content` attribute" % self)
 
     @LazyProperty
     def nodes(self):
         if self.commit:
             nodes = self.commit.get_nodes(self.path)
         else:
             nodes = self._nodes
         self._nodes_dict = dict((node.path, node) for node in nodes)
         return sorted(nodes)
 
     @LazyProperty
     def files(self):
         return sorted((node for node in self.nodes if node.is_file()))
 
     @LazyProperty
     def dirs(self):
         return sorted((node for node in self.nodes if node.is_dir()))
 
     def __iter__(self):
         for node in self.nodes:
             yield node
 
     def get_node(self, path):
         """
         Returns node from within this particular ``DirNode``, so it is now
         allowed to fetch, i.e. node located at 'docs/api/index.rst' from node
         'docs'. In order to access deeper nodes one must fetch nodes between
         them first - this would work::
 
            docs = root.get_node('docs')
            docs.get_node('api').get_node('index.rst')
 
         :param: path - relative to the current node
 
         .. note::
            To access lazily (as in example above) node have to be initialized
            with related commit object - without it node is out of
            context and may know nothing about anything else than nearest
            (located at same level) nodes.
         """
         try:
             path = path.rstrip('/')
             if path == '':
                 raise NodeError("Cannot retrieve node without path")
             self.nodes  # access nodes first in order to set _nodes_dict
             paths = path.split('/')
             if len(paths) == 1:
                 if not self.is_root():
                     path = '/'.join((self.path, paths[0]))
                 else:
                     path = paths[0]
                 return self._nodes_dict[path]
             elif len(paths) > 1:
                 if self.commit is None:
                     raise NodeError(
                         "Cannot access deeper nodes without commit")
                 else:
                     path1, path2 = paths[0], '/'.join(paths[1:])
                     return self.get_node(path1).get_node(path2)
             else:
                 raise KeyError
         except KeyError:
             raise NodeError("Node does not exist at %s" % path)
 
     @LazyProperty
     def state(self):
         raise NodeError("Cannot access state of DirNode")
 
     @LazyProperty
     def size(self):
         size = 0
         for root, dirs, files in self.commit.walk(self.path):
             for f in files:
                 size += f.size
 
         return size
 
     @LazyProperty
     def last_commit(self):
         if self.commit:
             pre_load = ["author", "date", "message", "parents"]
             return self.commit.get_path_commit(self.path, pre_load=pre_load)
         raise NodeError(
             "Cannot retrieve last commit of the file without "
             "related commit attribute")
 
     def __repr__(self):
         return '<%s %r @ %s>' % (self.__class__.__name__, self.path,
                                  getattr(self.commit, 'short_id', ''))
 
 
 class RootNode(DirNode):
     """
     DirNode being the root node of the repository.
     """
 
     def __init__(self, nodes=(), commit=None):
         super(RootNode, self).__init__(path='', nodes=nodes, commit=commit)
 
     def __repr__(self):
         return '<%s>' % self.__class__.__name__
 
 
 class SubModuleNode(Node):
     """
     represents a SubModule of Git or SubRepo of Mercurial
     """
     is_binary = False
     size = 0
 
     def __init__(self, name, url=None, commit=None, alias=None):
         self.path = name
         self.kind = NodeKind.SUBMODULE
         self.alias = alias
 
         # we have to use EmptyCommit here since this can point to svn/git/hg
         # submodules we cannot get from repository
         self.commit = EmptyCommit(str(commit), alias=alias)
         self.url = url or self._extract_submodule_url()
 
     def __repr__(self):
         return '<%s %r @ %s>' % (self.__class__.__name__, self.path,
                                  getattr(self.commit, 'short_id', ''))
 
     def _extract_submodule_url(self):
         # TODO: find a way to parse gits submodule file and extract the
         # linking URL
         return self.path
 
     @LazyProperty
     def name(self):
         """
         Returns name of the node so if its path
         then only last part is returned.
         """
         org = safe_unicode(self.path.rstrip('/').split('/')[-1])
         return u'%s @ %s' % (org, self.commit.short_id)
 
 
 class LargeFileNode(FileNode):
 
     def __init__(self, path, url=None, commit=None, alias=None, org_path=None):
         self.path = path
         self.org_path = org_path
         self.kind = NodeKind.LARGEFILE
         self.alias = alias
 
     def _validate_path(self, path):
         """
         we override check since the LargeFileNode path is system absolute
         """
         pass
 
     def __repr__(self):
         return '<%s %r>' % (self.__class__.__name__, self.path)
 
     @LazyProperty
     def size(self):
         return os.stat(self.path).st_size
 
     @LazyProperty
     def raw_bytes(self):
         with open(self.path, 'rb') as f:
             content = f.read()
         return content
 
     @LazyProperty
     def name(self):
         """
         Overwrites name to be the org lf path
         """
         return self.org_path
+
+    def stream_bytes(self):
+        with open(self.path, 'rb') as stream:
+            while True:
+                data = stream.read(16 * 1024)
+                if not data:
+                    break
+                yield data

rhodecode/tests/plugin.py

@@ -1,1902 +1,1826 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import collections
 import datetime
 import hashlib
 import os
 import re
 import pprint
 import shutil
 import socket
 import subprocess32
 import time
 import uuid
 import dateutil.tz
 import functools
 
 import mock
 import pyramid.testing
 import pytest
 import colander
 import requests
 import pyramid.paster
 
 import rhodecode
 from rhodecode.lib.utils2 import AttributeDict
 from rhodecode.model.changeset_status import ChangesetStatusModel
 from rhodecode.model.comment import CommentsModel
 from rhodecode.model.db import (
     PullRequest, Repository, RhodeCodeSetting, ChangesetStatus, RepoGroup,
     UserGroup, RepoRhodeCodeUi, RepoRhodeCodeSetting, RhodeCodeUi)
 from rhodecode.model.meta import Session
 from rhodecode.model.pull_request import PullRequestModel
 from rhodecode.model.repo import RepoModel
 from rhodecode.model.repo_group import RepoGroupModel
 from rhodecode.model.user import UserModel
 from rhodecode.model.settings import VcsSettingsModel
 from rhodecode.model.user_group import UserGroupModel
 from rhodecode.model.integration import IntegrationModel
 from rhodecode.integrations import integration_type_registry
 from rhodecode.integrations.types.base import IntegrationTypeBase
 from rhodecode.lib.utils import repo2db_mapper
-from rhodecode.lib.vcs import create_vcsserver_proxy
 from rhodecode.lib.vcs.backends import get_backend
 from rhodecode.lib.vcs.nodes import FileNode
 from rhodecode.tests import (
     login_user_session, get_new_dir, utils, TESTS_TMP_PATH,
     TEST_USER_ADMIN_LOGIN, TEST_USER_REGULAR_LOGIN, TEST_USER_REGULAR2_LOGIN,
     TEST_USER_REGULAR_PASS)
 from rhodecode.tests.utils import CustomTestApp, set_anonymous_access
 from rhodecode.tests.fixture import Fixture
 from rhodecode.config import utils as config_utils
 
 def _split_comma(value):
     return value.split(',')
 
 
 def pytest_addoption(parser):
     parser.addoption(
         '--keep-tmp-path', action='store_true',
         help="Keep the test temporary directories")
     parser.addoption(
         '--backends', action='store', type=_split_comma,
         default=['git', 'hg', 'svn'],
         help="Select which backends to test for backend specific tests.")
     parser.addoption(
         '--dbs', action='store', type=_split_comma,
         default=['sqlite'],
         help="Select which database to test for database specific tests. "
              "Possible options are sqlite,postgres,mysql")
     parser.addoption(
         '--appenlight', '--ae', action='store_true',
         help="Track statistics in appenlight.")
     parser.addoption(
         '--appenlight-api-key', '--ae-key',
         help="API key for Appenlight.")
     parser.addoption(
         '--appenlight-url', '--ae-url',
         default="https://ae.rhodecode.com",
         help="Appenlight service URL, defaults to https://ae.rhodecode.com")
     parser.addoption(
         '--sqlite-connection-string', action='store',
         default='', help="Connection string for the dbs tests with SQLite")
     parser.addoption(
         '--postgres-connection-string', action='store',
         default='', help="Connection string for the dbs tests with Postgres")
     parser.addoption(
         '--mysql-connection-string', action='store',
         default='', help="Connection string for the dbs tests with MySQL")
     parser.addoption(
         '--repeat', type=int, default=100,
         help="Number of repetitions in performance tests.")
 
 
 def pytest_configure(config):
     from rhodecode.config import patches
 
 
 def pytest_collection_modifyitems(session, config, items):
     # nottest marked, compare nose, used for transition from nose to pytest
     remaining = [
         i for i in items if getattr(i.obj, '__test__', True)]
     items[:] = remaining
 
 
 def pytest_generate_tests(metafunc):
     # Support test generation based on --backend parameter
     if 'backend_alias' in metafunc.fixturenames:
         backends = get_backends_from_metafunc(metafunc)
         scope = None
         if not backends:
             pytest.skip("Not enabled for any of selected backends")
         metafunc.parametrize('backend_alias', backends, scope=scope)
     elif hasattr(metafunc.function, 'backends'):
         backends = get_backends_from_metafunc(metafunc)
         if not backends:
             pytest.skip("Not enabled for any of selected backends")
 
 
 def get_backends_from_metafunc(metafunc):
     requested_backends = set(metafunc.config.getoption('--backends'))
     if hasattr(metafunc.function, 'backends'):
         # Supported backends by this test function, created from
         # pytest.mark.backends
         backends = metafunc.definition.get_closest_marker('backends').args
     elif hasattr(metafunc.cls, 'backend_alias'):
         # Support class attribute "backend_alias", this is mainly
         # for legacy reasons for tests not yet using pytest.mark.backends
         backends = [metafunc.cls.backend_alias]
     else:
         backends = metafunc.config.getoption('--backends')
     return requested_backends.intersection(backends)
 
 
 @pytest.fixture(scope='session', autouse=True)
 def activate_example_rcextensions(request):
     """
     Patch in an example rcextensions module which verifies passed in kwargs.
     """
     from rhodecode.config import rcextensions
 
     old_extensions = rhodecode.EXTENSIONS
     rhodecode.EXTENSIONS = rcextensions
     rhodecode.EXTENSIONS.calls = collections.defaultdict(list)
 
     @request.addfinalizer
     def cleanup():
         rhodecode.EXTENSIONS = old_extensions
 
 
 @pytest.fixture
 def capture_rcextensions():
     """
     Returns the recorded calls to entry points in rcextensions.
     """
     calls = rhodecode.EXTENSIONS.calls
     calls.clear()
     # Note: At this moment, it is still the empty dict, but that will
     # be filled during the test run and since it is a reference this
     # is enough to make it work.
     return calls
 
 
 @pytest.fixture(scope='session')
 def http_environ_session():
     """
     Allow to use "http_environ" in session scope.
     """
     return plain_http_environ()
 
 
 def plain_http_host_stub():
     """
     Value of HTTP_HOST in the test run.
     """
     return 'example.com:80'
 
 
 @pytest.fixture
 def http_host_stub():
     """
     Value of HTTP_HOST in the test run.
     """
     return plain_http_host_stub()
 
 
 def plain_http_host_only_stub():
     """
     Value of HTTP_HOST in the test run.
     """
     return plain_http_host_stub().split(':')[0]
 
 
 @pytest.fixture
 def http_host_only_stub():
     """
     Value of HTTP_HOST in the test run.
     """
     return plain_http_host_only_stub()
 
 
 def plain_http_environ():
     """
     HTTP extra environ keys.
 
     User by the test application and as well for setting up the pylons
     environment. In the case of the fixture "app" it should be possible
     to override this for a specific test case.
     """
     return {
         'SERVER_NAME': plain_http_host_only_stub(),
         'SERVER_PORT': plain_http_host_stub().split(':')[1],
         'HTTP_HOST': plain_http_host_stub(),
         'HTTP_USER_AGENT': 'rc-test-agent',
         'REQUEST_METHOD': 'GET'
     }
 
 
 @pytest.fixture
 def http_environ():
     """
     HTTP extra environ keys.
 
     User by the test application and as well for setting up the pylons
     environment. In the case of the fixture "app" it should be possible
     to override this for a specific test case.
     """
     return plain_http_environ()
 
 
 @pytest.fixture(scope='session')
 def baseapp(ini_config, vcsserver, http_environ_session):
     from rhodecode.lib.pyramid_utils import get_app_config
     from rhodecode.config.middleware import make_pyramid_app
 
     print("Using the RhodeCode configuration:{}".format(ini_config))
     pyramid.paster.setup_logging(ini_config)
 
     settings = get_app_config(ini_config)
     app = make_pyramid_app({'__file__': ini_config}, **settings)
 
     return app
 
 
 @pytest.fixture(scope='function')
 def app(request, config_stub, baseapp, http_environ):
     app = CustomTestApp(
         baseapp,
         extra_environ=http_environ)
     if request.cls:
         request.cls.app = app
     return app
 
 
 @pytest.fixture(scope='session')
 def app_settings(baseapp, ini_config):
     """
     Settings dictionary used to create the app.
 
     Parses the ini file and passes the result through the sanitize and apply
     defaults mechanism in `rhodecode.config.middleware`.
     """
     return baseapp.config.get_settings()
 
 
 @pytest.fixture(scope='session')
 def db_connection(ini_settings):
     # Initialize the database connection.
     config_utils.initialize_database(ini_settings)
 
 
 LoginData = collections.namedtuple('LoginData', ('csrf_token', 'user'))
 
 
 def _autologin_user(app, *args):
     session = login_user_session(app, *args)
     csrf_token = rhodecode.lib.auth.get_csrf_token(session)
     return LoginData(csrf_token, session['rhodecode_user'])
 
 
 @pytest.fixture
 def autologin_user(app):
     """
     Utility fixture which makes sure that the admin user is logged in
     """
     return _autologin_user(app)
 
 
 @pytest.fixture
 def autologin_regular_user(app):
     """
     Utility fixture which makes sure that the regular user is logged in
     """
     return _autologin_user(
         app, TEST_USER_REGULAR_LOGIN, TEST_USER_REGULAR_PASS)
 
 
 @pytest.fixture(scope='function')
 def csrf_token(request, autologin_user):
     return autologin_user.csrf_token
 
 
 @pytest.fixture(scope='function')
 def xhr_header(request):
     return {'HTTP_X_REQUESTED_WITH': 'XMLHttpRequest'}
 
 
 @pytest.fixture
 def real_crypto_backend(monkeypatch):
     """
     Switch the production crypto backend on for this test.
 
     During the test run the crypto backend is replaced with a faster
     implementation based on the MD5 algorithm.
     """
     monkeypatch.setattr(rhodecode, 'is_test', False)
 
 
 @pytest.fixture(scope='class')
 def index_location(request, baseapp):
     index_location = baseapp.config.get_settings()['search.location']
     if request.cls:
         request.cls.index_location = index_location
     return index_location
 
 
 @pytest.fixture(scope='session', autouse=True)
 def tests_tmp_path(request):
     """
     Create temporary directory to be used during the test session.
     """
     if not os.path.exists(TESTS_TMP_PATH):
         os.makedirs(TESTS_TMP_PATH)
 
     if not request.config.getoption('--keep-tmp-path'):
         @request.addfinalizer
         def remove_tmp_path():
             shutil.rmtree(TESTS_TMP_PATH)
 
     return TESTS_TMP_PATH
 
 
 @pytest.fixture
 def test_repo_group(request):
     """
     Create a temporary repository group, and destroy it after
     usage automatically
     """
     fixture = Fixture()
     repogroupid = 'test_repo_group_%s' % str(time.time()).replace('.', '')
     repo_group = fixture.create_repo_group(repogroupid)
 
     def _cleanup():
         fixture.destroy_repo_group(repogroupid)
 
     request.addfinalizer(_cleanup)
     return repo_group
 
 
 @pytest.fixture
 def test_user_group(request):
     """
     Create a temporary user group, and destroy it after
     usage automatically
     """
     fixture = Fixture()
     usergroupid = 'test_user_group_%s' % str(time.time()).replace('.', '')
     user_group = fixture.create_user_group(usergroupid)
 
     def _cleanup():
         fixture.destroy_user_group(user_group)
 
     request.addfinalizer(_cleanup)
     return user_group
 
 
 @pytest.fixture(scope='session')
 def test_repo(request):
     container = TestRepoContainer()
     request.addfinalizer(container._cleanup)
     return container
 
 
 class TestRepoContainer(object):
     """
     Container for test repositories which are used read only.
 
     Repositories will be created on demand and re-used during the lifetime
     of this object.
 
     Usage to get the svn test repository "minimal"::
 
        test_repo = TestContainer()
        repo = test_repo('minimal', 'svn')
 
     """
 
     dump_extractors = {
         'git': utils.extract_git_repo_from_dump,
         'hg': utils.extract_hg_repo_from_dump,
         'svn': utils.extract_svn_repo_from_dump,
     }
 
     def __init__(self):
         self._cleanup_repos = []
         self._fixture = Fixture()
         self._repos = {}
 
     def __call__(self, dump_name, backend_alias, config=None):
         key = (dump_name, backend_alias)
         if key not in self._repos:
             repo = self._create_repo(dump_name, backend_alias, config)
             self._repos[key] = repo.repo_id
         return Repository.get(self._repos[key])
 
     def _create_repo(self, dump_name, backend_alias, config):
         repo_name = '%s-%s' % (backend_alias, dump_name)
         backend = get_backend(backend_alias)
         dump_extractor = self.dump_extractors[backend_alias]
         repo_path = dump_extractor(dump_name, repo_name)
 
         vcs_repo = backend(repo_path, config=config)
         repo2db_mapper({repo_name: vcs_repo})
 
         repo = RepoModel().get_by_repo_name(repo_name)
         self._cleanup_repos.append(repo_name)
         return repo
 
     def _cleanup(self):
         for repo_name in reversed(self._cleanup_repos):
             self._fixture.destroy_repo(repo_name)
 
 
 def backend_base(request, backend_alias, baseapp, test_repo):
     if backend_alias not in request.config.getoption('--backends'):
         pytest.skip("Backend %s not selected." % (backend_alias, ))
 
     utils.check_xfail_backends(request.node, backend_alias)
     utils.check_skip_backends(request.node, backend_alias)
 
     repo_name = 'vcs_test_%s' % (backend_alias, )
     backend = Backend(
         alias=backend_alias,
         repo_name=repo_name,
         test_name=request.node.name,
         test_repo_container=test_repo)
     request.addfinalizer(backend.cleanup)
     return backend
 
 
 @pytest.fixture
 def backend(request, backend_alias, baseapp, test_repo):
     """
     Parametrized fixture which represents a single backend implementation.
 
     It respects the option `--backends` to focus the test run on specific
     backend implementations.
 
     It also supports `pytest.mark.xfail_backends` to mark tests as failing
     for specific backends. This is intended as a utility for incremental
     development of a new backend implementation.
     """
     return backend_base(request, backend_alias, baseapp, test_repo)
 
 
 @pytest.fixture
 def backend_git(request, baseapp, test_repo):
     return backend_base(request, 'git', baseapp, test_repo)
 
 
 @pytest.fixture
 def backend_hg(request, baseapp, test_repo):
     return backend_base(request, 'hg', baseapp, test_repo)
 
 
 @pytest.fixture
 def backend_svn(request, baseapp, test_repo):
     return backend_base(request, 'svn', baseapp, test_repo)
 
 
 @pytest.fixture
 def backend_random(backend_git):
     """
     Use this to express that your tests need "a backend.
 
     A few of our tests need a backend, so that we can run the code. This
     fixture is intended to be used for such cases. It will pick one of the
     backends and run the tests.
 
     The fixture `backend` would run the test multiple times for each
     available backend which is a pure waste of time if the test is
     independent of the backend type.
     """
     # TODO: johbo: Change this to pick a random backend
     return backend_git
 
 
 @pytest.fixture
 def backend_stub(backend_git):
     """
     Use this to express that your tests need a backend stub
 
     TODO: mikhail: Implement a real stub logic instead of returning
     a git backend
     """
     return backend_git
 
 
 @pytest.fixture
 def repo_stub(backend_stub):
     """
     Use this to express that your tests need a repository stub
     """
     return backend_stub.create_repo()
 
 
 class Backend(object):
     """
     Represents the test configuration for one supported backend
 
     Provides easy access to different test repositories based on
     `__getitem__`. Such repositories will only be created once per test
     session.
     """
 
     invalid_repo_name = re.compile(r'[^0-9a-zA-Z]+')
     _master_repo = None
     _commit_ids = {}
 
     def __init__(self, alias, repo_name, test_name, test_repo_container):
         self.alias = alias
         self.repo_name = repo_name
         self._cleanup_repos = []
         self._test_name = test_name
         self._test_repo_container = test_repo_container
         # TODO: johbo: Used as a delegate interim. Not yet sure if Backend or
         # Fixture will survive in the end.
         self._fixture = Fixture()
 
     def __getitem__(self, key):
         return self._test_repo_container(key, self.alias)
 
     def create_test_repo(self, key, config=None):
         return self._test_repo_container(key, self.alias, config)
 
     @property
     def repo(self):
         """
         Returns the "current" repository. This is the vcs_test repo or the
         last repo which has been created with `create_repo`.
         """
         from rhodecode.model.db import Repository
         return Repository.get_by_repo_name(self.repo_name)
 
     @property
     def default_branch_name(self):
         VcsRepository = get_backend(self.alias)
         return VcsRepository.DEFAULT_BRANCH_NAME
 
     @property
     def default_head_id(self):
         """
         Returns the default head id of the underlying backend.
 
         This will be the default branch name in case the backend does have a
         default branch. In the other cases it will point to a valid head
         which can serve as the base to create a new commit on top of it.
         """
         vcsrepo = self.repo.scm_instance()
         head_id = (
             vcsrepo.DEFAULT_BRANCH_NAME or
             vcsrepo.commit_ids[-1])
         return head_id
 
     @property
     def commit_ids(self):
         """
         Returns the list of commits for the last created repository
         """
         return self._commit_ids
 
     def create_master_repo(self, commits):
         """
         Create a repository and remember it as a template.
 
         This allows to easily create derived repositories to construct
         more complex scenarios for diff, compare and pull requests.
 
         Returns a commit map which maps from commit message to raw_id.
         """
         self._master_repo = self.create_repo(commits=commits)
         return self._commit_ids
 
     def create_repo(
             self, commits=None, number_of_commits=0, heads=None,
             name_suffix=u'', bare=False, **kwargs):
         """
         Create a repository and record it for later cleanup.
 
         :param commits: Optional. A sequence of dict instances.
            Will add a commit per entry to the new repository.
         :param number_of_commits: Optional. If set to a number, this number of
            commits will be added to the new repository.
         :param heads: Optional. Can be set to a sequence of of commit
            names which shall be pulled in from the master repository.
         :param name_suffix: adds special suffix to generated repo name
         :param bare: set a repo as bare (no checkout)
         """
         self.repo_name = self._next_repo_name() + name_suffix
         repo = self._fixture.create_repo(
             self.repo_name, repo_type=self.alias, bare=bare, **kwargs)
         self._cleanup_repos.append(repo.repo_name)
 
         commits = commits or [
             {'message': 'Commit %s of %s' % (x, self.repo_name)}
             for x in range(number_of_commits)]
         vcs_repo = repo.scm_instance()
         vcs_repo.count()
         self._add_commits_to_repo(vcs_repo, commits)
         if heads:
             self.pull_heads(repo, heads)
 
         return repo
 
     def pull_heads(self, repo, heads):
         """
         Make sure that repo contains all commits mentioned in `heads`
         """
         vcsmaster = self._master_repo.scm_instance()
         vcsrepo = repo.scm_instance()
         vcsrepo.config.clear_section('hooks')
         commit_ids = [self._commit_ids[h] for h in heads]
         vcsrepo.pull(vcsmaster.path, commit_ids=commit_ids)
 
     def create_fork(self):
         repo_to_fork = self.repo_name
         self.repo_name = self._next_repo_name()
         repo = self._fixture.create_fork(repo_to_fork, self.repo_name)
         self._cleanup_repos.append(self.repo_name)
         return repo
 
     def new_repo_name(self, suffix=u''):
         self.repo_name = self._next_repo_name() + suffix
         self._cleanup_repos.append(self.repo_name)
         return self.repo_name
 
     def _next_repo_name(self):
         return u"%s_%s" % (
             self.invalid_repo_name.sub(u'_', self._test_name), len(self._cleanup_repos))
 
     def ensure_file(self, filename, content='Test content\n'):
         assert self._cleanup_repos, "Avoid writing into vcs_test repos"
         commits = [
             {'added': [
                 FileNode(filename, content=content),
             ]},
         ]
         self._add_commits_to_repo(self.repo.scm_instance(), commits)
 
     def enable_downloads(self):
         repo = self.repo
         repo.enable_downloads = True
         Session().add(repo)
         Session().commit()
 
     def cleanup(self):
         for repo_name in reversed(self._cleanup_repos):
             self._fixture.destroy_repo(repo_name)
 
     def _add_commits_to_repo(self, repo, commits):
         commit_ids = _add_commits_to_repo(repo, commits)
         if not commit_ids:
             return
         self._commit_ids = commit_ids
 
         # Creating refs for Git to allow fetching them from remote repository
         if self.alias == 'git':
             refs = {}
             for message in self._commit_ids:
                 # TODO: mikhail: do more special chars replacements
                 ref_name = 'refs/test-refs/{}'.format(
                     message.replace(' ', ''))
                 refs[ref_name] = self._commit_ids[message]
             self._create_refs(repo, refs)
 
     def _create_refs(self, repo, refs):
         for ref_name in refs:
             repo.set_refs(ref_name, refs[ref_name])
 
 
 def vcsbackend_base(request, backend_alias, tests_tmp_path, baseapp, test_repo):
     if backend_alias not in request.config.getoption('--backends'):
         pytest.skip("Backend %s not selected." % (backend_alias, ))
 
     utils.check_xfail_backends(request.node, backend_alias)
     utils.check_skip_backends(request.node, backend_alias)
 
     repo_name = 'vcs_test_%s' % (backend_alias, )
     repo_path = os.path.join(tests_tmp_path, repo_name)
     backend = VcsBackend(
         alias=backend_alias,
         repo_path=repo_path,
         test_name=request.node.name,
         test_repo_container=test_repo)
     request.addfinalizer(backend.cleanup)
     return backend
 
 
 @pytest.fixture
 def vcsbackend(request, backend_alias, tests_tmp_path, baseapp, test_repo):
     """
     Parametrized fixture which represents a single vcs backend implementation.
 
     See the fixture `backend` for more details. This one implements the same
     concept, but on vcs level. So it does not provide model instances etc.
 
     Parameters are generated dynamically, see :func:`pytest_generate_tests`
     for how this works.
     """
     return vcsbackend_base(request, backend_alias, tests_tmp_path, baseapp, test_repo)
 
 
 @pytest.fixture
 def vcsbackend_git(request, tests_tmp_path, baseapp, test_repo):
     return vcsbackend_base(request, 'git', tests_tmp_path, baseapp, test_repo)
 
 
 @pytest.fixture
 def vcsbackend_hg(request, tests_tmp_path, baseapp, test_repo):
     return vcsbackend_base(request, 'hg', tests_tmp_path, baseapp, test_repo)
 
 
 @pytest.fixture
 def vcsbackend_svn(request, tests_tmp_path, baseapp, test_repo):
     return vcsbackend_base(request, 'svn', tests_tmp_path, baseapp, test_repo)
 
 
 @pytest.fixture
 def vcsbackend_stub(vcsbackend_git):
     """
     Use this to express that your test just needs a stub of a vcsbackend.
 
     Plan is to eventually implement an in-memory stub to speed tests up.
     """
     return vcsbackend_git
 
 
 class VcsBackend(object):
     """
     Represents the test configuration for one supported vcs backend.
     """
 
     invalid_repo_name = re.compile(r'[^0-9a-zA-Z]+')
 
     def __init__(self, alias, repo_path, test_name, test_repo_container):
         self.alias = alias
         self._repo_path = repo_path
         self._cleanup_repos = []
         self._test_name = test_name
         self._test_repo_container = test_repo_container
 
     def __getitem__(self, key):
         return self._test_repo_container(key, self.alias).scm_instance()
 
     @property
     def repo(self):
         """
         Returns the "current" repository. This is the vcs_test repo of the last
         repo which has been created.
         """
         Repository = get_backend(self.alias)
         return Repository(self._repo_path)
 
     @property
     def backend(self):
         """
         Returns the backend implementation class.
         """
         return get_backend(self.alias)
 
     def create_repo(self, commits=None, number_of_commits=0, _clone_repo=None,
                     bare=False):
         repo_name = self._next_repo_name()
         self._repo_path = get_new_dir(repo_name)
         repo_class = get_backend(self.alias)
         src_url = None
         if _clone_repo:
             src_url = _clone_repo.path
         repo = repo_class(self._repo_path, create=True, src_url=src_url, bare=bare)
         self._cleanup_repos.append(repo)
 
         commits = commits or [
             {'message': 'Commit %s of %s' % (x, repo_name)}
             for x in xrange(number_of_commits)]
         _add_commits_to_repo(repo, commits)
         return repo
 
     def clone_repo(self, repo):
         return self.create_repo(_clone_repo=repo)
 
     def cleanup(self):
         for repo in self._cleanup_repos:
             shutil.rmtree(repo.path)
 
     def new_repo_path(self):
         repo_name = self._next_repo_name()
         self._repo_path = get_new_dir(repo_name)
         return self._repo_path
 
     def _next_repo_name(self):
         return "%s_%s" % (
             self.invalid_repo_name.sub('_', self._test_name),
             len(self._cleanup_repos))
 
     def add_file(self, repo, filename, content='Test content\n'):
         imc = repo.in_memory_commit
         imc.add(FileNode(filename, content=content))
         imc.commit(
             message=u'Automatic commit from vcsbackend fixture',
             author=u'Automatic <[e-mail]>')
 
     def ensure_file(self, filename, content='Test content\n'):
         assert self._cleanup_repos, "Avoid writing into vcs_test repos"
         self.add_file(self.repo, filename, content)
 
 
 def _add_commits_to_repo(vcs_repo, commits):
     commit_ids = {}
     if not commits:
         return commit_ids
 
     imc = vcs_repo.in_memory_commit
     commit = None
 
     for idx, commit in enumerate(commits):
         message = unicode(commit.get('message', 'Commit %s' % idx))
 
         for node in commit.get('added', []):
             imc.add(FileNode(node.path, content=node.content))
         for node in commit.get('changed', []):
             imc.change(FileNode(node.path, content=node.content))
         for node in commit.get('removed', []):
             imc.remove(FileNode(node.path))
 
         parents = [
             vcs_repo.get_commit(commit_id=commit_ids[p])
             for p in commit.get('parents', [])]
 
         operations = ('added', 'changed', 'removed')
         if not any((commit.get(o) for o in operations)):
             imc.add(FileNode('file_%s' % idx, content=message))
 
         commit = imc.commit(
             message=message,
             author=unicode(commit.get('author', 'Automatic <[e-mail]>')),
             date=commit.get('date'),
             branch=commit.get('branch'),
             parents=parents)
 
         commit_ids[commit.message] = commit.raw_id
 
     return commit_ids
 
 
 @pytest.fixture
 def reposerver(request):
     """
     Allows to serve a backend repository
     """
 
     repo_server = RepoServer()
     request.addfinalizer(repo_server.cleanup)
     return repo_server
 
 
 class RepoServer(object):
     """
     Utility to serve a local repository for the duration of a test case.
 
     Supports only Subversion so far.
     """
 
     url = None
 
     def __init__(self):
         self._cleanup_servers = []
 
     def serve(self, vcsrepo):
         if vcsrepo.alias != 'svn':
             raise TypeError("Backend %s not supported" % vcsrepo.alias)
 
         proc = subprocess32.Popen(
             ['svnserve', '-d', '--foreground', '--listen-host', 'localhost',
              '--root', vcsrepo.path])
         self._cleanup_servers.append(proc)
         self.url = 'svn://localhost'
 
     def cleanup(self):
         for proc in self._cleanup_servers:
             proc.terminate()
 
 
 @pytest.fixture
 def pr_util(backend, request, config_stub):
     """
     Utility for tests of models and for functional tests around pull requests.
 
     It gives an instance of :class:`PRTestUtility` which provides various
     utility methods around one pull request.
 
     This fixture uses `backend` and inherits its parameterization.
     """
 
     util = PRTestUtility(backend)
     request.addfinalizer(util.cleanup)
 
     return util
 
 
 class PRTestUtility(object):
 
     pull_request = None
     pull_request_id = None
     mergeable_patcher = None
     mergeable_mock = None
     notification_patcher = None
 
     def __init__(self, backend):
         self.backend = backend
 
     def create_pull_request(
             self, commits=None, target_head=None, source_head=None,
             revisions=None, approved=False, author=None, mergeable=False,
             enable_notifications=True, name_suffix=u'', reviewers=None,
             title=u"Test", description=u"Description"):
         self.set_mergeable(mergeable)
         if not enable_notifications:
             # mock notification side effect
             self.notification_patcher = mock.patch(
                 'rhodecode.model.notification.NotificationModel.create')
             self.notification_patcher.start()
 
         if not self.pull_request:
             if not commits:
                 commits = [
                     {'message': 'c1'},
                     {'message': 'c2'},
                     {'message': 'c3'},
                 ]
                 target_head = 'c1'
                 source_head = 'c2'
                 revisions = ['c2']
 
             self.commit_ids = self.backend.create_master_repo(commits)
             self.target_repository = self.backend.create_repo(
                 heads=[target_head], name_suffix=name_suffix)
             self.source_repository = self.backend.create_repo(
                 heads=[source_head], name_suffix=name_suffix)
             self.author = author or UserModel().get_by_username(
                 TEST_USER_ADMIN_LOGIN)
 
             model = PullRequestModel()
             self.create_parameters = {
                 'created_by': self.author,
                 'source_repo': self.source_repository.repo_name,
                 'source_ref': self._default_branch_reference(source_head),
                 'target_repo': self.target_repository.repo_name,
                 'target_ref': self._default_branch_reference(target_head),
                 'revisions': [self.commit_ids[r] for r in revisions],
                 'reviewers': reviewers or self._get_reviewers(),
                 'title': title,
                 'description': description,
             }
             self.pull_request = model.create(**self.create_parameters)
             assert model.get_versions(self.pull_request) == []
 
             self.pull_request_id = self.pull_request.pull_request_id
 
             if approved:
                 self.approve()
 
             Session().add(self.pull_request)
             Session().commit()
 
         return self.pull_request
 
     def approve(self):
         self.create_status_votes(
             ChangesetStatus.STATUS_APPROVED,
             *self.pull_request.reviewers)
 
     def close(self):
         PullRequestModel().close_pull_request(self.pull_request, self.author)
 
     def _default_branch_reference(self, commit_message):
         reference = '%s:%s:%s' % (
             'branch',
             self.backend.default_branch_name,
             self.commit_ids[commit_message])
         return reference
 
     def _get_reviewers(self):
         return [
             (TEST_USER_REGULAR_LOGIN, ['default1'], False, []),
             (TEST_USER_REGULAR2_LOGIN, ['default2'], False, []),
         ]
 
     def update_source_repository(self, head=None):
         heads = [head or 'c3']
         self.backend.pull_heads(self.source_repository, heads=heads)
 
     def add_one_commit(self, head=None):
         self.update_source_repository(head=head)
         old_commit_ids = set(self.pull_request.revisions)
         PullRequestModel().update_commits(self.pull_request)
         commit_ids = set(self.pull_request.revisions)
         new_commit_ids = commit_ids - old_commit_ids
         assert len(new_commit_ids) == 1
         return new_commit_ids.pop()
 
     def remove_one_commit(self):
         assert len(self.pull_request.revisions) == 2
         source_vcs = self.source_repository.scm_instance()
         removed_commit_id = source_vcs.commit_ids[-1]
 
         # TODO: johbo: Git and Mercurial have an inconsistent vcs api here,
         # remove the if once that's sorted out.
         if self.backend.alias == "git":
             kwargs = {'branch_name': self.backend.default_branch_name}
         else:
             kwargs = {}
         source_vcs.strip(removed_commit_id, **kwargs)
 
         PullRequestModel().update_commits(self.pull_request)
         assert len(self.pull_request.revisions) == 1
         return removed_commit_id
 
     def create_comment(self, linked_to=None):
         comment = CommentsModel().create(
             text=u"Test comment",
             repo=self.target_repository.repo_name,
             user=self.author,
             pull_request=self.pull_request)
         assert comment.pull_request_version_id is None
 
         if linked_to:
             PullRequestModel()._link_comments_to_version(linked_to)
 
         return comment
 
     def create_inline_comment(
             self, linked_to=None, line_no=u'n1', file_path='file_1'):
         comment = CommentsModel().create(
             text=u"Test comment",
             repo=self.target_repository.repo_name,
             user=self.author,
             line_no=line_no,
             f_path=file_path,
             pull_request=self.pull_request)
         assert comment.pull_request_version_id is None
 
         if linked_to:
             PullRequestModel()._link_comments_to_version(linked_to)
 
         return comment
 
     def create_version_of_pull_request(self):
         pull_request = self.create_pull_request()
         version = PullRequestModel()._create_version_from_snapshot(
             pull_request)
         return version
 
     def create_status_votes(self, status, *reviewers):
         for reviewer in reviewers:
             ChangesetStatusModel().set_status(
                 repo=self.pull_request.target_repo,
                 status=status,
                 user=reviewer.user_id,
                 pull_request=self.pull_request)
 
     def set_mergeable(self, value):
         if not self.mergeable_patcher:
             self.mergeable_patcher = mock.patch.object(
                 VcsSettingsModel, 'get_general_settings')
             self.mergeable_mock = self.mergeable_patcher.start()
         self.mergeable_mock.return_value = {
             'rhodecode_pr_merge_enabled': value}
 
     def cleanup(self):
         # In case the source repository is already cleaned up, the pull
         # request will already be deleted.
         pull_request = PullRequest().get(self.pull_request_id)
         if pull_request:
             PullRequestModel().delete(pull_request, pull_request.author)
             Session().commit()
 
         if self.notification_patcher:
             self.notification_patcher.stop()
 
         if self.mergeable_patcher:
             self.mergeable_patcher.stop()
 
 
 @pytest.fixture
 def user_admin(baseapp):
     """
     Provides the default admin test user as an instance of `db.User`.
     """
     user = UserModel().get_by_username(TEST_USER_ADMIN_LOGIN)
     return user
 
 
 @pytest.fixture
 def user_regular(baseapp):
     """
     Provides the default regular test user as an instance of `db.User`.
     """
     user = UserModel().get_by_username(TEST_USER_REGULAR_LOGIN)
     return user
 
 
 @pytest.fixture
 def user_util(request, db_connection):
     """
     Provides a wired instance of `UserUtility` with integrated cleanup.
     """
     utility = UserUtility(test_name=request.node.name)
     request.addfinalizer(utility.cleanup)
     return utility
 
 
 # TODO: johbo: Split this up into utilities per domain or something similar
 class UserUtility(object):
 
     def __init__(self, test_name="test"):
         self._test_name = self._sanitize_name(test_name)
         self.fixture = Fixture()
         self.repo_group_ids = []
         self.repos_ids = []
         self.user_ids = []
         self.user_group_ids = []
         self.user_repo_permission_ids = []
         self.user_group_repo_permission_ids = []
         self.user_repo_group_permission_ids = []
         self.user_group_repo_group_permission_ids = []
         self.user_user_group_permission_ids = []
         self.user_group_user_group_permission_ids = []
         self.user_permissions = []
 
     def _sanitize_name(self, name):
         for char in ['[', ']']:
             name = name.replace(char, '_')
         return name
 
     def create_repo_group(
             self, owner=TEST_USER_ADMIN_LOGIN, auto_cleanup=True):
         group_name = "{prefix}_repogroup_{count}".format(
             prefix=self._test_name,
             count=len(self.repo_group_ids))
         repo_group = self.fixture.create_repo_group(
             group_name, cur_user=owner)
         if auto_cleanup:
             self.repo_group_ids.append(repo_group.group_id)
         return repo_group
 
     def create_repo(self, owner=TEST_USER_ADMIN_LOGIN, parent=None,
                     auto_cleanup=True, repo_type='hg', bare=False):
         repo_name = "{prefix}_repository_{count}".format(
             prefix=self._test_name,
             count=len(self.repos_ids))
 
         repository = self.fixture.create_repo(
             repo_name, cur_user=owner, repo_group=parent, repo_type=repo_type, bare=bare)
         if auto_cleanup:
             self.repos_ids.append(repository.repo_id)
         return repository
 
     def create_user(self, auto_cleanup=True, **kwargs):
         user_name = "{prefix}_user_{count}".format(
             prefix=self._test_name,
             count=len(self.user_ids))
         user = self.fixture.create_user(user_name, **kwargs)
         if auto_cleanup:
             self.user_ids.append(user.user_id)
         return user
 
     def create_additional_user_email(self, user, email):
         uem = self.fixture.create_additional_user_email(user=user, email=email)
         return uem
 
     def create_user_with_group(self):
         user = self.create_user()
         user_group = self.create_user_group(members=[user])
         return user, user_group
 
     def create_user_group(self, owner=TEST_USER_ADMIN_LOGIN, members=None,
                           auto_cleanup=True, **kwargs):
         group_name = "{prefix}_usergroup_{count}".format(
             prefix=self._test_name,
             count=len(self.user_group_ids))
         user_group = self.fixture.create_user_group(
             group_name, cur_user=owner, **kwargs)
 
         if auto_cleanup:
             self.user_group_ids.append(user_group.users_group_id)
         if members:
             for user in members:
                 UserGroupModel().add_user_to_group(user_group, user)
         return user_group
 
     def grant_user_permission(self, user_name, permission_name):
         self.inherit_default_user_permissions(user_name, False)
         self.user_permissions.append((user_name, permission_name))
 
     def grant_user_permission_to_repo_group(
             self, repo_group, user, permission_name):
         permission = RepoGroupModel().grant_user_permission(
             repo_group, user, permission_name)
         self.user_repo_group_permission_ids.append(
             (repo_group.group_id, user.user_id))
         return permission
 
     def grant_user_group_permission_to_repo_group(
             self, repo_group, user_group, permission_name):
         permission = RepoGroupModel().grant_user_group_permission(
             repo_group, user_group, permission_name)
         self.user_group_repo_group_permission_ids.append(
             (repo_group.group_id, user_group.users_group_id))
         return permission
 
     def grant_user_permission_to_repo(
             self, repo, user, permission_name):
         permission = RepoModel().grant_user_permission(
             repo, user, permission_name)
         self.user_repo_permission_ids.append(
             (repo.repo_id, user.user_id))
         return permission
 
     def grant_user_group_permission_to_repo(
             self, repo, user_group, permission_name):
         permission = RepoModel().grant_user_group_permission(
             repo, user_group, permission_name)
         self.user_group_repo_permission_ids.append(
             (repo.repo_id, user_group.users_group_id))
         return permission
 
     def grant_user_permission_to_user_group(
             self, target_user_group, user, permission_name):
         permission = UserGroupModel().grant_user_permission(
             target_user_group, user, permission_name)
         self.user_user_group_permission_ids.append(
             (target_user_group.users_group_id, user.user_id))
         return permission
 
     def grant_user_group_permission_to_user_group(
             self, target_user_group, user_group, permission_name):
         permission = UserGroupModel().grant_user_group_permission(
             target_user_group, user_group, permission_name)
         self.user_group_user_group_permission_ids.append(
             (target_user_group.users_group_id, user_group.users_group_id))
         return permission
 
     def revoke_user_permission(self, user_name, permission_name):
         self.inherit_default_user_permissions(user_name, True)
         UserModel().revoke_perm(user_name, permission_name)
 
     def inherit_default_user_permissions(self, user_name, value):
         user = UserModel().get_by_username(user_name)
         user.inherit_default_permissions = value
         Session().add(user)
         Session().commit()
 
     def cleanup(self):
         self._cleanup_permissions()
         self._cleanup_repos()
         self._cleanup_repo_groups()
         self._cleanup_user_groups()
         self._cleanup_users()
 
     def _cleanup_permissions(self):
         if self.user_permissions:
             for user_name, permission_name in self.user_permissions:
                 self.revoke_user_permission(user_name, permission_name)
 
         for permission in self.user_repo_permission_ids:
             RepoModel().revoke_user_permission(*permission)
 
         for permission in self.user_group_repo_permission_ids:
             RepoModel().revoke_user_group_permission(*permission)
 
         for permission in self.user_repo_group_permission_ids:
             RepoGroupModel().revoke_user_permission(*permission)
 
         for permission in self.user_group_repo_group_permission_ids:
             RepoGroupModel().revoke_user_group_permission(*permission)
 
         for permission in self.user_user_group_permission_ids:
             UserGroupModel().revoke_user_permission(*permission)
 
         for permission in self.user_group_user_group_permission_ids:
             UserGroupModel().revoke_user_group_permission(*permission)
 
     def _cleanup_repo_groups(self):
         def _repo_group_compare(first_group_id, second_group_id):
             """
             Gives higher priority to the groups with the most complex paths
             """
             first_group = RepoGroup.get(first_group_id)
             second_group = RepoGroup.get(second_group_id)
             first_group_parts = (
                 len(first_group.group_name.split('/')) if first_group else 0)
             second_group_parts = (
                 len(second_group.group_name.split('/')) if second_group else 0)
             return cmp(second_group_parts, first_group_parts)
 
         sorted_repo_group_ids = sorted(
             self.repo_group_ids, cmp=_repo_group_compare)
         for repo_group_id in sorted_repo_group_ids:
             self.fixture.destroy_repo_group(repo_group_id)
 
     def _cleanup_repos(self):
         sorted_repos_ids = sorted(self.repos_ids)
         for repo_id in sorted_repos_ids:
             self.fixture.destroy_repo(repo_id)
 
     def _cleanup_user_groups(self):
         def _user_group_compare(first_group_id, second_group_id):
             """
             Gives higher priority to the groups with the most complex paths
             """
             first_group = UserGroup.get(first_group_id)
             second_group = UserGroup.get(second_group_id)
             first_group_parts = (
                 len(first_group.users_group_name.split('/'))
                 if first_group else 0)
             second_group_parts = (
                 len(second_group.users_group_name.split('/'))
                 if second_group else 0)
             return cmp(second_group_parts, first_group_parts)
 
         sorted_user_group_ids = sorted(
             self.user_group_ids, cmp=_user_group_compare)
         for user_group_id in sorted_user_group_ids:
             self.fixture.destroy_user_group(user_group_id)
 
     def _cleanup_users(self):
         for user_id in self.user_ids:
             self.fixture.destroy_user(user_id)
 
 
 # TODO: Think about moving this into a pytest-pyro package and make it a
 # pytest plugin
 @pytest.hookimpl(tryfirst=True, hookwrapper=True)
 def pytest_runtest_makereport(item, call):
     """
     Adding the remote traceback if the exception has this information.
 
     VCSServer attaches this information as the attribute `_vcs_server_traceback`
     to the exception instance.
     """
     outcome = yield
     report = outcome.get_result()
     if call.excinfo:
         _add_vcsserver_remote_traceback(report, call.excinfo.value)
 
 
 def _add_vcsserver_remote_traceback(report, exc):
     vcsserver_traceback = getattr(exc, '_vcs_server_traceback', None)
 
     if vcsserver_traceback:
         section = 'VCSServer remote traceback ' + report.when
         report.sections.append((section, vcsserver_traceback))
 
 
 @pytest.fixture(scope='session')
 def testrun():
     return {
         'uuid': uuid.uuid4(),
         'start': datetime.datetime.utcnow().isoformat(),
         'timestamp': int(time.time()),
     }
 
 
[e-mail](autouse=True)
-def collect_appenlight_stats(request, testrun):
-    """
-    This fixture reports memory consumtion of single tests.
-
-    It gathers data based on `psutil` and sends them to Appenlight. The option
-    ``--ae`` has te be used to enable this fixture and the API key for your
-    application has to be provided in ``--ae-key``.
-    """
-    try:
-        # cygwin cannot have yet psutil support.
-        import psutil
-    except ImportError:
-        return
-
-    if not request.config.getoption('--appenlight'):
-        return
-    else:
-        # Only request the baseapp fixture if appenlight tracking is
-        # enabled. This will speed up a test run of unit tests by 2 to 3
-        # seconds if appenlight is not enabled.
-        baseapp = request.getfuncargvalue("baseapp")
-    url = '{}/api/logs'.format(request.config.getoption('--appenlight-url'))
-    client = AppenlightClient(
-        url=url,
-        api_key=request.config.getoption('--appenlight-api-key'),
-        namespace=request.node.nodeid,
-        request=str(testrun['uuid']),
-        testrun=testrun)
-
-    client.collect({
-        'message': "Starting",
-    })
-
-    server_and_port = baseapp.config.get_settings()['vcs.server']
-    protocol = baseapp.config.get_settings()['vcs.server.protocol']
-    server = create_vcsserver_proxy(server_and_port, protocol)
-    with server:
-        vcs_pid = server.get_pid()
-        server.run_gc()
-    vcs_process = psutil.Process(vcs_pid)
-    mem = vcs_process.memory_info()
-    client.tag_before('vcsserver.rss', mem.rss)
-    client.tag_before('vcsserver.vms', mem.vms)
-
-    test_process = psutil.Process()
-    mem = test_process.memory_info()
-    client.tag_before('test.rss', mem.rss)
-    client.tag_before('test.vms', mem.vms)
-
-    client.tag_before('time', time.time())
-
-    @request.addfinalizer
-    def send_stats():
-        client.tag_after('time', time.time())
-        with server:
-            gc_stats = server.run_gc()
-        for tag, value in gc_stats.items():
-            client.tag_after(tag, value)
-        mem = vcs_process.memory_info()
-        client.tag_after('vcsserver.rss', mem.rss)
-        client.tag_after('vcsserver.vms', mem.vms)
-
-        mem = test_process.memory_info()
-        client.tag_after('test.rss', mem.rss)
-        client.tag_after('test.vms', mem.vms)
-
-        client.collect({
-            'message': "Finished",
-        })
-        client.send_stats()
-
-    return client
-
-
-class AppenlightClient():
+class AppenlightClient(object):
 
     url_template = '{url}?protocol_version=0.5'
 
     def __init__(
             self, url, api_key, add_server=True, add_timestamp=True,
             namespace=None, request=None, testrun=None):
         self.url = self.url_template.format(url=url)
         self.api_key = api_key
         self.add_server = add_server
         self.add_timestamp = add_timestamp
         self.namespace = namespace
         self.request = request
         self.server = socket.getfqdn(socket.gethostname())
         self.tags_before = {}
         self.tags_after = {}
         self.stats = []
         self.testrun = testrun or {}
 
     def tag_before(self, tag, value):
         self.tags_before[tag] = value
 
     def tag_after(self, tag, value):
         self.tags_after[tag] = value
 
     def collect(self, data):
         if self.add_server:
             data.setdefault('server', self.server)
         if self.add_timestamp:
             data.setdefault('date', datetime.datetime.utcnow().isoformat())
         if self.namespace:
             data.setdefault('namespace', self.namespace)
         if self.request:
             data.setdefault('request', self.request)
         self.stats.append(data)
 
     def send_stats(self):
         tags = [
             ('testrun', self.request),
             ('testrun.start', self.testrun['start']),
             ('testrun.timestamp', self.testrun['timestamp']),
             ('test', self.namespace),
         ]
         for key, value in self.tags_before.items():
             tags.append((key + '.before', value))
             try:
                 delta = self.tags_after[key] - value
                 tags.append((key + '.delta', delta))
             except Exception:
                 pass
         for key, value in self.tags_after.items():
             tags.append((key + '.after', value))
         self.collect({
             'message': "Collected tags",
             'tags': tags,
         })
 
         response = requests.post(
             self.url,
             headers={
                 'X-appenlight-api-key': self.api_key},
             json=self.stats,
         )
 
         if not response.status_code == 200:
             pprint.pprint(self.stats)
             print(response.headers)
             print(response.text)
             raise Exception('Sending to appenlight failed')
 
 
 @pytest.fixture
 def gist_util(request, db_connection):
     """
     Provides a wired instance of `GistUtility` with integrated cleanup.
     """
     utility = GistUtility()
     request.addfinalizer(utility.cleanup)
     return utility
 
 
 class GistUtility(object):
     def __init__(self):
         self.fixture = Fixture()
         self.gist_ids = []
 
     def create_gist(self, **kwargs):
         gist = self.fixture.create_gist(**kwargs)
         self.gist_ids.append(gist.gist_id)
         return gist
 
     def cleanup(self):
         for id_ in self.gist_ids:
             self.fixture.destroy_gists(str(id_))
 
 
 @pytest.fixture
 def enabled_backends(request):
     backends = request.config.option.backends
     return backends[:]
 
 
 @pytest.fixture
 def settings_util(request, db_connection):
     """
     Provides a wired instance of `SettingsUtility` with integrated cleanup.
     """
     utility = SettingsUtility()
     request.addfinalizer(utility.cleanup)
     return utility
 
 
 class SettingsUtility(object):
     def __init__(self):
         self.rhodecode_ui_ids = []
         self.rhodecode_setting_ids = []
         self.repo_rhodecode_ui_ids = []
         self.repo_rhodecode_setting_ids = []
 
     def create_repo_rhodecode_ui(
             self, repo, section, value, key=None, active=True, cleanup=True):
         key = key or hashlib.sha1(
             '{}{}{}'.format(section, value, repo.repo_id)).hexdigest()
 
         setting = RepoRhodeCodeUi()
         setting.repository_id = repo.repo_id
         setting.ui_section = section
         setting.ui_value = value
         setting.ui_key = key
         setting.ui_active = active
         Session().add(setting)
         Session().commit()
 
         if cleanup:
             self.repo_rhodecode_ui_ids.append(setting.ui_id)
         return setting
 
     def create_rhodecode_ui(
             self, section, value, key=None, active=True, cleanup=True):
         key = key or hashlib.sha1('{}{}'.format(section, value)).hexdigest()
 
         setting = RhodeCodeUi()
         setting.ui_section = section
         setting.ui_value = value
         setting.ui_key = key
         setting.ui_active = active
         Session().add(setting)
         Session().commit()
 
         if cleanup:
             self.rhodecode_ui_ids.append(setting.ui_id)
         return setting
 
     def create_repo_rhodecode_setting(
             self, repo, name, value, type_, cleanup=True):
         setting = RepoRhodeCodeSetting(
             repo.repo_id, key=name, val=value, type=type_)
         Session().add(setting)
         Session().commit()
 
         if cleanup:
             self.repo_rhodecode_setting_ids.append(setting.app_settings_id)
         return setting
 
     def create_rhodecode_setting(self, name, value, type_, cleanup=True):
         setting = RhodeCodeSetting(key=name, val=value, type=type_)
         Session().add(setting)
         Session().commit()
 
         if cleanup:
             self.rhodecode_setting_ids.append(setting.app_settings_id)
 
         return setting
 
     def cleanup(self):
         for id_ in self.rhodecode_ui_ids:
             setting = RhodeCodeUi.get(id_)
             Session().delete(setting)
 
         for id_ in self.rhodecode_setting_ids:
             setting = RhodeCodeSetting.get(id_)
             Session().delete(setting)
 
         for id_ in self.repo_rhodecode_ui_ids:
             setting = RepoRhodeCodeUi.get(id_)
             Session().delete(setting)
 
         for id_ in self.repo_rhodecode_setting_ids:
             setting = RepoRhodeCodeSetting.get(id_)
             Session().delete(setting)
 
         Session().commit()
 
 
 @pytest.fixture
 def no_notifications(request):
     notification_patcher = mock.patch(
         'rhodecode.model.notification.NotificationModel.create')
     notification_patcher.start()
     request.addfinalizer(notification_patcher.stop)
 
 
 @pytest.fixture(scope='session')
 def repeat(request):
     """
     The number of repetitions is based on this fixture.
 
     Slower calls may divide it by 10 or 100. It is chosen in a way so that the
     tests are not too slow in our default test suite.
     """
     return request.config.getoption('--repeat')
 
 
 @pytest.fixture
 def rhodecode_fixtures():
     return Fixture()
 
 
 @pytest.fixture
 def context_stub():
     """
     Stub context object.
     """
     context = pyramid.testing.DummyResource()
     return context
 
 
 @pytest.fixture
 def request_stub():
     """
     Stub request object.
     """
     from rhodecode.lib.base import bootstrap_request
     request = bootstrap_request(scheme='https')
     return request
 
 
 @pytest.fixture
 def config_stub(request, request_stub):
     """
     Set up pyramid.testing and return the Configurator.
     """
     from rhodecode.lib.base import bootstrap_config
     config = bootstrap_config(request=request_stub)
 
     @request.addfinalizer
     def cleanup():
         pyramid.testing.tearDown()
 
     return config
 
 
 @pytest.fixture
 def StubIntegrationType():
     class _StubIntegrationType(IntegrationTypeBase):
         """ Test integration type class """
 
         key = 'test'
         display_name = 'Test integration type'
         description = 'A test integration type for testing'
 
         @classmethod
         def icon(cls):
             return 'test_icon_html_image'
 
         def __init__(self, settings):
             super(_StubIntegrationType, self).__init__(settings)
             self.sent_events = []  # for testing
 
         def send_event(self, event):
             self.sent_events.append(event)
 
         def settings_schema(self):
             class SettingsSchema(colander.Schema):
                 test_string_field = colander.SchemaNode(
                     colander.String(),
                     missing=colander.required,
                     title='test string field',
                 )
                 test_int_field = colander.SchemaNode(
                     colander.Int(),
                     title='some integer setting',
                 )
             return SettingsSchema()
 
 
     integration_type_registry.register_integration_type(_StubIntegrationType)
     return _StubIntegrationType
 
 @pytest.fixture
 def stub_integration_settings():
     return {
         'test_string_field': 'some data',
         'test_int_field': 100,
     }
 
 
 @pytest.fixture
 def repo_integration_stub(request, repo_stub, StubIntegrationType,
         stub_integration_settings):
     integration = IntegrationModel().create(
         StubIntegrationType, settings=stub_integration_settings, enabled=True,
         name='test repo integration',
         repo=repo_stub, repo_group=None, child_repos_only=None)
 
     @request.addfinalizer
     def cleanup():
         IntegrationModel().delete(integration)
 
     return integration
 
 
 @pytest.fixture
 def repogroup_integration_stub(request, test_repo_group, StubIntegrationType,
     stub_integration_settings):
     integration = IntegrationModel().create(
         StubIntegrationType, settings=stub_integration_settings, enabled=True,
         name='test repogroup integration',
         repo=None, repo_group=test_repo_group, child_repos_only=True)
 
     @request.addfinalizer
     def cleanup():
         IntegrationModel().delete(integration)
 
     return integration
 
 
 @pytest.fixture
 def repogroup_recursive_integration_stub(request, test_repo_group,
     StubIntegrationType, stub_integration_settings):
     integration = IntegrationModel().create(
         StubIntegrationType, settings=stub_integration_settings, enabled=True,
         name='test recursive repogroup integration',
         repo=None, repo_group=test_repo_group, child_repos_only=False)
 
     @request.addfinalizer
     def cleanup():
         IntegrationModel().delete(integration)
 
     return integration
 
 
 @pytest.fixture
 def global_integration_stub(request, StubIntegrationType,
     stub_integration_settings):
     integration = IntegrationModel().create(
         StubIntegrationType, settings=stub_integration_settings, enabled=True,
         name='test global integration',
         repo=None, repo_group=None, child_repos_only=None)
 
     @request.addfinalizer
     def cleanup():
         IntegrationModel().delete(integration)
 
     return integration
 
 
 @pytest.fixture
 def root_repos_integration_stub(request, StubIntegrationType,
     stub_integration_settings):
     integration = IntegrationModel().create(
         StubIntegrationType, settings=stub_integration_settings, enabled=True,
         name='test global integration',
         repo=None, repo_group=None, child_repos_only=True)
 
     @request.addfinalizer
     def cleanup():
         IntegrationModel().delete(integration)
 
     return integration
 
 
 @pytest.fixture
 def local_dt_to_utc():
     def _factory(dt):
         return dt.replace(tzinfo=dateutil.tz.tzlocal()).astimezone(
             dateutil.tz.tzutc()).replace(tzinfo=None)
     return _factory
 
 
 @pytest.fixture
 def disable_anonymous_user(request, baseapp):
     set_anonymous_access(False)
 
     @request.addfinalizer
     def cleanup():
         set_anonymous_access(True)
 
 
 @pytest.fixture(scope='module')
 def rc_fixture(request):
     return Fixture()
 
 
 @pytest.fixture
 def repo_groups(request):
     fixture = Fixture()
 
     session = Session()
     zombie_group = fixture.create_repo_group('zombie')
     parent_group = fixture.create_repo_group('parent')
     child_group = fixture.create_repo_group('parent/child')
     groups_in_db = session.query(RepoGroup).all()
     assert len(groups_in_db) == 3
     assert child_group.group_parent_id == parent_group.group_id
 
     @request.addfinalizer
     def cleanup():
         fixture.destroy_repo_group(zombie_group)
         fixture.destroy_repo_group(child_group)
         fixture.destroy_repo_group(parent_group)
 
     return zombie_group, parent_group, child_group
 
 
 @pytest.fixture(scope="session")
 def tmp_path_factory(request):
     """Return a :class:`_pytest.tmpdir.TempPathFactory` instance for the test session.
     """
 
     class TempPathFactory:
 
         def mktemp(self, basename):
             import tempfile
             return tempfile.mktemp(basename)
 
     return TempPathFactory()

rhodecode/tests/vcs/test_client_http.py

@@ -1,133 +1,133 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2019 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 import logging
 
 import mock
 import msgpack
 import pytest
 
 from rhodecode.lib import vcs
 from rhodecode.lib.vcs import client_http, exceptions
 
 
 def is_new_connection(logger, level, message):
     return (
         logger == 'requests.packages.urllib3.connectionpool' and
         message.startswith('Starting new HTTP'))
 
 
 @pytest.fixture
 def stub_session():
     """
     Stub of `requests.Session()`.
     """
     session = mock.Mock()
     post = session.post()
     post.content = msgpack.packb({})
     post.status_code = 200
 
     session.reset_mock()
     return session
 
 
 @pytest.fixture
 def stub_fail_session():
     """
     Stub of `requests.Session()`.
     """
     session = mock.Mock()
     post = session.post()
     post.content = msgpack.packb({'error': '500'})
     post.status_code = 500
 
     session.reset_mock()
     return session
 
 
 @pytest.fixture
 def stub_session_factory(stub_session):
     """
     Stub of `rhodecode.lib.vcs.client_http.ThreadlocalSessionFactory`.
     """
     session_factory = mock.Mock()
     session_factory.return_value = stub_session
     return session_factory
 
 
 @pytest.fixture
 def stub_session_failing_factory(stub_fail_session):
     """
     Stub of `rhodecode.lib.vcs.client_http.ThreadlocalSessionFactory`.
     """
     session_factory = mock.Mock()
     session_factory.return_value = stub_fail_session
     return session_factory
 
 
 def test_uses_persistent_http_connections(caplog, vcsbackend_hg):
     repo = vcsbackend_hg.repo
     remote_call = repo._remote.branches
 
     with caplog.at_level(logging.INFO):
         for x in range(5):
             remote_call(normal=True, closed=False)
 
     new_connections = [
         r for r in caplog.record_tuples if is_new_connection(*r)]
     assert len(new_connections) <= 1
 
 
 def test_repo_maker_uses_session_for_classmethods(stub_session_factory):
-    repo_maker = client_http.RepoMaker(
+    repo_maker = client_http.RemoteVCSMaker(
         'server_and_port', 'endpoint', 'test_dummy_scm', stub_session_factory)
     repo_maker.example_call()
     stub_session_factory().post.assert_called_with(
         'http://server_and_port/endpoint', data=mock.ANY)
 
 
 def test_repo_maker_uses_session_for_instance_methods(
         stub_session_factory, config):
-    repo_maker = client_http.RepoMaker(
+    repo_maker = client_http.RemoteVCSMaker(
         'server_and_port', 'endpoint', 'test_dummy_scm', stub_session_factory)
     repo = repo_maker('stub_path', 'stub_repo_id', config)
     repo.example_call()
     stub_session_factory().post.assert_called_with(
         'http://server_and_port/endpoint', data=mock.ANY)
 
 
 @mock.patch('rhodecode.lib.vcs.client_http.ThreadlocalSessionFactory')
 @mock.patch('rhodecode.lib.vcs.connection')
 def test_connect_passes_in_the_same_session(
         connection, session_factory_class, stub_session):
     session_factory = session_factory_class.return_value
     session_factory.return_value = stub_session
 
     vcs.connect_http('server_and_port')
 
 
 def test_repo_maker_uses_session_that_throws_error(
         stub_session_failing_factory, config):
-    repo_maker = client_http.RepoMaker(
+    repo_maker = client_http.RemoteVCSMaker(
         'server_and_port', 'endpoint', 'test_dummy_scm', stub_session_failing_factory)
     repo = repo_maker('stub_path', 'stub_repo_id', config)
 
     with pytest.raises(exceptions.HttpVCSCommunicationError):
         repo.example_call()