vcsserver: made binary content check be calculated on vcsserver

- now can use caching
- for git/hg there are faster dedicated functions instead of detecting the content

rhodecode/lib/vcs/backends/base.py

@@ -1,1890 +1,1896 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 is_node_binary(self, path):
+        """
+        Returns ``True`` is given path is a binary file
+        """
+        raise NotImplementedError
+
     def get_file_content(self, path):
         """
         Returns content of the file at the given `path`.
         """
         raise NotImplementedError
 
     def get_file_content_streamed(self, path):
         """
         returns a streaming response from vcsserver with file content
         """
         raise NotImplementedError
 
     def get_file_size(self, path):
         """
         Returns size of the file at the given `path`.
         """
         raise NotImplementedError
 
     def get_path_commit(self, path, pre_load=None):
         """
         Returns last commit of the file at the given `path`.
 
         :param pre_load: Optional. List of commit attributes to load.
         """
         commits = self.get_path_history(path, limit=1, pre_load=pre_load)
         if not commits:
             raise RepositoryError(
                 'Failed to fetch history for path {}. '
                 'Please check if such path exists in your repository'.format(
                     path))
         return commits[0]
 
     def get_path_history(self, path, limit=None, pre_load=None):
         """
         Returns history of file as reversed list of :class:`BaseCommit`
         objects for which file at given `path` has been modified.
 
         :param limit: Optional. Allows to limit the size of the returned
            history. This is intended as a hint to the underlying backend, so
            that it can apply optimizations depending on the limit.
         :param pre_load: Optional. List of commit attributes to load.
         """
         raise NotImplementedError
 
     def get_file_annotate(self, path, pre_load=None):
         """
         Returns a generator of four element tuples with
         lineno, sha, commit lazy loader and line
 
         :param pre_load: Optional. List of commit attributes to load.
         """
         raise NotImplementedError
 
     def get_nodes(self, path):
         """
         Returns combined ``DirNode`` and ``FileNode`` objects list representing
         state of commit at the given ``path``.
 
         :raises ``CommitError``: if node at the given ``path`` is not
           instance of ``DirNode``
         """
         raise NotImplementedError
 
     def get_node(self, path):
         """
         Returns ``Node`` object from the given ``path``.
 
         :raises ``NodeDoesNotExistError``: if there is no node at the given
           ``path``
         """
         raise NotImplementedError
 
     def get_largefile_node(self, path):
         """
         Returns the path to largefile from Mercurial/Git-lfs storage.
         or None if it's not a largefile node
         """
         return None
 
     def archive_repo(self, archive_dest_path, kind='tgz', subrepos=None,
                      prefix=None, write_metadata=False, mtime=None, archive_at_path='/'):
         """
         Creates an archive containing the contents of the repository.
 
         :param archive_dest_path: path to the file which to create the archive.
         :param kind: one of following: ``"tbz2"``, ``"tgz"``, ``"zip"``.
         :param prefix: name of root directory in archive.
             Default is repository name and commit's short_id joined with dash:
             ``"{repo_name}-{short_id}"``.
         :param write_metadata: write a metadata file into archive.
         :param mtime: custom modification time for archive creation, defaults
             to time.time() if not given.
         :param archive_at_path: pack files at this path (default '/')
 
         :raise VCSError: If prefix has a problem.
         """
         allowed_kinds = [x[0] for x in settings.ARCHIVE_SPECS]
         if kind not in allowed_kinds:
             raise ImproperArchiveTypeError(
                 'Archive kind (%s) not supported use one of %s' %
                 (kind, allowed_kinds))
 
         prefix = self._validate_archive_prefix(prefix)
 
         mtime = mtime is not None or time.mktime(self.date.timetuple())
 
         file_info = []
         cur_rev = self.repository.get_commit(commit_id=self.raw_id)
         for _r, _d, files in cur_rev.walk(archive_at_path):
             for f in files:
                 f_path = os.path.join(prefix, f.path)
                 file_info.append(
                     (f_path, f.mode, f.is_link(), f.raw_bytes))
 
         if write_metadata:
             metadata = [
                 ('repo_name', self.repository.name),
                 ('commit_id', self.raw_id),
                 ('mtime', mtime),
                 ('branch', self.branch),
                 ('tags', ','.join(self.tags)),
             ]
             meta = ["%s:%s" % (f_name, value) for f_name, value in metadata]
             file_info.append(('.archival.txt', 0o644, False, '\n'.join(meta)))
 
         connection.Hg.archive_repo(archive_dest_path, mtime, file_info, kind)
 
     def _validate_archive_prefix(self, prefix):
         if prefix is None:
             prefix = self._ARCHIVE_PREFIX_TEMPLATE.format(
                 repo_name=safe_str(self.repository.name),
                 short_id=self.short_id)
         elif not isinstance(prefix, str):
             raise ValueError("prefix not a bytes object: %s" % repr(prefix))
         elif prefix.startswith('/'):
             raise VCSError("Prefix cannot start with leading slash")
         elif prefix.strip() == '':
             raise VCSError("Prefix cannot be empty")
         return prefix
 
     @LazyProperty
     def root(self):
         """
         Returns ``RootNode`` object for this commit.
         """
         return self.get_node('')
 
     def next(self, branch=None):
         """
         Returns next commit from current, if branch is gives it will return
         next commit belonging to this branch
 
         :param branch: show commits within the given named branch
         """
         indexes = xrange(self.idx + 1, self.repository.count())
         return self._find_next(indexes, branch)
 
     def prev(self, branch=None):
         """
         Returns previous commit from current, if branch is gives it will
         return previous commit belonging to this branch
 
         :param branch: show commit within the given named branch
         """
         indexes = xrange(self.idx - 1, -1, -1)
         return self._find_next(indexes, branch)
 
     def _find_next(self, indexes, branch=None):
         if branch and self.branch != branch:
             raise VCSError('Branch option used on commit not belonging '
                            'to that branch')
 
         for next_idx in indexes:
             commit = self.repository.get_commit(commit_idx=next_idx)
             if branch and branch != commit.branch:
                 continue
             return commit
         raise CommitDoesNotExistError
 
     def diff(self, ignore_whitespace=True, context=3):
         """
         Returns a `Diff` object representing the change made by this commit.
         """
         parent = self.first_parent
         diff = self.repository.get_diff(
             parent, self,
             ignore_whitespace=ignore_whitespace,
             context=context)
         return diff
 
     @LazyProperty
     def added(self):
         """
         Returns list of added ``FileNode`` objects.
         """
         raise NotImplementedError
 
     @LazyProperty
     def changed(self):
         """
         Returns list of modified ``FileNode`` objects.
         """
         raise NotImplementedError
 
     @LazyProperty
     def removed(self):
         """
         Returns list of removed ``FileNode`` objects.
         """
         raise NotImplementedError
 
     @LazyProperty
     def size(self):
         """
         Returns total number of bytes from contents of all filenodes.
         """
         return sum((node.size for node in self.get_filenodes_generator()))
 
     def walk(self, topurl=''):
         """
         Similar to os.walk method. Insted of filesystem it walks through
         commit starting at given ``topurl``.  Returns generator of tuples
         (topnode, dirnodes, filenodes).
         """
         topnode = self.get_node(topurl)
         if not topnode.is_dir():
             return
         yield (topnode, topnode.dirs, topnode.files)
         for dirnode in topnode.dirs:
             for tup in self.walk(dirnode.path):
                 yield tup
 
     def get_filenodes_generator(self):
         """
         Returns generator that yields *all* file nodes.
         """
         for topnode, dirs, files in self.walk():
             for node in files:
                 yield node
 
     #
     # Utilities for sub classes to support consistent behavior
     #
 
     def no_node_at_path(self, path):
         return NodeDoesNotExistError(
             u"There is no file nor directory at the given path: "
             u"`%s` at commit %s" % (safe_unicode(path), self.short_id))
 
     def _fix_path(self, path):
         """
         Paths are stored without trailing slash so we need to get rid off it if
         needed.
         """
         return path.rstrip('/')
 
     #
     # Deprecated API based on changesets
     #
 
     @property
     def revision(self):
         warnings.warn("Use idx instead", DeprecationWarning)
         return self.idx
 
     @revision.setter
     def revision(self, value):
         warnings.warn("Use idx instead", DeprecationWarning)
         self.idx = value
 
     def get_file_changeset(self, path):
         warnings.warn("Use get_path_commit instead", DeprecationWarning)
         return self.get_path_commit(path)
 
 
 class BaseChangesetClass(type):
 
     def __instancecheck__(self, instance):
         return isinstance(instance, BaseCommit)
 
 
 class BaseChangeset(BaseCommit):
 
     __metaclass__ = BaseChangesetClass
 
     def __new__(cls, *args, **kwargs):
         warnings.warn(
             "Use BaseCommit instead of BaseChangeset", DeprecationWarning)
         return super(BaseChangeset, cls).__new__(cls, *args, **kwargs)
 
 
 class BaseInMemoryCommit(object):
     """
     Represents differences between repository's state (most recent head) and
     changes made *in place*.
 
     **Attributes**
 
         ``repository``
             repository object for this in-memory-commit
 
         ``added``
             list of ``FileNode`` objects marked as *added*
 
         ``changed``
             list of ``FileNode`` objects marked as *changed*
 
         ``removed``
             list of ``FileNode`` or ``RemovedFileNode`` objects marked to be
             *removed*
 
         ``parents``
             list of :class:`BaseCommit` instances representing parents of
             in-memory commit. Should always be 2-element sequence.
 
     """
 
     def __init__(self, repository):
         self.repository = repository
         self.added = []
         self.changed = []
         self.removed = []
         self.parents = []
 
     def add(self, *filenodes):
         """
         Marks given ``FileNode`` objects as *to be committed*.
 
         :raises ``NodeAlreadyExistsError``: if node with same path exists at
            latest commit
         :raises ``NodeAlreadyAddedError``: if node with same path is already
            marked as *added*
         """
         # Check if not already marked as *added* first
         for node in filenodes:
             if node.path in (n.path for n in self.added):
                 raise NodeAlreadyAddedError(
                     "Such FileNode %s is already marked for addition"
                     % node.path)
         for node in filenodes:
             self.added.append(node)
 
     def change(self, *filenodes):
         """
         Marks given ``FileNode`` objects to be *changed* in next commit.
 
         :raises ``EmptyRepositoryError``: if there are no commits yet
         :raises ``NodeAlreadyExistsError``: if node with same path is already
            marked to be *changed*
         :raises ``NodeAlreadyRemovedError``: if node with same path is already
            marked to be *removed*
         :raises ``NodeDoesNotExistError``: if node doesn't exist in latest
            commit
         :raises ``NodeNotChangedError``: if node hasn't really be changed
         """
         for node in filenodes:
             if node.path in (n.path for n in self.removed):
                 raise NodeAlreadyRemovedError(
                     "Node at %s is already marked as removed" % node.path)
         try:
             self.repository.get_commit()
         except EmptyRepositoryError:
             raise EmptyRepositoryError(
                 "Nothing to change - try to *add* new nodes rather than "
                 "changing them")
         for node in filenodes:
             if node.path in (n.path for n in self.changed):
                 raise NodeAlreadyChangedError(
                     "Node at '%s' is already marked as changed" % node.path)
             self.changed.append(node)
 
     def remove(self, *filenodes):
         """
         Marks given ``FileNode`` (or ``RemovedFileNode``) objects to be
         *removed* in next commit.
 
         :raises ``NodeAlreadyRemovedError``: if node has been already marked to
            be *removed*
         :raises ``NodeAlreadyChangedError``: if node has been already marked to
            be *changed*
         """
         for node in filenodes:
             if node.path in (n.path for n in self.removed):
                 raise NodeAlreadyRemovedError(
                     "Node is already marked to for removal at %s" % node.path)
             if node.path in (n.path for n in self.changed):
                 raise NodeAlreadyChangedError(
                     "Node is already marked to be changed at %s" % node.path)
             # We only mark node as *removed* - real removal is done by
             # commit method
             self.removed.append(node)
 
     def reset(self):
         """
         Resets this instance to initial state (cleans ``added``, ``changed``
         and ``removed`` lists).
         """
         self.added = []
         self.changed = []
         self.removed = []
         self.parents = []
 
     def get_ipaths(self):
         """
         Returns generator of paths from nodes marked as added, changed or
         removed.
         """
         for node in itertools.chain(self.added, self.changed, self.removed):
             yield node.path
 
     def get_paths(self):
         """
         Returns list of paths from nodes marked as added, changed or removed.
         """
         return list(self.get_ipaths())
 
     def check_integrity(self, parents=None):
         """
         Checks in-memory commit's integrity. Also, sets parents if not
         already set.
 
         :raises CommitError: if any error occurs (i.e.
           ``NodeDoesNotExistError``).
         """
         if not self.parents:
             parents = parents or []
             if len(parents) == 0:
                 try:
                     parents = [self.repository.get_commit(), None]
                 except EmptyRepositoryError:
                     parents = [None, None]
             elif len(parents) == 1:
                 parents += [None]
             self.parents = parents
 
         # Local parents, only if not None
         parents = [p for p in self.parents if p]
 
         # Check nodes marked as added
         for p in parents:
             for node in self.added:
                 try:
                     p.get_node(node.path)
                 except NodeDoesNotExistError:
                     pass
                 else:
                     raise NodeAlreadyExistsError(
                         "Node `%s` already exists at %s" % (node.path, p))
 
         # Check nodes marked as changed
         missing = set(self.changed)
         not_changed = set(self.changed)
         if self.changed and not parents:
             raise NodeDoesNotExistError(str(self.changed[0].path))
         for p in parents:
             for node in self.changed:
                 try:
                     old = p.get_node(node.path)
                     missing.remove(node)
                     # if content actually changed, remove node from not_changed
                     if old.content != node.content:
                         not_changed.remove(node)
                 except NodeDoesNotExistError:
                     pass
         if self.changed and missing:
             raise NodeDoesNotExistError(
                 "Node `%s` marked as modified but missing in parents: %s"
                 % (node.path, parents))
 
         if self.changed and not_changed:
             raise NodeNotChangedError(
                 "Node `%s` wasn't actually changed (parents: %s)"
                 % (not_changed.pop().path, parents))
 
         # Check nodes marked as removed
         if self.removed and not parents:
             raise NodeDoesNotExistError(
                 "Cannot remove node at %s as there "
                 "were no parents specified" % self.removed[0].path)
         really_removed = set()
         for p in parents:
             for node in self.removed:
                 try:
                     p.get_node(node.path)
                     really_removed.add(node)
                 except CommitError:
                     pass
         not_removed = set(self.removed) - really_removed
         if not_removed:
             # TODO: johbo: This code branch does not seem to be covered
             raise NodeDoesNotExistError(
                 "Cannot remove node at %s from "
                 "following parents: %s" % (not_removed, parents))
 
     def commit(self, message, author, parents=None, branch=None, date=None, **kwargs):
         """
         Performs in-memory commit (doesn't check workdir in any way) and
         returns newly created :class:`BaseCommit`. Updates repository's
         attribute `commits`.
 
         .. note::
 
            While overriding this method each backend's should call
            ``self.check_integrity(parents)`` in the first place.
 
         :param message: message of the commit
         :param author: full username, i.e. "Joe Doe <[e-mail]>"
         :param parents: single parent or sequence of parents from which commit
            would be derived
         :param date: ``datetime.datetime`` instance. Defaults to
            ``datetime.datetime.now()``.
         :param branch: branch name, as string. If none given, default backend's
            branch would be used.
 
         :raises ``CommitError``: if any error occurs while committing
         """
         raise NotImplementedError
 
 
 class BaseInMemoryChangesetClass(type):
 
     def __instancecheck__(self, instance):
         return isinstance(instance, BaseInMemoryCommit)
 
 
 class BaseInMemoryChangeset(BaseInMemoryCommit):
 
     __metaclass__ = BaseInMemoryChangesetClass
 
     def __new__(cls, *args, **kwargs):
         warnings.warn(
             "Use BaseCommit instead of BaseInMemoryCommit", DeprecationWarning)
         return super(BaseInMemoryChangeset, cls).__new__(cls, *args, **kwargs)
 
 
 class EmptyCommit(BaseCommit):
     """
     An dummy empty commit. It's possible to pass hash when creating
     an EmptyCommit
     """
 
     def __init__(
             self, commit_id=EMPTY_COMMIT_ID, repo=None, alias=None, idx=-1,
             message='', author='', date=None):
         self._empty_commit_id = commit_id
         # TODO: johbo: Solve idx parameter, default value does not make
         # too much sense
         self.idx = idx
         self.message = message
         self.author = author
         self.date = date or datetime.datetime.fromtimestamp(0)
         self.repository = repo
         self.alias = alias
 
     @LazyProperty
     def raw_id(self):
         """
         Returns raw string identifying this commit, useful for web
         representation.
         """
 
         return self._empty_commit_id
 
     @LazyProperty
     def branch(self):
         if self.alias:
             from rhodecode.lib.vcs.backends import get_backend
             return get_backend(self.alias).DEFAULT_BRANCH_NAME
 
     @LazyProperty
     def short_id(self):
         return self.raw_id[:12]
 
     @LazyProperty
     def id(self):
         return self.raw_id
 
     def get_path_commit(self, path):
         return self
 
     def get_file_content(self, path):
         return u''
 
     def get_file_content_streamed(self, path):
         yield self.get_file_content()
 
     def get_file_size(self, path):
         return 0
 
 
 class EmptyChangesetClass(type):
 
     def __instancecheck__(self, instance):
         return isinstance(instance, EmptyCommit)
 
 
 class EmptyChangeset(EmptyCommit):
 
     __metaclass__ = EmptyChangesetClass
 
     def __new__(cls, *args, **kwargs):
         warnings.warn(
             "Use EmptyCommit instead of EmptyChangeset", DeprecationWarning)
         return super(EmptyCommit, cls).__new__(cls, *args, **kwargs)
 
     def __init__(self, cs=EMPTY_COMMIT_ID, repo=None, requested_revision=None,
                  alias=None, revision=-1, message='', author='', date=None):
         if requested_revision is not None:
             warnings.warn(
                 "Parameter requested_revision not supported anymore",
                 DeprecationWarning)
         super(EmptyChangeset, self).__init__(
             commit_id=cs, repo=repo, alias=alias, idx=revision,
             message=message, author=author, date=date)
 
         @property
         def revision(self):
             warnings.warn("Use idx instead", DeprecationWarning)
             return self.idx
 
         @revision.setter
         def revision(self, value):
             warnings.warn("Use idx instead", DeprecationWarning)
             self.idx = value
 
 
 class EmptyRepository(BaseRepository):
     def __init__(self, repo_path=None, config=None, create=False, **kwargs):
         pass
 
     def get_diff(self, *args, **kwargs):
         from rhodecode.lib.vcs.backends.git.diff import GitDiff
         return GitDiff('')
 
 
 class CollectionGenerator(object):
 
     def __init__(self, repo, commit_ids, collection_size=None, pre_load=None, translate_tag=None):
         self.repo = repo
         self.commit_ids = commit_ids
         # TODO: (oliver) this isn't currently hooked up
         self.collection_size = None
         self.pre_load = pre_load
         self.translate_tag = translate_tag
 
     def __len__(self):
         if self.collection_size is not None:
             return self.collection_size
         return self.commit_ids.__len__()
 
     def __iter__(self):
         for commit_id in self.commit_ids:
             # TODO: johbo: Mercurial passes in commit indices or commit ids
             yield self._commit_factory(commit_id)
 
     def _commit_factory(self, commit_id):
         """
         Allows backends to override the way commits are generated.
         """
         return self.repo.get_commit(
             commit_id=commit_id, pre_load=self.pre_load,
             translate_tag=self.translate_tag)
 
     def __getslice__(self, i, j):
         """
         Returns an iterator of sliced repository
         """
         commit_ids = self.commit_ids[i:j]
         return self.__class__(
             self.repo, commit_ids, pre_load=self.pre_load,
             translate_tag=self.translate_tag)
 
     def __repr__(self):
         return '<CollectionGenerator[len:%s]>' % (self.__len__())
 
 
 class Config(object):
     """
     Represents the configuration for a repository.
 
     The API is inspired by :class:`ConfigParser.ConfigParser` from the
     standard library. It implements only the needed subset.
     """
 
     def __init__(self):
         self._values = {}
 
     def copy(self):
         clone = Config()
         for section, values in self._values.items():
             clone._values[section] = values.copy()
         return clone
 
     def __repr__(self):
         return '<Config(%s sections) at %s>' % (
             len(self._values), hex(id(self)))
 
     def items(self, section):
         return self._values.get(section, {}).iteritems()
 
     def get(self, section, option):
         return self._values.get(section, {}).get(option)
 
     def set(self, section, option, value):
         section_values = self._values.setdefault(section, {})
         section_values[option] = value
 
     def clear_section(self, section):
         self._values[section] = {}
 
     def serialize(self):
         """
         Creates a list of three tuples (section, key, value) representing
         this config object.
         """
         items = []
         for section in self._values:
             for option, value in self._values[section].items():
                 items.append(
                     (safe_str(section), safe_str(option), safe_str(value)))
         return items
 
 
 class Diff(object):
     """
     Represents a diff result from a repository backend.
 
     Subclasses have to provide a backend specific value for
     :attr:`_header_re` and :attr:`_meta_re`.
     """
     _meta_re = None
     _header_re = None
 
     def __init__(self, raw_diff):
         self.raw = raw_diff
 
     def chunks(self):
         """
         split the diff in chunks of separate --git a/file b/file chunks
         to make diffs consistent we must prepend with \n, and make sure
         we can detect last chunk as this was also has special rule
         """
 
         diff_parts = ('\n' + self.raw).split('\ndiff --git')
         header = diff_parts[0]
 
         if self._meta_re:
             match = self._meta_re.match(header)
 
         chunks = diff_parts[1:]
         total_chunks = len(chunks)
 
         return (
             DiffChunk(chunk, self, cur_chunk == total_chunks)
             for cur_chunk, chunk in enumerate(chunks, start=1))
 
 
 class DiffChunk(object):
 
     def __init__(self, chunk, diff, last_chunk):
         self._diff = diff
 
         # since we split by \ndiff --git that part is lost from original diff
         # we need to re-apply it at the end, EXCEPT ! if it's last chunk
         if not last_chunk:
             chunk += '\n'
 
         match = self._diff._header_re.match(chunk)
         self.header = match.groupdict()
         self.diff = chunk[match.end():]
         self.raw = chunk
 
 
 class BasePathPermissionChecker(object):
 
     @staticmethod
     def create_from_patterns(includes, excludes):
         if includes and '*' in includes and not excludes:
             return AllPathPermissionChecker()
         elif excludes and '*' in excludes:
             return NonePathPermissionChecker()
         else:
             return PatternPathPermissionChecker(includes, excludes)
 
     @property
     def has_full_access(self):
         raise NotImplemented()
 
     def has_access(self, path):
         raise NotImplemented()
 
 
 class AllPathPermissionChecker(BasePathPermissionChecker):
 
     @property
     def has_full_access(self):
         return True
 
     def has_access(self, path):
         return True
 
 
 class NonePathPermissionChecker(BasePathPermissionChecker):
 
     @property
     def has_full_access(self):
         return False
 
     def has_access(self, path):
         return False
 
 
 class PatternPathPermissionChecker(BasePathPermissionChecker):
 
     def __init__(self, includes, excludes):
         self.includes = includes
         self.excludes = excludes
         self.includes_re = [] if not includes else [
             re.compile(fnmatch.translate(pattern)) for pattern in includes]
         self.excludes_re = [] if not excludes else [
             re.compile(fnmatch.translate(pattern)) for pattern in excludes]
 
     @property
     def has_full_access(self):
         return '*' in self.includes and not self.excludes
 
     def has_access(self, path):
         for regex in self.excludes_re:
             if regex.match(path):
                 return False
         for regex in self.includes_re:
             if regex.match(path):
                 return True
         return False

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

@@ -1,479 +1,483 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 is_node_binary(self, path):
+        tree_id, _ = self._get_tree_id_for_path(path)
+        return self._remote.is_binary(tree_id)
+
     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/git/inmemory.py

@@ -1,104 +1,106 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 inmemory module
 """
 
 from rhodecode.lib.datelib import date_to_timestamp_plus_offset
 from rhodecode.lib.utils import safe_str
 from rhodecode.lib.vcs.backends import base
 
 
 class GitInMemoryCommit(base.BaseInMemoryCommit):
 
     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 `GitCommit`. Updates repository's
         `commit_ids`.
 
         :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
         """
         self.check_integrity(parents)
         if branch is None:
             branch = self.repository.DEFAULT_BRANCH_NAME
 
         ENCODING = "UTF-8"
 
         commit_tree = None
         if self.parents[0]:
             commit_tree = self.parents[0]._commit['tree']
 
         updated = []
         for node in self.added + self.changed:
-            if not node.is_binary:
+
+            if node.is_binary:
+                content = node.content
+            else:
                 content = node.content.encode(ENCODING)
-            else:
-                content = node.content
+
             updated.append({
                 'path': node.path,
                 'node_path': node.name.encode(ENCODING),
                 'content': content,
                 'mode': node.mode,
             })
 
         removed = [node.path for node in self.removed]
 
         date, tz = date_to_timestamp_plus_offset(date)
 
         # TODO: johbo: Make kwargs explicit and check if this is needed.
         author_time = kwargs.pop('author_time', date)
         author_tz = kwargs.pop('author_timezone', tz)
 
         commit_data = {
             'parents': [p._commit['id'] for p in self.parents if p],
             'author': safe_str(author),
             'committer': safe_str(author),
             'encoding': ENCODING,
             'message': safe_str(message),
             'commit_time': int(date),
             'author_time': int(author_time),
             'commit_timezone': tz,
             'author_timezone': author_tz,
         }
 
         commit_id = self.repository._remote.commit(
             commit_data, branch, commit_tree, updated, removed)
 
         # Update vcs repository object
         self.repository.append_commit_id(commit_id)
 
         # invalidate parsed refs after commit
         self.repository._refs = self.repository._get_refs()
         self.repository.branches = self.repository._get_branches()
         tip = self.repository.get_commit(commit_id)
 
         self.reset()
         return tip

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

@@ -1,385 +1,389 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 is_node_binary(self, path):
+        path = self._get_filectx(path)
+        return self._remote.is_binary(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

@@ -1,241 +1,245 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 is_node_binary(self, path):
+        path = self._fix_path(path)
+        return self._remote.is_binary(self._svn_rev, safe_str(path))
+
     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/nodes.py

@@ -1,870 +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")
+        raise NodeError("Cannot retrieve stream_bytes 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
+        if self.commit:
+            return self.commit.is_node_binary(self.path)
+        else:
+            raw_bytes = self._content
+            return raw_bytes and '\0' in raw_bytes
 
     @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")
+                    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