scm: change ._get_content() to .raw_bytes attribute to file nodes to

allow getting raw bytes instead of text content of a file and added
.md5 and .is_binary attributes to file nodes

rhodecode/lib/vcs/backends/base.py

@@ -1,1502 +1,1502 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2016  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 collections
 import datetime
 import itertools
 import logging
 import os
 import time
 import warnings
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from rhodecode.lib.utils2 import safe_str, safe_unicode
 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 = 0100644
 FILEMODE_EXECUTABLE = 0100755
 
 Reference = collections.namedtuple('Reference', ('type', 'name', 'commit_id'))
 MergeResponse = collections.namedtuple(
     'MergeResponse',
     ('possible', 'executed', 'merge_commit_id', 'failure_reason'))
 
 
 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
 
     # A involved commit could not be found.
     MISSING_COMMIT = 8
 
 
 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
 
     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)
 
     @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
         """
         raise NotImplementedError
 
     @LazyProperty
     def branches(self):
         """
         A `dict` which maps branch 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 is_empty(self):
         return not bool(self.commit_ids)
 
     @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
     # ==========================================================================
 
     def get_commit(self, commit_id=None, commit_idx=None, pre_load=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, pre_load=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 pre_load:
         """
         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`.
         """
         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, target_ref, source_repo, source_ref, workspace_id,
               user_name='', user_email='', message='', dry_run=False,
               use_rebase=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 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 workspace_id: `workspace_id` unique identifier.
         :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.
         """
         if dry_run:
             message = message or 'sample_message'
             user_email = user_email or '[e-mail]'
             user_name = user_name or 'user name'
         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')
 
         shadow_repository_path = self._maybe_prepare_merge_workspace(
             workspace_id, target_ref)
 
         try:
             return self._merge_repo(
                 shadow_repository_path, target_ref, source_repo,
                 source_ref, message, user_name, user_email, dry_run=dry_run,
                 use_rebase=use_rebase)
         except RepositoryError:
             log.exception(
                 'Unexpected failure when running merge, dry-run=%s',
                 dry_run)
             return MergeResponse(
                 False, False, None, MergeFailureReason.UNKNOWN)
 
     def _merge_repo(self, shadow_repository_path, target_ref,
                     source_repo, source_ref, merge_message,
                     merger_name, merger_email, dry_run=False):
         """Internal implementation of merge."""
         raise NotImplementedError
 
     def _maybe_prepare_merge_workspace(self, workspace_id, target_ref):
         """
         Create the merge workspace.
 
         :param workspace_id: `workspace_id` unique identifier.
         """
         raise NotImplementedError
 
     def cleanup_merge_workspace(self, 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.
         """
         raise NotImplementedError
 
     # ========== #
     # 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, basestring):
             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, basestring):
             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, basestring):
             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
 
 
 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):
         return 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
         }
 
     @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
 
     @property
     def merge(self):
         """
         Returns boolean if commit is a merge.
         """
         return len(self.parents) > 1
 
     @LazyProperty
     def children(self):
         """
         Returns list of child commits.
         """
         raise NotImplementedError
 
     @LazyProperty
     def id(self):
         """
         Returns string identifying this commit.
         """
         raise NotImplementedError
 
     @LazyProperty
     def raw_id(self):
         """
         Returns raw string identifying this commit.
         """
         raise NotImplementedError
 
     @LazyProperty
     def short_id(self):
         """
         Returns shortened version of ``raw_id`` attribute, as string,
         identifying this commit, useful for presentation to users.
         """
         raise NotImplementedError
 
     @LazyProperty
     def idx(self):
         """
         Returns integer identifying this commit.
         """
         raise NotImplementedError
 
     @LazyProperty
     def committer(self):
         """
         Returns committer for this commit
         """
         raise NotImplementedError
 
     @LazyProperty
     def committer_name(self):
         """
         Returns committer name for this commit
         """
 
         return author_name(self.committer)
 
     @LazyProperty
     def committer_email(self):
         """
         Returns committer email address for this commit
         """
 
         return author_email(self.committer)
 
     @LazyProperty
     def author(self):
         """
         Returns author for this commit
         """
 
         raise NotImplementedError
 
     @LazyProperty
     def author_name(self):
         """
         Returns author name for this commit
         """
 
         return author_name(self.author)
 
     @LazyProperty
     def author_email(self):
         """
         Returns author email address for this commit
         """
 
         return author_email(self.author)
 
     def get_file_mode(self, path):
         """
         Returns stat mode of the file at `path`.
         """
         raise NotImplementedError
 
     def is_link(self, path):
         """
         Returns ``True`` if given `path` is a symlink
         """
         raise NotImplementedError
 
     def get_file_content(self, path):
         """
         Returns content of the file at the given `path`.
         """
         raise NotImplementedError
 
     def get_file_size(self, path):
         """
         Returns size of the file at the given `path`.
         """
         raise NotImplementedError
 
     def get_file_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.
         """
         return self.get_file_history(path, limit=1, pre_load=pre_load)[0]
 
     def get_file_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 storage.
         """
         raise NotImplementedError
 
     def archive_repo(self, file_path, kind='tgz', subrepos=None,
                      prefix=None, write_metadata=False, mtime=None):
         """
         Creates an archive containing the contents of the repository.
 
         :param file_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.
 
         :raise VCSError: If prefix has a problem.
         """
         allowed_kinds = settings.ARCHIVE_SPECS.keys()
         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 or time.time()
 
         file_info = []
         cur_rev = self.repository.get_commit(commit_id=self.raw_id)
         for _r, _d, files in cur_rev.walk('/'):
             for f in files:
                 f_path = os.path.join(prefix, f.path)
                 file_info.append(
-                    (f_path, f.mode, f.is_link(), f._get_content()))
+                    (f_path, f.mode, f.is_link(), f.raw_bytes))
 
         if write_metadata:
             metadata = [
                 ('repo_name', self.repository.name),
                 ('rev', self.raw_id),
                 ('create_time', 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', 0644, False, '\n'.join(meta)))
 
         connection.Hg.archive_repo(file_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.parents[0] if self.parents else self.repository.EMPTY_COMMIT)
         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(
             "There is no file nor directory at the given path: "
             "'%s' at commit %s" % (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_file_commit instead", DeprecationWarning)
         return self.get_file_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='0' * 40, 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]
 
     def get_file_commit(self, path):
         return self
 
     def get_file_content(self, path):
         return u''
 
     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='0' * 40, 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 CollectionGenerator(object):
 
     def __init__(self, repo, commit_ids, collection_size=None, pre_load=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
 
     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)
 
     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)
 
     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 values) 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`.
     """
 
     _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
         """
         chunks = ('\n' + self.raw).split('\ndiff --git')[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

rhodecode/lib/vcs/nodes.py

@@ -1,738 +1,740 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2014-2016  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 stat
 
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 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.
     """
 
     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 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:
             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`.
     """
 
     def __init__(self, path, content=None, commit=None, mode=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
 
     @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
 
-    def _get_content(self):
+    @LazyProperty
+    def raw_bytes(self):
+        """
+        Returns lazily the raw bytes of the FileNode.
+        """
         if self.commit:
-            content = self.commit.get_file_content(self.path)
+            if self._content is None:
+                self._content = self.commit.get_file_content(self.path)
+            content = self._content
         else:
             content = self._content
         return content
 
-    @property
+    @LazyProperty
+    def md5(self):
+        """
+        Returns md5 of the file node.
+        """
+        return md5(self.raw_bytes)
+
+    @LazyProperty
     def content(self):
         """
         Returns lazily content of the FileNode. If possible, would try to
         decode content from UTF-8.
         """
-        content = self._get_content()
+        content = self.raw_bytes
 
-        if bool(content and '\0' in content):
+        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"]
             return self.commit.get_file_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]
 
     @LazyProperty
     def lexer(self):
         """
         Returns pygment's lexer class. Would try to guess lexer taking file's
         content, name and mimetype.
         """
         from pygments import lexers
         try:
             lexer = lexers.guess_lexer_for_filename(self.name, self.content, stripnl=False)
         except lexers.ClassNotFound:
             lexer = lexers.TextLexer(stripnl=False)
         # returns first alias
         return lexer
 
     @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_file_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"]
         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
 
-    @property
+    @LazyProperty
     def is_binary(self):
         """
         Returns True if file has binary content.
         """
-        _bin = '\0' in self._get_content()
+        _bin = self.raw_bytes and '\0' in self.raw_bytes
         return _bin
 
     @LazyProperty
     def extension(self):
         """Returns filenode extension"""
         return self.name.split('.')[-1]
 
     @property
     def is_executable(self):
         """
         Returns ``True`` if file has executable flag turned on.
         """
         return bool(self.mode & stat.S_IXUSR)
 
     def get_largefile_node(self):
         """
         Try to return a Mercurial FileNode from this node. It does internal
         checks inside largefile store, if that file exist there it will
         create special instance of LargeFileNode which can get content from
         LF store.
         """
         if self.commit and self.path.startswith(LARGEFILE_PREFIX):
             largefile_path = self.path.split(LARGEFILE_PREFIX)[-1].lstrip('/')
             return self.commit.get_largefile_node(largefile_path)
 
     def lines(self, count_empty=False):
         all_lines, empty_lines = 0, 0
 
         if not self.is_binary:
-            content = self._get_content()
+            content = self.content
             if count_empty:
                 all_lines = 0
                 empty_lines = 0
                 for line in content.splitlines(True):
                     if line == '\n':
                         empty_lines += 1
                     all_lines += 1
 
                 return all_lines, all_lines - empty_lines
             else:
                 # fast method
                 empty_lines = all_lines = content.count('\n')
                 if all_lines == 0 and content:
                     # one-line without a newline
                     empty_lines = all_lines = 1
 
         return all_lines, empty_lines
 
     def __repr__(self):
         return '<%s %r @ %s>' % (self.__class__.__name__, self.path,
                                  getattr(self.commit, 'short_id', ''))
 
 
 class RemovedFileNode(FileNode):
     """
     Dummy FileNode class - trying to access any public attribute except path,
     name, kind or state (or methods/attributes checking those two) would raise
     RemovedFileNodeError.
     """
     ALLOWED_ATTRIBUTES = [
         'name', 'path', 'state', 'is_root', 'is_file', 'is_dir', 'kind',
         'added', 'changed', 'not_changed', 'removed'
     ]
 
     def __init__(self, path):
         """
         :param path: relative path to the node
         """
         super(RemovedFileNode, self).__init__(path=path)
 
     def __getattribute__(self, attr):
         if attr.startswith('_') or attr in RemovedFileNode.ALLOWED_ATTRIBUTES:
             return super(RemovedFileNode, self).__getattribute__(attr)
         raise RemovedFileNodeError(
             "Cannot access attribute %s on RemovedFileNode" % attr)
 
     @LazyProperty
     def state(self):
         return NodeState.REMOVED
 
 
 class DirNode(Node):
     """
     DirNode stores list of files and directories within this node.
     Nodes may be used standalone but within repository context they
     lazily fetch data within same repositorty's commit.
     """
 
     def __init__(self, path, nodes=(), commit=None):
         """
         Only one of ``nodes`` and ``commit`` may be given. Passing both
         would raise ``NodeError`` exception.
 
         :param path: relative path to the node
         :param nodes: content may be passed to constructor
         :param commit: if given, will use it to lazily fetch content
         """
         if nodes and commit:
             raise NodeError("Cannot use both nodes and commit")
         super(DirNode, self).__init__(path, NodeKind.DIR)
         self.commit = commit
         self._nodes = nodes
 
     @LazyProperty
     def content(self):
         raise NodeError(
             "%s represents a dir and has no `content` attribute" % self)
 
     @LazyProperty
     def nodes(self):
         if self.commit:
             nodes = self.commit.get_nodes(self.path)
         else:
             nodes = self._nodes
         self._nodes_dict = dict((node.path, node) for node in nodes)
         return sorted(nodes)
 
     @LazyProperty
     def files(self):
         return sorted((node for node in self.nodes if node.is_file()))
 
     @LazyProperty
     def dirs(self):
         return sorted((node for node in self.nodes if node.is_dir()))
 
     def __iter__(self):
         for node in self.nodes:
             yield node
 
     def get_node(self, path):
         """
         Returns node from within this particular ``DirNode``, so it is now
         allowed to fetch, i.e. node located at 'docs/api/index.rst' from node
         'docs'. In order to access deeper nodes one must fetch nodes between
         them first - this would work::
 
            docs = root.get_node('docs')
            docs.get_node('api').get_node('index.rst')
 
         :param: path - relative to the current node
 
         .. note::
            To access lazily (as in example above) node have to be initialized
            with related commit object - without it node is out of
            context and may know nothing about anything else than nearest
            (located at same level) nodes.
         """
         try:
             path = path.rstrip('/')
             if path == '':
                 raise NodeError("Cannot retrieve node without path")
             self.nodes  # access nodes first in order to set _nodes_dict
             paths = path.split('/')
             if len(paths) == 1:
                 if not self.is_root():
                     path = '/'.join((self.path, paths[0]))
                 else:
                     path = paths[0]
                 return self._nodes_dict[path]
             elif len(paths) > 1:
                 if self.commit is None:
                     raise NodeError(
                         "Cannot access deeper nodes without commit")
                 else:
                     path1, path2 = paths[0], '/'.join(paths[1:])
                     return self.get_node(path1).get_node(path2)
             else:
                 raise KeyError
         except KeyError:
             raise NodeError("Node does not exist at %s" % path)
 
     @LazyProperty
     def state(self):
         raise NodeError("Cannot access state of DirNode")
 
     @LazyProperty
     def size(self):
         size = 0
         for root, dirs, files in self.commit.walk(self.path):
             for f in files:
                 size += f.size
 
         return size
 
     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 _validate_path(self, path):
         """
         we override check since the LargeFileNode path is system absolute
         """
 
-    def _get_content(self):
+    def raw_bytes(self):
         if self.commit:
             with open(self.path, 'rb') as f:
                 content = f.read()
         else:
             content = self._content
-        return content
-
-    @property
-    def content(self):
-        """
-        Returns lazily content of the `FileNode`. If possible, would try to
-        decode content from UTF-8.
-        """
-        content = self._get_content()
-
-        if bool(content and '\0' in content):
-            return content
-        return safe_unicode(content)
+        return content
  No newline at end of file

rhodecode/model/scm.py

@@ -1,1101 +1,1099 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2010-2016  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/
 
 """
 Scm model for RhodeCode
 """
 
 import os.path
 import re
 import sys
 import time
 import traceback
 import logging
 import cStringIO
 import pkg_resources
 
 import pylons
 from pylons.i18n.translation import _
 from sqlalchemy import func
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 import rhodecode
 from rhodecode.lib.vcs import get_backend
 from rhodecode.lib.vcs.exceptions import RepositoryError, NodeNotChangedError
 from rhodecode.lib.vcs.nodes import FileNode
 from rhodecode.lib.vcs.backends.base import EmptyCommit
 from rhodecode.lib import helpers as h
 
 from rhodecode.lib.auth import (
     HasRepoPermissionAny, HasRepoGroupPermissionAny,
     HasUserGroupPermissionAny)
 from rhodecode.lib.exceptions import NonRelativePathError, IMCCommitError
 from rhodecode.lib import hooks_utils, caches
 from rhodecode.lib.utils import (
     get_filesystem_repos, action_logger, make_db_config)
 from rhodecode.lib.utils2 import (
     safe_str, safe_unicode, get_server_url, md5)
 from rhodecode.model import BaseModel
 from rhodecode.model.db import (
     Repository, CacheKey, UserFollowing, UserLog, User, RepoGroup,
     PullRequest, DbMigrateVersion)
 from rhodecode.model.settings import VcsSettingsModel
 
 log = logging.getLogger(__name__)
 
 
 class UserTemp(object):
     def __init__(self, user_id):
         self.user_id = user_id
 
     def __repr__(self):
         return "<%s('id:%s')>" % (self.__class__.__name__, self.user_id)
 
 
 class RepoTemp(object):
     def __init__(self, repo_id):
         self.repo_id = repo_id
 
     def __repr__(self):
         return "<%s('id:%s')>" % (self.__class__.__name__, self.repo_id)
 
 
 class SimpleCachedRepoList(object):
     """
     Lighter version of of iteration of repos without the scm initialisation,
     and with cache usage
     """
     def __init__(self, db_repo_list, repos_path, order_by=None, perm_set=None):
         self.db_repo_list = db_repo_list
         self.repos_path = repos_path
         self.order_by = order_by
         self.reversed = (order_by or '').startswith('-')
         if not perm_set:
             perm_set = ['repository.read', 'repository.write',
                         'repository.admin']
         self.perm_set = perm_set
 
     def __len__(self):
         return len(self.db_repo_list)
 
     def __repr__(self):
         return '<%s (%s)>' % (self.__class__.__name__, self.__len__())
 
     def __iter__(self):
         for dbr in self.db_repo_list:
             # check permission at this level
             has_perm = HasRepoPermissionAny(*self.perm_set)(
                 dbr.repo_name, 'SimpleCachedRepoList check')
             if not has_perm:
                 continue
 
             tmp_d = {
                 'name': dbr.repo_name,
                 'dbrepo': dbr.get_dict(),
                 'dbrepo_fork': dbr.fork.get_dict() if dbr.fork else {}
             }
             yield tmp_d
 
 
 class _PermCheckIterator(object):
 
     def __init__(
             self, obj_list, obj_attr, perm_set, perm_checker,
             extra_kwargs=None):
         """
         Creates iterator from given list of objects, additionally
         checking permission for them from perm_set var
 
         :param obj_list: list of db objects
         :param obj_attr: attribute of object to pass into perm_checker
         :param perm_set: list of permissions to check
         :param perm_checker: callable to check permissions against
         """
         self.obj_list = obj_list
         self.obj_attr = obj_attr
         self.perm_set = perm_set
         self.perm_checker = perm_checker
         self.extra_kwargs = extra_kwargs or {}
 
     def __len__(self):
         return len(self.obj_list)
 
     def __repr__(self):
         return '<%s (%s)>' % (self.__class__.__name__, self.__len__())
 
     def __iter__(self):
         checker = self.perm_checker(*self.perm_set)
         for db_obj in self.obj_list:
             # check permission at this level
             name = getattr(db_obj, self.obj_attr, None)
             if not checker(name, self.__class__.__name__, **self.extra_kwargs):
                 continue
 
             yield db_obj
 
 
 class RepoList(_PermCheckIterator):
 
     def __init__(self, db_repo_list, perm_set=None, extra_kwargs=None):
         if not perm_set:
             perm_set = [
                 'repository.read', 'repository.write', 'repository.admin']
 
         super(RepoList, self).__init__(
             obj_list=db_repo_list,
             obj_attr='repo_name', perm_set=perm_set,
             perm_checker=HasRepoPermissionAny,
             extra_kwargs=extra_kwargs)
 
 
 class RepoGroupList(_PermCheckIterator):
 
     def __init__(self, db_repo_group_list, perm_set=None, extra_kwargs=None):
         if not perm_set:
             perm_set = ['group.read', 'group.write', 'group.admin']
 
         super(RepoGroupList, self).__init__(
             obj_list=db_repo_group_list,
             obj_attr='group_name', perm_set=perm_set,
             perm_checker=HasRepoGroupPermissionAny,
             extra_kwargs=extra_kwargs)
 
 
 class UserGroupList(_PermCheckIterator):
 
     def __init__(self, db_user_group_list, perm_set=None, extra_kwargs=None):
         if not perm_set:
             perm_set = ['usergroup.read', 'usergroup.write', 'usergroup.admin']
 
         super(UserGroupList, self).__init__(
             obj_list=db_user_group_list,
             obj_attr='users_group_name', perm_set=perm_set,
             perm_checker=HasUserGroupPermissionAny,
             extra_kwargs=extra_kwargs)
 
 
 class ScmModel(BaseModel):
     """
     Generic Scm Model
     """
 
     @LazyProperty
     def repos_path(self):
         """
         Gets the repositories root path from database
         """
 
         settings_model = VcsSettingsModel(sa=self.sa)
         return settings_model.get_repos_location()
 
     def repo_scan(self, repos_path=None):
         """
         Listing of repositories in given path. This path should not be a
         repository itself. Return a dictionary of repository objects
 
         :param repos_path: path to directory containing repositories
         """
 
         if repos_path is None:
             repos_path = self.repos_path
 
         log.info('scanning for repositories in %s', repos_path)
 
         config = make_db_config()
         config.set('extensions', 'largefiles', '')
         repos = {}
 
         for name, path in get_filesystem_repos(repos_path, recursive=True):
             # name need to be decomposed and put back together using the /
             # since this is internal storage separator for rhodecode
             name = Repository.normalize_repo_name(name)
 
             try:
                 if name in repos:
                     raise RepositoryError('Duplicate repository name %s '
                                           'found in %s' % (name, path))
                 elif path[0] in rhodecode.BACKENDS:
                     klass = get_backend(path[0])
                     repos[name] = klass(path[1], config=config)
             except OSError:
                 continue
         log.debug('found %s paths with repositories', len(repos))
         return repos
 
     def get_repos(self, all_repos=None, sort_key=None):
         """
         Get all repositories from db and for each repo create it's
         backend instance and fill that backed with information from database
 
         :param all_repos: list of repository names as strings
             give specific repositories list, good for filtering
 
         :param sort_key: initial sorting of repositories
         """
         if all_repos is None:
             all_repos = self.sa.query(Repository)\
                 .filter(Repository.group_id == None)\
                 .order_by(func.lower(Repository.repo_name)).all()
         repo_iter = SimpleCachedRepoList(
             all_repos, repos_path=self.repos_path, order_by=sort_key)
         return repo_iter
 
     def get_repo_groups(self, all_groups=None):
         if all_groups is None:
             all_groups = RepoGroup.query()\
                 .filter(RepoGroup.group_parent_id == None).all()
         return [x for x in RepoGroupList(all_groups)]
 
     def mark_for_invalidation(self, repo_name, delete=False):
         """
         Mark caches of this repo invalid in the database. `delete` flag
         removes the cache entries
 
         :param repo_name: the repo_name for which caches should be marked
             invalid, or deleted
         :param delete: delete the entry keys instead of setting bool
             flag on them
         """
         CacheKey.set_invalidate(repo_name, delete=delete)
         repo = Repository.get_by_repo_name(repo_name)
 
         if repo:
             config = repo._config
             config.set('extensions', 'largefiles', '')
             repo.update_commit_cache(config=config, cs_cache=None)
             caches.clear_repo_caches(repo_name)
 
     def toggle_following_repo(self, follow_repo_id, user_id):
 
         f = self.sa.query(UserFollowing)\
             .filter(UserFollowing.follows_repo_id == follow_repo_id)\
             .filter(UserFollowing.user_id == user_id).scalar()
 
         if f is not None:
             try:
                 self.sa.delete(f)
                 action_logger(UserTemp(user_id),
                               'stopped_following_repo',
                               RepoTemp(follow_repo_id))
                 return
             except Exception:
                 log.error(traceback.format_exc())
                 raise
 
         try:
             f = UserFollowing()
             f.user_id = user_id
             f.follows_repo_id = follow_repo_id
             self.sa.add(f)
 
             action_logger(UserTemp(user_id),
                           'started_following_repo',
                           RepoTemp(follow_repo_id))
         except Exception:
             log.error(traceback.format_exc())
             raise
 
     def toggle_following_user(self, follow_user_id, user_id):
         f = self.sa.query(UserFollowing)\
             .filter(UserFollowing.follows_user_id == follow_user_id)\
             .filter(UserFollowing.user_id == user_id).scalar()
 
         if f is not None:
             try:
                 self.sa.delete(f)
                 return
             except Exception:
                 log.error(traceback.format_exc())
                 raise
 
         try:
             f = UserFollowing()
             f.user_id = user_id
             f.follows_user_id = follow_user_id
             self.sa.add(f)
         except Exception:
             log.error(traceback.format_exc())
             raise
 
     def is_following_repo(self, repo_name, user_id, cache=False):
         r = self.sa.query(Repository)\
             .filter(Repository.repo_name == repo_name).scalar()
 
         f = self.sa.query(UserFollowing)\
             .filter(UserFollowing.follows_repository == r)\
             .filter(UserFollowing.user_id == user_id).scalar()
 
         return f is not None
 
     def is_following_user(self, username, user_id, cache=False):
         u = User.get_by_username(username)
 
         f = self.sa.query(UserFollowing)\
             .filter(UserFollowing.follows_user == u)\
             .filter(UserFollowing.user_id == user_id).scalar()
 
         return f is not None
 
     def get_followers(self, repo):
         repo = self._get_repo(repo)
 
         return self.sa.query(UserFollowing)\
             .filter(UserFollowing.follows_repository == repo).count()
 
     def get_forks(self, repo):
         repo = self._get_repo(repo)
         return self.sa.query(Repository)\
             .filter(Repository.fork == repo).count()
 
     def get_pull_requests(self, repo):
         repo = self._get_repo(repo)
         return self.sa.query(PullRequest)\
             .filter(PullRequest.target_repo == repo)\
             .filter(PullRequest.status != PullRequest.STATUS_CLOSED).count()
 
     def mark_as_fork(self, repo, fork, user):
         repo = self._get_repo(repo)
         fork = self._get_repo(fork)
         if fork and repo.repo_id == fork.repo_id:
             raise Exception("Cannot set repository as fork of itself")
 
         if fork and repo.repo_type != fork.repo_type:
             raise RepositoryError(
                 "Cannot set repository as fork of repository with other type")
 
         repo.fork = fork
         self.sa.add(repo)
         return repo
 
     def pull_changes(self, repo, username):
         dbrepo = self._get_repo(repo)
         clone_uri = dbrepo.clone_uri
         if not clone_uri:
             raise Exception("This repository doesn't have a clone uri")
 
         repo = dbrepo.scm_instance(cache=False)
         # TODO: marcink fix this an re-enable since we need common logic
         # for hg/git remove hooks so we don't trigger them on fetching
         # commits from remote
         repo.config.clear_section('hooks')
 
         repo_name = dbrepo.repo_name
         try:
             # TODO: we need to make sure those operations call proper hooks !
             repo.pull(clone_uri)
 
             self.mark_for_invalidation(repo_name)
         except Exception:
             log.error(traceback.format_exc())
             raise
 
     def commit_change(self, repo, repo_name, commit, user, author, message,
                       content, f_path):
         """
         Commits changes
 
         :param repo: SCM instance
 
         """
         user = self._get_user(user)
 
         # decoding here will force that we have proper encoded values
         # in any other case this will throw exceptions and deny commit
         content = safe_str(content)
         path = safe_str(f_path)
         # message and author needs to be unicode
         # proper backend should then translate that into required type
         message = safe_unicode(message)
         author = safe_unicode(author)
         imc = repo.in_memory_commit
         imc.change(FileNode(path, content, mode=commit.get_file_mode(f_path)))
         try:
             # TODO: handle pre-push action !
             tip = imc.commit(
                 message=message, author=author, parents=[commit],
                 branch=commit.branch)
         except Exception as e:
             log.error(traceback.format_exc())
             raise IMCCommitError(str(e))
         finally:
             # always clear caches, if commit fails we want fresh object also
             self.mark_for_invalidation(repo_name)
 
         # We trigger the post-push action
         hooks_utils.trigger_post_push_hook(
             username=user.username, action='push_local', repo_name=repo_name,
             repo_alias=repo.alias, commit_ids=[tip.raw_id])
         return tip
 
     def _sanitize_path(self, f_path):
         if f_path.startswith('/') or f_path.startswith('./') or '../' in f_path:
             raise NonRelativePathError('%s is not an relative path' % f_path)
         if f_path:
             f_path = os.path.normpath(f_path)
         return f_path
 
     def get_dirnode_metadata(self, commit, dir_node):
         if not dir_node.is_dir():
             return []
 
         data = []
         for node in dir_node:
             if not node.is_file():
                 # we skip file-nodes
                 continue
 
             last_commit = node.last_commit
             last_commit_date = last_commit.date
             data.append({
                 'name': node.name,
                 'size': h.format_byte_size_binary(node.size),
                 'modified_at': h.format_date(last_commit_date),
                 'modified_ts': last_commit_date.isoformat(),
                 'revision': last_commit.revision,
                 'short_id': last_commit.short_id,
                 'message': h.escape(last_commit.message),
                 'author': h.escape(last_commit.author),
                 'user_profile': h.gravatar_with_user(last_commit.author),
             })
 
         return data
 
     def get_nodes(self, repo_name, commit_id, root_path='/', flat=True,
                   extended_info=False, content=False):
         """
         recursive walk in root dir and return a set of all path in that dir
         based on repository walk function
 
         :param repo_name: name of repository
         :param commit_id: commit id for which to list nodes
         :param root_path: root path to list
-        :param flat: return as a list, if False returns a dict with decription
+        :param flat: return as a list, if False returns a dict with description
 
         """
         _files = list()
         _dirs = list()
         try:
             _repo = self._get_repo(repo_name)
             commit = _repo.scm_instance().get_commit(commit_id=commit_id)
             root_path = root_path.lstrip('/')
             for __, dirs, files in commit.walk(root_path):
                 for f in files:
                     _content = None
                     _data = f.unicode_path
+
                     if not flat:
                         _data = {
                             "name": f.unicode_path,
                             "type": "file",
                             }
                         if extended_info:
-                            _content = safe_str(f.content)
                             _data.update({
-                                "md5": md5(_content),
+                                "md5": f.md5,
                                 "binary": f.is_binary,
                                 "size": f.size,
                                 "extension": f.extension,
-
                                 "mimetype": f.mimetype,
                                 "lines": f.lines()[0]
                             })
+
                         if content:
                             full_content = None
                             if not f.is_binary:
-                                # in case we loaded the _content already
-                                # re-use it, or load from f[ile]
-                                full_content = _content or safe_str(f.content)
+                                full_content = safe_str(f.content)
 
                             _data.update({
-                                "content": full_content
+                                "content": full_content,
                             })
                     _files.append(_data)
                 for d in dirs:
                     _data = d.unicode_path
                     if not flat:
                         _data = {
                             "name": d.unicode_path,
                             "type": "dir",
                             }
                     if extended_info:
                         _data.update({
                             "md5": None,
                             "binary": None,
                             "size": None,
                             "extension": None,
                         })
                     if content:
                         _data.update({
                             "content": None
                         })
                     _dirs.append(_data)
         except RepositoryError:
             log.debug("Exception in get_nodes", exc_info=True)
             raise
 
         return _dirs, _files
 
     def create_nodes(self, user, repo, message, nodes, parent_commit=None,
                      author=None, trigger_push_hook=True):
         """
         Commits given multiple nodes into repo
 
         :param user: RhodeCode User object or user_id, the commiter
         :param repo: RhodeCode Repository object
         :param message: commit message
         :param nodes: mapping {filename:{'content':content},...}
         :param parent_commit: parent commit, can be empty than it's
            initial commit
         :param author: author of commit, cna be different that commiter
            only for git
         :param trigger_push_hook: trigger push hooks
 
         :returns: new commited commit
         """
 
         user = self._get_user(user)
         scm_instance = repo.scm_instance(cache=False)
 
         processed_nodes = []
         for f_path in nodes:
             f_path = self._sanitize_path(f_path)
             content = nodes[f_path]['content']
             f_path = safe_str(f_path)
             # decoding here will force that we have proper encoded values
             # in any other case this will throw exceptions and deny commit
             if isinstance(content, (basestring,)):
                 content = safe_str(content)
             elif isinstance(content, (file, cStringIO.OutputType,)):
                 content = content.read()
             else:
                 raise Exception('Content is of unrecognized type %s' % (
                     type(content)
                 ))
             processed_nodes.append((f_path, content))
 
         message = safe_unicode(message)
         commiter = user.full_contact
         author = safe_unicode(author) if author else commiter
 
         imc = scm_instance.in_memory_commit
 
         if not parent_commit:
             parent_commit = EmptyCommit(alias=scm_instance.alias)
 
         if isinstance(parent_commit, EmptyCommit):
             # EmptyCommit means we we're editing empty repository
             parents = None
         else:
             parents = [parent_commit]
         # add multiple nodes
         for path, content in processed_nodes:
             imc.add(FileNode(path, content=content))
         # TODO: handle pre push scenario
         tip = imc.commit(message=message,
                          author=author,
                          parents=parents,
                          branch=parent_commit.branch)
 
         self.mark_for_invalidation(repo.repo_name)
         if trigger_push_hook:
             hooks_utils.trigger_post_push_hook(
                 username=user.username, action='push_local',
                 repo_name=repo.repo_name, repo_alias=scm_instance.alias,
                 commit_ids=[tip.raw_id])
         return tip
 
     def update_nodes(self, user, repo, message, nodes, parent_commit=None,
                      author=None, trigger_push_hook=True):
         user = self._get_user(user)
         scm_instance = repo.scm_instance(cache=False)
 
         message = safe_unicode(message)
         commiter = user.full_contact
         author = safe_unicode(author) if author else commiter
 
         imc = scm_instance.in_memory_commit
 
         if not parent_commit:
             parent_commit = EmptyCommit(alias=scm_instance.alias)
 
         if isinstance(parent_commit, EmptyCommit):
             # EmptyCommit means we we're editing empty repository
             parents = None
         else:
             parents = [parent_commit]
 
         # add multiple nodes
         for _filename, data in nodes.items():
             # new filename, can be renamed from the old one, also sanitaze
             # the path for any hack around relative paths like ../../ etc.
             filename = self._sanitize_path(data['filename'])
             old_filename = self._sanitize_path(_filename)
             content = data['content']
 
             filenode = FileNode(old_filename, content=content)
             op = data['op']
             if op == 'add':
                 imc.add(filenode)
             elif op == 'del':
                 imc.remove(filenode)
             elif op == 'mod':
                 if filename != old_filename:
                     # TODO: handle renames more efficient, needs vcs lib
                     # changes
                     imc.remove(filenode)
                     imc.add(FileNode(filename, content=content))
                 else:
                     imc.change(filenode)
 
         try:
             # TODO: handle pre push scenario
             # commit changes
             tip = imc.commit(message=message,
                              author=author,
                              parents=parents,
                              branch=parent_commit.branch)
         except NodeNotChangedError:
             raise
         except Exception as e:
             log.exception("Unexpected exception during call to imc.commit")
             raise IMCCommitError(str(e))
         finally:
             # always clear caches, if commit fails we want fresh object also
             self.mark_for_invalidation(repo.repo_name)
 
         if trigger_push_hook:
             hooks_utils.trigger_post_push_hook(
                 username=user.username, action='push_local',
                 repo_name=repo.repo_name, repo_alias=scm_instance.alias,
                 commit_ids=[tip.raw_id])
 
     def delete_nodes(self, user, repo, message, nodes, parent_commit=None,
                      author=None, trigger_push_hook=True):
         """
         Deletes given multiple nodes into `repo`
 
         :param user: RhodeCode User object or user_id, the committer
         :param repo: RhodeCode Repository object
         :param message: commit message
         :param nodes: mapping {filename:{'content':content},...}
         :param parent_commit: parent commit, can be empty than it's initial
            commit
         :param author: author of commit, cna be different that commiter only
            for git
         :param trigger_push_hook: trigger push hooks
 
         :returns: new commit after deletion
         """
 
         user = self._get_user(user)
         scm_instance = repo.scm_instance(cache=False)
 
         processed_nodes = []
         for f_path in nodes:
             f_path = self._sanitize_path(f_path)
             # content can be empty but for compatabilty it allows same dicts
             # structure as add_nodes
             content = nodes[f_path].get('content')
             processed_nodes.append((f_path, content))
 
         message = safe_unicode(message)
         commiter = user.full_contact
         author = safe_unicode(author) if author else commiter
 
         imc = scm_instance.in_memory_commit
 
         if not parent_commit:
             parent_commit = EmptyCommit(alias=scm_instance.alias)
 
         if isinstance(parent_commit, EmptyCommit):
             # EmptyCommit means we we're editing empty repository
             parents = None
         else:
             parents = [parent_commit]
         # add multiple nodes
         for path, content in processed_nodes:
             imc.remove(FileNode(path, content=content))
 
         # TODO: handle pre push scenario
         tip = imc.commit(message=message,
                          author=author,
                          parents=parents,
                          branch=parent_commit.branch)
 
         self.mark_for_invalidation(repo.repo_name)
         if trigger_push_hook:
             hooks_utils.trigger_post_push_hook(
                 username=user.username, action='push_local',
                 repo_name=repo.repo_name, repo_alias=scm_instance.alias,
                 commit_ids=[tip.raw_id])
         return tip
 
     def strip(self, repo, commit_id, branch):
         scm_instance = repo.scm_instance(cache=False)
         scm_instance.config.clear_section('hooks')
         scm_instance.strip(commit_id, branch)
         self.mark_for_invalidation(repo.repo_name)
 
     def get_unread_journal(self):
         return self.sa.query(UserLog).count()
 
     def get_repo_landing_revs(self, repo=None):
         """
         Generates select option with tags branches and bookmarks (for hg only)
         grouped by type
 
         :param repo:
         """
 
         hist_l = []
         choices = []
         repo = self._get_repo(repo)
         hist_l.append(['rev:tip', _('latest tip')])
         choices.append('rev:tip')
         if not repo:
             return choices, hist_l
 
         repo = repo.scm_instance()
 
         branches_group = (
             [(u'branch:%s' % safe_unicode(b), safe_unicode(b))
                 for b in repo.branches],
             _("Branches"))
         hist_l.append(branches_group)
         choices.extend([x[0] for x in branches_group[0]])
 
         if repo.alias == 'hg':
             bookmarks_group = (
                 [(u'book:%s' % safe_unicode(b), safe_unicode(b))
                     for b in repo.bookmarks],
                 _("Bookmarks"))
             hist_l.append(bookmarks_group)
             choices.extend([x[0] for x in bookmarks_group[0]])
 
         tags_group = (
             [(u'tag:%s' % safe_unicode(t), safe_unicode(t))
                 for t in repo.tags],
             _("Tags"))
         hist_l.append(tags_group)
         choices.extend([x[0] for x in tags_group[0]])
 
         return choices, hist_l
 
     def install_git_hook(self, repo, force_create=False):
         """
         Creates a rhodecode hook inside a git repository
 
         :param repo: Instance of VCS repo
         :param force_create: Create even if same name hook exists
         """
 
         loc = os.path.join(repo.path, 'hooks')
         if not repo.bare:
             loc = os.path.join(repo.path, '.git', 'hooks')
         if not os.path.isdir(loc):
             os.makedirs(loc, mode=0777)
 
         tmpl_post = pkg_resources.resource_string(
             'rhodecode', '/'.join(
                 ('config', 'hook_templates', 'git_post_receive.py.tmpl')))
         tmpl_pre = pkg_resources.resource_string(
             'rhodecode', '/'.join(
                 ('config', 'hook_templates', 'git_pre_receive.py.tmpl')))
 
         for h_type, tmpl in [('pre', tmpl_pre), ('post', tmpl_post)]:
             _hook_file = os.path.join(loc, '%s-receive' % h_type)
             log.debug('Installing git hook in repo %s', repo)
             _rhodecode_hook = _check_rhodecode_hook(_hook_file)
 
             if _rhodecode_hook or force_create:
                 log.debug('writing %s hook file !', h_type)
                 try:
                     with open(_hook_file, 'wb') as f:
                         tmpl = tmpl.replace('_TMPL_', rhodecode.__version__)
                         tmpl = tmpl.replace('_ENV_', sys.executable)
                         f.write(tmpl)
                     os.chmod(_hook_file, 0755)
                 except IOError:
                     log.exception('error writing hook file %s', _hook_file)
             else:
                 log.debug('skipping writing hook file')
 
     def install_svn_hooks(self, repo, force_create=False):
         """
         Creates rhodecode hooks inside a svn repository
 
         :param repo: Instance of VCS repo
         :param force_create: Create even if same name hook exists
         """
         hooks_path = os.path.join(repo.path, 'hooks')
         if not os.path.isdir(hooks_path):
             os.makedirs(hooks_path)
         post_commit_tmpl = pkg_resources.resource_string(
             'rhodecode', '/'.join(
                 ('config', 'hook_templates', 'svn_post_commit_hook.py.tmpl')))
         pre_commit_template = pkg_resources.resource_string(
             'rhodecode', '/'.join(
                 ('config', 'hook_templates', 'svn_pre_commit_hook.py.tmpl')))
         templates = {
             'post-commit': post_commit_tmpl,
             'pre-commit': pre_commit_template
         }
         for filename in templates:
             _hook_file = os.path.join(hooks_path, filename)
             _rhodecode_hook = _check_rhodecode_hook(_hook_file)
             if _rhodecode_hook or force_create:
                 log.debug('writing %s hook file !', filename)
                 template = templates[filename]
                 try:
                     with open(_hook_file, 'wb') as f:
                         template = template.replace(
                             '_TMPL_', rhodecode.__version__)
                         template = template.replace('_ENV_', sys.executable)
                         f.write(template)
                     os.chmod(_hook_file, 0755)
                 except IOError:
                     log.exception('error writing hook file %s', filename)
             else:
                 log.debug('skipping writing hook file')
 
     def install_hooks(self, repo, repo_type):
         if repo_type == 'git':
             self.install_git_hook(repo)
         elif repo_type == 'svn':
             self.install_svn_hooks(repo)
 
     def get_server_info(self, environ=None):
         import platform
         import rhodecode
         import pkg_resources
         from rhodecode.model.meta import Base as sql_base, Session
         from sqlalchemy.engine import url
         from rhodecode.lib.base import get_server_ip_addr, get_server_port
         from rhodecode.lib.vcs.backends.git import discover_git_version
         from rhodecode.model.gist import GIST_STORE_LOC
 
         try:
             # cygwin cannot have yet psutil support.
             import psutil
         except ImportError:
             psutil = None
 
         environ = environ or {}
         _NA = 'NOT AVAILABLE'
         _memory = _NA
         _uptime = _NA
         _boot_time = _NA
         _cpu = _NA
         _disk = dict(percent=0, used=0, total=0, error='')
         _load = {'1_min': _NA, '5_min': _NA, '15_min': _NA}
 
         model = VcsSettingsModel()
         storage_path = model.get_repos_location()
         gist_storage_path = os.path.join(storage_path, GIST_STORE_LOC)
         archive_storage_path = rhodecode.CONFIG.get('archive_cache_dir', '')
         search_index_storage_path = rhodecode.CONFIG.get('search.location', '')
 
         if psutil:
             # disk storage
             try:
                 _disk = dict(psutil.disk_usage(storage_path)._asdict())
             except Exception as e:
                 log.exception('Failed to fetch disk info')
                 _disk = {'percent': 0, 'used': 0, 'total': 0, 'error': str(e)}
 
             # memory
             _memory = dict(psutil.virtual_memory()._asdict())
             _memory['percent2'] = psutil._common.usage_percent(
                 (_memory['total'] - _memory['free']),
                 _memory['total'], 1)
 
             # load averages
             if hasattr(psutil.os, 'getloadavg'):
                 _load = dict(zip(
                     ['1_min', '5_min', '15_min'], psutil.os.getloadavg()))
             _uptime = time.time() - psutil.boot_time()
             _boot_time = psutil.boot_time()
             _cpu = psutil.cpu_percent(0.5)
 
         mods = dict([(p.project_name, p.version)
                      for p in pkg_resources.working_set])
 
         def get_storage_size(storage_path):
             sizes = []
             for file_ in os.listdir(storage_path):
                 storage_file = os.path.join(storage_path, file_)
                 if os.path.isfile(storage_file):
                     try:
                         sizes.append(os.path.getsize(storage_file))
                     except OSError:
                         log.exception('Failed to get size of storage file %s',
                                       storage_file)
                         pass
 
             return sum(sizes)
 
         # archive cache storage
         _disk_archive = {'percent': 0, 'used': 0, 'total': 0}
         try:
             archive_storage_path_exists = os.path.isdir(
                 archive_storage_path)
             if archive_storage_path and archive_storage_path_exists:
                 used = get_storage_size(archive_storage_path)
                 _disk_archive.update({
                     'used': used,
                     'total': used,
                 })
         except Exception as e:
             log.exception('failed to fetch archive cache storage')
             _disk_archive['error'] = str(e)
 
         # search index storage
         _disk_index = {'percent': 0, 'used': 0, 'total': 0}
         try:
             search_index_storage_path_exists = os.path.isdir(
                 search_index_storage_path)
             if search_index_storage_path_exists:
                 used = get_storage_size(search_index_storage_path)
                 _disk_index.update({
                     'percent': 100,
                     'used': used,
                     'total': used,
                 })
         except Exception as e:
             log.exception('failed to fetch search index storage')
             _disk_index['error'] = str(e)
 
         # gist storage
         _disk_gist = {'percent': 0, 'used': 0, 'total': 0, 'items': 0}
         try:
             items_count = 0
             used = 0
             for root, dirs, files in os.walk(safe_str(gist_storage_path)):
                 if root == gist_storage_path:
                     items_count = len(dirs)
 
                 for f in files:
                     try:
                         used += os.path.getsize(os.path.join(root, f))
                     except OSError:
                         pass
             _disk_gist.update({
                 'percent': 100,
                 'used': used,
                 'total': used,
                 'items': items_count
             })
         except Exception as e:
             log.exception('failed to fetch gist storage items')
             _disk_gist['error'] = str(e)
 
         # GIT info
         git_ver = discover_git_version()
 
         # SVN info
         # TODO: johbo: Add discover_svn_version to replace this code.
         try:
             import svn.core
             svn_ver = svn.core.SVN_VERSION
         except ImportError:
             svn_ver = None
 
         # DB stuff
         db_info = url.make_url(rhodecode.CONFIG['sqlalchemy.db1.url'])
         db_type = db_info.__to_string__()
         try:
             engine = sql_base.metadata.bind
             db_server_info = engine.dialect._get_server_version_info(
                 Session.connection(bind=engine))
             db_version = '%s %s' % (db_info.drivername,
                                     '.'.join(map(str, db_server_info)))
         except Exception:
             log.exception('failed to fetch db version')
             db_version = '%s %s' % (db_info.drivername, '?')
 
         db_migrate = DbMigrateVersion.query().filter(
             DbMigrateVersion.repository_id == 'rhodecode_db_migrations').one()
         db_migrate_version = db_migrate.version
 
         info = {
             'py_version': ' '.join(platform._sys_version()),
             'py_path': sys.executable,
             'py_modules': sorted(mods.items(), key=lambda k: k[0].lower()),
 
             'platform': safe_unicode(platform.platform()),
             'storage': storage_path,
             'archive_storage': archive_storage_path,
             'index_storage': search_index_storage_path,
             'gist_storage': gist_storage_path,
 
 
             'db_type': db_type,
             'db_version': db_version,
             'db_migrate_version': db_migrate_version,
 
             'rhodecode_version': rhodecode.__version__,
             'rhodecode_config_ini': rhodecode.CONFIG.get('__file__'),
             'server_ip': '%s:%s' % (
                 get_server_ip_addr(environ, log_errors=False),
                 get_server_port(environ)
             ),
             'server_id': rhodecode.CONFIG.get('instance_id'),
 
             'git_version': safe_unicode(git_ver),
             'hg_version': mods.get('mercurial'),
             'svn_version': svn_ver,
 
             'uptime': _uptime,
             'boot_time': _boot_time,
             'load': _load,
             'cpu': _cpu,
             'memory': _memory,
             'disk': _disk,
             'disk_archive': _disk_archive,
             'disk_gist': _disk_gist,
             'disk_index': _disk_index,
         }
         return info
 
 
 def _check_rhodecode_hook(hook_path):
     """
     Check if the hook was created by RhodeCode
     """
     if not os.path.exists(hook_path):
         return True
 
     log.debug('hook exists, checking if it is from rhodecode')
     hook_content = _read_hook(hook_path)
     matches = re.search(r'(?:RC_HOOK_VER)\s*=\s*(.*)', hook_content)
     if matches:
         try:
             version = matches.groups()[0]
             log.debug('got %s, it is rhodecode', version)
             return True
         except Exception:
             log.exception("Exception while reading the hook version.")
 
     return False
 
 
 def _read_hook(hook_path):
     with open(hook_path, 'rb') as f:
         content = f.read()
-    return content
+    return content
  No newline at end of file