vcs: sanitize diff context values (Issue #306)

- Updated Git repository implementation to ensure context falls within
  0 to 2**31-1 range (inclusive) when fetching a diff.
- Added tests for Git repositories for checking passed-in negative and
  overflowing contexts (for the --unified option).
- Updated Mercurial repository implementation to ensure context is not
  negative when fetching a diff.
- Added tests for Mercurial repositories for checking passed-in
  negative context (for the --unified option).

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

@@ -1,734 +1,756 b''
 # -*- coding: utf-8 -*-
 """
     vcs.backends.git.repository
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
     Git repository implementation.
 
     :created_on: Apr 8, 2010
     :copyright: (c) 2010-2011 by Marcin Kuzminski, Lukasz Balcerzak.
 """
 
 import os
 import re
 import time
 import errno
 import urllib
 import urllib2
 import logging
 import posixpath
 
 from dulwich.objects import Tag
 from dulwich.repo import Repo, NotGitRepository
 from dulwich.config import ConfigFile
 
 from kallithea.lib.vcs import subprocessio
 from kallithea.lib.vcs.backends.base import BaseRepository, CollectionGenerator
 from kallithea.lib.vcs.conf import settings
 
 from kallithea.lib.vcs.exceptions import (
     BranchDoesNotExistError, ChangesetDoesNotExistError, EmptyRepositoryError,
     RepositoryError, TagAlreadyExistError, TagDoesNotExistError
 )
 from kallithea.lib.vcs.utils import safe_unicode, makedate, date_fromtimestamp
 from kallithea.lib.vcs.utils.lazy import LazyProperty
 from kallithea.lib.vcs.utils.ordered_dict import OrderedDict
 from kallithea.lib.vcs.utils.paths import abspath, get_user_home
 
 from kallithea.lib.vcs.utils.hgcompat import (
     hg_url, httpbasicauthhandler, httpdigestauthhandler
 )
 
 from .changeset import GitChangeset
 from .inmemory import GitInMemoryChangeset
 from .workdir import GitWorkdir
 
 SHA_PATTERN = re.compile(r'^[[0-9a-fA-F]{12}|[0-9a-fA-F]{40}]$')
 
 log = logging.getLogger(__name__)
 
 
 class GitRepository(BaseRepository):
     """
     Git repository backend.
     """
     DEFAULT_BRANCH_NAME = 'master'
     scm = 'git'
 
     def __init__(self, repo_path, create=False, src_url=None,
                  update_after_clone=False, bare=False):
 
         self.path = abspath(repo_path)
         repo = self._get_repo(create, src_url, update_after_clone, bare)
         self.bare = repo.bare
 
     @property
     def _config_files(self):
         return [
             self.bare and abspath(self.path, 'config')
                       or abspath(self.path, '.git', 'config'),
              abspath(get_user_home(), '.gitconfig'),
          ]
 
     @property
     def _repo(self):
         return Repo(self.path)
 
     @property
     def head(self):
         try:
             return self._repo.head()
         except KeyError:
             return None
 
     @property
     def _empty(self):
         """
         Checks if repository is empty ie. without any changesets
         """
 
         try:
             self.revisions[0]
         except (KeyError, IndexError):
             return True
         return False
 
     @LazyProperty
     def revisions(self):
         """
         Returns list of revisions' ids, in ascending order.  Being lazy
         attribute allows external tools to inject shas from cache.
         """
         return self._get_all_revisions()
 
     @classmethod
     def _run_git_command(cls, cmd, **opts):
         """
         Runs given ``cmd`` as git command and returns tuple
         (stdout, stderr).
 
         :param cmd: git command to be executed
         :param opts: env options to pass into Subprocess command
         """
 
         if '_bare' in opts:
             _copts = []
             del opts['_bare']
         else:
             _copts = ['-c', 'core.quotepath=false', ]
         safe_call = False
         if '_safe' in opts:
             #no exc on failure
             del opts['_safe']
             safe_call = True
 
         assert isinstance(cmd, list), cmd
 
         gitenv = os.environ
         # need to clean fix GIT_DIR !
         if 'GIT_DIR' in gitenv:
             del gitenv['GIT_DIR']
         gitenv['GIT_CONFIG_NOGLOBAL'] = '1'
 
         _git_path = settings.GIT_EXECUTABLE_PATH
         cmd = [_git_path] + _copts + cmd
 
         try:
             _opts = dict(
                 env=gitenv,
                 shell=False,
             )
             _opts.update(opts)
             p = subprocessio.SubprocessIOChunker(cmd, **_opts)
         except (EnvironmentError, OSError) as err:
             tb_err = ("Couldn't run git command (%s).\n"
                       "Original error was:%s\n" % (cmd, err))
             log.error(tb_err)
             if safe_call:
                 return '', err
             else:
                 raise RepositoryError(tb_err)
 
         return ''.join(p.output), ''.join(p.error)
 
     def run_git_command(self, cmd):
         opts = {}
         if os.path.isdir(self.path):
             opts['cwd'] = self.path
         return self._run_git_command(cmd, **opts)
 
     @classmethod
     def _check_url(cls, url):
         """
         Function will check given url and try to verify if it's a valid
         link. Sometimes it may happened that git will issue basic
         auth request that can cause whole API to hang when used from python
         or other external calls.
 
         On failures it'll raise urllib2.HTTPError, exception is also thrown
         when the return code is non 200
         """
 
         # check first if it's not an local url
         if os.path.isdir(url) or url.startswith('file:'):
             return True
 
         if url.startswith('git://'):
             return True
 
         if '+' in url[:url.find('://')]:
             url = url[url.find('+') + 1:]
 
         handlers = []
         url_obj = hg_url(url)
         test_uri, authinfo = url_obj.authinfo()
         url_obj.passwd = '*****'
         cleaned_uri = str(url_obj)
 
         if not test_uri.endswith('info/refs'):
             test_uri = test_uri.rstrip('/') + '/info/refs'
 
         if authinfo:
             #create a password manager
             passmgr = urllib2.HTTPPasswordMgrWithDefaultRealm()
             passmgr.add_password(*authinfo)
 
             handlers.extend((httpbasicauthhandler(passmgr),
                              httpdigestauthhandler(passmgr)))
 
         o = urllib2.build_opener(*handlers)
         o.addheaders = [('User-Agent', 'git/1.7.8.0')]  # fake some git
 
         q = {"service": 'git-upload-pack'}
         qs = '?%s' % urllib.urlencode(q)
         cu = "%s%s" % (test_uri, qs)
         req = urllib2.Request(cu, None, {})
 
         try:
             resp = o.open(req)
             if resp.code != 200:
                 raise Exception('Return Code is not 200')
         except Exception as e:
             # means it cannot be cloned
             raise urllib2.URLError("[%s] org_exc: %s" % (cleaned_uri, e))
 
         # now detect if it's proper git repo
         gitdata = resp.read()
         if not 'service=git-upload-pack' in gitdata:
             raise urllib2.URLError(
                 "url [%s] does not look like an git" % (cleaned_uri))
 
         return True
 
     def _get_repo(self, create, src_url=None, update_after_clone=False,
                   bare=False):
         if create and os.path.exists(self.path):
             raise RepositoryError("Location already exist")
         if src_url and not create:
             raise RepositoryError("Create should be set to True if src_url is "
                                   "given (clone operation creates repository)")
         try:
             if create and src_url:
                 GitRepository._check_url(src_url)
                 self.clone(src_url, update_after_clone, bare)
                 return Repo(self.path)
             elif create:
                 os.makedirs(self.path)
                 if bare:
                     return Repo.init_bare(self.path)
                 else:
                     return Repo.init(self.path)
             else:
                 return self._repo
         except (NotGitRepository, OSError) as err:
             raise RepositoryError(err)
 
     def _get_all_revisions(self):
         # we must check if this repo is not empty, since later command
         # fails if it is. And it's cheaper to ask than throw the subprocess
         # errors
         try:
             self._repo.head()
         except KeyError:
             return []
 
         rev_filter = settings.GIT_REV_FILTER
         cmd = ['rev-list', rev_filter, '--reverse', '--date-order']
         try:
             so, se = self.run_git_command(cmd)
         except RepositoryError:
             # Can be raised for empty repositories
             return []
         return so.splitlines()
 
     def _get_all_revisions2(self):
         #alternate implementation using dulwich
         includes = [x[1][0] for x in self._parsed_refs.iteritems()
                     if x[1][1] != 'T']
         return [c.commit.id for c in self._repo.get_walker(include=includes)]
 
     def _get_revision(self, revision):
         """
         For git backend we always return integer here. This way we ensure
         that changeset's revision attribute would become integer.
         """
 
         is_null = lambda o: len(o) == revision.count('0')
 
         if self._empty:
             raise EmptyRepositoryError("There are no changesets yet")
 
         if revision in (None, '', 'tip', 'HEAD', 'head', -1):
             return self.revisions[-1]
 
         is_bstr = isinstance(revision, (str, unicode))
         if ((is_bstr and revision.isdigit() and len(revision) < 12)
             or isinstance(revision, int) or is_null(revision)):
             try:
                 revision = self.revisions[int(revision)]
             except IndexError:
                 msg = ("Revision %s does not exist for %s" % (revision, self))
                 raise ChangesetDoesNotExistError(msg)
 
         elif is_bstr:
             # get by branch/tag name
             _ref_revision = self._parsed_refs.get(revision)
             if _ref_revision:  # and _ref_revision[1] in ['H', 'RH', 'T']:
                 return _ref_revision[0]
 
             _tags_shas = self.tags.values()
             # maybe it's a tag ? we don't have them in self.revisions
             if revision in _tags_shas:
                 return _tags_shas[_tags_shas.index(revision)]
 
             elif not SHA_PATTERN.match(revision) or revision not in self.revisions:
                 msg = ("Revision %s does not exist for %s" % (revision, self))
                 raise ChangesetDoesNotExistError(msg)
 
         # Ensure we return full id
         if not SHA_PATTERN.match(str(revision)):
             raise ChangesetDoesNotExistError("Given revision %s not recognized"
                 % revision)
         return revision
 
     def get_ref_revision(self, ref_type, ref_name):
         """
         Returns ``MercurialChangeset`` object representing repository's
         changeset at the given ``revision``.
         """
         return self._get_revision(ref_name)
 
     def _get_archives(self, archive_name='tip'):
 
         for i in [('zip', '.zip'), ('gz', '.tar.gz'), ('bz2', '.tar.bz2')]:
                 yield {"type": i[0], "extension": i[1], "node": archive_name}
 
     def _get_url(self, url):
         """
         Returns normalized url. If schema is not given, would fall to
         filesystem (``file:///``) schema.
         """
         url = str(url)
         if url != 'default' and not '://' in url:
             url = ':///'.join(('file', url))
         return url
 
     def get_hook_location(self):
         """
         returns absolute path to location where hooks are stored
         """
         loc = os.path.join(self.path, 'hooks')
         if not self.bare:
             loc = os.path.join(self.path, '.git', 'hooks')
         return loc
 
     @LazyProperty
     def name(self):
         return os.path.basename(self.path)
 
     @LazyProperty
     def last_change(self):
         """
         Returns last change made on this repository as datetime object
         """
         return date_fromtimestamp(self._get_mtime(), makedate()[1])
 
     def _get_mtime(self):
         try:
             return time.mktime(self.get_changeset().date.timetuple())
         except RepositoryError:
             idx_loc = '' if self.bare else '.git'
             # fallback to filesystem
             in_path = os.path.join(self.path, idx_loc, "index")
             he_path = os.path.join(self.path, idx_loc, "HEAD")
             if os.path.exists(in_path):
                 return os.stat(in_path).st_mtime
             else:
                 return os.stat(he_path).st_mtime
 
     @LazyProperty
     def description(self):
         undefined_description = u'unknown'
         _desc = self._repo.get_description()
         return safe_unicode(_desc or undefined_description)
 
     @LazyProperty
     def contact(self):
         undefined_contact = u'Unknown'
         return undefined_contact
 
     @property
     def branches(self):
         if not self.revisions:
             return {}
         sortkey = lambda ctx: ctx[0]
         _branches = [(x[0], x[1][0])
                      for x in self._parsed_refs.iteritems() if x[1][1] == 'H']
         return OrderedDict(sorted(_branches, key=sortkey, reverse=False))
 
     @LazyProperty
     def closed_branches(self):
         return {}
 
     @LazyProperty
     def tags(self):
         return self._get_tags()
 
     def _get_tags(self):
         if not self.revisions:
             return {}
 
         sortkey = lambda ctx: ctx[0]
         _tags = [(x[0], x[1][0])
                  for x in self._parsed_refs.iteritems() if x[1][1] == 'T']
         return OrderedDict(sorted(_tags, key=sortkey, reverse=True))
 
     def tag(self, name, user, revision=None, message=None, date=None,
             **kwargs):
         """
         Creates and returns a tag for the given ``revision``.
 
         :param name: name for new tag
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param revision: changeset id for which new tag would be created
         :param message: message of the tag's commit
         :param date: date of tag's commit
 
         :raises TagAlreadyExistError: if tag with same name already exists
         """
         if name in self.tags:
             raise TagAlreadyExistError("Tag %s already exists" % name)
         changeset = self.get_changeset(revision)
         message = message or "Added tag %s for commit %s" % (name,
             changeset.raw_id)
         self._repo.refs["refs/tags/%s" % name] = changeset._commit.id
 
         self._parsed_refs = self._get_parsed_refs()
         self.tags = self._get_tags()
         return changeset
 
     def remove_tag(self, name, user, message=None, date=None):
         """
         Removes tag with the given ``name``.
 
         :param name: name of the tag to be removed
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param message: message of the tag's removal commit
         :param date: date of tag's removal commit
 
         :raises TagDoesNotExistError: if tag with given name does not exists
         """
         if name not in self.tags:
             raise TagDoesNotExistError("Tag %s does not exist" % name)
         tagpath = posixpath.join(self._repo.refs.path, 'refs', 'tags', name)
         try:
             os.remove(tagpath)
             self._parsed_refs = self._get_parsed_refs()
             self.tags = self._get_tags()
         except OSError as e:
             raise RepositoryError(e.strerror)
 
     @LazyProperty
     def bookmarks(self):
         """
         Gets bookmarks for this repository
         """
         return {}
 
     @LazyProperty
     def _parsed_refs(self):
         return self._get_parsed_refs()
 
     def _get_parsed_refs(self):
         # cache the property
         _repo = self._repo
         refs = _repo.get_refs()
         keys = [('refs/heads/', 'H'),
                 ('refs/remotes/origin/', 'RH'),
                 ('refs/tags/', 'T')]
         _refs = {}
         for ref, sha in refs.iteritems():
             for k, type_ in keys:
                 if ref.startswith(k):
                     _key = ref[len(k):]
                     if type_ == 'T':
                         obj = _repo.get_object(sha)
                         if isinstance(obj, Tag):
                             sha = _repo.get_object(sha).object[1]
                     _refs[_key] = [sha, type_]
                     break
         return _refs
 
     def _heads(self, reverse=False):
         refs = self._repo.get_refs()
         heads = {}
 
         for key, val in refs.items():
             for ref_key in ['refs/heads/', 'refs/remotes/origin/']:
                 if key.startswith(ref_key):
                     n = key[len(ref_key):]
                     if n not in ['HEAD']:
                         heads[n] = val
 
         return heads if reverse else dict((y, x) for x, y in heads.iteritems())
 
     def get_changeset(self, revision=None):
         """
         Returns ``GitChangeset`` object representing commit from git repository
         at the given revision or head (most recent commit) if None given.
         """
         if isinstance(revision, GitChangeset):
             return revision
         revision = self._get_revision(revision)
         changeset = GitChangeset(repository=self, revision=revision)
         return changeset
 
     def get_changesets(self, start=None, end=None, start_date=None,
            end_date=None, branch_name=None, reverse=False):
         """
         Returns iterator of ``GitChangeset`` objects from start to end (both
         are inclusive), in ascending date order (unless ``reverse`` is set).
 
         :param start: changeset ID, as str; first returned changeset
         :param end: changeset ID, as str; last returned changeset
         :param start_date: if specified, changesets with commit date less than
           ``start_date`` would be filtered out from returned set
         :param end_date: if specified, changesets with commit date greater than
           ``end_date`` would be filtered out from returned set
         :param branch_name: if specified, changesets not reachable from given
           branch would be filtered out from returned set
         :param reverse: if ``True``, returned generator would be reversed
           (meaning that returned changesets would have descending date order)
 
         :raise BranchDoesNotExistError: If given ``branch_name`` does not
             exist.
         :raise ChangesetDoesNotExistError: If changeset for given ``start`` or
           ``end`` could not be found.
 
         """
         if branch_name and branch_name not in self.branches:
             raise BranchDoesNotExistError("Branch '%s' not found" \
                                           % branch_name)
         # actually we should check now if it's not an empty repo to not spaw
         # subprocess commands
         if self._empty:
             raise EmptyRepositoryError("There are no changesets yet")
 
         # %H at format means (full) commit hash, initial hashes are retrieved
         # in ascending date order
         cmd = ['log', '--date-order', '--reverse', '--pretty=format:%H']
         if start_date:
             cmd += ['--since', start_date.strftime('%m/%d/%y %H:%M:%S')]
         if end_date:
             cmd += ['--until', end_date.strftime('%m/%d/%y %H:%M:%S')]
         if branch_name:
             cmd.append(branch_name)
         else:
             cmd.append(settings.GIT_REV_FILTER)
 
         revs = self.run_git_command(cmd)[0].splitlines()
         start_pos = 0
         end_pos = len(revs)
         if start:
             _start = self._get_revision(start)
             try:
                 start_pos = revs.index(_start)
             except ValueError:
                 pass
 
         if end is not None:
             _end = self._get_revision(end)
             try:
                 end_pos = revs.index(_end)
             except ValueError:
                 pass
 
         if None not in [start, end] and start_pos > end_pos:
             raise RepositoryError('start cannot be after end')
 
         if end_pos is not None:
             end_pos += 1
 
         revs = revs[start_pos:end_pos]
         if reverse:
             revs = reversed(revs)
         return CollectionGenerator(self, revs)
 
     def get_diff(self, rev1, rev2, path=None, ignore_whitespace=False,
                  context=3):
         """
         Returns (git like) *diff*, as plain text. Shows changes introduced by
         ``rev2`` since ``rev1``.
 
         :param rev1: Entry point from which diff is shown. Can be
           ``self.EMPTY_CHANGESET`` - in this case, patch showing all
           the changes since empty state of the repository until ``rev2``
         :param rev2: Until which revision changes should be shown.
         :param ignore_whitespace: If set to ``True``, would not show whitespace
           changes. Defaults to ``False``.
         :param context: How many lines before/after changed lines should be
-          shown. Defaults to ``3``.
+          shown. Defaults to ``3``. Due to limitations in Git, if
+          value passed-in is greater than ``2**31-1``
+          (``2147483647``), it will be set to ``2147483647``
+          instead. If negative value is passed-in, it will be set to
+          ``0`` instead.
         """
+
+        # Git internally uses a signed long int for storing context
+        # size (number of lines to show before and after the
+        # differences). This can result in integer overflow, so we
+        # ensure the requested context is smaller by one than the
+        # number that would cause the overflow. It is highly unlikely
+        # that a single file will contain that many lines, so this
+        # kind of change should not cause any realistic consequences.
+        overflowed_long_int = 2**31
+
+        if context >= overflowed_long_int:
+            context = overflowed_long_int-1
+
+        # Negative context values make no sense, and will result in
+        # errors. Ensure this does not happen.
+        if context < 0:
+            context = 0
+
         flags = ['-U%s' % context, '--full-index', '--binary', '-p', '-M', '--abbrev=40']
         if ignore_whitespace:
             flags.append('-w')
 
         if hasattr(rev1, 'raw_id'):
             rev1 = getattr(rev1, 'raw_id')
 
         if hasattr(rev2, 'raw_id'):
             rev2 = getattr(rev2, 'raw_id')
 
         if rev1 == self.EMPTY_CHANGESET:
             rev2 = self.get_changeset(rev2).raw_id
             cmd = ['show'] + flags + [rev2]
         else:
             rev1 = self.get_changeset(rev1).raw_id
             rev2 = self.get_changeset(rev2).raw_id
             cmd = ['diff'] + flags + [rev1, rev2]
 
         if path:
             cmd += ['--', path]
 
         stdout, stderr = self.run_git_command(cmd)
         # TODO: don't ignore stderr
         # If we used 'show' command, strip first few lines (until actual diff
         # starts)
         if rev1 == self.EMPTY_CHANGESET:
             parts = stdout.split('\ndiff ', 1)
             if len(parts) > 1:
                 stdout = 'diff ' + parts[1]
         return stdout
 
     @LazyProperty
     def in_memory_changeset(self):
         """
         Returns ``GitInMemoryChangeset`` object for this repository.
         """
         return GitInMemoryChangeset(self)
 
     def clone(self, url, update_after_clone=True, bare=False):
         """
         Tries to clone changes from external location.
 
         :param update_after_clone: If set to ``False``, git won't checkout
           working directory
         :param bare: If set to ``True``, repository would be cloned into
           *bare* git repository (no working directory at all).
         """
         url = self._get_url(url)
         cmd = ['clone', '-q']
         if bare:
             cmd.append('--bare')
         elif not update_after_clone:
             cmd.append('--no-checkout')
         cmd += ['--', url, self.path]
         # If error occurs run_git_command raises RepositoryError already
         self.run_git_command(cmd)
 
     def pull(self, url):
         """
         Tries to pull changes from external location.
         """
         url = self._get_url(url)
         cmd = ['pull', '--ff-only', url]
         # If error occurs run_git_command raises RepositoryError already
         self.run_git_command(cmd)
 
     def fetch(self, url):
         """
         Tries to pull changes from external location.
         """
         url = self._get_url(url)
         so, se = self.run_git_command(['ls-remote', '-h', url])
         cmd = ['fetch', url, '--']
         for line in (x for x in so.splitlines()):
             sha, ref = line.split('\t')
             cmd.append('+%s:%s' % (ref, ref))
         self.run_git_command(cmd)
 
     def _update_server_info(self):
         """
         runs gits update-server-info command in this repo instance
         """
         from dulwich.server import update_server_info
         try:
             update_server_info(self._repo)
         except OSError as e:
             if e.errno not in [errno.ENOENT, errno.EROFS]:
                 raise
             # Workaround for dulwich crashing on for example its own dulwich/tests/data/repos/simple_merge.git/info/refs.lock
             log.error('Ignoring %s running update-server-info: %s', type(e).__name__, e)
 
     @LazyProperty
     def workdir(self):
         """
         Returns ``Workdir`` instance for this repository.
         """
         return GitWorkdir(self)
 
     def get_config_value(self, section, name, config_file=None):
         """
         Returns configuration value for a given [``section``] and ``name``.
 
         :param section: Section we want to retrieve value from
         :param name: Name of configuration we want to retrieve
         :param config_file: A path to file which should be used to retrieve
           configuration from (might also be a list of file paths)
         """
         if config_file is None:
             config_file = []
         elif isinstance(config_file, basestring):
             config_file = [config_file]
 
         def gen_configs():
             for path in config_file + self._config_files:
                 try:
                     yield ConfigFile.from_path(path)
                 except (IOError, OSError, ValueError):
                     continue
 
         for config in gen_configs():
             try:
                 return config.get(section, name)
             except KeyError:
                 continue
         return None
 
     def get_user_name(self, config_file=None):
         """
         Returns user's name from global configuration file.
 
         :param config_file: A path to file which should be used to retrieve
           configuration from (might also be a list of file paths)
         """
         return self.get_config_value('user', 'name', config_file)
 
     def get_user_email(self, config_file=None):
         """
         Returns user's email from global configuration file.
 
         :param config_file: A path to file which should be used to retrieve
           configuration from (might also be a list of file paths)
         """
         return self.get_config_value('user', 'email', config_file)

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

@@ -1,620 +1,627 b''
 # -*- coding: utf-8 -*-
 """
     vcs.backends.hg.repository
     ~~~~~~~~~~~~~~~~~~~~~~~~~~
 
     Mercurial repository implementation.
 
     :created_on: Apr 8, 2010
     :copyright: (c) 2010-2011 by Marcin Kuzminski, Lukasz Balcerzak.
 """
 
 import os
 import time
 import urllib
 import urllib2
 import logging
 import datetime
 
 from kallithea.lib.vcs.backends.base import BaseRepository, CollectionGenerator
 
 from kallithea.lib.vcs.exceptions import (
     BranchDoesNotExistError, ChangesetDoesNotExistError, EmptyRepositoryError,
     RepositoryError, VCSError, TagAlreadyExistError, TagDoesNotExistError
 )
 from kallithea.lib.vcs.utils import (
     author_email, author_name, date_fromtimestamp, makedate, safe_unicode, safe_str,
 )
 from kallithea.lib.vcs.utils.lazy import LazyProperty
 from kallithea.lib.vcs.utils.ordered_dict import OrderedDict
 from kallithea.lib.vcs.utils.paths import abspath
 from kallithea.lib.vcs.utils.hgcompat import (
     ui, nullid, match, patch, diffopts, clone, get_contact,
     localrepository, RepoLookupError, Abort, RepoError, hex, scmutil, hg_url,
     httpbasicauthhandler, httpdigestauthhandler, peer, httppeer, sshpeer
 )
 
 from .changeset import MercurialChangeset
 from .inmemory import MercurialInMemoryChangeset
 from .workdir import MercurialWorkdir
 
 log = logging.getLogger(__name__)
 
 
 class MercurialRepository(BaseRepository):
     """
     Mercurial repository backend
     """
     DEFAULT_BRANCH_NAME = 'default'
     scm = 'hg'
 
     def __init__(self, repo_path, create=False, baseui=None, src_url=None,
                  update_after_clone=False):
         """
         Raises RepositoryError if repository could not be find at the given
         ``repo_path``.
 
         :param repo_path: local path of the repository
         :param create=False: if set to True, would try to create repository if
            it does not exist rather than raising exception
         :param baseui=None: user data
         :param src_url=None: would try to clone repository from given location
         :param update_after_clone=False: sets update of working copy after
           making a clone
         """
 
         if not isinstance(repo_path, str):
             raise VCSError('Mercurial backend requires repository path to '
                            'be instance of <str> got %s instead' %
                            type(repo_path))
 
         self.path = abspath(repo_path)
         self.baseui = baseui or ui.ui()
         # We've set path and ui, now we can set _repo itself
         self._repo = self._get_repo(create, src_url, update_after_clone)
 
     @property
     def _empty(self):
         """
         Checks if repository is empty ie. without any changesets
         """
         # TODO: Following raises errors when using InMemoryChangeset...
         # return len(self._repo.changelog) == 0
         return len(self.revisions) == 0
 
     @LazyProperty
     def revisions(self):
         """
         Returns list of revisions' ids, in ascending order.  Being lazy
         attribute allows external tools to inject shas from cache.
         """
         return self._get_all_revisions()
 
     @LazyProperty
     def name(self):
         return os.path.basename(self.path)
 
     @LazyProperty
     def branches(self):
         return self._get_branches()
 
     @LazyProperty
     def closed_branches(self):
         return self._get_branches(normal=False, closed=True)
 
     @LazyProperty
     def allbranches(self):
         """
         List all branches, including closed branches.
         """
         return self._get_branches(closed=True)
 
     def _get_branches(self, normal=True, closed=False):
         """
         Gets branches for this repository
         Returns only not closed branches by default
 
         :param closed: return also closed branches for mercurial
         :param normal: return also normal branches
         """
 
         if self._empty:
             return {}
 
         bt = OrderedDict()
         for bn, _heads, tip, isclosed in sorted(self._repo.branchmap().iterbranches()):
             if isclosed:
                 if closed:
                     bt[safe_unicode(bn)] = hex(tip)
             else:
                 if normal:
                     bt[safe_unicode(bn)] = hex(tip)
 
         return bt
 
     @LazyProperty
     def tags(self):
         """
         Gets tags for this repository
         """
         return self._get_tags()
 
     def _get_tags(self):
         if self._empty:
             return {}
 
         sortkey = lambda ctx: ctx[0]  # sort by name
         _tags = [(safe_unicode(n), hex(h),) for n, h in
                  self._repo.tags().items()]
 
         return OrderedDict(sorted(_tags, key=sortkey, reverse=True))
 
     def tag(self, name, user, revision=None, message=None, date=None,
             **kwargs):
         """
         Creates and returns a tag for the given ``revision``.
 
         :param name: name for new tag
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param revision: changeset id for which new tag would be created
         :param message: message of the tag's commit
         :param date: date of tag's commit
 
         :raises TagAlreadyExistError: if tag with same name already exists
         """
         if name in self.tags:
             raise TagAlreadyExistError("Tag %s already exists" % name)
         changeset = self.get_changeset(revision)
         local = kwargs.setdefault('local', False)
 
         if message is None:
             message = "Added tag %s for changeset %s" % (name,
                 changeset.short_id)
 
         if date is None:
             date = datetime.datetime.now().ctime()
 
         try:
             self._repo.tag(name, changeset._ctx.node(), message, local, user,
                 date)
         except Abort as e:
             raise RepositoryError(e.message)
 
         # Reinitialize tags
         self.tags = self._get_tags()
         tag_id = self.tags[name]
 
         return self.get_changeset(revision=tag_id)
 
     def remove_tag(self, name, user, message=None, date=None):
         """
         Removes tag with the given ``name``.
 
         :param name: name of the tag to be removed
         :param user: full username, i.e.: "Joe Doe <[e-mail]>"
         :param message: message of the tag's removal commit
         :param date: date of tag's removal commit
 
         :raises TagDoesNotExistError: if tag with given name does not exists
         """
         if name not in self.tags:
             raise TagDoesNotExistError("Tag %s does not exist" % name)
         if message is None:
             message = "Removed tag %s" % name
         if date is None:
             date = datetime.datetime.now().ctime()
         local = False
 
         try:
             self._repo.tag(name, nullid, message, local, user, date)
             self.tags = self._get_tags()
         except Abort as e:
             raise RepositoryError(e.message)
 
     @LazyProperty
     def bookmarks(self):
         """
         Gets bookmarks for this repository
         """
         return self._get_bookmarks()
 
     def _get_bookmarks(self):
         if self._empty:
             return {}
 
         sortkey = lambda ctx: ctx[0]  # sort by name
         _bookmarks = [(safe_unicode(n), hex(h),) for n, h in
                  self._repo._bookmarks.items()]
         return OrderedDict(sorted(_bookmarks, key=sortkey, reverse=True))
 
     def _get_all_revisions(self):
 
         return [self._repo[x].hex() for x in self._repo.filtered('visible').changelog.revs()]
 
     def get_diff(self, rev1, rev2, path='', ignore_whitespace=False,
                   context=3):
         """
         Returns (git like) *diff*, as plain text. Shows changes introduced by
         ``rev2`` since ``rev1``.
 
         :param rev1: Entry point from which diff is shown. Can be
           ``self.EMPTY_CHANGESET`` - in this case, patch showing all
           the changes since empty state of the repository until ``rev2``
         :param rev2: Until which revision changes should be shown.
         :param ignore_whitespace: If set to ``True``, would not show whitespace
           changes. Defaults to ``False``.
         :param context: How many lines before/after changed lines should be
-          shown. Defaults to ``3``.
+          shown. Defaults to ``3``. If negative value is passed-in, it will be
+          set to ``0`` instead.
         """
+
+        # Negative context values make no sense, and will result in
+        # errors. Ensure this does not happen.
+        if context < 0:
+            context = 0
+
         if hasattr(rev1, 'raw_id'):
             rev1 = getattr(rev1, 'raw_id')
 
         if hasattr(rev2, 'raw_id'):
             rev2 = getattr(rev2, 'raw_id')
 
         # Check if given revisions are present at repository (may raise
         # ChangesetDoesNotExistError)
         if rev1 != self.EMPTY_CHANGESET:
             self.get_changeset(rev1)
         self.get_changeset(rev2)
         if path:
             file_filter = match(self.path, '', [path])
         else:
             file_filter = None
 
         return ''.join(patch.diff(self._repo, rev1, rev2, match=file_filter,
                           opts=diffopts(git=True,
                                         showfunc=True,
                                         ignorews=ignore_whitespace,
                                         context=context)))
 
     @classmethod
     def _check_url(cls, url, repoui=None):
         """
         Function will check given url and try to verify if it's a valid
         link. Sometimes it may happened that mercurial will issue basic
         auth request that can cause whole API to hang when used from python
         or other external calls.
 
         On failures it'll raise urllib2.HTTPError, exception is also thrown
         when the return code is non 200
         """
         # check first if it's not an local url
         if os.path.isdir(url) or url.startswith('file:'):
             return True
 
         if url.startswith('ssh:'):
             # in case of invalid uri or authentication issues, sshpeer will
             # throw an exception.
             sshpeer(repoui or ui.ui(), url).lookup('tip')
             return True
 
         url_prefix = None
         if '+' in url[:url.find('://')]:
             url_prefix, url = url.split('+', 1)
 
         handlers = []
         url_obj = hg_url(url)
         test_uri, authinfo = url_obj.authinfo()
         url_obj.passwd = '*****'
         cleaned_uri = str(url_obj)
 
         if authinfo:
             #create a password manager
             passmgr = urllib2.HTTPPasswordMgrWithDefaultRealm()
             passmgr.add_password(*authinfo)
 
             handlers.extend((httpbasicauthhandler(passmgr),
                              httpdigestauthhandler(passmgr)))
 
         o = urllib2.build_opener(*handlers)
         o.addheaders = [('Content-Type', 'application/mercurial-0.1'),
                         ('Accept', 'application/mercurial-0.1')]
 
         q = {"cmd": 'between'}
         q.update({'pairs': "%s-%s" % ('0' * 40, '0' * 40)})
         qs = '?%s' % urllib.urlencode(q)
         cu = "%s%s" % (test_uri, qs)
         req = urllib2.Request(cu, None, {})
 
         try:
             resp = o.open(req)
             if resp.code != 200:
                 raise Exception('Return Code is not 200')
         except Exception as e:
             # means it cannot be cloned
             raise urllib2.URLError("[%s] org_exc: %s" % (cleaned_uri, e))
 
         if not url_prefix: # skip svn+http://... (and git+... too)
             # now check if it's a proper hg repo
             try:
                 httppeer(repoui or ui.ui(), url).lookup('tip')
             except Exception as e:
                 raise urllib2.URLError(
                     "url [%s] does not look like an hg repo org_exc: %s"
                     % (cleaned_uri, e))
 
         return True
 
     def _get_repo(self, create, src_url=None, update_after_clone=False):
         """
         Function will check for mercurial repository in given path and return
         a localrepo object. If there is no repository in that path it will
         raise an exception unless ``create`` parameter is set to True - in
         that case repository would be created and returned.
         If ``src_url`` is given, would try to clone repository from the
         location at given clone_point. Additionally it'll make update to
         working copy accordingly to ``update_after_clone`` flag
         """
 
         try:
             if src_url:
                 url = str(self._get_url(src_url))
                 opts = {}
                 if not update_after_clone:
                     opts.update({'noupdate': True})
                 MercurialRepository._check_url(url, self.baseui)
                 clone(self.baseui, url, self.path, **opts)
 
                 # Don't try to create if we've already cloned repo
                 create = False
             return localrepository(self.baseui, self.path, create=create)
         except (Abort, RepoError) as err:
             if create:
                 msg = "Cannot create repository at %s. Original error was %s"\
                     % (self.path, err)
             else:
                 msg = "Not valid repository at %s. Original error was %s"\
                     % (self.path, err)
             raise RepositoryError(msg)
 
     @LazyProperty
     def in_memory_changeset(self):
         return MercurialInMemoryChangeset(self)
 
     @LazyProperty
     def description(self):
         undefined_description = u'unknown'
         _desc = self._repo.ui.config('web', 'description', None, untrusted=True)
         return safe_unicode(_desc or undefined_description)
 
     @LazyProperty
     def contact(self):
         undefined_contact = u'Unknown'
         return safe_unicode(get_contact(self._repo.ui.config)
                             or undefined_contact)
 
     @LazyProperty
     def last_change(self):
         """
         Returns last change made on this repository as datetime object
         """
         return date_fromtimestamp(self._get_mtime(), makedate()[1])
 
     def _get_mtime(self):
         try:
             return time.mktime(self.get_changeset().date.timetuple())
         except RepositoryError:
             #fallback to filesystem
             cl_path = os.path.join(self.path, '.hg', "00changelog.i")
             st_path = os.path.join(self.path, '.hg', "store")
             if os.path.exists(cl_path):
                 return os.stat(cl_path).st_mtime
             else:
                 return os.stat(st_path).st_mtime
 
     def _get_revision(self, revision):
         """
         Gets an ID revision given as str. This will always return a fill
         40 char revision number
 
         :param revision: str or int or None
         """
         if isinstance(revision, unicode):
             revision = safe_str(revision)
 
         if self._empty:
             raise EmptyRepositoryError("There are no changesets yet")
 
         if revision in [-1, 'tip', None]:
             revision = 'tip'
 
         try:
             revision = hex(self._repo.lookup(revision))
         except (LookupError, ):
             msg = ("Ambiguous identifier `%s` for %s" % (revision, self))
             raise ChangesetDoesNotExistError(msg)
         except (IndexError, ValueError, RepoLookupError, TypeError):
             msg = ("Revision %s does not exist for %s" % (revision, self))
             raise ChangesetDoesNotExistError(msg)
 
         return revision
 
     def get_ref_revision(self, ref_type, ref_name):
         """
         Returns revision number for the given reference.
         """
         ref_name = safe_str(ref_name)
         if ref_type == 'rev' and not ref_name.strip('0'):
             return self.EMPTY_CHANGESET
         # lookup up the exact node id
         _revset_predicates = {
                 'branch': 'branch',
                 'book': 'bookmark',
                 'tag': 'tag',
                 'rev': 'id',
             }
         # avoid expensive branch(x) iteration over whole repo
         rev_spec = "%%s & %s(%%s)" % _revset_predicates[ref_type]
         try:
             revs = self._repo.revs(rev_spec, ref_name, ref_name)
         except LookupError:
             msg = ("Ambiguous identifier %s:%s for %s" % (ref_type, ref_name, self.name))
             raise ChangesetDoesNotExistError(msg)
         except RepoLookupError:
             msg = ("Revision %s:%s does not exist for %s" % (ref_type, ref_name, self.name))
             raise ChangesetDoesNotExistError(msg)
         if revs:
             try:
                 revision = revs.last()
             except AttributeError:
                 # removed in hg 3.2
                 revision = revs[-1]
         else:
             # TODO: just report 'not found'?
             revision = ref_name
 
         return self._get_revision(revision)
 
     def _get_archives(self, archive_name='tip'):
         allowed = self.baseui.configlist("web", "allow_archive",
                                          untrusted=True)
         for i in [('zip', '.zip'), ('gz', '.tar.gz'), ('bz2', '.tar.bz2')]:
             if i[0] in allowed or self._repo.ui.configbool("web",
                                                            "allow" + i[0],
                                                            untrusted=True):
                 yield {"type": i[0], "extension": i[1], "node": archive_name}
 
     def _get_url(self, url):
         """
         Returns normalized url. If schema is not given, would fall
         to filesystem
         (``file:///``) schema.
         """
         url = str(url)
         if url != 'default' and not '://' in url:
             url = "file:" + urllib.pathname2url(url)
         return url
 
     def get_hook_location(self):
         """
         returns absolute path to location where hooks are stored
         """
         return os.path.join(self.path, '.hg', '.hgrc')
 
     def get_changeset(self, revision=None):
         """
         Returns ``MercurialChangeset`` object representing repository's
         changeset at the given ``revision``.
         """
         revision = self._get_revision(revision)
         changeset = MercurialChangeset(repository=self, revision=revision)
         return changeset
 
     def get_changesets(self, start=None, end=None, start_date=None,
                        end_date=None, branch_name=None, reverse=False):
         """
         Returns iterator of ``MercurialChangeset`` objects from start to end
         (both are inclusive)
 
         :param start: None, str, int or mercurial lookup format
         :param end:  None, str, int or mercurial lookup format
         :param start_date:
         :param end_date:
         :param branch_name:
         :param reversed: return changesets in reversed order
         """
 
         start_raw_id = self._get_revision(start)
         start_pos = self.revisions.index(start_raw_id) if start else None
         end_raw_id = self._get_revision(end)
         end_pos = self.revisions.index(end_raw_id) if end else None
 
         if None not in [start, end] and start_pos > end_pos:
             raise RepositoryError("Start revision '%s' cannot be "
                                   "after end revision '%s'" % (start, end))
 
         if branch_name and branch_name not in self.allbranches.keys():
             msg = ("Branch %s not found in %s" % (branch_name, self))
             raise BranchDoesNotExistError(msg)
         if end_pos is not None:
             end_pos += 1
         #filter branches
         filter_ = []
         if branch_name:
             filter_.append('branch("%s")' % (branch_name))
 
         if start_date and not end_date:
             filter_.append('date(">%s")' % start_date)
         if end_date and not start_date:
             filter_.append('date("<%s")' % end_date)
         if start_date and end_date:
             filter_.append('date(">%s") and date("<%s")' % (start_date, end_date))
         if filter_:
             revisions = scmutil.revrange(self._repo, filter_)
         else:
             revisions = self.revisions
 
         # this is very much a hack to turn this into a list; a better solution
         # would be to get rid of this function entirely and use revsets
         revs = list(revisions)[start_pos:end_pos]
         if reverse:
             revs = reversed(revs)
 
         return CollectionGenerator(self, revs)
 
     def pull(self, url):
         """
         Tries to pull changes from external location.
         """
         url = self._get_url(url)
         other = peer(self._repo, {}, url)
         try:
             # hg 3.2 moved push / pull to exchange module
             from mercurial import exchange
             exchange.pull(self._repo, other, heads=None, force=None)
         except ImportError:
             self._repo.pull(other, heads=None, force=None)
         except Abort as err:
             # Propagate error but with vcs's type
             raise RepositoryError(str(err))
 
     @LazyProperty
     def workdir(self):
         """
         Returns ``Workdir`` instance for this repository.
         """
         return MercurialWorkdir(self)
 
     def get_config_value(self, section, name=None, config_file=None):
         """
         Returns configuration value for a given [``section``] and ``name``.
 
         :param section: Section we want to retrieve value from
         :param name: Name of configuration we want to retrieve
         :param config_file: A path to file which should be used to retrieve
           configuration from (might also be a list of file paths)
         """
         if config_file is None:
             config_file = []
         elif isinstance(config_file, basestring):
             config_file = [config_file]
 
         config = self._repo.ui
         for path in config_file:
             config.readconfig(path)
         return config.config(section, name)
 
     def get_user_name(self, config_file=None):
         """
         Returns user's name from global configuration file.
 
         :param config_file: A path to file which should be used to retrieve
           configuration from (might also be a list of file paths)
         """
         username = self.get_config_value('ui', 'username')
         if username:
             return author_name(username)
         return None
 
     def get_user_email(self, config_file=None):
         """
         Returns user's email from global configuration file.
 
         :param config_file: A path to file which should be used to retrieve
           configuration from (might also be a list of file paths)
         """
         username = self.get_config_value('ui', 'username')
         if username:
             return author_email(username)
         return None

kallithea/tests/vcs/test_git.py

@@ -1,850 +1,890 b''
 
 import os
 import sys
 import mock
 import datetime
 import urllib2
 from kallithea.lib.vcs.backends.git import GitRepository, GitChangeset
 from kallithea.lib.vcs.exceptions import RepositoryError, VCSError, NodeDoesNotExistError
 from kallithea.lib.vcs.nodes import NodeKind, FileNode, DirNode, NodeState
 from kallithea.lib.vcs.utils.compat import unittest
 from kallithea.model.scm import ScmModel
 from kallithea.tests.vcs.base import _BackendTestMixin
 from kallithea.tests.vcs.conf import TEST_GIT_REPO, TEST_GIT_REPO_CLONE, get_new_dir
 
 
 class GitRepositoryTest(unittest.TestCase):
 
     def __check_for_existing_repo(self):
         if os.path.exists(TEST_GIT_REPO_CLONE):
             self.fail('Cannot test git clone repo as location %s already '
                       'exists. You should manually remove it first.'
                       % TEST_GIT_REPO_CLONE)
 
     def setUp(self):
         self.repo = GitRepository(TEST_GIT_REPO)
 
     def test_wrong_repo_path(self):
         wrong_repo_path = '/tmp/errorrepo'
         self.assertRaises(RepositoryError, GitRepository, wrong_repo_path)
 
     def test_git_cmd_injection(self):
         repo_inject_path = TEST_GIT_REPO + '; echo "Cake";'
         with self.assertRaises(urllib2.URLError):
             # Should fail because URL will contain the parts after ; too
             urlerror_fail_repo = GitRepository(get_new_dir('injection-repo'), src_url=repo_inject_path, update_after_clone=True, create=True)
 
         with self.assertRaises(RepositoryError):
             # Should fail on direct clone call, which as of this writing does not happen outside of class
             clone_fail_repo = GitRepository(get_new_dir('injection-repo'), create=True)
             clone_fail_repo.clone(repo_inject_path, update_after_clone=True,)
 
         # Verify correct quoting of evil characters that should work on posix file systems
         if sys.platform == 'win32':
             # windows does not allow '"' in dir names
             tricky_path = get_new_dir("tricky-path-repo-$'`")
         else:
             tricky_path = get_new_dir("tricky-path-repo-$'\"`")
         successfully_cloned = GitRepository(tricky_path, src_url=TEST_GIT_REPO, update_after_clone=True, create=True)
         # Repo should have been created
         self.assertFalse(successfully_cloned._repo.bare)
 
         if sys.platform == 'win32':
             # windows does not allow '"' in dir names
             tricky_path_2 = get_new_dir("tricky-path-2-repo-$'`")
         else:
             tricky_path_2 = get_new_dir("tricky-path-2-repo-$'\"`")
         successfully_cloned2 = GitRepository(tricky_path_2, src_url=tricky_path, bare=True, create=True)
         # Repo should have been created and thus used correct quoting for clone
         self.assertTrue(successfully_cloned2._repo.bare)
 
         # Should pass because URL has been properly quoted
         successfully_cloned.pull(tricky_path_2)
         successfully_cloned2.fetch(tricky_path)
 
     def test_repo_create_with_spaces_in_path(self):
         repo_path = get_new_dir("path with spaces")
         repo = GitRepository(repo_path, src_url=None, bare=True, create=True)
         # Repo should have been created
         self.assertTrue(repo._repo.bare)
 
     def test_repo_clone(self):
         self.__check_for_existing_repo()
         repo = GitRepository(TEST_GIT_REPO)
         repo_clone = GitRepository(TEST_GIT_REPO_CLONE,
             src_url=TEST_GIT_REPO, create=True, update_after_clone=True)
         self.assertEqual(len(repo.revisions), len(repo_clone.revisions))
         # Checking hashes of changesets should be enough
         for changeset in repo.get_changesets():
             raw_id = changeset.raw_id
             self.assertEqual(raw_id, repo_clone.get_changeset(raw_id).raw_id)
 
     def test_repo_clone_with_spaces_in_path(self):
         repo_path = get_new_dir("path with spaces")
         successfully_cloned = GitRepository(repo_path, src_url=TEST_GIT_REPO, update_after_clone=True, create=True)
         # Repo should have been created
         self.assertFalse(successfully_cloned._repo.bare)
 
         successfully_cloned.pull(TEST_GIT_REPO)
         self.repo.fetch(repo_path)
 
     def test_repo_clone_without_create(self):
         self.assertRaises(RepositoryError, GitRepository,
             TEST_GIT_REPO_CLONE + '_wo_create', src_url=TEST_GIT_REPO)
 
     def test_repo_clone_with_update(self):
         repo = GitRepository(TEST_GIT_REPO)
         clone_path = TEST_GIT_REPO_CLONE + '_with_update'
         repo_clone = GitRepository(clone_path,
             create=True, src_url=TEST_GIT_REPO, update_after_clone=True)
         self.assertEqual(len(repo.revisions), len(repo_clone.revisions))
 
         #check if current workdir was updated
         fpath = os.path.join(clone_path, 'MANIFEST.in')
         self.assertEqual(True, os.path.isfile(fpath),
             'Repo was cloned and updated but file %s could not be found'
             % fpath)
 
     def test_repo_clone_without_update(self):
         repo = GitRepository(TEST_GIT_REPO)
         clone_path = TEST_GIT_REPO_CLONE + '_without_update'
         repo_clone = GitRepository(clone_path,
             create=True, src_url=TEST_GIT_REPO, update_after_clone=False)
         self.assertEqual(len(repo.revisions), len(repo_clone.revisions))
         #check if current workdir was *NOT* updated
         fpath = os.path.join(clone_path, 'MANIFEST.in')
         # Make sure it's not bare repo
         self.assertFalse(repo_clone._repo.bare)
         self.assertEqual(False, os.path.isfile(fpath),
             'Repo was cloned and updated but file %s was found'
             % fpath)
 
     def test_repo_clone_into_bare_repo(self):
         repo = GitRepository(TEST_GIT_REPO)
         clone_path = TEST_GIT_REPO_CLONE + '_bare.git'
         repo_clone = GitRepository(clone_path, create=True,
             src_url=repo.path, bare=True)
         self.assertTrue(repo_clone._repo.bare)
 
     def test_create_repo_is_not_bare_by_default(self):
         repo = GitRepository(get_new_dir('not-bare-by-default'), create=True)
         self.assertFalse(repo._repo.bare)
 
     def test_create_bare_repo(self):
         repo = GitRepository(get_new_dir('bare-repo'), create=True, bare=True)
         self.assertTrue(repo._repo.bare)
 
     def test_revisions(self):
         # there are 112 revisions (by now)
         # so we can assume they would be available from now on
         subset = set([
             'c1214f7e79e02fc37156ff215cd71275450cffc3',
             '38b5fe81f109cb111f549bfe9bb6b267e10bc557',
             'fa6600f6848800641328adbf7811fd2372c02ab2',
             '102607b09cdd60e2793929c4f90478be29f85a17',
             '49d3fd156b6f7db46313fac355dca1a0b94a0017',
             '2d1028c054665b962fa3d307adfc923ddd528038',
             'd7e0d30fbcae12c90680eb095a4f5f02505ce501',
             'ff7ca51e58c505fec0dd2491de52c622bb7a806b',
             'dd80b0f6cf5052f17cc738c2951c4f2070200d7f',
             '8430a588b43b5d6da365400117c89400326e7992',
             'd955cd312c17b02143c04fa1099a352b04368118',
             'f67b87e5c629c2ee0ba58f85197e423ff28d735b',
             'add63e382e4aabc9e1afdc4bdc24506c269b7618',
             'f298fe1189f1b69779a4423f40b48edf92a703fc',
             'bd9b619eb41994cac43d67cf4ccc8399c1125808',
             '6e125e7c890379446e98980d8ed60fba87d0f6d1',
             'd4a54db9f745dfeba6933bf5b1e79e15d0af20bd',
             '0b05e4ed56c802098dfc813cbe779b2f49e92500',
             '191caa5b2c81ed17c0794bf7bb9958f4dcb0b87e',
             '45223f8f114c64bf4d6f853e3c35a369a6305520',
             'ca1eb7957a54bce53b12d1a51b13452f95bc7c7e',
             'f5ea29fc42ef67a2a5a7aecff10e1566699acd68',
             '27d48942240f5b91dfda77accd2caac94708cc7d',
             '622f0eb0bafd619d2560c26f80f09e3b0b0d78af',
             'e686b958768ee96af8029fe19c6050b1a8dd3b2b'])
         self.assertTrue(subset.issubset(set(self.repo.revisions)))
 
 
 
     def test_slicing(self):
         #4 1 5 10 95
         for sfrom, sto, size in [(0, 4, 4), (1, 2, 1), (10, 15, 5),
                                  (10, 20, 10), (5, 100, 95)]:
             revs = list(self.repo[sfrom:sto])
             self.assertEqual(len(revs), size)
             self.assertEqual(revs[0], self.repo.get_changeset(sfrom))
             self.assertEqual(revs[-1], self.repo.get_changeset(sto - 1))
 
 
     def test_branches(self):
         # TODO: Need more tests here
         # Removed (those are 'remotes' branches for cloned repo)
         #self.assertTrue('master' in self.repo.branches)
         #self.assertTrue('gittree' in self.repo.branches)
         #self.assertTrue('web-branch' in self.repo.branches)
         for name, id in self.repo.branches.items():
             self.assertTrue(isinstance(
                 self.repo.get_changeset(id), GitChangeset))
 
     def test_tags(self):
         # TODO: Need more tests here
         self.assertTrue('v0.1.1' in self.repo.tags)
         self.assertTrue('v0.1.2' in self.repo.tags)
         for name, id in self.repo.tags.items():
             self.assertTrue(isinstance(
                 self.repo.get_changeset(id), GitChangeset))
 
     def _test_single_changeset_cache(self, revision):
         chset = self.repo.get_changeset(revision)
         self.assertTrue(revision in self.repo.changesets)
         self.assertTrue(chset is self.repo.changesets[revision])
 
     def test_initial_changeset(self):
         id = self.repo.revisions[0]
         init_chset = self.repo.get_changeset(id)
         self.assertEqual(init_chset.message, 'initial import\n')
         self.assertEqual(init_chset.author,
             'Marcin Kuzminski <[e-mail]>')
         for path in ('vcs/__init__.py',
                      'vcs/backends/BaseRepository.py',
                      'vcs/backends/__init__.py'):
             self.assertTrue(isinstance(init_chset.get_node(path), FileNode))
         for path in ('', 'vcs', 'vcs/backends'):
             self.assertTrue(isinstance(init_chset.get_node(path), DirNode))
 
         self.assertRaises(NodeDoesNotExistError, init_chset.get_node, path='foobar')
 
         node = init_chset.get_node('vcs/')
         self.assertTrue(hasattr(node, 'kind'))
         self.assertEqual(node.kind, NodeKind.DIR)
 
         node = init_chset.get_node('vcs')
         self.assertTrue(hasattr(node, 'kind'))
         self.assertEqual(node.kind, NodeKind.DIR)
 
         node = init_chset.get_node('vcs/__init__.py')
         self.assertTrue(hasattr(node, 'kind'))
         self.assertEqual(node.kind, NodeKind.FILE)
 
     def test_not_existing_changeset(self):
         self.assertRaises(RepositoryError, self.repo.get_changeset,
             'f' * 40)
 
     def test_changeset10(self):
 
         chset10 = self.repo.get_changeset(self.repo.revisions[9])
         README = """===
 VCS
 ===
 
 Various Version Control System management abstraction layer for Python.
 
 Introduction
 ------------
 
 TODO: To be written...
 
 """
         node = chset10.get_node('README.rst')
         self.assertEqual(node.kind, NodeKind.FILE)
         self.assertEqual(node.content, README)
 
 
 class GitChangesetTest(unittest.TestCase):
 
     def setUp(self):
         self.repo = GitRepository(TEST_GIT_REPO)
 
     def test_default_changeset(self):
         tip = self.repo.get_changeset()
         self.assertEqual(tip, self.repo.get_changeset(None))
         self.assertEqual(tip, self.repo.get_changeset('tip'))
 
     def test_root_node(self):
         tip = self.repo.get_changeset()
         self.assertTrue(tip.root is tip.get_node(''))
 
     def test_lazy_fetch(self):
         """
         Test if changeset's nodes expands and are cached as we walk through
         the revision. This test is somewhat hard to write as order of tests
         is a key here. Written by running command after command in a shell.
         """
         hex = '2a13f185e4525f9d4b59882791a2d397b90d5ddc'
         self.assertTrue(hex in self.repo.revisions)
         chset = self.repo.get_changeset(hex)
         self.assertTrue(len(chset.nodes) == 0)
         root = chset.root
         self.assertTrue(len(chset.nodes) == 1)
         self.assertTrue(len(root.nodes) == 8)
         # accessing root.nodes updates chset.nodes
         self.assertTrue(len(chset.nodes) == 9)
 
         docs = root.get_node('docs')
         # we haven't yet accessed anything new as docs dir was already cached
         self.assertTrue(len(chset.nodes) == 9)
         self.assertTrue(len(docs.nodes) == 8)
         # accessing docs.nodes updates chset.nodes
         self.assertTrue(len(chset.nodes) == 17)
 
         self.assertTrue(docs is chset.get_node('docs'))
         self.assertTrue(docs is root.nodes[0])
         self.assertTrue(docs is root.dirs[0])
         self.assertTrue(docs is chset.get_node('docs'))
 
     def test_nodes_with_changeset(self):
         hex = '2a13f185e4525f9d4b59882791a2d397b90d5ddc'
         chset = self.repo.get_changeset(hex)
         root = chset.root
         docs = root.get_node('docs')
         self.assertTrue(docs is chset.get_node('docs'))
         api = docs.get_node('api')
         self.assertTrue(api is chset.get_node('docs/api'))
         index = api.get_node('index.rst')
         self.assertTrue(index is chset.get_node('docs/api/index.rst'))
         self.assertTrue(index is chset.get_node('docs')\
             .get_node('api')\
             .get_node('index.rst'))
 
     def test_branch_and_tags(self):
         """
         rev0 = self.repo.revisions[0]
         chset0 = self.repo.get_changeset(rev0)
         self.assertEqual(chset0.branch, 'master')
         self.assertEqual(chset0.tags, [])
 
         rev10 = self.repo.revisions[10]
         chset10 = self.repo.get_changeset(rev10)
         self.assertEqual(chset10.branch, 'master')
         self.assertEqual(chset10.tags, [])
 
         rev44 = self.repo.revisions[44]
         chset44 = self.repo.get_changeset(rev44)
         self.assertEqual(chset44.branch, 'web-branch')
 
         tip = self.repo.get_changeset('tip')
         self.assertTrue('tip' in tip.tags)
         """
         # Those tests would fail - branches are now going
         # to be changed at main API in order to support git backend
         pass
 
     def _test_slices(self, limit, offset):
         count = self.repo.count()
         changesets = self.repo.get_changesets(limit=limit, offset=offset)
         idx = 0
         for changeset in changesets:
             rev = offset + idx
             idx += 1
             rev_id = self.repo.revisions[rev]
             if idx > limit:
                 self.fail("Exceeded limit already (getting revision %s, "
                     "there are %s total revisions, offset=%s, limit=%s)"
                     % (rev_id, count, offset, limit))
             self.assertEqual(changeset, self.repo.get_changeset(rev_id))
         result = list(self.repo.get_changesets(limit=limit, offset=offset))
         start = offset
         end = limit and offset + limit or None
         sliced = list(self.repo[start:end])
         self.failUnlessEqual(result, sliced,
             msg="Comparison failed for limit=%s, offset=%s"
             "(get_changeset returned: %s and sliced: %s"
             % (limit, offset, result, sliced))
 
     def _test_file_size(self, revision, path, size):
         node = self.repo.get_changeset(revision).get_node(path)
         self.assertTrue(node.is_file())
         self.assertEqual(node.size, size)
 
     def test_file_size(self):
         to_check = (
             ('c1214f7e79e02fc37156ff215cd71275450cffc3',
                 'vcs/backends/BaseRepository.py', 502),
             ('d7e0d30fbcae12c90680eb095a4f5f02505ce501',
                 'vcs/backends/hg.py', 854),
             ('6e125e7c890379446e98980d8ed60fba87d0f6d1',
                 'setup.py', 1068),
 
             ('d955cd312c17b02143c04fa1099a352b04368118',
                 'vcs/backends/base.py', 2921),
             ('ca1eb7957a54bce53b12d1a51b13452f95bc7c7e',
                 'vcs/backends/base.py', 3936),
             ('f50f42baeed5af6518ef4b0cb2f1423f3851a941',
                 'vcs/backends/base.py', 6189),
         )
         for revision, path, size in to_check:
             self._test_file_size(revision, path, size)
 
     def _test_dir_size(self, revision, path, size):
         node = self.repo.get_changeset(revision).get_node(path)
         self.assertEqual(node.size, size)
 
     def test_dir_size(self):
         to_check = (
             ('5f2c6ee195929b0be80749243c18121c9864a3b3', '/', 674076),
             ('7ab37bc680b4aa72c34d07b230c866c28e9fc204', '/', 674049),
             ('6892503fb8f2a552cef5f4d4cc2cdbd13ae1cd2f', '/', 671830),
         )
         for revision, path, size in to_check:
             self._test_dir_size(revision, path, size)
 
     def test_repo_size(self):
         self.assertEqual(self.repo.size, 674076)
 
     def test_file_history(self):
         # we can only check if those revisions are present in the history
         # as we cannot update this test every time file is changed
         files = {
             'setup.py': [
                 '54386793436c938cff89326944d4c2702340037d',
                 '51d254f0ecf5df2ce50c0b115741f4cf13985dab',
                 '998ed409c795fec2012b1c0ca054d99888b22090',
                 '5e0eb4c47f56564395f76333f319d26c79e2fb09',
                 '0115510b70c7229dbc5dc49036b32e7d91d23acd',
                 '7cb3fd1b6d8c20ba89e2264f1c8baebc8a52d36e',
                 '2a13f185e4525f9d4b59882791a2d397b90d5ddc',
                 '191caa5b2c81ed17c0794bf7bb9958f4dcb0b87e',
                 'ff7ca51e58c505fec0dd2491de52c622bb7a806b',
             ],
             'vcs/nodes.py': [
                 '33fa3223355104431402a888fa77a4e9956feb3e',
                 'fa014c12c26d10ba682fadb78f2a11c24c8118e1',
                 'e686b958768ee96af8029fe19c6050b1a8dd3b2b',
                 'ab5721ca0a081f26bf43d9051e615af2cc99952f',
                 'c877b68d18e792a66b7f4c529ea02c8f80801542',
                 '4313566d2e417cb382948f8d9d7c765330356054',
                 '6c2303a793671e807d1cfc70134c9ca0767d98c2',
                 '54386793436c938cff89326944d4c2702340037d',
                 '54000345d2e78b03a99d561399e8e548de3f3203',
                 '1c6b3677b37ea064cb4b51714d8f7498f93f4b2b',
                 '2d03ca750a44440fb5ea8b751176d1f36f8e8f46',
                 '2a08b128c206db48c2f0b8f70df060e6db0ae4f8',
                 '30c26513ff1eb8e5ce0e1c6b477ee5dc50e2f34b',
                 'ac71e9503c2ca95542839af0ce7b64011b72ea7c',
                 '12669288fd13adba2a9b7dd5b870cc23ffab92d2',
                 '5a0c84f3e6fe3473e4c8427199d5a6fc71a9b382',
                 '12f2f5e2b38e6ff3fbdb5d722efed9aa72ecb0d5',
                 '5eab1222a7cd4bfcbabc218ca6d04276d4e27378',
                 'f50f42baeed5af6518ef4b0cb2f1423f3851a941',
                 'd7e390a45f6aa96f04f5e7f583ad4f867431aa25',
                 'f15c21f97864b4f071cddfbf2750ec2e23859414',
                 'e906ef056cf539a4e4e5fc8003eaf7cf14dd8ade',
                 'ea2b108b48aa8f8c9c4a941f66c1a03315ca1c3b',
                 '84dec09632a4458f79f50ddbbd155506c460b4f9',
                 '0115510b70c7229dbc5dc49036b32e7d91d23acd',
                 '2a13f185e4525f9d4b59882791a2d397b90d5ddc',
                 '3bf1c5868e570e39569d094f922d33ced2fa3b2b',
                 'b8d04012574729d2c29886e53b1a43ef16dd00a1',
                 '6970b057cffe4aab0a792aa634c89f4bebf01441',
                 'dd80b0f6cf5052f17cc738c2951c4f2070200d7f',
                 'ff7ca51e58c505fec0dd2491de52c622bb7a806b',
             ],
             'vcs/backends/git.py': [
                 '4cf116ad5a457530381135e2f4c453e68a1b0105',
                 '9a751d84d8e9408e736329767387f41b36935153',
                 'cb681fb539c3faaedbcdf5ca71ca413425c18f01',
                 '428f81bb652bcba8d631bce926e8834ff49bdcc6',
                 '180ab15aebf26f98f714d8c68715e0f05fa6e1c7',
                 '2b8e07312a2e89e92b90426ab97f349f4bce2a3a',
                 '50e08c506174d8645a4bb517dd122ac946a0f3bf',
                 '54000345d2e78b03a99d561399e8e548de3f3203',
             ],
         }
         for path, revs in files.items():
             node = self.repo.get_changeset(revs[0]).get_node(path)
             node_revs = [chset.raw_id for chset in node.history]
             self.assertTrue(set(revs).issubset(set(node_revs)),
                 "We assumed that %s is subset of revisions for which file %s "
                 "has been changed, and history of that node returned: %s"
                 % (revs, path, node_revs))
 
     def test_file_annotate(self):
         files = {
             'vcs/backends/__init__.py': {
                 'c1214f7e79e02fc37156ff215cd71275450cffc3': {
                     'lines_no': 1,
                     'changesets': [
                         'c1214f7e79e02fc37156ff215cd71275450cffc3',
                     ],
                 },
                 '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647': {
                     'lines_no': 21,
                     'changesets': [
                         '49d3fd156b6f7db46313fac355dca1a0b94a0017',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                     ],
                 },
                 'e29b67bd158580fc90fc5e9111240b90e6e86064': {
                     'lines_no': 32,
                     'changesets': [
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '5eab1222a7cd4bfcbabc218ca6d04276d4e27378',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '54000345d2e78b03a99d561399e8e548de3f3203',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '78c3f0c23b7ee935ec276acb8b8212444c33c396',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '2a13f185e4525f9d4b59882791a2d397b90d5ddc',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '78c3f0c23b7ee935ec276acb8b8212444c33c396',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '992f38217b979d0b0987d0bae3cc26dac85d9b19',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                         '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
                     ],
                 },
             },
         }
 
         for fname, revision_dict in files.items():
             for rev, data in revision_dict.items():
                 cs = self.repo.get_changeset(rev)
 
                 l1_1 = [x[1] for x in cs.get_file_annotate(fname)]
                 l1_2 = [x[2]().raw_id for x in cs.get_file_annotate(fname)]
                 self.assertEqual(l1_1, l1_2)
                 l1 = l1_1
                 l2 = files[fname][rev]['changesets']
                 self.assertTrue(l1 == l2 , "The lists of revision for %s@rev %s"
                                 "from annotation list should match each other, "
                                 "got \n%s \nvs \n%s " % (fname, rev, l1, l2))
 
     def test_files_state(self):
         """
         Tests state of FileNodes.
         """
         node = self.repo\
             .get_changeset('e6ea6d16e2f26250124a1f4b4fe37a912f9d86a0')\
             .get_node('vcs/utils/diffs.py')
         self.assertTrue(node.state, NodeState.ADDED)
         self.assertTrue(node.added)
         self.assertFalse(node.changed)
         self.assertFalse(node.not_changed)
         self.assertFalse(node.removed)
 
         node = self.repo\
             .get_changeset('33fa3223355104431402a888fa77a4e9956feb3e')\
             .get_node('.hgignore')
         self.assertTrue(node.state, NodeState.CHANGED)
         self.assertFalse(node.added)
         self.assertTrue(node.changed)
         self.assertFalse(node.not_changed)
         self.assertFalse(node.removed)
 
         node = self.repo\
             .get_changeset('e29b67bd158580fc90fc5e9111240b90e6e86064')\
             .get_node('setup.py')
         self.assertTrue(node.state, NodeState.NOT_CHANGED)
         self.assertFalse(node.added)
         self.assertFalse(node.changed)
         self.assertTrue(node.not_changed)
         self.assertFalse(node.removed)
 
         # If node has REMOVED state then trying to fetch it would raise
         # ChangesetError exception
         chset = self.repo.get_changeset(
             'fa6600f6848800641328adbf7811fd2372c02ab2')
         path = 'vcs/backends/BaseRepository.py'
         self.assertRaises(NodeDoesNotExistError, chset.get_node, path)
         # but it would be one of ``removed`` (changeset's attribute)
         self.assertTrue(path in [rf.path for rf in chset.removed])
 
         chset = self.repo.get_changeset(
             '54386793436c938cff89326944d4c2702340037d')
         changed = ['setup.py', 'tests/test_nodes.py', 'vcs/backends/hg.py',
             'vcs/nodes.py']
         self.assertEqual(set(changed), set([f.path for f in chset.changed]))
 
     def test_commit_message_is_unicode(self):
         for cs in self.repo:
             self.assertEqual(type(cs.message), unicode)
 
     def test_changeset_author_is_unicode(self):
         for cs in self.repo:
             self.assertEqual(type(cs.author), unicode)
 
     def test_repo_files_content_is_unicode(self):
         changeset = self.repo.get_changeset()
         for node in changeset.get_node('/'):
             if node.is_file():
                 self.assertEqual(type(node.content), unicode)
 
     def test_wrong_path(self):
         # There is 'setup.py' in the root dir but not there:
         path = 'foo/bar/setup.py'
         tip = self.repo.get_changeset()
         self.assertRaises(VCSError, tip.get_node, path)
 
     def test_author_email(self):
         self.assertEqual('[e-mail]',
           self.repo.get_changeset('c1214f7e79e02fc37156ff215cd71275450cffc3')\
           .author_email)
         self.assertEqual('[e-mail]',
           self.repo.get_changeset('ff7ca51e58c505fec0dd2491de52c622bb7a806b')\
           .author_email)
         self.assertEqual('none@none',
           self.repo.get_changeset('8430a588b43b5d6da365400117c89400326e7992')\
           .author_email)
 
     def test_author_username(self):
         self.assertEqual('Marcin Kuzminski',
           self.repo.get_changeset('c1214f7e79e02fc37156ff215cd71275450cffc3')\
           .author_name)
         self.assertEqual('Lukasz Balcerzak',
           self.repo.get_changeset('ff7ca51e58c505fec0dd2491de52c622bb7a806b')\
           .author_name)
         self.assertEqual('marcink',
           self.repo.get_changeset('8430a588b43b5d6da365400117c89400326e7992')\
           .author_name)
 
 
 class GitSpecificTest(unittest.TestCase):
 
     def test_error_is_raised_for_added_if_diff_name_status_is_wrong(self):
         repo = mock.MagicMock()
         changeset = GitChangeset(repo, 'foobar')
         changeset._diff_name_status = 'foobar'
         with self.assertRaises(VCSError):
             changeset.added
 
     def test_error_is_raised_for_changed_if_diff_name_status_is_wrong(self):
         repo = mock.MagicMock()
         changeset = GitChangeset(repo, 'foobar')
         changeset._diff_name_status = 'foobar'
         with self.assertRaises(VCSError):
             changeset.added
 
     def test_error_is_raised_for_removed_if_diff_name_status_is_wrong(self):
         repo = mock.MagicMock()
         changeset = GitChangeset(repo, 'foobar')
         changeset._diff_name_status = 'foobar'
         with self.assertRaises(VCSError):
             changeset.added
 
 
 class GitSpecificWithRepoTest(_BackendTestMixin, unittest.TestCase):
     backend_alias = 'git'
 
     @classmethod
     def _get_commits(cls):
         return [
             {
                 'message': 'Initial',
                 'author': 'Joe Doe <[e-mail]>',
                 'date': datetime.datetime(2010, 1, 1, 20),
                 'added': [
                     FileNode('foobar/static/js/admin/base.js', content='base'),
                     FileNode('foobar/static/admin', content='admin',
                         mode=0120000), # this is a link
                     FileNode('foo', content='foo'),
                 ],
             },
             {
                 'message': 'Second',
                 'author': 'Joe Doe <[e-mail]>',
                 'date': datetime.datetime(2010, 1, 1, 22),
                 'added': [
                     FileNode('foo2', content='foo2'),
                 ],
             },
         ]
 
     def test_paths_slow_traversing(self):
         cs = self.repo.get_changeset()
         self.assertEqual(cs.get_node('foobar').get_node('static').get_node('js')
             .get_node('admin').get_node('base.js').content, 'base')
 
     def test_paths_fast_traversing(self):
         cs = self.repo.get_changeset()
         self.assertEqual(cs.get_node('foobar/static/js/admin/base.js').content,
             'base')
 
     def test_workdir_get_branch(self):
         self.repo.run_git_command(['checkout', '-b', 'production'])
         # Regression test: one of following would fail if we don't check
         # .git/HEAD file
         self.repo.run_git_command(['checkout', 'production'])
         self.assertEqual(self.repo.workdir.get_branch(), 'production')
         self.repo.run_git_command(['checkout', 'master'])
         self.assertEqual(self.repo.workdir.get_branch(), 'master')
 
     def test_get_diff_runs_git_command_with_hashes(self):
         self.repo.run_git_command = mock.Mock(return_value=['', ''])
         self.repo.get_diff(0, 1)
         self.repo.run_git_command.assert_called_once_with(
             ['diff', '-U3', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
              self.repo._get_revision(0), self.repo._get_revision(1)])
 
     def test_get_diff_runs_git_command_with_str_hashes(self):
         self.repo.run_git_command = mock.Mock(return_value=['', ''])
         self.repo.get_diff(self.repo.EMPTY_CHANGESET, 1)
         self.repo.run_git_command.assert_called_once_with(
             ['show', '-U3', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
              self.repo._get_revision(1)])
 
     def test_get_diff_runs_git_command_with_path_if_its_given(self):
         self.repo.run_git_command = mock.Mock(return_value=['', ''])
         self.repo.get_diff(0, 1, 'foo')
         self.repo.run_git_command.assert_called_once_with(
             ['diff', '-U3', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
              self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
 
+    def test_get_diff_does_not_sanitize_valid_context(self):
+        almost_overflowed_long_int = 2**31-1
+
+        self.repo.run_git_command = mock.Mock(return_value=['', ''])
+        self.repo.get_diff(0, 1, 'foo', context=almost_overflowed_long_int)
+        self.repo.run_git_command.assert_called_once_with(
+            ['diff', '-U' + str(almost_overflowed_long_int), '--full-index', '--binary', '-p', '-M', '--abbrev=40',
+             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
+
+    def test_get_diff_sanitizes_overflowing_context(self):
+        overflowed_long_int = 2**31
+        sanitized_overflowed_long_int = overflowed_long_int-1
+
+        self.repo.run_git_command = mock.Mock(return_value=['', ''])
+        self.repo.get_diff(0, 1, 'foo', context=overflowed_long_int)
+
+        self.repo.run_git_command.assert_called_once_with(
+            ['diff', '-U' + str(sanitized_overflowed_long_int), '--full-index', '--binary', '-p', '-M', '--abbrev=40',
+             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
+
+    def test_get_diff_does_not_sanitize_zero_context(self):
+        zero_context = 0
+
+        self.repo.run_git_command = mock.Mock(return_value=['', ''])
+        self.repo.get_diff(0, 1, 'foo', context=zero_context)
+
+        self.repo.run_git_command.assert_called_once_with(
+            ['diff', '-U' + str(zero_context), '--full-index', '--binary', '-p', '-M', '--abbrev=40',
+             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
+
+    def test_get_diff_sanitizes_negative_context(self):
+        negative_context = -10
+
+        self.repo.run_git_command = mock.Mock(return_value=['', ''])
+        self.repo.get_diff(0, 1, 'foo', context=negative_context)
+
+        self.repo.run_git_command.assert_called_once_with(
+            ['diff', '-U0', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
+             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
+
 
 class GitRegressionTest(_BackendTestMixin, unittest.TestCase):
     backend_alias = 'git'
 
     @classmethod
     def _get_commits(cls):
         return [
             {
                 'message': 'Initial',
                 'author': 'Joe Doe <[e-mail]>',
                 'date': datetime.datetime(2010, 1, 1, 20),
                 'added': [
                     FileNode('bot/__init__.py', content='base'),
                     FileNode('bot/templates/404.html', content='base'),
                     FileNode('bot/templates/500.html', content='base'),
                 ],
             },
             {
                 'message': 'Second',
                 'author': 'Joe Doe <[e-mail]>',
                 'date': datetime.datetime(2010, 1, 1, 22),
                 'added': [
                     FileNode('bot/build/migrations/1.py', content='foo2'),
                     FileNode('bot/build/migrations/2.py', content='foo2'),
                     FileNode('bot/build/static/templates/f.html', content='foo2'),
                     FileNode('bot/build/static/templates/f1.html', content='foo2'),
                     FileNode('bot/build/templates/err.html', content='foo2'),
                     FileNode('bot/build/templates/err2.html', content='foo2'),
                 ],
             },
         ]
 
     def test_similar_paths(self):
         cs = self.repo.get_changeset()
         paths = lambda *n:[x.path for x in n]
         self.assertEqual(paths(*cs.get_nodes('bot')), ['bot/build', 'bot/templates', 'bot/__init__.py'])
         self.assertEqual(paths(*cs.get_nodes('bot/build')), ['bot/build/migrations', 'bot/build/static', 'bot/build/templates'])
         self.assertEqual(paths(*cs.get_nodes('bot/build/static')), ['bot/build/static/templates'])
         # this get_nodes below causes troubles !
         self.assertEqual(paths(*cs.get_nodes('bot/build/static/templates')), ['bot/build/static/templates/f.html', 'bot/build/static/templates/f1.html'])
         self.assertEqual(paths(*cs.get_nodes('bot/build/templates')), ['bot/build/templates/err.html', 'bot/build/templates/err2.html'])
         self.assertEqual(paths(*cs.get_nodes('bot/templates/')), ['bot/templates/404.html', 'bot/templates/500.html'])
 
 
 class GitHooksTest(unittest.TestCase):
     """
     Tests related to hook functionality of Git repositories.
     """
 
     def setUp(self):
         # For each run we want a fresh repo.
         self.repo_directory = get_new_dir("githookrepo")
         self.repo = GitRepository(self.repo_directory, create=True)
 
         # Create a dictionary where keys are hook names, and values are paths to
         # them. Deduplicates code in tests a bit.
         self.hook_directory = self.repo.get_hook_location()
         self.kallithea_hooks = {h: os.path.join(self.hook_directory, h) for h in ("pre-receive", "post-receive")}
 
     def test_hooks_created_if_missing(self):
         """
         Tests if hooks are installed in repository if they are missing.
         """
 
         for hook, hook_path in self.kallithea_hooks.iteritems():
             if os.path.exists(hook_path):
                 os.remove(hook_path)
 
         ScmModel().install_git_hooks(repo=self.repo)
 
         for hook, hook_path in self.kallithea_hooks.iteritems():
             self.assertTrue(os.path.exists(hook_path))
 
     def test_kallithea_hooks_updated(self):
         """
         Tests if hooks are updated if they are Kallithea hooks already.
         """
 
         for hook, hook_path in self.kallithea_hooks.iteritems():
             with open(hook_path, "w") as f:
                 f.write("KALLITHEA_HOOK_VER=0.0.0\nJUST_BOGUS")
 
         ScmModel().install_git_hooks(repo=self.repo)
 
         for hook, hook_path in self.kallithea_hooks.iteritems():
             with open(hook_path) as f:
                 self.assertNotIn("JUST_BOGUS", f.read())
 
     def test_custom_hooks_untouched(self):
         """
         Tests if hooks are left untouched if they are not Kallithea hooks.
         """
 
         for hook, hook_path in self.kallithea_hooks.iteritems():
             with open(hook_path, "w") as f:
                 f.write("#!/bin/bash\n#CUSTOM_HOOK")
 
         ScmModel().install_git_hooks(repo=self.repo)
 
         for hook, hook_path in self.kallithea_hooks.iteritems():
             with open(hook_path) as f:
                 self.assertIn("CUSTOM_HOOK", f.read())
 
     def test_custom_hooks_forced_update(self):
         """
         Tests if hooks are forcefully updated even though they are custom hooks.
         """
 
         for hook, hook_path in self.kallithea_hooks.iteritems():
             with open(hook_path, "w") as f:
                 f.write("#!/bin/bash\n#CUSTOM_HOOK")
 
         ScmModel().install_git_hooks(repo=self.repo, force_create=True)
 
         for hook, hook_path in self.kallithea_hooks.iteritems():
             with open(hook_path) as f:
                 self.assertIn("KALLITHEA_HOOK_VER", f.read())
 
 
 if __name__ == '__main__':
     unittest.main()

kallithea/tests/vcs/test_hg.py

@@ -1,567 +1,587 b''
 
 import os
+
+import mock
+
 from kallithea.lib.vcs.backends.hg import MercurialRepository, MercurialChangeset
 from kallithea.lib.vcs.exceptions import RepositoryError, VCSError, NodeDoesNotExistError
 from kallithea.lib.vcs.nodes import NodeKind, NodeState
 from kallithea.tests.vcs.conf import TEST_HG_REPO, TEST_HG_REPO_CLONE, \
     TEST_HG_REPO_PULL
 from kallithea.lib.vcs.utils.compat import unittest
 
 
 class MercurialRepositoryTest(unittest.TestCase):
 
     def __check_for_existing_repo(self):
         if os.path.exists(TEST_HG_REPO_CLONE):
             self.fail('Cannot test mercurial clone repo as location %s already '
                       'exists. You should manually remove it first.'
                       % TEST_HG_REPO_CLONE)
 
     def setUp(self):
         self.repo = MercurialRepository(TEST_HG_REPO)
 
     def test_wrong_repo_path(self):
         wrong_repo_path = '/tmp/errorrepo'
         self.assertRaises(RepositoryError, MercurialRepository, wrong_repo_path)
 
     def test_unicode_path_repo(self):
         self.assertRaises(VCSError,lambda:MercurialRepository(u'iShouldFail'))
 
     def test_repo_clone(self):
         self.__check_for_existing_repo()
         repo = MercurialRepository(TEST_HG_REPO)
         repo_clone = MercurialRepository(TEST_HG_REPO_CLONE,
             src_url=TEST_HG_REPO, update_after_clone=True)
         self.assertEqual(len(repo.revisions), len(repo_clone.revisions))
         # Checking hashes of changesets should be enough
         for changeset in repo.get_changesets():
             raw_id = changeset.raw_id
             self.assertEqual(raw_id, repo_clone.get_changeset(raw_id).raw_id)
 
     def test_repo_clone_with_update(self):
         repo = MercurialRepository(TEST_HG_REPO)
         repo_clone = MercurialRepository(TEST_HG_REPO_CLONE + '_w_update',
             src_url=TEST_HG_REPO, update_after_clone=True)
         self.assertEqual(len(repo.revisions), len(repo_clone.revisions))
 
         #check if current workdir was updated
         self.assertEqual(os.path.isfile(os.path.join(TEST_HG_REPO_CLONE \
                                                     + '_w_update',
                                                     'MANIFEST.in')), True,)
 
     def test_repo_clone_without_update(self):
         repo = MercurialRepository(TEST_HG_REPO)
         repo_clone = MercurialRepository(TEST_HG_REPO_CLONE + '_wo_update',
             src_url=TEST_HG_REPO, update_after_clone=False)
         self.assertEqual(len(repo.revisions), len(repo_clone.revisions))
         self.assertEqual(os.path.isfile(os.path.join(TEST_HG_REPO_CLONE \
                                                     + '_wo_update',
                                                     'MANIFEST.in')), False,)
 
     def test_pull(self):
         if os.path.exists(TEST_HG_REPO_PULL):
             self.fail('Cannot test mercurial pull command as location %s '
                       'already exists. You should manually remove it first'
                       % TEST_HG_REPO_PULL)
         repo_new = MercurialRepository(TEST_HG_REPO_PULL, create=True)
         self.assertTrue(len(self.repo.revisions) > len(repo_new.revisions))
 
         repo_new.pull(self.repo.path)
         repo_new = MercurialRepository(TEST_HG_REPO_PULL)
         self.assertTrue(len(self.repo.revisions) == len(repo_new.revisions))
 
     def test_revisions(self):
         # there are 21 revisions at bitbucket now
         # so we can assume they would be available from now on
         subset = set(['b986218ba1c9b0d6a259fac9b050b1724ed8e545',
                  '3d8f361e72ab303da48d799ff1ac40d5ac37c67e',
                  '6cba7170863a2411822803fa77a0a264f1310b35',
                  '56349e29c2af3ac913b28bde9a2c6154436e615b',
                  '2dda4e345facb0ccff1a191052dd1606dba6781d',
                  '6fff84722075f1607a30f436523403845f84cd9e',
                  '7d4bc8ec6be56c0f10425afb40b6fc315a4c25e7',
                  '3803844fdbd3b711175fc3da9bdacfcd6d29a6fb',
                  'dc5d2c0661b61928834a785d3e64a3f80d3aad9c',
                  'be90031137367893f1c406e0a8683010fd115b79',
                  'db8e58be770518cbb2b1cdfa69146e47cd481481',
                  '84478366594b424af694a6c784cb991a16b87c21',
                  '17f8e105dddb9f339600389c6dc7175d395a535c',
                  '20a662e756499bde3095ffc9bc0643d1def2d0eb',
                  '2e319b85e70a707bba0beff866d9f9de032aa4f9',
                  '786facd2c61deb9cf91e9534735124fb8fc11842',
                  '94593d2128d38210a2fcd1aabff6dda0d6d9edf8',
                  'aa6a0de05b7612707db567078e130a6cd114a9a7',
                  'eada5a770da98ab0dd7325e29d00e0714f228d09'
                 ])
         self.assertTrue(subset.issubset(set(self.repo.revisions)))
 
 
         # check if we have the proper order of revisions
         org = ['b986218ba1c9b0d6a259fac9b050b1724ed8e545',
                 '3d8f361e72ab303da48d799ff1ac40d5ac37c67e',
                 '6cba7170863a2411822803fa77a0a264f1310b35',
                 '56349e29c2af3ac913b28bde9a2c6154436e615b',
                 '2dda4e345facb0ccff1a191052dd1606dba6781d',
                 '6fff84722075f1607a30f436523403845f84cd9e',
                 '7d4bc8ec6be56c0f10425afb40b6fc315a4c25e7',
                 '3803844fdbd3b711175fc3da9bdacfcd6d29a6fb',
                 'dc5d2c0661b61928834a785d3e64a3f80d3aad9c',
                 'be90031137367893f1c406e0a8683010fd115b79',
                 'db8e58be770518cbb2b1cdfa69146e47cd481481',
                 '84478366594b424af694a6c784cb991a16b87c21',
                 '17f8e105dddb9f339600389c6dc7175d395a535c',
                 '20a662e756499bde3095ffc9bc0643d1def2d0eb',
                 '2e319b85e70a707bba0beff866d9f9de032aa4f9',
                 '786facd2c61deb9cf91e9534735124fb8fc11842',
                 '94593d2128d38210a2fcd1aabff6dda0d6d9edf8',
                 'aa6a0de05b7612707db567078e130a6cd114a9a7',
                 'eada5a770da98ab0dd7325e29d00e0714f228d09',
                 '2c1885c735575ca478bf9e17b0029dca68824458',
                 'd9bcd465040bf869799b09ad732c04e0eea99fe9',
                 '469e9c847fe1f6f7a697b8b25b4bc5b48780c1a7',
                 '4fb8326d78e5120da2c7468dcf7098997be385da',
                 '62b4a097164940bd66030c4db51687f3ec035eed',
                 '536c1a19428381cfea92ac44985304f6a8049569',
                 '965e8ab3c44b070cdaa5bf727ddef0ada980ecc4',
                 '9bb326a04ae5d98d437dece54be04f830cf1edd9',
                 'f8940bcb890a98c4702319fbe36db75ea309b475',
                 'ff5ab059786ebc7411e559a2cc309dfae3625a3b',
                 '6b6ad5f82ad5bb6190037671bd254bd4e1f4bf08',
                 'ee87846a61c12153b51543bf860e1026c6d3dcba', ]
         self.assertEqual(org, self.repo.revisions[:31])
 
     def test_iter_slice(self):
         sliced = list(self.repo[:10])
         itered = list(self.repo)[:10]
         self.assertEqual(sliced, itered)
 
     def test_slicing(self):
         #4 1 5 10 95
         for sfrom, sto, size in [(0, 4, 4), (1, 2, 1), (10, 15, 5),
                                  (10, 20, 10), (5, 100, 95)]:
             revs = list(self.repo[sfrom:sto])
             self.assertEqual(len(revs), size)
             self.assertEqual(revs[0], self.repo.get_changeset(sfrom))
             self.assertEqual(revs[-1], self.repo.get_changeset(sto - 1))
 
     def test_branches(self):
         # TODO: Need more tests here
 
         #active branches
         self.assertTrue('default' in self.repo.branches)
         self.assertTrue('stable' in self.repo.branches)
 
         # closed
         self.assertTrue('git' in self.repo._get_branches(closed=True))
         self.assertTrue('web' in self.repo._get_branches(closed=True))
 
         for name, id in self.repo.branches.items():
             self.assertTrue(isinstance(
                 self.repo.get_changeset(id), MercurialChangeset))
 
     def test_tip_in_tags(self):
         # tip is always a tag
         self.assertIn('tip', self.repo.tags)
 
     def test_tip_changeset_in_tags(self):
         tip = self.repo.get_changeset()
         self.assertEqual(self.repo.tags['tip'], tip.raw_id)
 
     def test_initial_changeset(self):
 
         init_chset = self.repo.get_changeset(0)
         self.assertEqual(init_chset.message, 'initial import')
         self.assertEqual(init_chset.author,
             'Marcin Kuzminski <[e-mail]>')
         self.assertEqual(sorted(init_chset._file_paths),
             sorted([
                 'vcs/__init__.py',
                 'vcs/backends/BaseRepository.py',
                 'vcs/backends/__init__.py',
             ])
         )
         self.assertEqual(sorted(init_chset._dir_paths),
             sorted(['', 'vcs', 'vcs/backends']))
 
         self.assertRaises(NodeDoesNotExistError, init_chset.get_node, path='foobar')
 
         node = init_chset.get_node('vcs/')
         self.assertTrue(hasattr(node, 'kind'))
         self.assertEqual(node.kind, NodeKind.DIR)
 
         node = init_chset.get_node('vcs')
         self.assertTrue(hasattr(node, 'kind'))
         self.assertEqual(node.kind, NodeKind.DIR)
 
         node = init_chset.get_node('vcs/__init__.py')
         self.assertTrue(hasattr(node, 'kind'))
         self.assertEqual(node.kind, NodeKind.FILE)
 
     def test_not_existing_changeset(self):
         #rawid
         self.assertRaises(RepositoryError, self.repo.get_changeset,
             'abcd' * 10)
         #shortid
         self.assertRaises(RepositoryError, self.repo.get_changeset,
             'erro' * 4)
         #numeric
         self.assertRaises(RepositoryError, self.repo.get_changeset,
             self.repo.count() + 1)
 
 
         # Small chance we ever get to this one
         revision = pow(2, 30)
         self.assertRaises(RepositoryError, self.repo.get_changeset, revision)
 
     def test_changeset10(self):
 
         chset10 = self.repo.get_changeset(10)
         README = """===
 VCS
 ===
 
 Various Version Control System management abstraction layer for Python.
 
 Introduction
 ------------
 
 TODO: To be written...
 
 """
         node = chset10.get_node('README.rst')
         self.assertEqual(node.kind, NodeKind.FILE)
         self.assertEqual(node.content, README)
 
+    @mock.patch('kallithea.lib.vcs.backends.hg.repository.diffopts')
+    def test_get_diff_does_not_sanitize_zero_context(self, mock_diffopts):
+        zero_context = 0
+
+        self.repo.get_diff(0, 1, 'foo', context=zero_context)
+
+        mock_diffopts.assert_called_once_with(git=True, showfunc=True, ignorews=False, context=zero_context)
+
+    @mock.patch('kallithea.lib.vcs.backends.hg.repository.diffopts')
+    def test_get_diff_sanitizes_negative_context(self, mock_diffopts):
+        negative_context = -10
+        zero_context = 0
+
+        self.repo.get_diff(0, 1, 'foo', context=negative_context)
+
+        mock_diffopts.assert_called_once_with(git=True, showfunc=True, ignorews=False, context=zero_context)
+
 
 class MercurialChangesetTest(unittest.TestCase):
 
     def setUp(self):
         self.repo = MercurialRepository(TEST_HG_REPO)
 
     def _test_equality(self, changeset):
         revision = changeset.revision
         self.assertEqual(changeset, self.repo.get_changeset(revision))
 
     def test_equality(self):
         self.setUp()
         revs = [0, 10, 20]
         changesets = [self.repo.get_changeset(rev) for rev in revs]
         for changeset in changesets:
             self._test_equality(changeset)
 
     def test_default_changeset(self):
         tip = self.repo.get_changeset('tip')
         self.assertEqual(tip, self.repo.get_changeset())
         self.assertEqual(tip, self.repo.get_changeset(revision=None))
         self.assertEqual(tip, list(self.repo[-1:])[0])
 
     def test_root_node(self):
         tip = self.repo.get_changeset('tip')
         self.assertTrue(tip.root is tip.get_node(''))
 
     def test_lazy_fetch(self):
         """
         Test if changeset's nodes expands and are cached as we walk through
         the revision. This test is somewhat hard to write as order of tests
         is a key here. Written by running command after command in a shell.
         """
         self.setUp()
         chset = self.repo.get_changeset(45)
         self.assertTrue(len(chset.nodes) == 0)
         root = chset.root
         self.assertTrue(len(chset.nodes) == 1)
         self.assertTrue(len(root.nodes) == 8)
         # accessing root.nodes updates chset.nodes
         self.assertTrue(len(chset.nodes) == 9)
 
         docs = root.get_node('docs')
         # we haven't yet accessed anything new as docs dir was already cached
         self.assertTrue(len(chset.nodes) == 9)
         self.assertTrue(len(docs.nodes) == 8)
         # accessing docs.nodes updates chset.nodes
         self.assertTrue(len(chset.nodes) == 17)
 
         self.assertTrue(docs is chset.get_node('docs'))
         self.assertTrue(docs is root.nodes[0])
         self.assertTrue(docs is root.dirs[0])
         self.assertTrue(docs is chset.get_node('docs'))
 
     def test_nodes_with_changeset(self):
         self.setUp()
         chset = self.repo.get_changeset(45)
         root = chset.root
         docs = root.get_node('docs')
         self.assertTrue(docs is chset.get_node('docs'))
         api = docs.get_node('api')
         self.assertTrue(api is chset.get_node('docs/api'))
         index = api.get_node('index.rst')
         self.assertTrue(index is chset.get_node('docs/api/index.rst'))
         self.assertTrue(index is chset.get_node('docs')\
             .get_node('api')\
             .get_node('index.rst'))
 
     def test_branch_and_tags(self):
         chset0 = self.repo.get_changeset(0)
         self.assertEqual(chset0.branch, 'default')
         self.assertEqual(chset0.tags, [])
 
         chset10 = self.repo.get_changeset(10)
         self.assertEqual(chset10.branch, 'default')
         self.assertEqual(chset10.tags, [])
 
         chset44 = self.repo.get_changeset(44)
         self.assertEqual(chset44.branch, 'web')
 
         tip = self.repo.get_changeset('tip')
         self.assertTrue('tip' in tip.tags)
 
     def _test_file_size(self, revision, path, size):
         node = self.repo.get_changeset(revision).get_node(path)
         self.assertTrue(node.is_file())
         self.assertEqual(node.size, size)
 
     def test_file_size(self):
         to_check = (
             (10, 'setup.py', 1068),
             (20, 'setup.py', 1106),
             (60, 'setup.py', 1074),
 
             (10, 'vcs/backends/base.py', 2921),
             (20, 'vcs/backends/base.py', 3936),
             (60, 'vcs/backends/base.py', 6189),
         )
         for revision, path, size in to_check:
             self._test_file_size(revision, path, size)
 
     def _test_dir_size(self, revision, path, size):
         node = self.repo.get_changeset(revision).get_node(path)
         self.assertFalse(node.is_file())
         self.assertEqual(node.size, size)
 
     def test_dir_size(self):
         to_check = (
             ('96507bd11ecc', '/', 682421),
             ('a53d9201d4bc', '/', 682410),
             ('90243de06161', '/', 682006),
         )
         for revision, path, size in to_check:
             self._test_dir_size(revision, path, size)
 
     def test_repo_size(self):
         self.assertEqual(self.repo.size, 682421)
 
     def test_file_history(self):
         # we can only check if those revisions are present in the history
         # as we cannot update this test every time file is changed
         files = {
             'setup.py': [7, 18, 45, 46, 47, 69, 77],
             'vcs/nodes.py': [7, 8, 24, 26, 30, 45, 47, 49, 56, 57, 58, 59, 60,
                 61, 73, 76],
             'vcs/backends/hg.py': [4, 5, 6, 11, 12, 13, 14, 15, 16, 21, 22, 23,
                 26, 27, 28, 30, 31, 33, 35, 36, 37, 38, 39, 40, 41, 44, 45, 47,
                 48, 49, 53, 54, 55, 58, 60, 61, 67, 68, 69, 70, 73, 77, 78, 79,
                 82],
         }
         for path, revs in files.items():
             tip = self.repo.get_changeset(revs[-1])
             node = tip.get_node(path)
             node_revs = [chset.revision for chset in node.history]
             self.assertTrue(set(revs).issubset(set(node_revs)),
                 "We assumed that %s is subset of revisions for which file %s "
                 "has been changed, and history of that node returned: %s"
                 % (revs, path, node_revs))
 
     def test_file_annotate(self):
         files = {
                  'vcs/backends/__init__.py':
                   {89: {'lines_no': 31,
                         'changesets': [32, 32, 61, 32, 32, 37, 32, 32, 32, 44,
                                        37, 37, 37, 37, 45, 37, 44, 37, 37, 37,
                                        32, 32, 32, 32, 37, 32, 37, 37, 32,
                                        32, 32]},
                    20: {'lines_no': 1,
                         'changesets': [4]},
                    55: {'lines_no': 31,
                         'changesets': [32, 32, 45, 32, 32, 37, 32, 32, 32, 44,
                                        37, 37, 37, 37, 45, 37, 44, 37, 37, 37,
                                        32, 32, 32, 32, 37, 32, 37, 37, 32,
                                        32, 32]}},
                  'vcs/exceptions.py':
                  {89: {'lines_no': 18,
                        'changesets': [16, 16, 16, 16, 16, 16, 16, 16, 16, 16,
                                       16, 16, 17, 16, 16, 18, 18, 18]},
                   20: {'lines_no': 18,
                        'changesets': [16, 16, 16, 16, 16, 16, 16, 16, 16, 16,
                                       16, 16, 17, 16, 16, 18, 18, 18]},
                   55: {'lines_no': 18, 'changesets': [16, 16, 16, 16, 16, 16,
                                                       16, 16, 16, 16, 16, 16,
                                                       17, 16, 16, 18, 18, 18]}},
                  'MANIFEST.in': {89: {'lines_no': 5,
                                       'changesets': [7, 7, 7, 71, 71]},
                                  20: {'lines_no': 3,
                                       'changesets': [7, 7, 7]},
                                  55: {'lines_no': 3,
                                      'changesets': [7, 7, 7]}}}
 
         for fname, revision_dict in files.items():
             for rev, data in revision_dict.items():
                 cs = self.repo.get_changeset(rev)
                 l1_1 = [x[1] for x in cs.get_file_annotate(fname)]
                 l1_2 = [x[2]().raw_id for x in cs.get_file_annotate(fname)]
                 self.assertEqual(l1_1, l1_2)
                 l1 = l1_2 = [x[2]().revision for x in cs.get_file_annotate(fname)]
                 l2 = files[fname][rev]['changesets']
                 self.assertTrue(l1 == l2 , "The lists of revision for %s@rev%s"
                                 "from annotation list should match each other,"
                                 "got \n%s \nvs \n%s " % (fname, rev, l1, l2))
 
     def test_changeset_state(self):
         """
         Tests which files have been added/changed/removed at particular revision
         """
 
         # rev 46ad32a4f974:
         # hg st --rev 46ad32a4f974
         #    changed: 13
         #    added:   20
         #    removed: 1
         changed = set(['.hgignore'
             , 'README.rst' , 'docs/conf.py' , 'docs/index.rst' , 'setup.py'
             , 'tests/test_hg.py' , 'tests/test_nodes.py' , 'vcs/__init__.py'
             , 'vcs/backends/__init__.py' , 'vcs/backends/base.py'
             , 'vcs/backends/hg.py' , 'vcs/nodes.py' , 'vcs/utils/__init__.py'])
 
         added = set(['docs/api/backends/hg.rst'
             , 'docs/api/backends/index.rst' , 'docs/api/index.rst'
             , 'docs/api/nodes.rst' , 'docs/api/web/index.rst'
             , 'docs/api/web/simplevcs.rst' , 'docs/installation.rst'
             , 'docs/quickstart.rst' , 'setup.cfg' , 'vcs/utils/baseui_config.py'
             , 'vcs/utils/web.py' , 'vcs/web/__init__.py' , 'vcs/web/exceptions.py'
             , 'vcs/web/simplevcs/__init__.py' , 'vcs/web/simplevcs/exceptions.py'
             , 'vcs/web/simplevcs/middleware.py' , 'vcs/web/simplevcs/models.py'
             , 'vcs/web/simplevcs/settings.py' , 'vcs/web/simplevcs/utils.py'
             , 'vcs/web/simplevcs/views.py'])
 
         removed = set(['docs/api.rst'])
 
         chset64 = self.repo.get_changeset('46ad32a4f974')
         self.assertEqual(set((node.path for node in chset64.added)), added)
         self.assertEqual(set((node.path for node in chset64.changed)), changed)
         self.assertEqual(set((node.path for node in chset64.removed)), removed)
 
         # rev b090f22d27d6:
         # hg st --rev b090f22d27d6
         #    changed: 13
         #    added:   20
         #    removed: 1
         chset88 = self.repo.get_changeset('b090f22d27d6')
         self.assertEqual(set((node.path for node in chset88.added)), set())
         self.assertEqual(set((node.path for node in chset88.changed)),
             set(['.hgignore']))
         self.assertEqual(set((node.path for node in chset88.removed)), set())
 #
         # 85:
         #    added:   2 ['vcs/utils/diffs.py', 'vcs/web/simplevcs/views/diffs.py']
         #    changed: 4 ['vcs/web/simplevcs/models.py', ...]
         #    removed: 1 ['vcs/utils/web.py']
         chset85 = self.repo.get_changeset(85)
         self.assertEqual(set((node.path for node in chset85.added)), set([
             'vcs/utils/diffs.py',
             'vcs/web/simplevcs/views/diffs.py']))
         self.assertEqual(set((node.path for node in chset85.changed)), set([
             'vcs/web/simplevcs/models.py',
             'vcs/web/simplevcs/utils.py',
             'vcs/web/simplevcs/views/__init__.py',
             'vcs/web/simplevcs/views/repository.py',
             ]))
         self.assertEqual(set((node.path for node in chset85.removed)),
             set(['vcs/utils/web.py']))
 
 
     def test_files_state(self):
         """
         Tests state of FileNodes.
         """
         chset = self.repo.get_changeset(85)
         node = chset.get_node('vcs/utils/diffs.py')
         self.assertTrue(node.state, NodeState.ADDED)
         self.assertTrue(node.added)
         self.assertFalse(node.changed)
         self.assertFalse(node.not_changed)
         self.assertFalse(node.removed)
 
         chset = self.repo.get_changeset(88)
         node = chset.get_node('.hgignore')
         self.assertTrue(node.state, NodeState.CHANGED)
         self.assertFalse(node.added)
         self.assertTrue(node.changed)
         self.assertFalse(node.not_changed)
         self.assertFalse(node.removed)
 
         chset = self.repo.get_changeset(85)
         node = chset.get_node('setup.py')
         self.assertTrue(node.state, NodeState.NOT_CHANGED)
         self.assertFalse(node.added)
         self.assertFalse(node.changed)
         self.assertTrue(node.not_changed)
         self.assertFalse(node.removed)
 
         # If node has REMOVED state then trying to fetch it would raise
         # ChangesetError exception
         chset = self.repo.get_changeset(2)
         path = 'vcs/backends/BaseRepository.py'
         self.assertRaises(NodeDoesNotExistError, chset.get_node, path)
         # but it would be one of ``removed`` (changeset's attribute)
         self.assertTrue(path in [rf.path for rf in chset.removed])
 
     def test_commit_message_is_unicode(self):
         for cm in self.repo:
             self.assertEqual(type(cm.message), unicode)
 
     def test_changeset_author_is_unicode(self):
         for cm in self.repo:
             self.assertEqual(type(cm.author), unicode)
 
     def test_repo_files_content_is_unicode(self):
         test_changeset = self.repo.get_changeset(100)
         for node in test_changeset.get_node('/'):
             if node.is_file():
                 self.assertEqual(type(node.content), unicode)
 
     def test_wrong_path(self):
         # There is 'setup.py' in the root dir but not there:
         path = 'foo/bar/setup.py'
         self.assertRaises(VCSError, self.repo.get_changeset().get_node, path)
 
 
     def test_archival_file(self):
         #TODO:
         pass
 
     def test_archival_as_generator(self):
         #TODO:
         pass
 
     def test_archival_wrong_kind(self):
         tip = self.repo.get_changeset()
         self.assertRaises(VCSError, tip.fill_archive, kind='error')
 
     def test_archival_empty_prefix(self):
         #TODO:
         pass
 
 
     def test_author_email(self):
         self.assertEqual('[e-mail]',
                          self.repo.get_changeset('b986218ba1c9').author_email)
         self.assertEqual('[e-mail]',
                          self.repo.get_changeset('3803844fdbd3').author_email)
         self.assertEqual('',
                          self.repo.get_changeset('84478366594b').author_email)
 
     def test_author_username(self):
         self.assertEqual('Marcin Kuzminski',
                          self.repo.get_changeset('b986218ba1c9').author_name)
         self.assertEqual('Lukasz Balcerzak',
                          self.repo.get_changeset('3803844fdbd3').author_name)
         self.assertEqual('marcink',
                          self.repo.get_changeset('84478366594b').author_name)