Session().commit()

    click.echo('Updated database with information about latest change in the following %s repositories:' % (len(repo_list)))

    click.echo('\n'.join(repo.repo_name for repo in repo_list))

@cli_base.register_command(config_file_initialize_app=True)

@click.option('--ask/--no-ask', default=True, help='Ask for confirmation or not. Default is --ask.')

@click.option('--older-than',

        help="""Only purge repositories that have been removed at least the given time ago.

        For example, '--older-than=30d' purges repositories deleted 30 days ago or longer.

        Possible suffixes: d (days), h (hours), m (minutes), s (seconds).""")

def repo_purge_deleted(ask, older_than):

    """Purge backups of deleted repositories.

    When a repository is deleted via the Kallithea web interface, the actual

    data is still present on the filesystem but set aside using a special name.

    This command allows to delete these files permanently.

    """

    def _parse_older_than(val):

        regex = re.compile(r'((?P<days>\d+?)d)?((?P<hours>\d+?)h)?((?P<minutes>\d+?)m)?((?P<seconds>\d+?)s)?')

        parts = regex.match(val)

        if not parts:

            return

        parts = parts.groupdict()

        time_params = {}

        for (name, param) in parts.iteritems():

            if param:

                time_params[name] = int(param)

        return datetime.timedelta(**time_params)

    def _extract_date(name):

        """

        Extract the date part from rm__<date> pattern of removed repos,

        and convert it to datetime object

        :param name:

        """

        date_part = name[4:19]  # 4:19 since we don't parse milliseconds

        return datetime.datetime.strptime(date_part, '%Y%m%d_%H%M%S')

    repos_location = Ui.get_repos_location()

    to_remove = []

    for dn_, dirs, f in os.walk(safe_str(repos_location)):

        alldirs = list(dirs)

        del dirs[:]

        if ('.hg' in alldirs or

            '.git' in alldirs or

            '.svn' in alldirs or

            continue

        for loc in alldirs:

            if REMOVED_REPO_PAT.match(loc):

                to_remove.append([os.path.join(dn_, loc),

                                  _extract_date(loc)])

            else:

                dirs.append(loc)

        if dirs:

            click.echo('Scanning: %s' % dn_)

    if not to_remove:

        click.echo('There are no deleted repositories.')

        return

    # filter older than (if present)!

    if older_than:

        now = datetime.datetime.now()

        to_remove_filtered = []

        older_than_date = _parse_older_than(older_than)

        for name, date_ in to_remove:

            repo_age = now - date_

            if repo_age > older_than_date:

                to_remove_filtered.append([name, date_])

        to_remove = to_remove_filtered

        if not to_remove:

            click.echo('There are no deleted repositories older than %s (%s)'

                    % (older_than, older_than_date))

            return

        click.echo('Considering %s deleted repositories older than %s (%s).'

            % (len(to_remove), older_than, older_than_date))

    else:

        click.echo('Considering %s deleted repositories.' % len(to_remove))

    if not ask:

        remove = True

    else:

        remove = ask_ok('The following repositories will be removed completely:\n%s\n'

                'Do you want to proceed? [y/n] '

                % '\n'.join(['%s deleted on %s' % (safe_str(x[0]), safe_str(x[1]))

                                     for x in to_remove]))

    if remove:

        for path, date_ in to_remove:

            click.echo('Purging repository %s' % path)

            shutil.rmtree(path)

from kallithea.lib.vcs.exceptions import ChangesetDoesNotExistError, ChangesetError, EmptyRepositoryError, NodeDoesNotExistError, RepositoryError

log = logging.getLogger(__name__)

class ChangelogController(BaseRepoController):

    def _before(self, *args, **kwargs):

        super(ChangelogController, self)._before(*args, **kwargs)

        c.affected_files_cut_off = 60

    @staticmethod

    def __get_cs(rev, repo):

        """

        Safe way to get changeset. If error occur fail with error message.

        :param rev: revision to fetch

        :param repo: repo instance

        """

        try:

            return c.db_repo_scm_instance.get_changeset(rev)

        except EmptyRepositoryError as e:

            h.flash(_('There are no changesets yet'), category='error')

        except RepositoryError as e:

            log.error(traceback.format_exc())

            h.flash(safe_str(e), category='error')

        raise HTTPBadRequest()

    @LoginRequired(allow_default_user=True)

    @HasRepoPermissionLevelDecorator('read')

    def index(self, repo_name, revision=None, f_path=None):

        limit = 2000

        default = 100

        if request.GET.get('size'):

            c.size = max(min(safe_int(request.GET.get('size')), limit), 1)

            session['changelog_size'] = c.size

            session.save()

        else:

            c.size = int(session.get('changelog_size', default))

        # min size must be 1

        c.size = max(c.size, 1)

        p = safe_int(request.GET.get('page'), 1)

        branch_name = request.GET.get('branch', None)

        if (branch_name and

            branch_name not in c.db_repo_scm_instance.branches and

            branch_name not in c.db_repo_scm_instance.closed_branches and

            raise HTTPFound(location=url('changelog_file_home', repo_name=c.repo_name,

                                    revision=branch_name, f_path=f_path or ''))

        if revision == 'tip':

            revision = None

        c.changelog_for_path = f_path

        try:

            if f_path:

                log.debug('generating changelog for path %s', f_path)

                # get the history for the file !

                tip_cs = c.db_repo_scm_instance.get_changeset()

                try:

                    collection = tip_cs.get_file_history(f_path)

                except (NodeDoesNotExistError, ChangesetError):

                    # this node is not present at tip !

                    try:

                        cs = self.__get_cs(revision, repo_name)

                        collection = cs.get_file_history(f_path)

                    except RepositoryError as e:

                        h.flash(safe_str(e), category='warning')

                        raise HTTPFound(location=h.url('changelog_home', repo_name=repo_name))

                collection = list(reversed(collection))

            else:

                collection = c.db_repo_scm_instance.get_changesets(start=0, end=revision,

                                                        branch_name=branch_name)

            c.total_cs = len(collection)

            c.cs_pagination = RepoPage(collection, page=p, item_count=c.total_cs,

                                    items_per_page=c.size, branch=branch_name,)

            page_revisions = [x.raw_id for x in c.cs_pagination]

            c.cs_comments = c.db_repo.get_comments(page_revisions)

            c.cs_statuses = c.db_repo.statuses(page_revisions)

        except EmptyRepositoryError as e:

            h.flash(safe_str(e), category='warning')

            raise HTTPFound(location=url('summary_home', repo_name=c.repo_name))

        except (RepositoryError, ChangesetDoesNotExistError, Exception) as e:

            log.error(traceback.format_exc())

            h.flash(safe_str(e), category='error')

            raise HTTPFound(location=url('changelog_home', repo_name=c.repo_name))

        c.branch_name = branch_name

        c.branch_filters = [('', _('None'))] + \

            [(k, k) for k in c.db_repo_scm_instance.branches.keys()]

        if c.db_repo_scm_instance.closed_branches:

            prefix = _('(closed)') + ' '

from kallithea.model.meta import Session

from kallithea.model.pull_request import CreatePullRequestAction, CreatePullRequestIterationAction, PullRequestModel

log = logging.getLogger(__name__)

def _get_reviewer(user_id):

    """Look up user by ID and validate it as a potential reviewer."""

    try:

        user = User.get(int(user_id))

    except ValueError:

        user = None

    if user is None or user.is_default_user:

        h.flash(_('Invalid reviewer "%s" specified') % user_id, category='error')

        raise HTTPBadRequest()

    return user

class PullrequestsController(BaseRepoController):

    def _get_repo_refs(self, repo, rev=None, branch=None, branch_rev=None):

        """return a structure with repo's interesting changesets, suitable for

        the selectors in pullrequest.html

        rev: a revision that must be in the list somehow and selected by default

        branch: a branch that must be in the list and selected by default - even if closed

        branch_rev: a revision of which peers should be preferred and available."""

        # list named branches that has been merged to this named branch - it should probably merge back

        peers = []

        if rev:

            rev = safe_str(rev)

        if branch:

            branch = safe_str(branch)

        if branch_rev:

            branch_rev = safe_str(branch_rev)

            # a revset not restricting to merge() would be better

            # (especially because it would get the branch point)

            # ... but is currently too expensive

            # including branches of children could be nice too

            peerbranches = set()

            for i in repo._repo.revs(

                "sort(parents(branch(id(%s)) and merge()) - branch(id(%s)), -rev)",

                for abranch in repo.get_changeset(i).branches:

                    if abranch not in peerbranches:

                        n = 'branch:%s:%s' % (abranch, repo.get_changeset(abranch).raw_id)

                        peers.append((n, abranch))

                        peerbranches.add(abranch)

        selected = None

        tiprev = repo.tags.get('tip')

        tipbranch = None

        branches = []

        for abranch, branchrev in repo.branches.iteritems():

            n = 'branch:%s:%s' % (abranch, branchrev)

            desc = abranch

            if branchrev == tiprev:

                tipbranch = abranch

                desc = '%s (current tip)' % desc

            branches.append((n, desc))

            if rev == branchrev:

                selected = n

            if branch == abranch:

                if not rev:

                    selected = n

                branch = None

        if branch:  # branch not in list - it is probably closed

            branchrev = repo.closed_branches.get(branch)

            if branchrev:

                n = 'branch:%s:%s' % (branch, branchrev)

                branches.append((n, _('%s (closed)') % branch))

                selected = n

                branch = None

            if branch:

                log.debug('branch %r not found in %s', branch, repo)

        bookmarks = []

        for bookmark, bookmarkrev in repo.bookmarks.iteritems():

            n = 'book:%s:%s' % (bookmark, bookmarkrev)

            bookmarks.append((n, bookmark))

            if rev == bookmarkrev:

                selected = n

        tags = []

        for tag, tagrev in repo.tags.iteritems():

            if tag == 'tip':

                continue

            n = 'tag:%s:%s' % (tag, tagrev)

            tags.append((n, tag))

            # note: even if rev == tagrev, don't select the static tag - it must be chosen explicitly

                password = get_password()

                if not password:

                    # second try

                    password = get_password()

                    if not password:

                        sys.exit()

            if email is None:

                email = raw_input('Specify admin email:')

            self.create_user(username, password, email, True)

        else:

            log.info('creating admin and regular test users')

            from kallithea.tests.base import TEST_USER_ADMIN_LOGIN, \

            TEST_USER_ADMIN_PASS, TEST_USER_ADMIN_EMAIL, \

            TEST_USER_REGULAR_LOGIN, TEST_USER_REGULAR_PASS, \

            TEST_USER_REGULAR_EMAIL, TEST_USER_REGULAR2_LOGIN, \

            TEST_USER_REGULAR2_PASS, TEST_USER_REGULAR2_EMAIL

            self.create_user(TEST_USER_ADMIN_LOGIN, TEST_USER_ADMIN_PASS,

                             TEST_USER_ADMIN_EMAIL, True)

            self.create_user(TEST_USER_REGULAR_LOGIN, TEST_USER_REGULAR_PASS,

                             TEST_USER_REGULAR_EMAIL, False)

            self.create_user(TEST_USER_REGULAR2_LOGIN, TEST_USER_REGULAR2_PASS,

                             TEST_USER_REGULAR2_EMAIL, False)

    def create_auth_plugin_options(self, skip_existing=False):

        """

        Create default auth plugin settings, and make it active

        :param skip_existing:

        """

        for k, v, t in [('auth_plugins', 'kallithea.lib.auth_modules.auth_internal', 'list'),

                        ('auth_internal_enabled', 'True', 'bool')]:

            if skip_existing and Setting.get_by_name(k) is not None:

                log.debug('Skipping option %s', k)

                continue

            setting = Setting(k, v, t)

            self.sa.add(setting)

    def create_default_options(self, skip_existing=False):

        """Creates default settings"""

        for k, v, t in [

            ('default_repo_enable_downloads', False, 'bool'),

            ('default_repo_enable_statistics', False, 'bool'),

            ('default_repo_private', False, 'bool'),

            if skip_existing and Setting.get_by_name(k) is not None:

                log.debug('Skipping option %s', k)

                continue

            setting = Setting(k, v, t)

            self.sa.add(setting)

    def fixup_groups(self):

        def_usr = User.get_default_user()

        for g in RepoGroup.query().all():

            g.group_name = g.get_new_name(g.name)

            # get default perm

            default = UserRepoGroupToPerm.query() \

                .filter(UserRepoGroupToPerm.group == g) \

                .filter(UserRepoGroupToPerm.user == def_usr) \

                .scalar()

            if default is None:

                log.debug('missing default permission for group %s adding', g)

                RepoGroupModel()._create_default_perms(g)

    def reset_permissions(self, username):

        """

        Resets permissions to default state, useful when old systems had

        bad permissions, we must clean them up

        :param username:

        """

        default_user = User.get_by_username(username)

        if not default_user:

            return

        u2p = UserToPerm.query() \

            .filter(UserToPerm.user == default_user).all()

        fixed = False

        if len(u2p) != len(Permission.DEFAULT_USER_PERMISSIONS):

            for p in u2p:

                Session().delete(p)

            fixed = True

            self.populate_default_permissions()

        return fixed

    def update_repo_info(self):

        for repo in Repository.query():

            repo.update_changeset_cache()

    def prompt_repo_root_path(self, test_repo_path='', retries=3):

        _path = self.cli_args.get('repos_location')

        if retries == 3:

                    return Repo.init_bare(self.path)

                else:

                    return Repo.init(self.path)

            else:

                return Repo(self.path)

        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)

            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 = safe_str(url)

        if url != 'default' and '://' not in url:

            url = ':///'.join(('file', url))

        return url

    #        raise e

    #

    #    return answer

    """

    def __init__(self, cmd, inputstream=None, buffer_size=65536,

                 chunk_size=4096, starting_values=None, **kwargs):

        """

        Initializes SubprocessIOChunker

        :param cmd: A Subprocess.Popen style "cmd". Can be string or array of strings

        :param inputstream: (Default: None) A file-like, string, or file pointer.

        :param buffer_size: (Default: 65536) A size of total buffer per stream in bytes.

        :param chunk_size: (Default: 4096) A max size of a chunk. Actual chunk may be smaller.

        :param starting_values: (Default: []) An array of strings to put in front of output que.

        """

        starting_values = starting_values or []

        if inputstream:

            input_streamer = StreamFeeder(inputstream)

            input_streamer.start()

            inputstream = input_streamer.output

        # Note: fragile cmd mangling has been removed for use in Kallithea

        assert isinstance(cmd, list), cmd

        _p = subprocess.Popen(cmd, bufsize=-1,

                              stdin=inputstream,

                              stdout=subprocess.PIPE,

                              stderr=subprocess.PIPE,

                              **kwargs)

        bg_out = BufferedGenerator(_p.stdout, buffer_size, chunk_size,

                                   starting_values)

        bg_err = BufferedGenerator(_p.stderr, 16000, 1, bottomless=True)

        while not bg_out.done_reading and not bg_out.reading_paused:

            # doing this until we reach either end of file, or end of buffer.

            bg_out.data_added_event.wait(1)

            bg_out.data_added_event.clear()

        # at this point it's still ambiguous if we are done reading or just full buffer.

        # Either way, if error (returned by ended process, or implied based on

        # presence of stuff in stderr output) we error out.

        # Else, we are happy.

        returncode = _p.poll()

        if (returncode is not None # process has terminated

            bg_out.stop()

            out = ''.join(bg_out)

            bg_err.stop()

            err = ''.join(bg_err)

            if (err.strip() == 'fatal: The remote end hung up unexpectedly' and

                # hack inspired by https://github.com/schacon/grack/pull/7

                bg_out = iter([out])

                _p = None

            elif err:

                raise EnvironmentError(

                    "Subprocess exited due to an error:\n" + err)

            else:

                raise EnvironmentError(

                    "Subprocess exited with non 0 ret code: %s" % returncode)

        self.process = _p

        self.output = bg_out

        self.error = bg_err

        self.inputstream = inputstream

    def __iter__(self):

        return self

    def next(self):

        if self.process:

            returncode = self.process.poll()

            if (returncode is not None # process has terminated

                self.output.stop()

                self.error.stop()

                err = ''.join(self.error)

                raise EnvironmentError("Subprocess exited due to an error:\n" + err)

        return self.output.next()

    def throw(self, type, value=None, traceback=None):

        if self.output.length or not self.output.done_reading:

            raise type(value)

    def close(self):

        try:

            self.process.terminate()

        except:

            pass

        try:

            self.output.close()

        except:

            pass

        try:

            self.error.close()

        except:

            pass

        try:

            os.close(self.inputstream)

        except:

            pass

    def __del__(self):

        self.close()

    def remove_reviewers(self, user, pull_request, reviewers):

        """Remove specified users from being reviewers of the PR."""

        if not reviewers:

            return # avoid SQLAlchemy warning about empty sequence for IN-predicate

        PullRequestReviewer.query() \

            .filter_by(pull_request=pull_request) \

            .filter(PullRequestReviewer.user_id.in_(r.user_id for r in reviewers)) \

            .delete(synchronize_session='fetch') # the default of 'evaluate' is not available

    def delete(self, pull_request):

        pull_request = PullRequest.guess_instance(pull_request)

        Session().delete(pull_request)

        if pull_request.org_repo.scm_instance.alias == 'git':

            # remove a ref under refs/pull/ so that commits can be garbage-collected

            try:

                del pull_request.org_repo.scm_instance._repo["refs/pull/%d/head" % pull_request.pull_request_id]

            except KeyError:

                pass

    def close_pull_request(self, pull_request):

        pull_request = PullRequest.guess_instance(pull_request)

        pull_request.status = PullRequest.STATUS_CLOSED

        pull_request.updated_on = datetime.datetime.now()

class CreatePullRequestAction(object):

    class ValidationError(Exception):

        pass

    class Empty(ValidationError):

        pass

    class AmbiguousAncestor(ValidationError):

        pass

    class Unauthorized(ValidationError):

        pass

    @staticmethod

    def is_user_authorized(org_repo, other_repo):

        """Performs authorization check with only the minimum amount of

        information needed for such a check, rather than a full command

        object.

        """

        if (h.HasRepoPermissionLevel('read')(org_repo.repo_name) and

            return True

        return False

    def __init__(self, org_repo, other_repo, org_ref, other_ref, title, description, owner, reviewers):

        from kallithea.controllers.compare import CompareController

        reviewers = set(reviewers)

        _assert_valid_reviewers(reviewers)

        (org_ref_type,

         org_ref_name,

         org_rev) = org_ref.split(':')

        org_display = h.short_ref(org_ref_type, org_ref_name)

        if org_ref_type == 'rev':

            cs = org_repo.scm_instance.get_changeset(org_rev)

            org_ref = 'branch:%s:%s' % (cs.branch, cs.raw_id)

        (other_ref_type,

         other_ref_name,

         other_rev) = other_ref.split(':')

        if other_ref_type == 'rev':

            cs = other_repo.scm_instance.get_changeset(other_rev)

            other_ref_name = cs.raw_id[:12]

            other_ref = '%s:%s:%s' % (other_ref_type, other_ref_name, cs.raw_id)

        other_display = h.short_ref(other_ref_type, other_ref_name)

        cs_ranges, _cs_ranges_not, ancestor_revs = \

            CompareController._get_changesets(org_repo.scm_instance.alias,

                                              other_repo.scm_instance, other_rev, # org and other "swapped"

                                              org_repo.scm_instance, org_rev,

                                              )

        if not cs_ranges:

            raise self.Empty(_('Cannot create empty pull request'))

        if not ancestor_revs:

            ancestor_rev = org_repo.scm_instance.EMPTY_CHANGESET

        elif len(ancestor_revs) == 1:

            ancestor_rev = ancestor_revs[0]

        else:

            raise self.AmbiguousAncestor(

                _('Cannot create pull request - criss cross merge detected, please merge a later %s revision to %s')

                % (other_ref_name, org_ref_name))

        self.revisions = [cs_.raw_id for cs_ in cs_ranges]

        # hack: ancestor_rev is not an other_rev but we want to show the

        # requested destination and have the exact ancestor

        other_ref = '%s:%s:%s' % (other_ref_type, other_ref_name, ancestor_rev)