from shutil import rmtree

from time import mktime

from os.path import dirname as dn

from os.path import join as jn

#to get the rhodecode import

project_path = dn(dn(dn(dn(os.path.realpath(__file__)))))

sys.path.append(project_path)

from rhodecode.config.conf import INDEX_EXTENSIONS

from rhodecode.model.scm import ScmModel

from rhodecode.lib.utils2 import safe_unicode

from rhodecode.lib.indexers import SCHEMA, IDX_NAME, CHGSETS_SCHEMA, \

    CHGSET_IDX_NAME

from rhodecode.lib.vcs.exceptions import ChangesetError, RepositoryError, \

    NodeDoesNotExistError

from whoosh.index import create_in, open_dir, exists_in

from whoosh.query import *

from whoosh.qparser import QueryParser

log = logging.getLogger('whoosh_indexer')

class WhooshIndexingDaemon(object):

    """

    Daemon for atomic indexing jobs

    """

    def __init__(self, indexname=IDX_NAME, index_location=None,

                 repo_location=None, sa=None, repo_list=None,

                 repo_update_list=None):

        self.indexname = indexname

        self.index_location = index_location

        if not index_location:

            raise Exception('You have to provide index location')

        self.repo_location = repo_location

        if not repo_location:

            raise Exception('You have to provide repositories location')

        self.repo_paths = ScmModel(sa).repo_scan(self.repo_location)

        #filter repo list

        if repo_list:

            self.filtered_repo_paths = {}

            for repo_name, repo in self.repo_paths.items():

                if repo_name in repo_list:

                    self.filtered_repo_paths[repo_name] = repo

            self.repo_paths = self.filtered_repo_paths

        #filter update repo list

        self.filtered_repo_update_paths = {}

        if repo_update_list:

            self.filtered_repo_update_paths = {}

            for repo_name, repo in self.repo_paths.items():

                if repo_name in repo_update_list:

                    self.filtered_repo_update_paths[repo_name] = repo

            self.repo_paths = self.filtered_repo_update_paths

        self.initial = True

        if not os.path.isdir(self.index_location):

            os.makedirs(self.index_location)

            log.info('Cannot run incremental index since it does not'

                     ' yet exist running full build')

        elif not exists_in(self.index_location, IDX_NAME):

            log.info('Running full index build as the file content'

                     ' index does not exist')

        elif not exists_in(self.index_location, CHGSET_IDX_NAME):

            log.info('Running full index build as the changeset'

                     ' index does not exist')

        else:

            self.initial = False

    def get_paths(self, repo):

        """

        recursive walk in root dir and return a set of all path in that dir

        based on repository walk function

        """

        index_paths_ = set()

        try:

            tip = repo.get_changeset('tip')

            for _topnode, _dirs, files in tip.walk('/'):

                for f in files:

                    index_paths_.add(jn(repo.path, f.path))

        except RepositoryError:

            log.debug(traceback.format_exc())

            pass

        return index_paths_

    def get_node(self, repo, path):