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:

            #Fix non-ascii repo names to unicode

            repo_list = map(safe_unicode, 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_index_revision(self, repo):

        db_repo = Repository.get_by_repo_name(repo.name)

        landing_rev = 'tip'

        if db_repo:

            landing_rev = db_repo.landing_rev

        return landing_rev

    def _get_index_changeset(self, repo):

        index_rev = self._get_index_revision(repo)

        cs = repo.get_changeset(index_rev)

        return cs

    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:

            cs = self._get_index_changeset(repo)

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

                for f in files:

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

        except RepositoryError:

            log.debug(traceback.format_exc())

            pass

        return index_paths_

    def get_node(self, repo, path):

        cs = self._get_index_changeset(repo)

        return node

    def get_node_mtime(self, node):

        return mktime(node.last_changeset.date.timetuple())

    def add_doc(self, writer, path, repo, repo_name):

        """

        Adding doc to writer this function itself fetches data from

        the instance of vcs backend

        """

        node = self.get_node(repo, path)

        indexed = indexed_w_content = 0

        # we just index the content of chosen files, and skip binary files

        if node.extension in INDEX_EXTENSIONS and not node.is_binary:

            u_content = node.content

            if not isinstance(u_content, unicode):

                log.warning('  >> %s Could not get this content as unicode '

                            'replacing with empty content' % path)

                u_content = u''

            else:

                log.debug('    >> %s [WITH CONTENT]' % path)

                indexed_w_content += 1

        else:

            log.debug('    >> %s' % path)

            # just index file name without it's content

            u_content = u''

            indexed += 1

        p = safe_unicode(path)

        writer.add_document(

            fileid=p,

            owner=unicode(repo.contact),

            repository=safe_unicode(repo_name),

            path=p,

            content=u_content,

            modtime=self.get_node_mtime(node),

            extension=node.extension

        )

        return indexed, indexed_w_content

    def index_changesets(self, writer, repo_name, repo, start_rev=None):

        """

        Add all changeset in the vcs repo starting at start_rev

        to the index writer

        :param writer: the whoosh index writer to add to

        :param repo_name: name of the repository from whence the

          changeset originates including the repository group

        :param repo: the vcs repository instance to index changesets for,

          the presumption is the repo has changesets to index

        :param start_rev=None: the full sha id to start indexing from

          if start_rev is None then index from the first changeset in

          the repo

        """

        if start_rev is None:

            start_rev = repo[0].raw_id

        log.debug('indexing changesets in %s starting at rev: %s' %

                  (repo_name, start_rev))

        indexed = 0

        for cs in repo.get_changesets(start=start_rev):

            log.debug('    >> %s' % cs)

            writer.add_document(

                raw_id=unicode(cs.raw_id),

                owner=unicode(repo.contact),

                date=cs._timestamp,

                repository=safe_unicode(repo_name),

                author=cs.author,

                message=cs.message,

                last=cs.last,

                added=u' '.join([safe_unicode(node.path) for node in cs.added]).lower(),

                removed=u' '.join([safe_unicode(node.path) for node in cs.removed]).lower(),

                changed=u' '.join([safe_unicode(node.path) for node in cs.changed]).lower(),

                parents=u' '.join([cs.raw_id for cs in cs.parents]),

            )

            indexed += 1

        log.debug('indexed %d changesets for repo %s' % (indexed, repo_name))

        return indexed

    def index_files(self, file_idx_writer, repo_name, repo):

        """

        Index files for given repo_name

        :param file_idx_writer: the whoosh index writer to add to

        :param repo_name: name of the repository we're indexing

        :param repo: instance of vcs repo

        """

        i_cnt = iwc_cnt = 0

        log.debug('building index for %s @revision:%s' % (repo.path,

                                                self._get_index_revision(repo)))

        for idx_path in self.get_paths(repo):