mygpo/search/index.py

   1
   2 """ Contains code for indexing other objects """
   3
   4 from pyes import ES, QueryStringQuery, FunctionScoreQuery
   5 from pyes.exceptions import IndexAlreadyExistsException, NoServerAvailable
   6
   7 from django.conf import settings
   8
   9 from mygpo.search.json import podcast_to_json
  10 from mygpo.search.models import PodcastResult
  11
  12 import logging
  13 logger = logging.getLogger(__name__)
  14
  15
  16 def get_connection():
  17     """ Create a connection from Django settings """
  18     conn = ES([settings.ELASTICSEARCH_SERVER],
  19               timeout=settings.ELASTICSEARCH_TIMEOUT)
  20     return conn
  21
  22
  23 def index_podcast(sender, **kwargs):
  24     """ Indexes a podcast """
  25
  26     conn = get_connection()
  27     podcast = kwargs['instance']
  28     logger.info('Indexing podcast %s', podcast)
  29
  30     document = podcast_to_json(podcast)
  31
  32     try:
  33         conn.index(document, settings.ELASTICSEARCH_INDEX,
  34                    'podcast', podcast.id.hex)
  35     except NoServerAvailable:
  36         logger.exception('Indexing podcast failed')
  37
  38
  39 def create_index():
  40     """ Creates the Elasticsearch index """
  41     conn = get_connection()
  42
  43     logger.info('Creating index %s' % settings.ELASTICSEARCH_INDEX)
  44     try:
  45         conn.indices.create_index(settings.ELASTICSEARCH_INDEX)
  46
  47     except IndexAlreadyExistsException as ex:
  48         logger.info(str(ex))
  49
  50
  51 def search_podcasts(query):
  52     """ Search for podcasts according to 'query' """
  53     conn = get_connection()
  54
  55     q = {
  56         "function_score" : {
  57             "boost_mode": 'replace',
  58             "query" : {
  59                  'simple_query_string': {'query': query}
  60             },
  61             "script_score" : {
  62                'script': "_score * (doc.subscribers.value / 4000)"
  63             }
  64         }
  65     }
  66     results = conn.search(query=q, indices=settings.ELASTICSEARCH_INDEX,
  67                           doc_types='podcast',
  68                           model=lambda conn, doc: PodcastResult.from_doc(doc))
  69     return results