Alternative indexation pattern for Addok, based on trigrams.

pip install addok-trigrams

In your local configuration file:

• remove unwanted RESULTS_COLLECTORS_PYPATHS:

    from addok.config.default import RESULTS_COLLECTORS_PYPATHS
    RESULTS_COLLECTORS_PYPATHS.remove('addok.helpers.collectors.extend_results_reducing_tokens')
    RESULTS_COLLECTORS_PYPATHS.remove('addok.autocomplete.only_commons_but_geohash_try_autocomplete_collector')
    RESULTS_COLLECTORS_PYPATHS.remove('addok.autocomplete.no_meaningful_but_common_try_autocomplete_collector')
    RESULTS_COLLECTORS_PYPATHS.remove('addok.autocomplete.only_commons_try_autocomplete_collector')
    RESULTS_COLLECTORS_PYPATHS.remove('addok.autocomplete.autocomplete_meaningful_collector')
    RESULTS_COLLECTORS_PYPATHS.remove('addok.fuzzy.fuzzy_collector')
  

• remove all autocomplete and fuzzy RESULTS_COLLECTORS_PYPATHS, add new ones:

    RESULTS_COLLECTORS_PYPATHS += [
        'addok_trigrams.extend_results_removing_numbers',
        'addok_trigrams.extend_results_removing_one_whole_word',
        'addok_trigrams.extend_results_removing_successive_trigrams',
    ]
  

• add trigramize to PROCESSORS_PYPATHS:

    from addok.config.default import PROCESSORS_PYPATHS
    PROCESSORS_PYPATHS += [
        'addok_trigrams.trigramize',
    ]
  

• remove pairs and autocomplete indexers from INDEXERS_PYPATHS:

    from addok.config.default import INDEXERS_PYPATHS
    INDEXERS_PYPATHS.remove('addok.pairs.PairsIndexer')
    INDEXERS_PYPATHS.remove('addok.autocomplete.EdgeNgramIndexer')
  

By default, digit only words are not turned into trigrams. To prevent this, set TRIGRAM_SKIP_DIGIT=False.

Use addok batch just like with genuine addok for importing documents, but no need for running addok ngrams, given they are already part of the index strategy.