How can I do a fuzzy search using django-haystack and the elasticsearch backend?

It looks as if elasticsearch supports fuzzy queries (http://www.elasticsearch.org/guide/reference/query-dsl/fuzzy-query/) but I can't figure out a way to have django-haystack pass in that option.

I dug into the django-haystack search and it looks as if it's using the 'match_all' query when using the elasticsearch backend. Is it possible to get the fuzzy match behavior without having to modify the django-haystack source code?

Haystack Source: https://github.com/toastdriven/django-haystack/blob/master/haystack/backends/elasticsearch_backend.py (the build_search_kwargs method is what I suspect I need to change)

Solution

No need to fork Haystack, you can update that method in your own backend (for more details, see Stretching Haystack's ElasticSearch Backend). The build_search_kwargs method returns a dictionary so you can just modify the original return value.

Disclaimer: this code is just an example of how you could update your own backend, not how to implement fuzzy search.

class FuzzyBackend(ElasticsearchSearchBackend):
    def build_search_kwargs(self, query_string, **kwargs):
        fuzzy = kwargs.pop('fuzzy', False)
        fuzzy_field = kwargs.pop('min_similarity', '')
        search_kwargs = super(FuzzyBackend, self).build_search_kwargs(
                query_string, kwargs)
        if fuzzy:
            search_kwargs = {'fuzzy': {fuzzy_field: query_string}}
        return search_kwargs