Django Haystack Elasticsearch Autocomplete Backend
Custom Elasticsearch backend for django-haystack which makes it easier to customize Elasticsearch backend settings.
Install
$ pip install django-hesab
Why?
I find myself having to subclass and customize the Elasticsearch backend in Haystack for almost every single project I'm using it in. Specially, the autocomplete functionality bothered me, and I struggled to figure out why it didn't "feel" like autocomplete. Well, it's because it really isn't. The EdgeNGram analyzer does what it should, but to make it "feel" a bit more natural there's some tweaking that has to be done.
- Create a filter that triggers at the first character
- Create an analyzer that use the filter
- Create a field mappings that use the new analyzer to index, and the standard analyzer to search.
How?
Simply use the default tweaks shipped in this package, or write your own.
Declare it in settings.py under the ES_BACKEND_SETTINGS namespace like this:
If you don't need to do any special configuration, simply ignore this.
ES_BACKEND_SETTINGS = {
'INDEX_SETTINGS: {
'settings: {
'number_of_shards': 2,
'analysis': {
'analyzer': {
'my_analyzer': {
'type': 'custom',
'filter': ['my_filter'],
...
}
},
'filter: {
'my_filter': {
'type': 'edgeNgram',
...
}
}
}
}
},
'FIELD_MAPPINGS': {
'my_field_mapping': {
'type': 'string',
'index_analyzer': 'my_analyzer',
...
}
}
}Start using it
#
# settings.py
#
HAYSTACK_CONNECTIONS = {
'default': {
'ENGINE': 'hesab.backends.ElasticsearchSearchEngine',
'URL': 'http://localhost:9200/',
'INDEX_NAME': 'my-index',
'TIMEOUT': 300
}
}
#
# search_indexes.py
#
from haystack import indexes
from hesab.fields import AutocompleteEdgeNgramField
class MySearchIndex(indexes.SearchIndex, indexes.Indexable):
text = indexes.Charfield(document=True, use_template=True)
autocomplete = AutocompleteEdgeNgramField()
def get_model(self):
return Address
@staticmethod
def prepare_autocomplete(obj):
autocomplete = " ".join((obj.street_address, obj.city, obj.zip_code))
return " ".join(set(autocomplete.split()))
