How to retrieve spelling suggestions in the serializer?

[pymarco]

I would like to return spelling suggestions in my serializer. How would I do that? Thanks

Haystack method: http://django-haystack.readthedocs.io/en/latest/searchqueryset_api.html?#SearchQuerySet.spelling_suggestion

[rhblind]

Hi.
Support for spelling suggestion is not currently implemented.
It's probably quite easy to add support for, so I'll see what I can do.

[pymarco]

Any chance there is an update on this?

[rhblind]

Hi,
I have started looking into it, but I got busy with other work so it is currently a bit on hold.
I'll see if I have som time soon-ish..

[ChadTaljaardt]

Any news on this?

[jpatel3]

@rhblind any pointer? I am also looking for same, if you are busy I can add it. please share if you have already done an analysis of it.

[pymarco]

@jpatel3 have you made any progress on this implementation?

[jpatel3]

@pymarco I was using ListAPIView, so I did override on list method to add a suggestion in it if any.

def list(self, request, *args, **kwargs):
    response = super(SearchList, self).list(request, args, kwargs)

    q = request.query_params.get('q', None)
    suggestion_query = SearchQuerySet().auto_query(q.strip())

    if suggestion_query.count() == 0:
        suggestion = suggestion_query.spelling_suggestion()
        if suggestion:
            suggestion = re.sub(r"[\(.*?\)]", "", suggestion)

        response.data['suggestion'] = suggestion

    response.data['searchQuery'] = q.strip()

    return response

Where suggestion is spelling suggestion, and searchQuery is actual query. The reason why I extracted the value using regular expression is that the spelling suggestion string was coming in bracket e.g. "(government)".

Hope it helps!

[rhblind]

Hello guys.
Sorry for not replying, been insanely busy lately.

I did start working on this, but didn't get very far. I started by adding this as a filter feature, but I think that's going down the wrong path. I quickly ran into issues where spelling suggestions duplicated themselves (probably because of stacked filters or something), and didn't have any more time investigating.

If anybody find a clean and generic way of implementing this, I'm always open of merging pull requests!