damoti / django-tsvector-field Goto Github PK

Pending readonly field attribute implementation in Django:

django/deps#36

fields.py imports django.utils.encoding.force_text which has been deprecated in favor of force_str. Django 4.0 has now removed the function entirely: https://docs.djangoproject.com/en/4.0/releases/4.0/#features-removed-in-4-0

Any chance for a simple update on this really handy package so I can keep using it with the latest django versions?

Following along with the tutorial I've added a 'search' field to an existing model and generated the migration file. But when I run 'migrate' the migrations fails with a ProgrammingError 'name-of-the-field_tsvector' already exists.

Running django-tsvector-field 0.9.3, django 1.11, and PostgreSQL 9.5.6

Hey, awesome library! I was disappointed to see that Django does not offer any support out of the box for FTS for real applications - using triggers and store procedures instead of indexing on the fly - which obviously does not scale on even a modest level of traffic.

For our use case, we'd like to index fields on related models through joins. One simple example to allude to is the ability to search for orders by line item descriptions. Briefly digging into the codebase, it does not look like this is currently supported.

I'd be happy to contribute something to make this work. Has this feature been thought about before? Any recommendations for how this should be tackled?

I believe there is a bug in the generated trigger logic. If a NULL column is given a non-NULL value, this won't trigger an update to the tsvector field to include the new text.

The issue is with logic like:

     ELSIF (NEW."some_indexed_field" <> OLD."some_indexed_field") THEN do_update = true;

Since NULL <> 'something' is not true in SQL. The simplest solution I could come up with is to replace the <> tests with IS DISTINCT FROM.

Let me know if you'd like a pull request.

Update the README with some instructions on how to use the Headline class and present matched excerpts from the result set....

https://github.com/damoti/django-tsvector-field/blob/master/tsvector_field/query.py#L5

from django.db.models.expressions import F
from django.contrib.postgres.search import SearchRank, SearchQuery
import tsvector_field

query = SearchQuery(self.form['text'].value())
qs = (
  qs.filter(text__search=query)
    .annotate(
      match_title=tsvector_field.Headline(F('text__title'), query),
      match_text=tsvector_field.Headline(F('text__text'), query)
    )
    .annotate(rank=SearchRank(F('text__search'), query))
    .order_by('-rank')
)

The contents of the tsvector is an implementation detail in postgres, it's kind of pointless to keep shuttling this back and forth between Postgres and Django.