How to create GIN index in Django migration

Question

In Django, since version 1.11 we have a class for PostgreSQL GinIndex (https://docs.djangoproject.com/en/1.11/ref/contrib/postgres/indexes/). I'd like to create a migration that constructs such index on a VectorSearchField I added to one of my tables. So far, I've tried to simply add db_index=True to the VectorSearchField, but that fails, because it tries to create a B-Tree index (I think) and the VectorSearchField values are too long.

I managed to create the index I want by running a migrations.RunSQL() migration with:

CREATE INDEX entryline_sv_index ON entryline USING GIN (sv);

However, I guess, since there is a special GinIndex class in Django, maybe there is a way to create such index without executing raw SQL?

Here's a model class:

import django.contrib.postgres.search as pg_search

class EntryLine(models.Model):
    speaker = models.CharField(max_length=512, db_index=True)
    text = models.TextField()
    sv = pg_search.SearchVectorField(null=True)  # I want a GIN index on this field.

Any idea how to properly create an index for sv field in a migration? Or is executing the CREATE INDEX ... query the best way?

Fictionalize

Answer 1

Haven't yet had a chance to migrate my old manual CREATE INDEX codes to the new system introduced in 1.11 but my understanding is

from django.contrib.postgres.indexes import GinIndex
import django.contrib.postgres.search as pg_search
    
class EntryLine(models.Model):
    speaker = models.CharField(max_length=512, db_index=True)
    text = models.TextField()
    sv = pg_search.SearchVectorField(null=True) 
    class Meta:
        indexes = [GinIndex(fields=['sv'])]

Is what's required. Raw SQL CREATE INDEX statements need not be used any more.

Metametabel

This is exactly what I wanted! :) Just didn't know the proper way to use the GinIndex class. Thank you! – Fictionalize

Glad I could help. All the best with your project – Metametabel