Postgresql tsvector field and query support

Have you considered support for Postgresql’s tsvector type? ( https://www.postgresql.org/docs/12/textsearch.html )

I realize this would be entirely Postgresql specific so perhaps it would be more appropriate as a contrib module than a core feature.

I have been looking at implementing it and had a few different ideas; something like this perhaps:

• custom Field which creates a tsvector column type
• meta-field listing in a Model to describe which columns to full-text index
• Add a CREATE TRIGGER to the schema generation for each tsvector field. For each insert to the table, invoke tsvector_update_trigger ( https://www.postgresql.org/docs/12/functions-textsearch.html ) for the new row to create the vector on the fields listed by your meta-field. This is the thing I’m least sure about; should this be implemented in Tortoise, or just use the database trigger to manage it?
• custom query Function which allows for the ‘@@ tsquery’ syntax

One thing that Tortoise is missing is that it does not appear to be possible to tell Tortoise what index type you’d like when specifying to index a column; you would want to index the tsvector columns with either GIST or GIN depending on user requirements.

[grigi]

A special filter makes sense. I’m quite sick at the moment (and thinking is difficult), so I won’t get to anything soon. Possibly @abondar can help in the mean time if you want to attempt it?

[parkournerd]

I found that Postgres 12 added auto-generated and saved tsvectors: https://pgdash.io/blog/postgres-12-generated-columns.html. This could keep TsVectorField updated instead of using triggers. Seems like it almost fit into the generated_sql, but needs fields to compute from