I'm using postgres >= 9.6. I have tasks, tags and task_tags tables, for a typical n:m relationship between tasks and tags.

I'd like to be able to programmatically build queries against the tasks table that supports queries against the actual fields of tasks, but also on the tags (tag names) of a task.

Queries on the task fields themselves are straight-forward. Queries on the tags for a "does it have tag A?" are also straight-forward. What I am struggling with is coming up with a select/query structure that would allow me to also match things like "does it have tag A and tag B"?

The best I've come up with is a lateral join on a subquery with an array aggregation, and then using array matching functions, e.g.:

   SELECT array_agg(tags.name) AS tags FROM task_tags INNER JOIN tags ON task_tags.tag_id = tags.id WHERE task_tags.task_id = tasks.id GROUP BY task_tags.task_id) tt
WHERE tt.tags @> array['tag1'::varchar, 'tag3'::varchar];

That way, it should be possible to programmatically build a WHERE clause (using tasks.* and tt.tags) satisfying all of the conditions of the user-provided "query".

However, I'm not sure if this is the best way of doing it - thoughts?

Similarly, is there any way at all of making it work with wildcards against the tag names? Normal array matching wouldn't allow that, and solutions I've seen suggest using unnest (or, well, not using arrays in the first place), but then I'd lose the ability of saying "it needs to have both tagA and tagB".

Is there any other way of building a query on these relationships that would allow that kind of "both tagA and tagB" matching?