-- This sets up the `r` schema, which contains things that can be safely dropped and replaced instead of being -- changed using migrations. -- -- Statements in this file may not create or modify things outside of the `r` schema (indicated by the `r.` prefix), -- except for these things, which are associated with something other than a schema (usually a table): -- * A trigger if the function name after `EXECUTE FUNCTION` is in `r` (dropping `r` drops the trigger) -- -- The default schema is not temporarily set to `r` because it would not affect some things (such as triggers) which -- makes it hard to tell if the rule above is being followed. -- -- If you add something here that depends on something (such as a table) created in a new migration, then down.sql must use -- `CASCADE` when dropping it. This doesn't need to be fixed in old migrations because the "replaceable-schema" migration -- runs `DROP SCHEMA IF EXISTS r CASCADE` in down.sql. BEGIN; DROP SCHEMA IF EXISTS r CASCADE; CREATE SCHEMA r; -- Rank calculations CREATE OR REPLACE FUNCTION r.controversy_rank (upvotes numeric, downvotes numeric) RETURNS float AS $$ BEGIN IF downvotes <= 0 OR upvotes <= 0 THEN RETURN 0; ELSE RETURN (upvotes + downvotes) * CASE WHEN upvotes > downvotes THEN downvotes::float / upvotes::float ELSE upvotes::float / downvotes::float END; END IF; END; $$ LANGUAGE plpgsql IMMUTABLE; -- Selects both old and new rows in a trigger and allows using `sum(count_diff)` to get the number to add to a count CREATE FUNCTION r.combine_transition_tables () RETURNS SETOF record LANGUAGE sql AS $$ SELECT -1 AS count_diff, * FROM old_table UNION ALL SELECT 1 AS count_diff, * FROM new_table; $$; -- These triggers resolve an item's reports when the item is marked as removed. CREATE PROCEDURE r.resolve_reports_when_target_removed (target_name text) LANGUAGE plpgsql AS $a$ BEGIN EXECUTE format($b$ CREATE FUNCTION r.resolve_reports_when_%1$s_removed ( ) RETURNS TRIGGER LANGUAGE plpgsql AS $$ BEGIN UPDATE %1$s_report AS report SET resolved = TRUE, resolver_id = mod_person_id, updated = now() FROM new_removal WHERE report.%1$s_id = new_removal.%1$s_id AND new_removal.removed; RETURN NULL; END $$; CREATE TRIGGER resolve_reports AFTER INSERT ON mod_remove_%1$s REFERENCING NEW TABLE AS new_removal FOR EACH STATEMENT EXECUTE FUNCTION r.resolve_reports_when_%1$s_removed ( ); $b$, target_name); END $a$; CALL r.resolve_reports_when_target_removed ('comment'); CALL r.resolve_reports_when_target_removed ('post'); -- These triggers create and update rows in each aggregates table to match its associated table's rows. -- Deleting rows and updating IDs are already handled by `CASCADE` in foreign key constraints. CREATE FUNCTION r.comment_aggregates_from_comment () RETURNS TRIGGER LANGUAGE plpgsql AS $$ BEGIN INSERT INTO comment_aggregates (comment_id, published) SELECT id, published FROM new_comment; RETURN NULL; END $$; CREATE TRIGGER aggregates AFTER INSERT ON comment REFERENCING NEW TABLE AS new_comment FOR EACH STATEMENT EXECUTE FUNCTION r.comment_aggregates_from_comment (); CREATE FUNCTION r.community_aggregates_from_community () RETURNS TRIGGER LANGUAGE plpgsql AS $$ BEGIN INSERT INTO community_aggregates (community_id, published) SELECT community_id, published FROM new_community; RETURN NULL; END $$; CREATE TRIGGER aggregates AFTER INSERT ON community REFERENCING NEW TABLE AS new_community FOR EACH STATEMENT EXECUTE FUNCTION r.community_aggregates_from_community (); CREATE FUNCTION r.person_aggregates_from_person () RETURNS TRIGGER LANGUAGE plpgsql AS $$ BEGIN INSERT INTO person_aggregates (person_id) SELECT id, FROM new_person; RETURN NULL; END $$; CREATE TRIGGER aggregates AFTER INSERT ON person REFERENCING NEW TABLE AS new_person FOR EACH STATEMENT EXECUTE FUNCTION r.person_aggregates_from_person (); CREATE FUNCTION r.post_aggregates_from_post () RETURNS TRIGGER LANGUAGE plpgsql AS $$ BEGIN INSERT INTO post_aggregates (post_id, published, newest_comment_time, newest_comment_time_necro, community_id, creator_id, instance_id, featured_community, featured_local) SELECT new_post.id, new_post.published, new_post.published, new_post.published, new_post.community_id, new_post.creator_id, community.instance_id, new_post.featured_community, new_post.featured_local FROM new_post INNER JOIN community ON community.id = new_post.community_id ON CONFLICT DO UPDATE SET featured_community = excluded.featured_community, featured_local = excluded.featured_local; RETURN NULL; END $$; CREATE TRIGGER aggregates AFTER INSERT OR UPDATE OF featured_community, featured_local ON post REFERENCING NEW TABLE AS new_post FOR EACH STATEMENT EXECUTE FUNCTION r.post_aggregates_from_post (); CREATE FUNCTION r.site_aggregates_from_site () RETURNS TRIGGER LANGUAGE plpgsql AS $$ BEGIN -- we only ever want to have a single value in site_aggregate because the site_aggregate triggers update all rows in that table. -- a cleaner check would be to insert it for the local_site but that would break assumptions at least in the tests IF NOT EXISTS ( SELECT 1 FROM site_aggregates) THEN INSERT INTO site_aggregates (site_id) VALUES (NEW.id); RETURN NULL; END $$; CREATE TRIGGER aggregates AFTER INSERT ON site FOR EACH ROW EXECUTE FUNCTION r.site_aggregates_from_site (); -- These triggers update aggregates in response to votes. CREATE PROCEDURE r.aggregates_from_like (target_name text, creator_id_getter text) LANGUAGE plpgsql AS $a$ BEGIN EXECUTE format($b$ CREATE FUNCTION r.%1$s_aggregates_from_like ( ) RETURNS TRIGGER LANGUAGE plpgsql AS $$ BEGIN -- Update aggregates for target, then update aggregates for target's creator WITH target_diff AS ( UPDATE %1$s_aggregates SET (score, upvotes, downvotes, controversy_rank) = (score + diff.upvotes - diff.downvotes, upvotes + diff.upvotes, downvotes + diff.downvotes, controversy_rank ((upvotes + diff.upvotes)::numeric, (downvotes + diff.downvotes)::numeric)) FROM ( SELECT %1$s_id, sum(count_diff) FILTER (WHERE score = 1) AS upvotes, sum(count_diff) FILTER (WHERE score <> 1) AS downvotes FROM r.combine_transition_tables () GROUP BY %1$s_id) AS diff WHERE %1$s_aggregates.%1 $ s_id = diff.%1$s_id RETURNING %2$s AS creator_id, diff.upvotes - diff.downvotes AS score) UPDATE person_aggregates SET %1$s_score = %1$s_score + diff.sum FROM ( SELECT creator_id, sum(score) FROM target_diff GROUP BY creator_id) AS diff WHERE person_aggregates.person_id = diff.creator_id; RETURN NULL; END $$; CREATE TRIGGER aggregates AFTER INSERT OR DELETE OR UPDATE OF score ON %1$s_like REFERENCING OLD TABLE AS old_table NEW TABLE AS new_table FOR EACH STATEMENT EXECUTE FUNCTION r.%1$s_aggregates_from_like; $b$, target_name, creator_id_getter); END $a$; CALL r.aggregates_from_like ('comment', '(SELECT creator_id FROM comment WHERE comment.id = target_aggregates.comment_id LIMIT 1)'); CALL r.aggregates_from_like ('post', 'target_aggregates.creator_id'); COMMIT;