jokull/supabase/migrations/20240410070316_scheduled-jobs.sql

-- downloads data async into the http response table.
SELECT
  cron.schedule(
    'download-bus-data',
    '5 seconds',
    $$
    select gather_bus_data();
    $$
  );

-- copies data into a more permanent tabe/more useful format. note: in
-- supabase, you cannot delete triggers on net._http_response
-- directly. but a cascading delete of the trigger function also
-- removes the trigger itself.
CREATE FUNCTION copy_to_raw_table() RETURNS trigger LANGUAGE plpgsql AS $$
  BEGIN
    -- NOTE: the http_response sequence resets on DB restart, so there
    -- is potential for old responses to have duplicated ids. we use a
    -- constraint to only update newer entries (within the last 30
    -- seconds). this should make accidentally overwriting older data
    -- with newer values difficult. The API also provides a
    -- lastUpdated value. we require that raw.created is at or after
    -- that value to be updated.
    UPDATE raw_bus_positions raw
       SET response_json = NEW.content::jsonb->'data'->'BusLocationByRoute',
           response_status = NEW.status_code
     WHERE raw.request_id = NEW.id
       -- fairly generous constraint to account for long requests.
       AND raw.created >= (NEW.created - '30 seconds'::interval)
       -- the response must be at or after we actually sent the request.
       AND raw.created >= (NEW.content::jsonb->'data'->'BusLocationByRoute'->>'lastUpdate')::timestamptz
       AND NEW.status_code = 200
       AND raw.response_json IS NULL;

      -- what if we want to stream other data? we can do multiple updates in
      -- different tables, where the request id is.
    RETURN NULL; -- this is an AFTER trigger
  END;
$$;

CREATE CONSTRAINT TRIGGER copy_http_response
  AFTER INSERT ON net._http_response
  DEFERRABLE INITIALLY DEFERRED
  FOR EACH ROW
  EXECUTE FUNCTION copy_to_raw_table();