Postgres rule to help with CSV import

Question

I need to import a csv file that has 10's of thousands of rows into a Postgres database daily. I'm looking into the most efficient way to do that as each line in the csv file can be a new record or an existing one that should be updated if it's there. After many searches, I stumbled upon a solution, which I used:

CREATE OR REPLACE RULE insert_on_duplicate_update_advertiser_campaign_keywords_table AS
  ON INSERT TO advertiser_campaign_keywords
  WHERE (new.phrase, new.match_type, new.advertiser_campaign_id) IN (
    SELECT phrase, match_type, advertiser_campaign_id
    FROM advertiser_campaign_keywords
    WHERE phrase = new.phrase AND match_type = new.match_type AND advertiser_campaign_id = new.advertiser_campaign_id AND state != 'deleted')
DO INSTEAD
  UPDATE advertiser_campaign_keywords
  SET bid_price_cpc = new.bid_price_cpc
  WHERE phrase = new.phrase AND match_type = new.match_type AND advertiser_campaign_id = new.advertiser_campaign_id;

This is the closest I've come to a working solution, but it's not complete. It fails on inserts that look like this:

INSERT INTO advertiser_campaign_keywords (phrase, bid_price_cpc, match_type, advertiser_campaign_id) VALUES 
('dollar', 1::text::money, 'Broad', 1450),
('two words', 1.2::text::money, 'Broad', 1450),
('two words', 1.0::text::money, 'Broad', 1450),
('three words exact', 2.5::text::money, 'Exact', 1450),
('four words broad match', 1.1::text::money, 'Exclusive', 1450),
('three words exact', 2.1::text::money, 'Exact', 1450);

The error message is:

duplicate key value violates unique constraint "unique_phrase_campaign_combo"

unique_phrase_campaign_combo looks like:

CONSTRAINT "unique_phrase_campaign_combo" UNIQUE ("phrase", "advertiser_campaign_id", "match_type", "deleted_at")

deleted_at is null unless the record is marked as deleted.

Anyone know how I can solve this problem?

Thanks

Answer 1

The best way to do this is to add a staging table. Use copy to populate the staging table. Then use that to do your inserts and updates.

UPDATE target_table t
  SET ...
FROM staging_table s
WHERE t.id = s.id

INSERT INTO target_table
SELECT * FROM staging_table s
WHERE s.id NOT EXISTS (
  SELECT id FROM target_table
)