Reputation: 7005
I have a PostgeresDB with the following constraint:
CONSTRAINT "Car_Data_3PM_pkey" PRIMARY KEY ("F_ID", "Date"),
CONSTRAINT "Car_Data_3PM_F_ID_fkey" FOREIGN KEY ("F_ID")
REFERENCES "Bike_Data" ("F_ID") MATCH SIMPLE
ON UPDATE NO ACTION ON DELETE NO ACTION
When I try to insert multiple values using:
INSERT INTO "Car_Data_3PM" ("F_ID","Date","Price_Type","O","H","L","LT","EQ","V","NAD") VALUES (38,'2016-10-02 08:19:40.056679','x',0,0,0,112.145,0,0,112.145),(14,'2016-10-02 08:19:40.056679','x',0,0,0,5476,0,0,5476),(13,'2016-10-02
I get this error:
ERROR: insert or update on table "Car_Data_3PM" violates foreign key constraint "Car_Data_3PM_F_ID_fkey" SQL state: 23503 Detail: Key (F_ID)=(38) is not present in table "Bike_Data".
NO ROWS are inserted.
How can I make Postgres ONLY miss out the rows where the constraint is an issue? i.e Insert most of them?
Upvotes: 2
Views: 3500
Reputation: 246413
One way is to write a trigger that filters out the bad values, like this:
CREATE FUNCTION car_insert_filter() RETURNS trigger
LANGUAGE plpgsql AS
$$BEGIN
IF EXISTS(SELECT 1 FROM "Bike_Data" WHERE "F_ID" = NEW."F_ID")
THEN
RETURN NEW;
ELSE
RAISE NOTICE 'Skipping row with "F_ID"=% and "Date"=%',
NEW."F_ID", NEW."Date";
RETURN NULL;
END IF;
END;$$;
CREATE TRIGGER car_insert_filter
BEFORE INSERT ON "Car_Data_3PM" FOR EACH ROW
EXECUTE PROCEDURE car_insert_filter();
Upvotes: 3
Reputation:
You can't make Postgres ignore the values, but you can rewrite your statement to not insert those rows:
INSERT INTO "Car_Data_3PM" ("F_ID","Date","Price_Type","O","H","L","LT","EQ","V","NAD")
select *
from (
VALUES
(38,'2016-10-02 08:19:40.056679','x',0,0,0,112.145,0,0,112.145),
(14,'2016-10-02 08:19:40.056679','x',0,0,0,5476,0,0,5476),
... -- all other rows
) as x (id, date, price_type, o, h, l, lt, eq, v nad)
where exists (select 1
from "Bike_Data" bd
where bd."F_ID" = x .id)
Upvotes: 3