Reputation: 808
What is the best way to solve the following issue with Postgresql?
For each row I INSERT into an article Table based on a SELECT from another table I want to UPDATE certain columns of the inserted articles.
This is my current solution:
-- temporarily alter table to avoid not null issues
ALTER TABLE article ALTER COLUMN fk_article_unit DROP NOT NULL;
(...)
--create article and return inserted pks, store these in a temporary table so they can be used for all following updates
WITH articles AS (
insert into article
(
...
)
select
...
from other_table
where some_condition
RETURNING pk
)
SELECT pk INTO temporary temp_articles
FROM articles;
-- update various fk for all newly created articles
UPDATE article
SET fk_article_type =
(SELECT pk
FROM article_type
WHERE unique_id = 'service')
WHERE pk in (select pk from temp_articles);
UPDATE article
SET fk_article_type =
(SELECT min(pk)
FROM vat_code)
WHERE fk_article_type is null;
(... several more updates)
--readd no null constraint
ALTER TABLE article ALTER COLUMN fk_article_type SET NOT NULL;
(...)
Upvotes: 1
Views: 47
Reputation: 123
I don't see why this can't be done with a single insert
query. If the solution below doesn't apply please provide some additional information on your data model.
insert into article
(
...,
fk_article_type
)
select
...,
coalesce -- if first query is null, then the result of second will be used
(
( -- 1st query
select pk from article_type where unique_id= 'service'
),
( -- 2nd query
select min(pk) from vat_code
)
)
from other_table
where some_condition
returning pk;
Upvotes: 1