I wander around can I transfer the data gathered in memory table to actual one by sql query only.
The two tables have the same structure and pk is products_id(int, AI)
The problem is the criteria is completely different than the pk. The products are identified by 2 columns - barcode and company.
So ignoring the pk in whole, I need to update the data if in actual table there is a row with the same barcode and company, and insert new record if there is none.
Tried this:
INSERT INTO products (products_sku, ...)
SELECT products_sku... FROM temp_products
WHERE (temp_products.products_barcode = products.products_barcode) AND (temp_products.products_comp = products.products_comp)
But i dont have access in the select to products table so to make the filtering
CodePudding user response:
I think you need to add a unique key on products_barcode
and products_comp
:
ALTER TABLE products ADD UNIQUE KEY (products_barcode, products_comp);
Once you have it you can perform insert-or-update in one statement:
INSERT INTO products (/* all columns except the id */)
SELECT /* all columns except the id */
FROM products_sku
ON DUPLICATE KEY UPDATE some_field = VALUES(some_field), ...
/* list all columns except the id / barcode / comp */;
So when it meets a duplicate barcode/comp pair it will fall into the ON DUPLICATE KEY UPDATE and won't insert. Read more how it works: https://dev.mysql.com/doc/refman/8.0/en/insert-on-duplicate.html