Combine two columns together from separate tables

Question

Let's say for instance:

• I have two tables: old_data and new_data.

• Both old_data and new_data have one column called this_is_col.

• Both old_data and new_data have various (hundreds) of rows of dates (2010-02-06, 2010-01-09, 2007-06-02, etc.). Both tables don't necessarily have the same dates, but they both have the same format.

• The fields of both tables are various integers.

---

My task:

• Copy the fields from old_data to new_data.

• If a date exists in both tables, the field in new_data will be replaced.

• If the date doesn't exist in new_data, then the correct row will be added and the field will be copied over.

---

Here is how far I've gotten:

Create a temporary column:

ALTER TABLE `new_data` ADD `tempColumn` TEXT NULL;

Copy over data from old_data:

INSERT INTO `new_data` (`tempColumn`) SELECT `this_is_col` FROM `old_data`;

Combine temporary column and new_data . this_is_col. (I haven't really figured this step out since I haven't gotten this far).

MERGE? `tempColumn` `this_is_col`;

Delete temporary table

ALTER TABLE `new_data` DROP `tempColumn`;

---

Upon performing the second action (transferring the data over to the temporary column) I get this error:

#1062 - Duplicate entry '0000-00-00' for key 1

And now I'm stuck. Any help would be appreciated. I'm using MySQL and phpMyAdmin to test the SQL commands.

Answer 1

you want INSERT ... ON DUPLICATE KEY UPDATE. your solution already satisfies steps 1 and 3 of your task, ON DUPLICATE KEY UPDATE will take care of step 2.

Answer 2

How about just doing an update and insert

UPDATE new_data SET col=col FROM new_data a join old_data b on a.this_is_col = b.this_is_col

Then

INSERT INTO new_data (cols) SELECT cols FROM old_data WHERE this_is_col NOT IN (SELECT this_is_col FROM new_data)

Unless I misunderstood...

Answer 3

Assuming your dates are indexed as unique keys:

INSERT INTO newtable
SELECT *
FROM oldtable
ON DUPLICATE KEY column1=oldcolumn1, ...

Answer 4

If you'd rather delete the row first, instead of updating: REPLACE

It'd be just one line too, so: REPLACE data SELECT, you wouldn't have to do the weirdness with adding a text column.