MySQL: how to convert to EAV - Part 3? - mysql

Question

Previous Related Posts:
MySQL: how to convert to EAV?
MySQL: how to convert to EAV - Part 2?
Given a table:
TABLE: foo
===============================
| id | first_name | last_name |
===============================
| 1 | John | Doe |
| 2 | Jane | Smith |
| 3 | Ronald | McDonald |
-------------------------------
How do I take this table and convert it to these tables (an EAV implementation)?:
TABLE: attribute
===========================
| id | fk_id | attribute |
===========================
| 1 | 100 | first_name |
| 2 | 100 | last_name |
---------------------------
TABLE: value
=========================================
| id | attribute_id | row_id | value |
=========================================
| 1 | 1 | 1 | John |
| 2 | 2 | 1 | Doe |
| 3 | 1 | 2 | Jane |
| 4 | 2 | 2 | Smith |
| 5 | 1 | 3 | Ronald |
| 6 | 2 | 3 | McDonald |
-----------------------------------------
NOTES:
attribute.fk_id will be provided.
value.row_id is used to identify how the values are grouped as records in the original table.
UPDATE: Also, how do I query the EAV tables so that I can make it look like table foo again.

Answer 1

I give +1 to #Phil's solution for populating the EAV table. Insert one attribute at a time.
Here's another solution to reverse an EAV transformation:
SELECT v.row_id AS id,
MAX(IF(a.attribute='first_name',v.value,NULL)) AS first_name,
MAX(IF(a.attribute='last_name',v.value,NULL)) AS last_name
FROM value v INNER JOIN attribute a
ON v.attribute_id = a.id
GROUP BY v.row_id
Except that by using EAV, you've put all your values into a column of VARCHAR(255) or whatever, so you have lost information about the respective data types of the original columns.
There's really no way to do it "dynamically" without hard-coding the attribute names, either as joins as #Phil shows, or as columns as I show. It's essentially the same problem as trying to write dynamic pivot queries.
I have written more about EAV in my presentation Practical Object-Oriented Models in SQL and in my book, SQL Antipatterns Volume 1: Avoiding the Pitfalls of Database Programming.

Answer 2

I think your only hope is if you use the foo table. bar is essentially useless without the ID.
Try something like this (assuming attribute.id is an auto-increment primary key)
INSERT INTO `attribute` (`fk_id`, `attribute`)
VALUES (100, 'first_name');
INSERT INTO `value` (`attribute_id`, `row_id`, `value`)
SELECT LAST_INSERT_ID(), `id`, `first_name`
FROM `foo`;
INSERT INTO `attribute` (`fk_id`, `attribute`)
VALUES (100, 'last_name');
INSERT INTO `value` (`attribute_id`, `row_id`, `value`)
SELECT LAST_INSERT_ID(), `id`, `last_name`
FROM `foo`;
To reconstruct the foo table, try this
SELECT
`fn`.`row_id` AS `id`,
`fn`.`value` AS `first_name`,
`ln`.`value` AS `last_name`
FROM `value` `fn`
INNER JOIN `attribute` `fn_a`
ON `fn`.`attribute_id` = `fn_a`.`id`
AND `fn_a`.`attribute` = 'first_name'
INNER JOIN `value` `ln`
ON `fn`.`row_id` = `ln`.`row_id`
INNER JOIN `attribute` `ln_a`
ON `ln`.`attribute_id` = `ln_a`.`id`
AND `ln_a`.`attribute` = 'last_name'
Ergh, thanks for reminding me why I hate this pattern