How to split / unpivot group_concat in mysql? - mysql

Question

I currently have a view that has a column called alt_email_contact and I used the group_concat function in order to get multiple emails associated with one contact. However I want to be able to extract each email and create a separate column for each.
Example:
id email
1 SkyW#gmail.com, SW#gmail.com, WW#gmail.com, WalterW#gmail.com
the amount of emails is subject to change from one user to another therefore there wont always be four emails for each user. I want to create a new column per each email like so:
id email_1 email_2 email_3 email_4
1 SkyW#gmail.com SW#gmail.com WW#gmail.com WalterW#gmail.com
(I am using phpmyadmin) I would like to be able to modify my view to contain the variable amount of emails per user.

Answer 1

You can use substring_index() to achieve what you want. It is not really pretty, but it will work:
select id,
substring_index(emails, ', ', 1) as email_1
(case when length(emails) - length(replace(emails, ',', '')) >= 1
then substring_index(substring_index(emails, ', ', 2), ', ', -1)
end) as email_2,
(case when length(emails) - length(replace(emails, ',', '')) >= 2
then substring_index(substring_index(emails, ', ', 3), ', ', -1)
end) as email_3,
(case when length(emails) - length(replace(emails, ',', '')) >= 3
then substring_index(substring_index(emails, ', ', 4), ', ', -1)
end) as email_4,
(case when length(emails) - length(replace(emails, ',', '')) >= 4
then substring_index(substring_index(emails, ', ', 5), ', ', -1)
end) as email_5
from table t;
You can insert these values into another table, if you like.

Answer 2

Don't use that view, since GROUP_CONCAT() has already ruined the normalization. You want to simulate a pivot table using the limited SQL capabilities of MySQL.
Let's assume that your view is based on a Contacts table that looks like this:
CREATE TABLE Contacts
( id INTEGER NOT NULL
, alt_email_contact VARCHAR(256) NOT NULL
);
Create this helper view instead (which is basically RANK() OVER (PARTITION BY id ORDER BY alt_email_contact), except that MySQL doesn't support RANK()):
CREATE VIEW NumberedContacts AS
SELECT c1.id, c1.alt_email_contact, COUNT(*) AS rank
FROM Contacts c1
INNER JOIN Contacts c2
ON c2.id = c1.id AND
c1.alt_email_contact >= c2.alt_email_contact
GROUP BY c1.id, c1.alt_email_contact;
Then you can write this query or view, which gives you up to 5 alternate e-mail addresses, ordered alphabetically:
CREATE VIEW ContactsForImport AS
SELECT c1.id
, c1.alt_email_contact AS email_1
, c2.alt_email_contact AS email_2
, c3.alt_email_contact AS email_3
, c4.alt_email_contact AS email_4
, c5.alt_email_contact AS email_5
FROM NumberedContacts AS c1
LEFT OUTER JOIN NumberedContacts AS c2
ON c1.id = c2.id AND c2.rank = 2
LEFT OUTER JOIN NumberedContacts AS c3
ON c1.id = c3.id AND c3.rank = 3
LEFT OUTER JOIN NumberedContacts AS c4
ON c1.id = c4.id AND c4.rank = 4
LEFT OUTER JOIN NumberedContacts AS c5
ON c1.id = c5.id AND c5.rank = 5
WHERE c1.rank = 1;
SQL Fiddle