Postgres - Transpose Rows to Columns

Question

I have the following table, which gives multiple email addresses for each user.

I need to flatten this out to columns on a user query. To give me the "newest" 3 email addresses based on the creation date.

user.name | user.id | email1          | email2           | email3**

Mary      | 123     | mary@gmail.com  | mary@yahoo.co.uk | mary@test.com

Joe       | 345     | joe@gmail.com   | [NULL]           | [NULL]

Answer 1

Use crosstab() from the tablefunc module.

SELECT * FROM crosstab(
   $$SELECT user_id, user_name, rn, email_address
     FROM  (
        SELECT u.user_id, u.user_name, e.email_address
             , row_number() OVER (PARTITION BY u.user_id
                            ORDER BY e.creation_date DESC NULLS LAST) AS rn
        FROM   usr u
        LEFT   JOIN email_tbl e USING (user_id)
        ) sub
     WHERE  rn < 4
     ORDER  BY user_id
   $$
  , 'VALUES (1),(2),(3)'
   ) AS t (user_id int, user_name text, email1 text, email2 text, email3 text);

I used dollar-quoting for the first parameter, which has no special meaning. It's just convenient if you have to escape single quotes in the query string which is a common case:

• Insert text with single quotes in PostgreSQL

Detailed explanation and instructions here:

• PostgreSQL Crosstab Query

And in particular, for "extra columns":

• Pivot on Multiple Columns using Tablefunc

The special difficulties here are:

• The lack of key names.
  -> We substitute with row_number() in a subquery.

• The varying number of emails.
  -> We limit to a max. of three in the outer SELECT
  and use crosstab() with two parameters, providing a list of possible keys.

Pay attention to NULLS LAST in the ORDER BY.

Answer 2

If anyone else that finds this question and needs a dynamic solution for this where you have an undefined number of columns to transpose to and not exactly 3, you can find a nice solution here: https://github.com/jumpstarter-io/colpivot

Answer 3

the above code has really helped me. Now, I am trying to add more fields to the pivot table.

    SELECT * INTO temp_pivot_table FROM crosstab(
    $$SELECT lead_id, magazine_id, days_between_mailers, rn
    FROM  (
    SELECT u.lead_id, u.magazine_id, u.days_between_mailers
         , row_number() OVER (PARTITION BY u.lead_id
                        ORDER BY u.id ASC NULLS LAST) AS rn
    FROM   temp_mailer_stats u
    ) sub
    WHERE  rn <= 3
    ORDER  BY lead_id
   $$

  , 'VALUES (1),(2),(3),(4),(5),(6)'
  ) AS t (lead_id int, magazine1 text, magazine2 text, magazine3 text, 
   days_between1 integer, days_between2 integer, days_between3 integer);

Of course I can make separate tables and then join them together, but if I can save a step that would be ideal.

Thanks in advance for the help.