SQL: ORDER BY using a substring within a specific column… possible?

Question

I have a database whose columns are npID, title, URL, and issue.

Here is an example of two years\' entries:

npID               title               UR         


        

Answer 1

Standard SQL

Try a CASE statement in the ORDER BY:

SELECT npID, title, URL, issue
FROM   tbl
ORDER  BY substring(issue, 1, 4) DESC
      ,CASE 
          WHEN substring(issue, 6, 100) IN ('Winter','First_Quarter')  THEN 1
          WHEN substring(issue, 6, 100) IN ('Summer','Second_Quarter') THEN 2
          WHEN substring(issue, 6, 100) IN ('Spring','Third_Quarter')  THEN 3
          WHEN substring(issue, 6, 100) IN ('Fall',  'Fourth Quarter') THEN 4
          ELSE 5 
       END;

Don't ask, why Winter -> Summer -> Spring - it's what the client wants! :)

Optimize performance

A "simple" CASE should perform better since the expression is only evaluated once.
And right(issue, -5) is equivalent to substring(issue, 6, 100), but a bit faster:

SELECT npid, title, url, issue
FROM   tbl
ORDER  BY left(issue, 4) DESC
      ,CASE right(issue, -5)
          WHEN 'Winter'         THEN 1
          WHEN 'First_Quarter'  THEN 1
          WHEN 'Summer'         THEN 2
          WHEN 'Second_Quarter' THEN 2
          WHEN 'Spring'         THEN 3
          WHEN 'Third_Quarter'  THEN 3
          WHEN 'Fall'           THEN 4
          WHEN 'Fourth Quarter' THEN 4
          ELSE 5 
       END;

left() and right() have been added with PostgreSQL 9.1. The trick with right() is to use a negative number to trim a constant number of characters from the left.

Syntax variants

These are equivalent (for strings of <= 100 characters):

SELECT substring(issue from 6 for 100) AS substring1
      ,substring(issue, 6, 100)        AS substring2
      ,substring(issue, 6)             AS substring3
      ,substr(issue, 6, 100)           AS substr1
      ,substr(issue, 6)                AS substr2
      ,right(issue, -5)                AS right0
FROM tbl

-> sqlfiddle