Counting regexp matches with MySQL?

[Tipem]

Is there any way to COUNT regexp matches in MySQL and store it in a column in the results set?

e.g.,

SELECT name, COUNT(REGEXP '[ae]') as matches FROM table

If so, could you provide a couple of examples? I need to do something like this for sorting purposes.

[wibble wobble]

Cant tell what you are trying to match with regex, but this should work:
SELECT SUM(name REGEXP '[ae]') AS matches FROM table

(I use 'SUM' because 'name REGEXP exp' returns 1 or 0.)

[r937]

actually, the fact that REGEXP returns 0 or 1 (instead of TRUE or FALSE) is in my opinion not sufficient reason to sum these values directly

it's better to write it as a CASE expression -- it's more obvious, and easier to swap with some other relevancy expression should you ever want something other than a REGEXP test

SELECT SUM(
        CASE WHEN name REGEXP '[ae]'
             THEN 1 
             ELSE 0 END
          ) AS matches 
  FROM ...

tipem, you might also be interested in this, if you're looking to sort by relevance...
Simple Keyword Relevance

[wibble wobble]

actually, the fact that REGEXP returns 0 or 1 (instead of TRUE or FALSE) is in my opinion not sufficient reason to sum these values directly

it's better to write it as a CASE expression -- it's more obvious, and easier to swap with some other relevancy expression should you ever want something other than a REGEXP test

SELECT SUM(
        CASE WHEN name REGEXP '[ae]'
             THEN 1 
             ELSE 0 END
          ) AS matches 
  FROM ...

tipem, you might also be interested in this, if you're looking to sort by relevance...
Simple Keyword Relevance

Tipem: This is why r937 deserves your vote in the SP Awards for SQL guru

One thing I'm wondering: you want to use CASE for clarity and maintainability. Why not just give the query a comment saying "Sums because REGEXP returns 1 or 0". Over a largish data set, surely the case statements would slow it down quite a bit?

[Tipem]

You guys truly rock my socks! Awesomesauce!

I have run into a problem though... I am trying to find all of the following instances and then count them (yes, with the square brackets included): [a], [e], [p], [l], [d], [w], and [f]. The regexp that I am using is \[[aefdwpl]\] to create this query (obviously this query is just to test on one row):

SELECT w_attack, SUM(
        CASE WHEN w_attack REGEXP '\[[aefdwpl]\]'
             THEN 1 
             ELSE 0 END
          ) AS attack_icons 
  FROM weapons WHERE w_id = 1 GROUP BY w_id

But unfortuately it is returning only 1 for attack_icons (when in fact, there are 23 matches). Does the MySQL REGEXP function match everything? Or just find one match and then stop? Let me know... how should I go about fixing this problem?

Thanks in advanced! Again, you guys rock! I should nominate r937 for SQL guru...

[r937]

I am trying to find all of the following instances and then count them (yes, with the square brackets included): [a], [e], [p], [l], [d], [w], and [f].

try this:

, SUM(
   CASE WHEN w_attack IN ( '[a]', '[e]', '[p]', '[l]', '[d]', '[w]', '[f]' )
        THEN 1 
        ELSE 0 END ) AS attack_icons

[wibble wobble]

MySQL doesnt support regex replacements (which is what you would need: replace other chars with nothing, then count the resulting length). Do this in your programming/scripting language and store the result in a new field ('attack_icons_count' maybe).

Edit: If you need help with it, let me know and I'll do a blog post on it.