Question

I’m using regex_replace in postgreSQL and trying to strip out any character in a string that is not a letter or number. However, using this regex:

select * from regexp_replace('blink-182', '[^a-zA-Z0-9]*$', '')

returns ‘blink-182’. The hyphen is not being removed and replaced with nothing (”) as I would expect.

How do I modify this regex to also replace the hypen – I’ve tested with many other characters (!,.#) and they are all replaced correctly.

Any ideas?

Answer 1

You currently replace a run of non-alphanumeric characters at the end of the string only. I guess your tests were mainly strings of the form foobar!# which worked because the characters to remove were at the end of the string.

To replace every occurrence of such a character in the string remove the $ from the regex:

[^a-zA-Z0-9]+

(also I changed the * into a + to prevent zero-length replaces between every character.

If you want to retain whitespace as well you need to add it to the character class:

[^a-zA-Z0-9 ]+

or possibly

[^a-zA-Z0-9\s]+

If the regex in the beginning was in fact correct in that you only want to remove non-alphanumeric characters from the end of the string but you also want to remove hyphen-minus in the middle of a string (while retaining other non-alphanumeric characters in the middle of the string), then the following should work:

[^a-zA-Z0-9]+$|-

maniek points out that you need to add an argument to regexp_replace so it will replace more than once match:

regexp_replace('blink-182', '[^a-zA-Z0-9]+$|-', '', 'g')