Test if a string is regex

Question

Is there a good way of test if a string is a regex or normal string in PHP?

Ideally I want to write a function to run a string through, that returns true or false.

I had a look at preg_last_error():

<?php
preg_match('/[a-z]/', 'test');
var_dump(preg_last_error());
preg_match('invalid regex', 'test');
var_dump(preg_last_error());
?>

Where obviously first one is not an error, and second one is. But preg_last_error() returns int 0 both times.

Any ideas?

`'invalid regex'` is a _valid_ regex - in fact, any string that doesn't involve the regex special characters will be a valid regex. Try `'['` instead. — Aasmund Eldhuset, May 28 '12 at 00:52
@greg0ire: The regex specification is an irregular language. *runs* — Ignacio Vazquez-Abrams, May 28 '12 at 00:53
Any normal string can be used as "regex" to do exact match. What are you trying to achieve? — nhahtdh, May 28 '12 at 00:53
To those that care: `'['` is not a valid regex, therefore not all strings are valid regexs. — Ignacio Vazquez-Abrams, May 28 '12 at 00:54
@AasmundEldhuset, `'invalid regex'` is actually not a valid regex, PHP returns `Warning: preg_match() [function.preg-match]: Delimiter must not be alphanumeric or backslash in D:\xampp\htdocs\overheard\test.php on line 5`. — Hosh Sadiq, May 28 '12 at 01:01
@HoshSadiq: Because the delimiters are missing. However, your code would most likely do something like `'/'.$regex.'/'`, i.e. always add the delimiters. — ThiefMaster, May 28 '12 at 01:02
@ThiefMaster: Yes, but that's the thing, I need to be able to completely check if the regex is valid, and if it's valid if I add the delimiters. — Hosh Sadiq, May 28 '12 at 01:13

score 16 · Answer 1 · edited Jul 17 '14 at 06:35

16

The simplest way to test if a string is a regex is:

if( preg_match("/^\/.+\/[a-z]*$/i",$regex))

This will tell you if a string has a good chance of being intended to be as a regex. However there are many string that would pass that check but fail being a regex. Unescaped slashes in the middle, unknown modifiers at the end, mismatched parentheses etc. could all cause problems.

The reason preg_last_error returned 0 is because the "invalid regex" is not:

PREG_INTERNAL_ERROR (an internal error)
PREG_BACKTRACK_LIMIT_ERROR (excessively forcing backtracking)
PREG_RECURSION_LIMIT_ERROR (excessively recursing)
PREG_BAD_UTF8_ERROR (badly formatted UTF-8)
PREG_BAD_UTF8_OFFSET_ERROR (offset to the middle of a UTF-8 character)

edited Jul 17 '14 at 06:35

kingjeffrey

14,316
6
41
47

answered May 28 '12 at 00:57

Niet the Dark Absol

311,322
76
447
566

Yes this is correct! Had this kind of figured out, just didn't know how to do it properly! However, using `track_errors` worked! – Hosh Sadiq May 28 '12 at 01:09
2

Note also, this only works for regexes with `/` delimiters. – Phil Tune Mar 04 '16 at 18:34

score 15 · Answer 2 · edited May 23 '17 at 12:02

15

Here is a good answer how to:

https://stackoverflow.com/a/12941133/2519073

if(@preg_match($yourPattern, null) === false){
    //pattern is broken
}else{
    //pattern is real
}

edited May 23 '17 at 12:02

Community

1
1

answered Apr 07 '15 at 14:39

ya_dimon

2,835
2
26
39

Better solutions the rest 2, as `track_errors` is deprecated with PHP 7.2 – Stanislav Stankov Feb 04 '21 at 21:39

score 10 · Accepted Answer · answered May 28 '12 at 00:59

10

The only easy way to test if a regex is valid in PHP is to use it and check if a warning is thrown.

ini_set('track_errors', 'on');
$php_errormsg = '';
@preg_match('/[blah/', '');
if($php_errormsg) echo 'regex is invalid';

However, using arbitrary user input as a regex is a bad idea. There were security holes (buffer overflow => remote code execution) in the PCRE engine before and it might be possible to create specially crafted long regexes which require lots of cpu/memory to compile/execute.

answered May 28 '12 at 00:59

ThiefMaster

298,938
77
579
623

It's not arbitrary user input, rather, rows taken from a database table that will only be added through direct code (for link generation). But yes, this worked :) Thanks! – Hosh Sadiq May 28 '12 at 01:08
2

Why is this better than just checking the output of preg_match? – Veda Aug 24 '15 at 09:55
`track_errors ` is deprecated as of PHP 7.2 – Stanislav Stankov Feb 04 '21 at 21:38

score 9 · Answer 4 · answered Apr 19 '13 at 05:52

9

Why not just use...another regex? Three lines, no @ kludges or anything:

// Test this string
$str = "/^[A-Za-z ]+$/";

// Compare it to a regex pattern that simulates any regex
$regex = "/^\/[\s\S]+\/$/";

// Will it blend?
echo (preg_match($regex, $str) ? "TRUE" : "FALSE");

Or, in function form, even more pretty:

public static function isRegex($str0) {
    $regex = "/^\/[\s\S]+\/$/";
    return preg_match($regex, $str0);
}

This doesn't test validity; but it looks like the question is Is there a good way of test if a string is a regex or normal string in PHP? and it does do that.

answered Apr 19 '13 at 05:52

Ben

51,262
47
169
217

1

I presume it's because you can't detect all regexes with a regex - the best this can do is decide whether it looks *kinda like* a regex or not. – Winfield Trail Jul 27 '13 at 04:56
For the regex-lovers, the answer to [this question](http://stackoverflow.com/questions/172303/is-there-a-regular-expression-to-detect-a-valid-regular-expression) might be interesting... – Mr Tsjolder Sep 16 '16 at 18:14
1

Does not work with flags? Sample '/(\n)}[\s\n\r]*$/s' – ajthinking May 13 '18 at 18:13
1

this answer does not support modifiers – alexwenzel Nov 15 '20 at 17:23

Test if a string is regex

4 Answers4

Linked

Related