PHP Regexp捕获重复的字符组，例如hahaha jajajaja hihihi

Question

As title, is there a way in PHP, with preg_match_all to catch all the repetitions of chars group? For instante catch

1. hahahaha
2. jajajaj
3. hihihi

It's fine to catch repetition of any char, like abababab, acacacacac. Also, is there a way to count the number of repetition?

The idea is to catch all this "forms" of smiling on social media. I figured out that there are also other cases, such as misspelled instances like ahahhahaah (where you have two consecutive a or h). Any ideas?

Answer 1

How about this:

preg_match_all('/((?i)[a-z])((?i)[a-z])(\1\2)+/', $str, $m);
$matches = $m[0]; //$matches will contain an array of matches

A bit complicated, but it does work. To explain, the first subpattern ( ((?i)[az]) ) matches any character between a and z, no matter the case. The second subpattern ( ((?i)[az]) ) does the same thing. The third subpattern ( (\\1\\2)+ ) matches one or more repetitions of the first two letters, in the same case as they were originally put. This regular expression also assumes that there's an even number of repetitions. If you don't want that, you can add \\1? at the end, meaning that (as long as it contains one or more repetitions), it can end with the first character (for instance, hahah and ikikikik would both be valid, but not asa ).

To retrieve the number of repetitions for a specific match, you can do:

$numb = strlen($matches[$index])/2 - 1; //-1 because the first two letters aren't repetitions

Answer 2

For the shortest repetition (eg ha gets repeated multiple times in hahahaha ):

(.+?)\1+

See demo .

For the longest repetition (eg haha gets repeated in hahahaha ):

(.+)\1+

Counting Repetitions

The non-regex solution is to compare the lengths of Group 1 (the repteated token) and the overall match.

With pure regex, in .NET, you could simply do (.+?)(\\1)+ and look at the number of captures in the Group 1 CaptureCollection object.

In PHP, that's not possible, but there are some hacks. See, for instance, this question about matching a line number —it's the same technique. This is for "study purposes" only—you wouldn't want to use that in real life.