为什么这个正则表达式在PHP中并不贪心

Question

This regex should match lists just like in Markdown:

/((?:(?:(?:^[\+\*\-] )(?:[^
]+))(?:|
?))+)/m

It works in Javascript (with g flag added) but I have problems porting it to PHP. It does not behave greedy. Here's my example code:

$string = preg_replace_callback('`((?:(?:(?:^\* )(?:[^
]+))(?:|
?))+)`m', array(&$this, 'bullet_list'), $string);

function bullet_list($matches) { var_dump($matches) }

When I feed to it a list of three lines it displays this:

array(2) { [0]=> string(6) "* one " [1]=> string(6) "* one " } array(2) { [0]=> string(6) "* two " [1]=> string(6) "* two " } array(2) { [0]=> string(8) "* three " [1]=> string(8) "* three " } 

Apparently var_dump is being called three times instead of just once as I expect from it since the regex is greedy and must match as many lines as possible. I have tested it on regex101.com. How do I make it work properly?

Answer 1

This regex won't work correctly if you have newlines in your input text.

The part (?:| ?) matches either an or an , but not both. (regex101 treats newlines as only, so it works there).

Does the following work?

/(?:(?:(?:^[+*-] )(?:[^
]+))[
]*)+/m

(or, after removal of all the unnecessary non-capturing groups - thanks @M42!)

/(?:^[+*-] [^
]+[
]*)+/m