php preg_match_all backreference

Question

for the following input string , pattern and :

$str1 = 'span class="outline">Iron Man butts heads with Nick Fury and Shield after HYDRA attacks a meeting of the United Nations.</span>
<span class="credit">
    Dir: <a href="/name/nm0381817/">Vinton Heuck</a>, <a href="/name/nm1367649/">Ciro Nieli</a>, <a href="/name/nm1367649/">Aditya Parikh</a>'

$pattern='/class=&quot;credit&quot;&gt;[\s]+?Dir:([,\s]+?&lt;a[\s]+?href=&quot;\/name\/nm\d{7}\/&quot;&gt;([\/\(\)-:@!%*#=_|?$&;.\w\s]+?)&lt;\/a&gt;)+/um';

preg_match_all($pattern,$str1,$dir);

Output is as follows for print_r:

Array ( [0] => Array ( [0] => class="credit"> Dir: <a href="/name/nm0381817/">Vinton Heuck</a>, <a href="/name/nm1367649/">Ciro Nieli</a>, <a href="/name/nm1367649/">Aditya Parikh</a> ) [1] => Array ( [0] => , <a href="/name/nm1367649/">Aditya Parikh</a> ) [2] => Array ( [0] => Aditya Parikh ) )

As you can see the Array[2] gives Aditya Parikh, I was hoping to receieve Vinton Heuck and Ciro Nieli also. But didn't.

Any solution ??

There are several other questions here pertaining to using regex on HTML. Read some of those and see why this is not recommended. — Asad Saeeduddin, Oct 22 '12 at 08:17

score 1 · Answer 1 · answered Oct 22 '12 at 08:43

The logic behind the matching array returned by preg_match_all is not that obvious.

First of all, don't use regex to parse html. With that said:

The result you're getting is on the form of $array[paren_num][match_num].

A basic example: abc ran against the regex /(.)/ would return the following matches array:

Array
(
    [0] => Array
        (
            [0] => a
            [1] => b
            [2] => c
        )

    [1] => Array
        (
            [0] => a
            [1] => b
            [2] => c
        )

)

the index 0 contains all the consumed data. Index 1 means it's the first backreference (we only have 1 parenthesis). The 0-2 index inside of that corresponds to each match. The expression was in other words ran 3 times until it finished.

I hope this helps.

score 1 · Answer 2 · edited May 23 '17 at 12:12

You should really consider using a DOM parser. For example, this one. Regular expressions just cannot properly parse HTML.

However, here is why your approach does not work as expected:

You are using the same capturing group for all 3 names. But a capturing group has only one number, so you will always just get the last thing that was captured (the right-most name). But even if you did just match one name (arbitrarily far into the span tag) you would get a different problem:

Matches cannot overlap. Since all three matches you want would contain at least class="credit"> Dir: and some more common text, you cannot get all of them. You could solve this with a lookbehind assertion (because it is not part of match), but unfortunately PHP does not allow variable-length lookbehinds (which would be required). There are workarounds to solve this, but at the end of the day, you are best of using a DOM parser.

Here is just a basic example using the parser I linked above:

require "simple_html_dom.php";

$html = str_get_html($str1);

$names = array();
foreach($html->find('span[class=credit] a') as $link)
    $names[] = $link->innertext;

print_r($names);

Resulting in:

Array
(
    [0] => Vinton Heuck
    [1] => Ciro Nieli
    [2] => Aditya Parikh
)

php preg_match_all backreference

2 Answers2