preg match text between tags excluding same tag in between

Question

Well I know there several questions similar but could not find any with this specific case.

I took one code and tweak it to my needs but now I'm founding a bug on it that I can't correct.

Code:

$tag = 'namespace';
$match = Tags::get($f, $tag);
var_dump($match); 

  static function get( $xml, $tag) { // http://stackoverflow.com/questions/3404433/get-content-within-a-html-tag-using-7-processing
// bug case      string(56) "<namespaces>
//      <namespace key="-2">Media</namespace>"
      $tag_ini = "<{$tag}[^\>]*?>"; $tag_end = "<\\/{$tag}>";
      $tag_regex = '/' . $tag_ini . '(.*?)' . $tag_end . '/si';

      preg_match_all($tag_regex,
      $xml,
      $matches,
      PREG_OFFSET_CAPTURE);
      return $matches;
  }

As you can see, there is a bug if the tag is nested:

<namespaces> <namespace key="-2">Media</namespace>

When it should return 'Media', or even the outer '<namespaces>' and then the inside ones.

I tried to add " <{$tag}[^\\>|^\\r\\n ]*?> ", ^\\s+ , changing the * to *?, and other few things that in best case turned to recognize only the bugged case.

Also tried "<{$tag}[^{$tag}]*?>" which gives blank, I suppose it nullifies itself.

I'm a newb on regex, I can tell that to fix this just is needed to add don't let open a new tag of the same type. Or I could even use a hack answer for my use case, that excludes if the inside text has new line carriage.

Can anyone get the right syntax for this?

---

You can check an extract of the text here: http://pastebin.com/f2naN2S3

---

After the proposed change: $tag_ini = "<{$tag}\\\\b[^>]*>"; $tag_end = "<\\\\/{$tag}>"; $tag_ini = "<{$tag}\\\\b[^>]*>"; $tag_end = "<\\\\/{$tag}>"; it does work for the the example case, but not for this one:

<namespace key="0" />
      <namespace key="1">Talk</namespace>

As it results in:

<namespace key="1">Talk"

It's because numbers and " and letters are considered inside word boundary. How could I address that?

Answer 1

This is probably not the idea answer, but I was messing with a regex generator:

<?php
# URL that generated this code:
# http://txt2re.com/index-php.php3?s=%3Cnamespace%3E%3Cnamespace%20key=%22-2%22%3EMedia%3C/namespace%3E&12&11

$txt='arstarstarstarstarstarst<namespace key="-2">Media</namespace>arstarstarstarstarst';

$re1='.*?'; # Non-greedy match on filler
$re2='(?:[a-z][a-z]+)'; # Uninteresting: word
$re3='.*?'; # Non-greedy match on filler
$re4='(?:[a-z][a-z]+)'; # Uninteresting: word
$re5='.*?'; # Non-greedy match on filler
$re6='(?:[a-z][a-z]+)'; # Uninteresting: word
$re7='.*?'; # Non-greedy match on filler
$re8='((?:[a-z][a-z]+))';   # Word 1

if ($c=preg_match_all ("/".$re1.$re2.$re3.$re4.$re5.$re6.$re7.$re8."/is", $txt, $matches))
{
    $word1=$matches[1][0];
    print "($word1) \n";
}

#-----
# Paste the code into a new php file. Then in Unix:
# $ php x.php
#-----
?>

Answer 2

The main problem is that you did not use a word boundary after the opening tag and thus, namespace in the pattern could also match namespaces tag, and many others.

The subsequent issue is that the <${tag}\\b[^>]*>(.*?)<\\/${tag}> pattern would overfire if there is a self-closing namespace tag followed with a "normal" paired open/close namespace tag. So, you need to either use a negative lookbehind (?<!\\/) before the > (see demo ), or use a (?![^>]*\\/>) negative lookahead after \\b (see demo ).

So, you can use

$tag_ini = "<{$tag}\\b[^>]*(?<!\\/)>"; $tag_end = "<\\/{$tag}>";

Answer 3

This line is what I needed

   $tag_ini = "<{$tag}\\b[^>|^\\/>]*>"; $tag_end = "<\\/{$tag}>";

Thank you very much you @Alison and @Wictor for your help and directions