Question

I’m replacing some text using a regex that I’ve found here.

$items = array(
  ':)'   => 'smile',
  ':('   => 'sad',
  '=))'  => 'laugh',
  ':p'   => 'tongue',      
); 

foreach($items as $key => $class)
  $regex[] = preg_quote($key, '#');

$regex = '#(?!<\w)('.implode('|', $regex).')(?!\w)#';

$string = preg_replace_callback($regex, function($matches) use($items){

  if(isset($items[$matches[0]])) 
    return '<span class="'.$items[$matches[0]].'">'.$matches[0].'</span>';

  return $matches[0];

}, $string);

It works but how can ignore matches within HTML tag definitions (like within tag attributes) ?

For example:

$string = 'Hello :) <a title="Hello :)"> Bye :( </a>';

=> The second :) should be ignored.

Answer 1

Here’s a DOMDocument-based implementation that does a by-the-book string replacement for your HTML:

$string = 'Hello :) <a title="Hello :)"> Bye :( </a>';

$items = array(
  ':)'   => 'smile',
  ':('   => 'sad',
  '=))'  => 'laugh',
  ':p'   => 'tongue',      
); 

foreach($items as $key => $class) $regex[] = preg_quote($key);

$regex = '#(?!<\w)('.implode('|', $regex).')(?!\w)#';

$doc = new DOMDocument();
$doc->loadHTML($string);

$xp = new DOMXPath($doc);

$text_nodes = $xp->query('//text()');

foreach ($text_nodes as $text_node)
{
  $parent  = $text_node->parentNode;
  $context = $text_node->nextSibling;
  $text    = $text_node->nodeValue;
  $matches = array();
  $offset  = 0;

  $parent->removeChild($text_node);

  while ( preg_match($regex, $text, $matches, PREG_OFFSET_CAPTURE, $offset) > 0 )
  {
    $match  = $matches[0];
    $smiley = $match[0];
    $pos    = $match[1];
    $prefix = substr($text, $offset, $pos - $offset);
    $offset = $pos + strlen($smiley);

    $span = $doc->createElement('span', $smiley);
    $span->setAttribute('class', $items[$smiley]);

    $parent->insertBefore( $doc->createTextNode($prefix), $context );
    $parent->insertBefore( $span, $context );
  }

  $suffix = substr($text, $offset);
  $parent->insertBefore( $doc->createTextNode($suffix), $context );
}

$body = $doc->getElementsByTagName('body');
$html = $doc->saveHTML( $body[0] );

Wrap it in a function and you’re good to go. It may be more lines of code than regex, but it’s not an ugly, bug-ridden maintenance nightmare (like any regex-based solution would be).