These code find first occurance $word in $text, and replace it by something:
<?php
$text = preg_replace("/\b($word)\b/i", 'something', $text, 1);
?>
But i want ignore if this word surrounded by "a" tag, for example, searching should find only second "word" here:
<a href="something">text text word text</a>. text2 text2 word text2...
I think to do this with just a regular expression is possible, but cumbersome. So here's a programmatical way, that is, however, dirty.
I would first replace every occurance of word by an auxiliary string that doesn't occur in the original string (such as e.g. @jska_x). Then I would do a regular expression replacement for @jska_x inside an a-tag in order to restore the words you do not want to replace.
After all, I would replace @jska_x by target_word.
@\b(word\d+)\b(?![^<>]*</|[^><]*>)@i
<a href="something">text text word1 text</a>. text2 \ (cont. on next line)
<a asdasd> text2 word2 text2... fwefw fwe few fw <a>word3</a> \
<a href="/word5.html">asdada</a>
// don't mind the numbers after word. Used them for detection which word matches
Something like this could do the trick, but I advice you not to go with regular expressions on this task. May be you could use DOM and check if word is not in allowed tags, then replace it.
Use a DOM Parser to find all text nodes that contain the needle and which do not have a a parent element with a name of "a":
$html = <<< HTML
<p>
. text2 text2 word text2...
<a href="something">text text word <span> word </span> text</a>
. text2 text2 word text2...
<p>
HTML;
Code:
$dom = new DOMDocument;
$dom->loadHTML($html);
$xp = new DOMXPath($dom);
$nodes = $xp->query('//*[name() != "a"]/text()[contains(.,"word")]');
foreach($nodes as $node) {
// can use a Regex in here too if you are after word boundaries
$node->nodeValue = str_replace('word', 'something', $node->nodeValue);
}
echo $dom->saveXML($dom->documentElement);
Outputs:
<html><body><p>
. text2 text2 something text2...
<a href="something">text text word <span> something </span> text</a>
. text2 text2 something text2...
</p><p/></body></html>
Note how this will also replace word inside the span inside the a. If you want to exclude those too, you have to adjust the XPath to:
'//text()[not(ancestor::a) and contains(., "word")]'
to find all text nodes containing the needle that are not nested anywhere inside an a element.
There is a number of third party parsers worth mentioning that aim to enhance DOM: phpQuery, Zend_Dom, QueryPath and FluentDom.