Highlight keywords in a paragraph

Question

I need to highlight a keyword in a paragraph, as google does in its search results. Let\'s assume that I have a MySQL db with blog posts. When a user searches for a certain

Answer 1

If it contains html (note that this is a pretty robust solution):

$string = '
foobar
';
$keyword = 'foo';
$dom = new DomDocument();
$dom->loadHtml($string);
$xpath = new DomXpath($dom);
$elements = $xpath->query('//*[contains(.,"'.$keyword.'")]');
foreach ($elements as $element) {
    foreach ($element->childNodes as $child) {
        if (!$child instanceof DomText) continue;
        $fragment = $dom->createDocumentFragment();
        $text = $child->textContent;
        $stubs = array();
        while (($pos = stripos($text, $keyword)) !== false) {
            $fragment->appendChild(new DomText(substr($text, 0, $pos)));
            $word = substr($text, $pos, strlen($keyword));
            $highlight = $dom->createElement('span');
            $highlight->appendChild(new DomText($word));
            $highlight->setAttribute('class', 'highlight');
            $fragment->appendChild($highlight);
            $text = substr($text, $pos + strlen($keyword));
        }
        if (!empty($text)) $fragment->appendChild(new DomText($text));
        $element->replaceChild($fragment, $child);
    }
}
$string = $dom->saveXml($dom->getElementsByTagName('body')->item(0)->firstChild);

Results in:

foobar

And with:

$string = '
foobarbazbar
';
$keyword = 'bar';

You get (broken onto multiple lines for readability):

foo
    bar
    baz
    
        bar
    

Beware of non-dom solutions (like regex or str_replace) since highlighting something like "div" has a tendency of completely destroying your HTML... This will only ever "highlight" strings in the body, never inside of a tag...

---

Edit Since you want Google style results, here's one way of doing it:

function getKeywordStubs($string, array $keywords, $maxStubSize = 10) {
    $dom = new DomDocument();
    $dom->loadHtml($string);
    $xpath = new DomXpath($dom);
    $results = array();
    $maxStubHalf = ceil($maxStubSize / 2);
    foreach ($keywords as $keyword) {
        $elements = $xpath->query('//*[contains(.,"'.$keyword.'")]');
        $replace = ''.$keyword.'';
        foreach ($elements as $element) {
            $stub = $element->textContent;
            $regex = '#^.*?((\w*\W*){'.
                 $maxStubHalf.'})('.
                 preg_quote($keyword, '#').
                 ')((\w*\W*){'.
                 $maxStubHalf.'}).*?$#ims';
            preg_match($regex, $stub, $match);
            var_dump($regex, $match);
            $stub = preg_replace($regex, '\\1\\3\\4', $stub);
            $stub = str_ireplace($keyword, $replace, $stub);
            $results[] = $stub;
        }
    }
    $results = array_unique($results);
    return $results;
}

Ok, so what that does is return an array of matches with $maxStubSize words around it (namely up to half that number before, and half after)...

So, given a string:

a whole 
    bunch of text 
    here for 
    us to foo bar baz replace out from this string
    bar

Calling getKeywordStubs($string, array('bar', 'bunch')) will result in:

array(4) {
  [0]=>
  string(75) "here for us to foo bar baz replace out from "
  [3]=>
  string(34) "bar"
  [4]=>
  string(62) "a whole bunch of text here for "
  [7]=>
  string(39) "bunch of"
}

So, then you could build your result blurb by sorting the list by strlen and then picking the two longest matches... (assuming php 5.3+):

usort($results, function($str1, $str2) { 
    return strlen($str2) - strlen($str1);
});
$description = implode('...', array_slice($results, 0, 2));

Which results in:

here for us to foo bar baz replace out...a whole bunch of text here for 

I hope that helps... (I do feel this is a bit... bloated... I'm sure there are better ways to do this, but here's one way)...