Inserting multiple links into text, ignoring matches that happen to be inserted

Question

The site I'm working on has a database table filled with glossary terms. I am building a function that will take some HTML and replace the first instances of the glossary terms with tooltip links.

I am running into a problem though. Since it's not just one replace, the function is replacing text that has been inserted in previous iterations, so the HTML is getting mucked up.

I guess the bottom line is, I need to ignore text if it:

• Appears within the < and > of any HTML tag, or
• Appears within the text of an <a></a> tag.

Here's what I have so far. I was hoping someone out there would have a clever solution.

function insertGlossaryLinks($html)
{
    // Get glossary terms from database, once per request
    static $terms;
    if (is_null($terms)) {
        $query = Doctrine_Query::create()
            ->select('gt.title, gt.alternate_spellings, gt.description')
            ->from('GlossaryTerm gt');
        $glossaryTerms = $query->rows();

        // Create whole list in $terms, including alternate spellings
        $terms = array();
        foreach ($glossaryTerms as $glossaryTerm) {

            // Initialize with title
            $term = array(
                'wordsHtml' => array(
                    h(trim($glossaryTerm['title']))
                    ),
                'descriptionHtml' => h($glossaryTerm['description'])
                );

            // Add alternate spellings
            foreach (explode(',', $glossaryTerm['alternate_spellings']) as $alternateSpelling) {
                $alternateSpelling = h(trim($alternateSpelling));
                if (empty($alternateSpelling)) {
                    continue;
                }
                $term['wordsHtml'][] = $alternateSpelling;
            }

            $terms[] = $term;
        }
    }

    // Do replacements on this HTML
    $newHtml = $html;
    foreach ($terms as $term) {
        $callback = create_function('$m', 'return \'<a href="javascript:void(0);" class="glossary-term" title="'.$term['descriptionHtml'].'"><span>\'.$m[0].\'</span></a>\';');
        $term['wordsHtmlPreg'] = array_map('preg_quote', $term['wordsHtml']);
        $pattern = '/\b('.implode('|', $term['wordsHtmlPreg']).')\b/i';
        $newHtml = preg_replace_callback($pattern, $callback, $newHtml, 1);
    }

    return $newHtml;
}

Answer 1

Using Regexes to process HTML is always risky business. You will spend a long time fiddling with the greediness and laziness of your Regexes to only capture text that is not in a tag, and not in a tag name itself. My recommendation would be to ditch the method you are currently using and parse your HTML with an HTML parser, like this one: http://simplehtmldom.sourceforge.net/. I have used it before and have recommended it to others. It is a much simpler way of dealing with complex HTML.

Answer 2

I ended up using preg_replace_callback to replace all existing links with placeholders. Then I inserted the new glossary term links. Then I put back the links that I had replaced.

It's working great!