Tag aware truncation of HTML by words

truncate_html.js

function truncate_html(html, words_left, depth) {
    if (typeof html == 'string') {
        html = html.replace(/</g, ' <').replace(/>/g, '> ').trim().split(/\s+/);
    }
    if (!depth) { depth = 0; }
    var opening_tag = /^<\s*\w+[^>]*>/;
    var closing_tag = /^<\/\s*\w+[^>]*>/;
    var void_tag = /^<\s*(area|base|br|col|command|embed|hr|img|input|keygen|link|meta|param|source|track|wbr)[^>]*>/;
    var word = html[0];
    if (!word) { return []; }
    var html = html.slice(1);
    var is_void_tag = word.match(void_tag);
    var is_opening_tag = !is_void_tag && word.match(opening_tag);
    var is_closing_tag = word.match(closing_tag);
    if (is_opening_tag) { depth++; }
    if (is_closing_tag) { depth--; }
    if (words_left) {
        if (!(is_opening_tag || is_void_tag || is_closing_tag)) {
            words_left--;
        }
        return [word].concat(truncate_html(html, words_left, depth));
    }
    if (is_closing_tag) {
        if (depth == 0) {
            return [word];
        } else {
            return [word].concat(truncate_html(html, words_left, depth));
        }
    }
    return truncate_html(html, words_left, depth);
}