Function parseHTMLToFragment

Pagetual/pagetual.user.js:4817–4880

(html, fragment, doc)

        });
    }
    function parseHTMLToFragment(html, fragment, doc) {
        const stack = [fragment];
        const tokenRe = /<!--[\s\S]*?-->|<!doctype[^>]*>|<\/?[a-zA-Z][^>]*>|[^<]+/gi;
        let match;
        while ((match = tokenRe.exec(html))) {
            const token = match[0];
            if (token[0] !== '<') {
                const text = decodeEntities(token);
                if (text) stack[stack.length - 1].appendChild(doc.createTextNode(text));
                continue;
            }
            if (token.indexOf('<!--') === 0) {
                continue;
            }
            if (/^<!doctype/i.test(token)) {
                continue;
            }
            if (token[1] === '/') {
                const tag = token.slice(2, -1).trim().toLowerCase();
                for (let i = stack.length - 1; i > 0; i--) {
                    const node = stack[i];
                    if (node.nodeType === 1 && node.nodeName.toLowerCase() === tag) {
                        stack.length = i;
                        break;
                    }
                }
                continue;
            }
            const tagMatch = /^<\s*([^\s/>]+)/.exec(token);
            if (!tagMatch) continue;
            const rawName = tagMatch[1];
            const tagName = rawName.toLowerCase();
            const parent = stack[stack.length - 1];
            const parentIsSvg = parent.nodeType === 1 && parent.namespaceURI === SVG_NS;
            const isSvg = parentIsSvg || tagName === 'svg';
            const el = isSvg ? doc.createElementNS(SVG_NS, rawName) : doc.createElement(tagName);
            const attrPart = token.replace(/^<\s*[^\s/>]+/, '').replace(/\/?>$/, '');
            if (attrPart) {
                const attrRe = /([^\s=]+)(?:\s*=\s*(?:"([^"]*)"|'([^']*)'|([^\s"'=<>`]+)))?/g;
                let attrMatch;
                while ((attrMatch = attrRe.exec(attrPart))) {
                    const name = attrMatch[1];
                    const value = decodeEntities(attrMatch[2] || attrMatch[3] || attrMatch[4] || '');
                    el.setAttribute(name, value);
                }
            }
            parent.appendChild(el);
            const selfClosing = token.endsWith('/>');
            if (!selfClosing && !VOID_TAGS[tagName]) {
                stack.push(el);
                if (RAW_TEXT_TAGS[tagName]) {
                    const closeRe = new RegExp('<\\/\\s*' + tagName + '\\s*>', 'ig');
                    closeRe.lastIndex = tokenRe.lastIndex;
                    const closeMatch = closeRe.exec(html);
                    if (closeMatch) {
                        const rawText = html.slice(tokenRe.lastIndex, closeMatch.index);
                        if (rawText) el.appendChild(doc.createTextNode(rawText));
                        tokenRe.lastIndex = closeMatch.index + closeMatch[0].length;