const urlUtils = require('lib/urlUtils.js'); const Entities = require('html-entities').AllHtmlEntities; const htmlentities = new Entities().encode; // [\s\S] instead of . for multiline matching // https://stackoverflow.com/a/16119722/561309 const imageRegex = /<img([\s\S]*?)src=["']([\s\S]*?)["']([\s\S]*?)>/gi; const anchorRegex = /<a([\s\S]*?)href=["']([\s\S]*?)["']([\s\S]*?)>/gi; class HtmlUtils { headAndBodyHtml(doc) { const output = []; if (doc.head) output.push(doc.head.innerHTML); if (doc.body) output.push(doc.body.innerHTML); return output.join('\n'); } extractImageUrls(html) { if (!html) return []; const output = []; let matches; while ((matches = imageRegex.exec(html))) { output.push(matches[2]); } return output; } replaceImageUrls(html, callback) { return this.processImageTags(html, data => { const newSrc = callback(data.src); return { type: 'replaceSource', src: newSrc, }; }); } processImageTags(html, callback) { if (!html) return ''; return html.replace(imageRegex, (v, before, src, after) => { const action = callback({ src: src }); console.info('src', src); if (!action) return `<img${before}src="${src}"${after}>`; if (action.type === 'replaceElement') { return action.html; } if (action.type === 'replaceSource') { return `<img${before}src="${action.src}"${after}>`; } if (action.type === 'setAttributes') { const attrHtml = this.attributesHtml(action.attrs); return `<img${before}${attrHtml}${after}>`; } throw new Error(`Invalid action: ${action.type}`); }); } prependBaseUrl(html, baseUrl) { if (!html) return ''; return html.replace(anchorRegex, (v, before, href, after) => { const newHref = urlUtils.prependBaseUrl(href, baseUrl); return `<a${before}href="${newHref}"${after}>`; }); } attributesHtml(attr) { const output = []; for (const n in attr) { if (!attr.hasOwnProperty(n)) continue; output.push(`${n}="${htmlentities(attr[n])}"`); } return output.join(' '); } } const htmlUtils = new HtmlUtils(); module.exports = htmlUtils;