Updated Dorcel scraper, added movie support.

src/utils/qu.js

@@ -242,6 +242,42 @@ function urls(context, selector = 'a', attr = 'href', { origin, protocol = 'http
 	return attr ? urlEls.map(urlEl => prefixUrl(urlEl, origin, protocol)) : urlEls;
 }
 
+function sourceSet(context, selector, attr, options = {}) {
+	const srcset = q(context, selector, attr);
+
+	if (!srcset) {
+		return null;
+	}
+
+	const sources = srcset
+		.split(/\s*,\s*/)
+		.map((source) => {
+			const [link, descriptor] = source.split(' ');
+
+			return {
+				descriptor: descriptor || 'fallback',
+				url: prefixUrl(link, options.origin, options.protocol),
+			};
+		})
+		.sort((sourceA, sourceB) => {
+			if (sourceB.descriptor === 'fallback' || parseInt(sourceA.descriptor, 10) > parseInt(sourceB.descriptor, 10)) {
+				return -1;
+			}
+
+			if (parseInt(sourceA.descriptor, 10) < parseInt(sourceB.descriptor, 10)) {
+				return 1;
+			}
+
+			return 0;
+		});
+
+	if (options.includeDescriptor) {
+		return sources;
+	}
+
+	return sources.map(source => source.url);
+}
+
 function poster(context, selector = 'video', attr = 'poster', { origin, protocol = 'https' } = {}) {
 	const posterEl = q(context, selector, attr);
 
@@ -267,17 +303,17 @@ function duration(context, selector, match, attr = 'textContent') {
 	const durationMatch = durationString.match(match || /(\d+:)?\d+:\d+/);
 
 	if (durationMatch) {
-		const segments = ['00'].concat(durationMatch[0].split(':')).slice(-3);
+		const segments = ['00'].concat(durationMatch[0].split(/[:hm]/)).slice(-3);
 
 		return moment.duration(segments.join(':')).asSeconds();
 	}
 
-	const timestampMatch = durationString.match(/T(\d+H)?(\d+M)?\d+S/);
+	const timestampMatch = durationString.match(/(\d+H)?\s*(\d+M)?\s*\d+S?/i);
 
 	if (timestampMatch) {
-		const hours = timestampMatch[0].match(/(\d+)H/)?.[1] || 0;
-		const minutes = timestampMatch[0].match(/(\d+)M/)?.[1] || 0;
-		const seconds = timestampMatch[0].match(/(\d+)S/)?.[1] || 0;
+		const hours = timestampMatch[0].match(/(\d+)H/i)?.[1] || 0;
+		const minutes = timestampMatch[0].match(/(\d+)M/i)?.[1] || 0;
+		const seconds = timestampMatch[0].match(/(\d+)(S|$)/i)?.[1] || 0;
 
 		return (Number(hours) * 3600) + (Number(minutes) * 60) + Number(seconds);
 	}
@@ -345,6 +381,10 @@ const quFuncs = {
 	num: number,
 	poster,
 	q,
+	sourceSet,
+	sources: sourceSet,
+	srcs: sourceSet,
+	srcset: sourceSet,
 	style,
 	styles,
 	text,
@@ -415,10 +455,12 @@ function extractAll(htmlValue, selector) {
 	return initAll(window.document, selector, window);
 }
 
-async function get(urlValue, selector, headers, options, queryAll = false) {
-	const res = await http.get(urlValue, headers, options);
+async function request(method = 'get', urlValue, body, selector, headers, options, queryAll = false) {
+	const res = await (method === 'post'
+		? http.post(urlValue, body, headers, options)
+		: http[method](urlValue, headers, options));
 
-	if (res.statusCode === 200) {
+	if (res.ok) {
 		const item = queryAll
 			? extractAll(res.body.toString(), selector)
 			: extract(res.body.toString(), selector);
@@ -443,8 +485,20 @@ async function get(urlValue, selector, headers, options, queryAll = false) {
 	};
 }
 
+async function get(urlValue, selector, headers, options) {
+	return request('get', urlValue, null, selector, headers, options, false);
+}
+
+async function post(urlValue, body, selector, headers, options) {
+	return request('post', urlValue, body, selector, headers, options, false);
+}
+
 async function getAll(urlValue, selector, headers, options) {
-	return get(urlValue, selector, headers, options, true);
+	return request('get,', urlValue, selector, headers, options, true);
+}
+
+async function postAll(urlValue, body, selector, headers, options) {
+	return request('post', urlValue, body, selector, headers, options, true);
 }
 
 module.exports = {
@@ -470,6 +524,8 @@ module.exports = {
 	geta: getAll,
 	qu: quFuncs,
 	query: quFuncs,
+	post,
+	postAll,
 	prefixUrl,
 	...legacyFuncs,
 };