unprint 0.5.0 → 0.6.1

package/README.md

@@ -122,6 +122,14 @@ Options
 
 Returns a Date object.
 
+#### Query a duration
+`query.duration(selector, format, [options])` or `query.dur`
+
+Options
+* `match` (RegExp): The text to extract before attempting to parse it as a duration. The default expression will attempt to extract `(hh:)mm:ss` and `PT##H##M##S`.
+
+Returns the duration in seconds as a number.
+
 #### Query JSON
 `query.json([selector], [options])`
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "unprint",
-  "version": "0.5.0",
+  "version": "0.6.1",
   "description": "Simplify common web scraping tasks while staying in control of the data.",
   "main": "src/app.js",
   "scripts": {

package/src/app.js

@@ -435,6 +435,49 @@ function queryDates(context, selector, format, customOptions) {
 	}));
 }
 
+function extractDuration(durationString, match) {
+	const durationMatch = durationString.match(match || /(\d+:)?\d+:\d+/);
+
+	if (durationMatch) {
+		const segments = ['00'].concat(durationMatch[0].split(/[:hm]/)).slice(-3);
+
+		return moment.duration(segments.join(':')).asSeconds();
+	}
+
+	return null;
+}
+
+function extractTimestamp(durationString) {
+	const timestampMatch = durationString.match(/(\d+H)?\s*(\d+M)?\s*\d+S?/i);
+
+	if (timestampMatch) {
+		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);
+	}
+
+	return null;
+}
+
+function queryDuration(context, selector, customOptions) {
+	const options = { ...customOptions };
+	const durationString = queryContent(context, selector, customOptions);
+
+	if (!durationString) {
+		return null;
+	}
+
+	if (options.match) {
+		return extractDuration(durationString, options.match);
+	}
+
+	return extractDuration(durationString)
+		|| extractTimestamp(durationString)
+		|| null;
+}
+
 const queryFns = {
 	element: queryElement,
 	elements: queryElements,
@@ -459,6 +502,8 @@ const queryFns = {
 	jsons: queryJsons,
 	date: queryDate,
 	dates: queryDates,
+	duration: queryDuration,
+	dur: queryDuration,
 	sourceSet: querySourceSet,
 	srcSet: querySourceSet,
 	url: queryUrl,
@@ -638,6 +683,7 @@ module.exports = {
 	init,
 	initAll,
 	extractDate,
+	extractDuration,
 	options: configure,
 	query: initQueryFns(queryFns),
 };

package/tests/index.html

@@ -21,6 +21,10 @@
 		<div id="date">Date: 22-07-2022 02:00</div>
 		<div id="date2">Date: 13-05-2022 18:00</div>
 
+		<div id="duration">01:15:33</div>
+		<div id="timestamp">PT1H34M18S</div>
+		<div id="timestamp">PT34M18S</div>
+
 		<img class="image" src="https://i.redd.it/vn9h981hlx281.png">
 		<img class="image" src="https://i.redd.it/1s22dsrqy0181.jpg">
 		<img class="image" src="https://i.redd.it/e91oo4ueyeb71.jpg">

package/tests/init.js

@@ -18,6 +18,8 @@ async function initTest() {
 
 	console.log('title', res.context.query.content('//*[contains(text(), "Test")]'));
 	console.log('date', res.context.query.date('#date', 'DD-MM-YYYY HH:mm'));
+	console.log('duration', res.context.query.duration('#duration'));
+	console.log('timestamp', res.context.query.duration('#timestamp'));
 	console.log('data', res.context.query.json('#json'));
 	console.log('items', res.context.query.contents('.item'));
 	console.log('link', res.context.query.url('#link'));