unprint 0.6.1 → 0.7.0

package/README.md

@@ -71,6 +71,14 @@ Return the number of elements that match the selector.
 
 Return the text contents of an element (`.textContent`).
 
+#### Query a number
+`query.number([selector], [options])`
+
+Options
+* `match`: The regular expression to use to extract a number from text, default `/\d+(\.\d*)?/` for decimal numbers.
+
+Return the contents of the element or attribute as a Number primitive.
+
 #### Query the HTML
 `query.content([selector], [options])`
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "unprint",
-  "version": "0.6.1",
+  "version": "0.7.0",
   "description": "Simplify common web scraping tasks while staying in control of the data.",
   "main": "src/app.js",
   "scripts": {

package/src/app.js

@@ -177,6 +177,48 @@ function queryAttributes(context, selector, attribute, customOptions) {
 	});
 }
 
+const defaultNumberRegexp = /\d+(\.\d*)?/;
+
+function matchNumberString(numberString, match) {
+	if (numberString && match) {
+		return Number(numberString.match(match)?.[0]);
+	}
+
+	if (numberString) {
+		return Number(numberString);
+	}
+
+	return null;
+}
+
+function queryNumber(context, selector, customOptions) {
+	const numberString = queryContent(context, selector, customOptions);
+
+	const options = {
+		match: defaultNumberRegexp,
+		...customOptions,
+	};
+
+	return matchNumberString(numberString, options.match);
+}
+
+function queryNumbers(context, selector, customOptions) {
+	const numberStrings = queryContents(context, selector, customOptions);
+
+	const options = {
+		match: defaultNumberRegexp,
+		...customOptions,
+	};
+
+	if (!numberStrings) {
+		return null;
+	}
+
+	return numberStrings
+		.map((numberString) => matchNumberString(numberString, options.match))
+		.filter(Boolean);
+}
+
 function queryHtml(context, selector, customOptions) {
 	const target = queryElement(context, selector, customOptions);
 
@@ -436,7 +478,7 @@ function queryDates(context, selector, format, customOptions) {
 }
 
 function extractDuration(durationString, match) {
-	const durationMatch = durationString.match(match || /(\d+:)?\d+:\d+/);
+	const durationMatch = durationString?.match(match || /(\d+:)?\d+:\d+/);
 
 	if (durationMatch) {
 		const segments = ['00'].concat(durationMatch[0].split(/[:hm]/)).slice(-3);
@@ -448,7 +490,7 @@ function extractDuration(durationString, match) {
 }
 
 function extractTimestamp(durationString) {
-	const timestampMatch = durationString.match(/(\d+H)?\s*(\d+M)?\s*\d+S?/i);
+	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;
@@ -500,6 +542,10 @@ const queryFns = {
 	imgs: queryImages,
 	json: queryJson,
 	jsons: queryJsons,
+	number: queryNumber,
+	num: queryNumber,
+	numbers: queryNumbers,
+	nums: queryNumbers,
 	date: queryDate,
 	dates: queryDates,
 	duration: queryDuration,
@@ -684,6 +730,7 @@ module.exports = {
 	initAll,
 	extractDate,
 	extractDuration,
+	extractTimestamp,
 	options: configure,
 	query: initQueryFns(queryFns),
 };

package/tests/index.html

@@ -14,16 +14,22 @@
 			<li class="item">Item 3</li>
 		</ul>
 
+		<ul id="numbers">
+			<li class="number">123</li>
+			<li class="number">234.56</li>
+			<li class="number">789.0</li>
+		</ul>
+
 		<a id="link" href="http://localhost:3101/html">Get HTML</a>
 		<a id="path" href="/json">Get data</a>
 		<a id="relativePath" href="./json">Get data</a>
 
 		<div id="date">Date: 22-07-2022 02:00</div>
 		<div id="date2">Date: 13-05-2022 18:00</div>
+		<div>Today: Nov 14, 2022</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">

package/tests/init.js

@@ -18,8 +18,11 @@ 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('date xpath', res.context.query.date('//div[contains(text(), "Today:")]', 'MMM DD, YYYY'));
 	console.log('duration', res.context.query.duration('#duration'));
 	console.log('timestamp', res.context.query.duration('#timestamp'));
+	console.log('number', res.context.query.number('.number'));
+	console.log('numbers', res.context.query.numbers('.number'));
 	console.log('data', res.context.query.json('#json'));
 	console.log('items', res.context.query.contents('.item'));
 	console.log('link', res.context.query.url('#link'));