unprint 0.18.6 → 0.18.8

package/README.md

@@ -168,6 +168,22 @@ Options
 
 Returns a Date object.
 
+#### Query a relative date
+`query.dateAgo(selector, [options])`
+
+Parses relative timestamps such as '8 weeks ago' and '1 year ago'.
+
+Options
+* `match` (RegExp): The text to extract before attempting to parse it as a period. Expects two capture groups, one for the duration and one for the unit.
+
+Returns
+```
+{
+	date: '2026-01-01', // Date object
+	precision: 'day', // year, month, week, day
+}
+```
+
 #### Query a duration
 `query.duration(selector, format, [options])` or `query.dur`
 
@@ -217,7 +233,7 @@ Options
 * `apiUserAgent`: The default user agent header for raw requests (`get` interface `request`)
 
 Use Playwright with Chromium (experimental)
-* `unprint.browserRequest(url, [options])` or `unprint.browser(url, [options])`
+* `unprint.browser(url, [options])`
 * `unprint.closeAllBrowsers()`: Close reused browser instances.
 
 Additional options

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "unprint",
-  "version": "0.18.6",
+  "version": "0.18.8",
   "description": "Simplify common web scraping tasks while staying in control of the data.",
   "main": "src/app.js",
   "scripts": {},

package/src/app.js

@@ -775,6 +775,50 @@ function queryDates(context, selector, format, customOptions) {
 	}));
 }
 
+const periodRegex = /years?|months?|weeks?|days?/;
+
+function extractDateAgo(dateString, customOptions) {
+	if (!dateString) {
+		return null;
+	}
+
+	const options = {
+		match: /(\d+)\s*(\w+)/,
+		...customOptions,
+	};
+
+	const timeMatch = dateString.match(options.match);
+
+	if (timeMatch) {
+		const [n, period] = timeMatch.slice(1);
+
+		if (periodRegex.test(period)) {
+			const thenDate = moment.utc().subtract(Number(n), period);
+
+			return {
+				date: thenDate.toDate(),
+				precision: period.replace(/s$/, ''),
+			};
+		}
+	}
+
+	return null;
+}
+
+function queryDateAgo(context, selector, customOptions) {
+	const dateString = queryContent(context, selector, customOptions);
+
+	return extractDateAgo(dateString, customOptions);
+}
+
+function queryDatesAgo(context, selector, customOptions) {
+	const dateStrings = queryContents(context, selector, customOptions);
+
+	return dateStrings
+		.map((dateString) => extractDateAgo(dateString, customOptions))
+		.filter(Boolean);
+}
+
 function formatDate(dateValue, format, inputFormat) {
 	if (inputFormat) {
 		return moment(dateValue, inputFormat).format(format);
@@ -866,6 +910,9 @@ const queryFns = {
 	posters: queryPosters,
 	date: queryDate,
 	dates: queryDates,
+	dateAgo: queryDateAgo,
+	datesAgo: queryDatesAgo,
+	dateAgos: queryDatesAgo,
 	duration: queryDuration,
 	dur: queryDuration,
 	sourceSet: querySourceSet,
@@ -1434,6 +1481,7 @@ async function request(url, body, customOptions = {}, method = 'GET', redirects
 		method,
 		body: curatedBody.body,
 		headers,
+		redirect: options.followRedirects ? 'follow' : 'manual',
 		signal: options.abortSignal,
 	})).catch((error) => ({ // tends to happen when proxy can't reach host
 		status: 500,
@@ -1519,6 +1567,7 @@ module.exports = {
 	init,
 	initAll,
 	extractDate,
+	extractDateAgo,
 	extractDuration,
 	extractNumber,
 	extractTimestamp,

package/tests/index.html

@@ -31,6 +31,11 @@
 		<div id="date2">Date: 13-05-2022 18:00</div>
 		<div>Today: Nov 14, 2022</div>
 
+		<div class="date-ago">3 weeks ago</div>
+		<div class="date-ago">2 days ago</div>
+		<div class="date-ago">1 year ago</div>
+		<div class="date-ago">November 10th</div>
+
 		<div id="duration">01:15:33</div>
 		<div id="timestamp">PT1H34M18S</div>
 		<div id="timestring">1 hour 40 minutes 5 seconds</div>

package/tests/init.js

@@ -60,6 +60,8 @@ async function initTest() {
 	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('date grouped xpath', res.context.query.date('(//div[contains(text(), "Today:")])', 'MMM DD, YYYY'));
+	console.log('date ago', res.context.query.dateAgo('.date-ago'));
+	console.log('dates ago', res.context.query.dateAgos('.date-ago'));
 	console.log('duration', res.context.query.duration('#duration'));
 	console.log('timestamp', res.context.query.duration('#timestamp'));
 	console.log('timestring', res.context.query.duration('#timestring'));