unprint 0.9.0 → 0.9.2

package/README.md

@@ -41,7 +41,7 @@ query.element('h1#title'); // HTMLHeadingElement
 **From here on, the query methods will be described in their initialized form.** The API for the *uninitialized* methods is identical, except for the element passed as the first argument
 
 #### Selector
-The selector can be a CSS selector, an XPath selector starting with `//`, or an array of either or both acting as fallbacks. If the selector is falsy, the input element will be used.
+The selector can be a CSS selector, an XPath selector starting with `/`, or an array of either or both acting as fallbacks. If the selector is falsy, the input element will be used.
 
 #### Querying multiple elements
 Most methods can be used in plural, returning an array of results, i.e. `query.elements()`, `query.dates()`.
@@ -75,7 +75,8 @@ Return the text contents of an element (`.textContent`).
 `query.number([selector], [options])`
 
 Options
-* `match`: The regular expression to use to extract a number from text, default `/\d+(\.\d*)?/` for decimal numbers.
+* `match`: The regular expression to use to extract a number from text, default `/\d+(\.\d+)?/` for decimal numbers.
+* `matchIndex`: The index of the match result, useful for expressions containing groups or a global flag, default `0`.
 
 Return the contents of the element or attribute as a Number primitive.
 
@@ -131,6 +132,7 @@ Arguments
 
 Options
 * `match` (RegExp): The text to extract before attempting to parse it as a date. The default expression will attempt to extract any of 01-01-1970, 1970-01-01, 01/01/1970 or January 1, 1970 with optional 00:00[:00] time.
+* `matchIndex`: The index of the match result, useful for expressions containing groups or a global flag, default `0`.
 * `timezone` (string): The name of the input timezone, defaults to 'UTC'.
 
 Returns a Date object.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "unprint",
-  "version": "0.9.0",
+  "version": "0.9.2",
   "description": "Simplify common web scraping tasks while staying in control of the data.",
   "main": "src/app.js",
   "scripts": {

package/src/app.js

@@ -189,11 +189,11 @@ function queryDatasets(context, selector, dataAttribute, customOptions) {
 	return targets.map((target) => target.dataset[dataAttribute]);
 }
 
-const defaultNumberRegexp = /\d+(\.\d*)?/;
+const defaultNumberRegexp = /\d+(\.\d+)?/;
 
-function matchNumberString(numberString, match) {
-	if (numberString && match) {
-		return Number(numberString.match(match)?.[0]);
+function matchNumberString(numberString, options) {
+	if (numberString && options.match) {
+		return Number(numberString.match(options.match)?.[options.matchIndex]);
 	}
 
 	if (numberString) {
@@ -208,10 +208,11 @@ function queryNumber(context, selector, customOptions) {
 
 	const options = {
 		match: defaultNumberRegexp,
+		matchIndex: 0,
 		...customOptions,
 	};
 
-	return matchNumberString(numberString, options.match);
+	return matchNumberString(numberString, options);
 }
 
 function queryNumbers(context, selector, customOptions) {
@@ -219,6 +220,7 @@ function queryNumbers(context, selector, customOptions) {
 
 	const options = {
 		match: defaultNumberRegexp,
+		matchIndex: 0,
 		...customOptions,
 	};
 
@@ -227,7 +229,7 @@ function queryNumbers(context, selector, customOptions) {
 	}
 
 	return numberStrings
-		.map((numberString) => matchNumberString(numberString, options.match))
+		.map((numberString) => matchNumberString(numberString, options))
 		.filter(Boolean);
 }
 
@@ -450,28 +452,28 @@ function queryPosters(context, selector = 'video', customOptions) {
 	return posterUrls.map((posterUrl) => prefixUrl(posterUrl, options.origin, options));
 }
 
-function extractJson(element) {
-	if (!element) {
+function extractJson(dataString) {
+	if (!dataString) {
 		return null;
 	}
 
 	try {
-		return JSON.parse(element.innerHTML);
+		return JSON.parse(dataString);
 	} catch (error) {
 		return null;
 	}
 }
 
 function queryJson(context, selector, customOptions) {
-	const target = queryElement(context, selector, customOptions);
+	const dataString = queryContent(context, selector, customOptions);
 
-	return extractJson(target);
+	return extractJson(dataString);
 }
 
 function queryJsons(context, selector, customOptions) {
-	const targets = queryElements(context, selector, customOptions);
+	const dataStrings = queryContents(context, selector, customOptions);
 
-	return targets.map((target) => extractJson(target)).filter(Boolean);
+	return dataStrings.map((dataString) => extractJson(dataString)).filter(Boolean);
 }
 
 function extractDate(dateString, format, customOptions) {
@@ -485,6 +487,7 @@ function extractDate(dateString, format, customOptions) {
 
 	const options = {
 		match: /((\d{1,4}[/-]\d{1,2}[/-]\d{1,4})|(\w+\s+\d{1,2},?\s+\d{4}))(\s+\d{1,2}:\d{2}(:\d{2})?)?/g, // matches any of 01-01-1970, 1970-01-01 and January 1, 1970 with optional 00:00[:00] time
+		matchIndex: 0,
 		timezone: 'UTC',
 		...customOptions,
 	};
@@ -494,7 +497,7 @@ function extractDate(dateString, format, customOptions) {
 		: trim(dateString);
 
 	if (dateStamp) {
-		const dateValue = moment.tz(options.match ? dateStamp[0] : dateStamp, format, options.timezone);
+		const dateValue = moment.tz(options.match ? dateStamp[options.matchIndex] : dateStamp, format, options.timezone);
 
 		if (dateValue.isValid()) {
 			return dateValue.toDate();

package/tests/init.js

@@ -23,6 +23,7 @@ async function initTest() {
 	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('number indexed', res.context.query.number('.number', { match: /(\d+)/, matchIndex: 1 }));
 	console.log('data', res.context.query.json('#json'));
 	console.log('items', res.context.query.contents('.item'));
 	console.log('link', res.context.query.url('#link'));