unprint 0.16.1 → 0.16.3

package/README.md

@@ -52,11 +52,16 @@ 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 `(`, or an array of either or both acting as fallbacks. If the selector is falsy, the input element will be used.
+
+* XPath Caveat: `//` and `(//` at the *start* of the selector are converted to `.//` and `(.//` for more intuitive relative selection, but any consecutive `//` will be absolute.
 
 #### Querying multiple elements
 Most methods can be used in plural, returning an array of results, i.e. `query.elements()`, `query.dates()`.
 
+Options
+* `filterDuplicates`: When an array of selectors results in the same element being selected multiple times, ensure each element is only returned once, default `true`.
+
 #### Query an element
 * `query.element([selector], [options])`
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "unprint",
-  "version": "0.16.1",
+  "version": "0.16.3",
   "description": "Simplify common web scraping tasks while staying in control of the data.",
   "main": "src/app.js",
   "scripts": {

package/src/app.js

@@ -74,9 +74,9 @@ function getElements(context, selector, firstOnly = false) {
 		return context.element;
 	}
 
-	if (/^\//.test(selector)) {
-		// XPath selector
-		const iterator = globalWindow.document.evaluate(`.${selector}`, context.element, null, globalWindow.XPathResult.ORDERED_NODE_ITERATOR_TYPE, null);
+	if (selector.startsWith('/') || selector.startsWith('(')) {
+		// XPath selector, . prefix ensures selector is relative to current node, won't work for deeper selections
+		const iterator = globalWindow.document.evaluate(selector.replace(/^\//, './').replace(/^\(\//, '(./'), context.element, null, globalWindow.XPathResult.ORDERED_NODE_ITERATOR_TYPE, null);
 
 		if (firstOnly) {
 			return iterator.iterateNext();
@@ -102,14 +102,20 @@ function queryElement(context, selectors, _customOptions) {
 	return target || null;
 }
 
-function queryElements(context, selectors, _customOptions) {
+function queryElements(context, selectors, customOptions = {}) {
 	if (!selectors) {
 		return context.element;
 	}
 
-	const targets = [].concat(selectors).reduce((acc, selector) => acc || getElements(context, selector, false), null);
+	const options = customOptions;
+	const targets = [].concat(selectors).reduce((acc, selector) => acc.concat(getElements(context, selector, false)), []).filter(Boolean);
 
-	return targets || [];
+	if (options.filterDuplicates === false) {
+		return targets || [];
+	}
+
+	// findIndex always finds first index, if current index is not the first index, it's a dupe
+	return targets.filter((target, index, array) => index === array.findIndex((dupe) => target === dupe));
 }
 
 function queryExistence(context, selector, customOptions) {

package/tests/init.js

@@ -37,6 +37,7 @@ 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('date grouped 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('timestring', res.context.query.duration('#timestring'));
@@ -46,6 +47,7 @@ async function initTest() {
 	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('items css xpath array', res.context.query.contents(['.item', '//li[contains(@class, "number")]']));
 	console.log('link', res.context.query.url('#link'));
 	console.log('links', res.context.query.urls('.link'));
 	console.log('text', res.context.query.text('.text'));