unprint 0.11.9 → 0.11.11

package/README.md

@@ -166,6 +166,7 @@ Returns the parsed JSON content of an element as an object.
 
 Options
 * `styleAttribute`: the CSS style attribute to extract, returns an object with all properties by default.
+* `attemptBugfix`: Attempts to fix/bypass JSDOM quirks related in particular to style attributes containing `url()`, at the risk of losing some surrounding definitions (e.g. `url() 0 0 no-repeat;` may become `url()`. Try disabling this property if you require those definitions; it may break the attribute entirely, though.
 
 Returns the CSS style attributes of an element as an object.
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "unprint",
-  "version": "0.11.9",
+  "version": "0.11.11",
   "description": "Simplify common web scraping tasks while staying in control of the data.",
   "main": "src/app.js",
   "scripts": {

package/src/app.js

@@ -479,19 +479,24 @@ function querySourceSets(context, selector, attr = 'srcset', customOptions = {})
 
 function removeStyleFunctionSpaces(el) {
 	// jsdom appears to have a bug where it ignores inline CSS attributes set to a function() containing spaces, e.g. url( image.png )
-	el.setAttribute('style', el.getAttribute('style').replace(/\(\s+(.*)\s+\)/g, (match, cssArgs) => `(${cssArgs})`));
+	el.setAttribute('style', el.getAttribute('style')
+		.replace(/\(\s+(.*)\s+\)/g, (match, cssArgs) => `(${cssArgs})`)
+		.replace(/\)[\w\s-]+;/g, ');'));
 }
 
 function queryStyle(context, selector, customOptions) {
 	const options = {
-		...customOptions,
 		attribute: 'style',
+		attemptBugfix: true,
+		...customOptions,
 	};
 
 	const element = queryElement(context, selector, options);
 
 	if (element) {
-		removeStyleFunctionSpaces(element);
+		if (options.attemptBugfix) {
+			removeStyleFunctionSpaces(element, options);
+		}
 
 		if (element.style) {
 			return options.styleAttribute
@@ -858,8 +863,8 @@ async function request(url, body, customOptions = {}, method = 'GET') {
 		...options,
 		timeout: options.timeout,
 		signal: options.abortSignal,
-		httpAgent: new http.Agent({ ...options.agent }),
-		httpsAgent: new https.Agent({ ...options.agent }),
+		httpAgent: options.httpAgent || new http.Agent({ ...options.agent }),
+		httpsAgent: options.httpsAgent || new https.Agent({ ...options.agent }),
 	});
 
 	if (!(res.status >= 200 && res.status < 300)) {

package/tests/index.html

@@ -44,6 +44,7 @@
 
 		<!-- deliberate space in url( ) to test JSDOM's quirky handling of this -->
 		<div class="style" style="background-image: url( https://i.imgur.com/eDQmLys.jpg ); color: red;"></div>
+		<div class="background" style="background: url(https://i.imgur.com/xFHbuDL.jpeg) 0 0 no-repeat; color: green;"></div>
 		<div class="style" style="margin: 1rem; color: blue;"></div>
 
 		<video id="video" poster="https://i.imgur.com/eDQmLys.jpg"><source src="https://i.imgur.com/eDQmLys.mp4"></video>

package/tests/init.js

@@ -38,7 +38,8 @@ async function initTest() {
 	console.log('srcset', res.context.query.sourceSet('.srcset'));
 	console.log('srcsets', res.context.query.sourceSets('.srcset'));
 	console.log('style', res.context.query.style('.style'));
-	console.log('style background', res.context.query.style('.style', { styleAttribute: 'background-image' }));
+	console.log('style background-image', res.context.query.style('.style', { styleAttribute: 'background-image' }));
+	console.log('style background', res.context.query.style('.background'));
 	console.log('styles', res.context.query.styles('.style'));
 	console.log('styles color', res.context.query.styles('.style', { styleAttribute: 'color' }));
 	console.log('path', res.context.query.url('#path'));