unprint 0.18.24 → 0.18.25

package/README.md

@@ -121,7 +121,8 @@ Options
 
 Options
 * `origin`: The hostname to prefix when it is not included in the URL (`/path`).
-* `protocol`: The protocol to use when it is not included in the URL (`:www.example.com`, default `http`).
+* `protocol`: The protocol to use when it is not included in the URL (`//www.example.com`, default `http`). Set to `null` to disable this behavior.
+* `forceProtocol`: Overwrite the input or origin protocol with the specified `protocol`.
 
 Returns the `href` from an anchor element (or any other specified target) as a string.
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "unprint",
-  "version": "0.18.24",
+  "version": "0.18.25",
   "description": "Simplify common web scraping tasks while staying in control of the data.",
   "main": "src/app.js",
   "scripts": {},

package/src/app.js

@@ -365,25 +365,35 @@ function queryTexts(context, selector, customOptions) {
 }
 
 function prefixUrl(urlPath, originUrl, customOptions) {
+	const options = {
+		protocol: 'https',
+		forceProtocol: false,
+		...customOptions,
+	};
+
 	if (!urlPath) {
 		return null;
 	}
 
 	if (/^http/.test(urlPath)) {
 		// this is already a complete URL
-		return urlPath;
+		return options.forceProtocol && options.protocol
+			? urlPath?.replace(/^.*?:/, `${options.protocol}:`)
+			: urlPath;
 	}
 
 	if (!originUrl) {
+		// path without protocol, i.e. //www.example.com
+		if (options.protocol && /^\/\//.test(urlPath)) {
+			return `${options.protocol}:${urlPath}`;
+		}
+
 		return null;
 	}
 
-	const options = {
-		protocol: 'https',
-		...customOptions,
-	};
-
-	const origin = originUrl?.replace(/^.*?:/, `${options.protocol}:`);
+	const origin = options.forceProtocol && options.protocol
+		? originUrl?.replace(/^.*?:/, `${options.protocol}:`)
+		: originUrl;
 
 	try {
 		return new URL(urlPath, origin).href;