Cleanup & docs

danburzo · danburzo · commit 9038fcc9137b · 2025-08-29T17:40:37.000+03:00
diff --git a/index.js b/index.js
@@ -36,7 +36,7 @@ import { resolveSequence, resolveParallel } from './src/util/promises.js';
 import addExif from './src/exif.js';
 
 import { isURL } from './src/util/url.js';
-import { isFeed, processFeed } from './src/util/feeds.js';
+import { isFeed, processFeed } from './src/feeds.js';
 
 import get_style_attribute_value from './src/get-style-attribute-value.js';
 
diff --git a/src/cleanup-item.js b/src/cleanup-item.js
@@ -5,7 +5,7 @@ import createDOMPurify from 'dompurify';
 
 import { hyphenateDom } from './hyphenate.js';
 import { textToIso6391, getLanguageAttribute } from './util/language.js';
-import { getUrlOrigin } from './util/url-origin.js';
+import { getURLOrigin } from './util/url.js';
 import { setIdsAndReturnHeadings, nestHeadings } from './headings.js';
 
 import {
@@ -120,7 +120,7 @@ export default async function cleanupItem(
 					stripping the URL down to its origin, 
 					but just in case, let’s strip it ourselves.
 				*/
-				referrer: getUrlOrigin(url),
+				referrer: getURLOrigin(url),
 				referrerPolicy: 'strict-origin-when-cross-origin',
 				timeout: 10 * 1000
 			},
diff --git a/src/feeds.js b/src/feeds.js
@@ -1,12 +1,24 @@
-import { isURL } from './url.js';
+import { isURL } from './util/url.js';
 
+/*
+	Specification:
+	
+	RFC 4287: The Atom Syndication Format
+	https://datatracker.ietf.org/doc/html/rfc4287
+*/
 function isAtomFeed(doc) {
 	return (
 		doc.documentElement.localName === 'feed' &&
 		doc.documentElement.namespaceURI === 'http://www.w3.org/2005/Atom'
 	);
 }
 
+/*
+	Specification:
+
+	RSS 2.0 Specification
+	https://www.rssboard.org/rss-specification
+*/
 function isRssFeed(doc) {
 	return doc.documentElement.localName === 'rss';
 }
diff --git a/src/remote-resources.js b/src/remote-resources.js
@@ -5,7 +5,7 @@ import {
 	extForMimetype,
 	isImageURL
 } from './util/file-mimetype.js';
-import { getUrlOrigin } from './util/url-origin.js';
+import { getURLOrigin } from './util/url.js';
 
 export default function remoteResources(doc) {
 	let srcs = new Map();
@@ -31,7 +31,7 @@ export default function remoteResources(doc) {
 			srcs.set(src, {
 				original: src,
 				mapped: `rr-${uuid()}${ext}`,
-				origin: getUrlOrigin(doc.baseURI),
+				origin: getURLOrigin(doc.baseURI),
 				mimetype: mime
 			});
 		}
diff --git a/src/util/url-origin.js b/src/util/url-origin.js
diff --git a/src/util/url.js b/src/util/url.js
@@ -1,3 +1,4 @@
+// Polyfill for URL.canParse(), which was added in later Node.js versions
 export function isURL(ref) {
 	try {
 		new URL(ref);
@@ -7,3 +8,13 @@ export function isURL(ref) {
 	}
 	return false;
 }
+
+export function getURLOrigin(str) {
+	let origin;
+	try {
+		origin = new URL(str).origin;
+	} catch (err) {
+		// ignore
+	}
+	return origin && origin !== 'null' ? origin : undefined;
+}
diff --git a/src/util/wrap-html-fragment.js b/src/util/wrap-html-fragment.js
@@ -1,9 +1,18 @@
+/* 
+	A minimal HTML shell for a DOM fragment,
+	such as the content of a XML feed entry,
+	to pass to the JSDOM constructor.
+*/
 export default function wrapHTMLFragment(item) {
-	return `<!doctype html>
-<html>
-	<head>
-		<title>${item.title}</title>
-	</head>
-	<body><article>${item.content}</article></body>
-</html>`;
+	return `
+		<!doctype html>
+		<html>
+			<head>
+				<title>${item.title || ''}</title>
+			</head>
+			<body>
+				<article>${item.content || ''}</article>
+			</body>
+		</html>
+	`;
 }
diff --git a/test/url-origin.test.js b/test/url-origin.test.js
diff --git a/test/url.test.js b/test/url.test.js
@@ -0,0 +1,16 @@
+import tape from 'tape';
+import { getURLOrigin, isURL } from '../src/util/url.js';
+
+tape('getURLOrigin', t => {
+	t.equal(getURLOrigin('invalid'), undefined);
+	t.equal(getURLOrigin('file:///Users/myuser/'), undefined);
+	t.equal(getURLOrigin('https://github.com/user/repo'), 'https://github.com');
+	t.end();
+});
+
+tape('isURL', t => {
+	t.equal(isURL('invalid'), false);
+	t.equal(isURL('file:///Users/myuser/'), true);
+	t.equal(isURL('https://github.com/user/repo'), true);
+	t.end();
+});
