From a1c13d907e170bc49232d4bffba32461f48a80e6 Mon Sep 17 00:00:00 2001
From: Lukas Mager <7467162+lkmgr@users.noreply.github.com>
Date: Thu, 30 Oct 2025 11:51:48 +0100
Subject: [PATCH] fix: handle errors during `getLinks` parsing

---
 src/crawler.ts     | 18 +++++++++++++++---
 src/links.ts       | 12 ++++++++----
 test/test.index.ts | 20 ++++++++++++++++++++
 test/test.links.ts | 22 ++++++++++++++++++++++
 4 files changed, 65 insertions(+), 7 deletions(-)
 create mode 100644 test/test.links.ts

diff --git a/src/crawler.ts b/src/crawler.ts
index 7df26db6..8f82c501 100644
--- a/src/crawler.ts
+++ b/src/crawler.ts
@@ -186,10 +186,22 @@ export class LinkChecker extends EventEmitter {
 
 		const state =
 			status >= 200 && status < 300 ? LinkState.OK : LinkState.BROKEN;
-		this.emitResult(opts, state, status, failures);
 
-		// Recurse if body is HTML and crawling is enabled
-		await this.maybeRecurse(opts, response);
+		if (state === LinkState.BROKEN) {
+			this.emitResult(opts, state, status, failures);
+			return;
+		}
+
+		try {
+			// Recurse if body is HTML and crawling is enabled
+			await this.maybeRecurse(opts, response);
+			this.emitResult(opts, state, status, failures);
+		} catch (error) {
+			// Report as a broken link when parsing body failed
+			this.emitResult(opts, LinkState.BROKEN, 0, [
+				{ cause: (error as Error).cause, message: (error as Error).message },
+			]);
+		}
 	}
 
 	// Perform fetch, handle retry on 429, collect failures
diff --git a/src/links.ts b/src/links.ts
index bd58938e..0de14599 100644
--- a/src/links.ts
+++ b/src/links.ts
@@ -183,10 +183,14 @@ export async function getLinks(
 		},
 	});
 	await new Promise((resolve, reject) => {
-		Stream.Readable.fromWeb(source as import('stream/web').ReadableStream)
-			.pipe(parser)
-			.on('finish', resolve)
-			.on('error', reject);
+		const rs = Stream.Readable.fromWeb(
+			source as import('stream/web').ReadableStream,
+		);
+
+		// Reject on Readable error
+		rs.on('error', reject);
+
+		rs.pipe(parser).on('finish', resolve).on('error', reject);
 	});
 	return links;
 }
diff --git a/test/test.index.ts b/test/test.index.ts
index be7619dc..57061679 100644
--- a/test/test.index.ts
+++ b/test/test.index.ts
@@ -8,6 +8,7 @@ import {
 	LinkState,
 	check,
 } from '../src/index.js';
+import * as linksMethods from '../src/links.ts';
 import { DEFAULT_OPTIONS } from '../src/options.ts';
 import { invertedPromise } from './utils.ts';
 
@@ -629,6 +630,25 @@ describe('linkinator', () => {
 		scope.done();
 	});
 
+	it('should treat link as broken when getLinks throws', async () => {
+		const parseErr = new Error('Parsing failure');
+		const spy = vi.spyOn(linksMethods, 'getLinks').mockRejectedValue(parseErr);
+
+		const checker = new LinkChecker();
+		const results = await checker.check({
+			path: 'test/fixtures/basic',
+		});
+
+		assert.ok(!results.passed);
+		assert.strictEqual(results.links[0].state, LinkState.BROKEN);
+		assert.strictEqual(
+			(results.links[0]?.failureDetails?.[0] as Error).message,
+			'Parsing failure',
+		);
+
+		spy.mockRestore();
+	});
+
 	describe('element metadata', () => {
 		it('should provide <a> text in results', async () => {
 			const scope = nock('http://example.invalid').head('/').reply(404);
diff --git a/test/test.links.ts b/test/test.links.ts
new file mode 100644
index 00000000..4fbcfa29
--- /dev/null
+++ b/test/test.links.ts
@@ -0,0 +1,22 @@
+import { describe, expect, it } from 'vitest';
+import { getLinks } from '../src/links.js';
+
+describe('getLinks', () => {
+	it('should reject when the HTML stream emits an error', async () => {
+		const body = new ReadableStream({
+			start(controller) {
+				setTimeout(() => controller.error(new Error('StreamError')), 0);
+			},
+		});
+
+		const response = {
+			body,
+			headers: new Headers({ 'content-type': 'text/html' }),
+		} as unknown as Response;
+
+		// Expect getLinks to reject with our error,
+		await expect(getLinks(response, 'http://example.invalid')).rejects.toThrow(
+			'StreamError',
+		);
+	});
+});