Fix annotation error due to URI encoding

src/annotator/integrations/html-metadata.ts

@@ -47,16 +47,23 @@ export class HTMLMetadata {
    * Returns the primary URI for the document being annotated
    */
   uri(): string {
-    let uri = decodeURIComponent(this._getDocumentHref());
+    let uri = this._getDocumentHref(); // Get the URI without decoding it first
 
-    // Use the `link[rel=canonical]` element's href as the URL if present.
+    // Attempt to decode the URI, handle exceptions if the URI is malformed
+    try {
+      uri = decodeURIComponent(uri);
+    } catch (error) {
+      // Log error for debugging. After this point we fall back to the original URI
+      console.error('Error decoding URI:', error);
+    }
+
+    // Use the `link[rel=canonical]` element's href as the URI if present.
     const links = this._getLinks();
     for (const link of links) {
       if (link.rel === 'canonical') {
-        uri = link.href;
+        uri = link.href; // Assuming canonical hrefs are correctly encoded
       }
     }
-
     return uri;
   }
 

src/annotator/integrations/test/html-metadata-test.js

@@ -405,5 +405,24 @@ describe('HTMLMetadata', () => {
 
       assert.equal(doc.uri(), canonicalLink.href);
     });
+
+    it('should log an error if URI is not decodable', () => {
+      // '%%CUST_ID%%' is an invalid escape sequence, so it will throw a URIError when decoded
+      const badURI = 'https://example.com/?foo=%%CUST_ID%%';
+      const doc = createDoc(badURI);
+      const consoleErrorSpy = sinon.stub(console, 'error');
+
+      try {
+        assert.equal(badURI, doc.uri());
+        assert.calledOnce(consoleErrorSpy);
+        assert.calledWith(
+          consoleErrorSpy,
+          'Error decoding URI:',
+          sinon.match.instanceOf(URIError),
+        );
+      } finally {
+        console.error.restore();
+      }
+    });
   });
 });