fix(core): stop the chat link regex backtracking on long words

.changeset/quick-link-backtracking.md

@@ -0,0 +1,5 @@
+---
+'@livekit/components-core': patch
+---
+
+Stop the chat link regex from freezing the page on a long word with no dot in it.

packages/core/src/helper/url-regex.test.ts

@@ -33,4 +33,14 @@ describe('url-regex', () => {
     // Invalid URLs
     expect(createUrlRegExp({}).test('livekitcom')).toBe(false);
   });
+
+  test('should not backtrack for long on a long word with no url in it', () => {
+    // the backtracking pattern takes 8s+ on each input, the current one tens of ms, so the
+    // bound leaves room for a slow CI worker
+    for (const text of ['a'.repeat(3000), 'x:'.repeat(1500)]) {
+      const start = performance.now();
+      expect(text.match(createUrlRegExp({}))).toBeNull();
+      expect(performance.now() - start).toBeLessThan(5000);
+    }
+  });
 });

packages/core/src/helper/url-regex.ts

@@ -34,13 +34,15 @@ export function createUrlRegExp(options: RegExOptions) {
   };
 
   const protocol = `(?:(?:[a-z]+:)?//)?`;
-  const auth = '(?:\\S+(?::\\S*)?@)?';
+  const auth = '(?:\\S+@)?';
   const ip = new RegExp(
     '(?:25[0-5]|2[0-4]\\d|1\\d\\d|[1-9]\\d|\\d)(?:\\.(?:25[0-5]|2[0-4]\\d|1\\d\\d|[1-9]\\d|\\d)){3}',
     'g',
   ).source;
-  const host = '(?:(?:[a-z\\u00a1-\\uffff0-9][-_]*)*[a-z\\u00a1-\\uffff0-9]+)';
-  const domain = '(?:\\.(?:[a-z\\u00a1-\\uffff0-9]-*)*[a-z\\u00a1-\\uffff0-9]+)*';
+  // each label is written so that a run of letters can be split only one way; the nested
+  // `(?:x[-_]*)*x+` form backtracks for seconds on a long word with no dot
+  const host = '(?:[a-z\\u00a1-\\uffff0-9]+(?:[-_]+[a-z\\u00a1-\\uffff0-9]+)*)';
+  const domain = '(?:\\.[a-z\\u00a1-\\uffff0-9]+(?:-+[a-z\\u00a1-\\uffff0-9]+)*)*';
   const tld = `(?:\\.(?:[a-z\\u00a1-\\uffff]{2,}))(?:\\.(?=[:/?#]))?`;
   const port = '(?::\\d{2,5})?';
   // a path keeps balanced parentheses but not the punctuation that ends a sentence around it