fix: remove unicode soft hyphen in normalizeWhitespace (#34920)

packages/playwright-core/src/server/injected/roleUtils.ts

@@ -337,7 +337,7 @@ function trimFlatString(s: string): string {
 function asFlatString(s: string): string {
   // "Flat string" at https://w3c.github.io/accname/#terminology
   // Note that non-breaking spaces are preserved.
-  return s.split('\u00A0').map(chunk => chunk.replace(/\r\n/g, '\n').replace(/\s\s*/g, ' ')).join('\u00A0').trim();
+  return s.split('\u00A0').map(chunk => chunk.replace(/\r\n/g, '\n').replace(/[\u200b\u00ad]/g, '').replace(/\s\s*/g, ' ')).join('\u00A0').trim();
 }
 
 function queryInAriaOwned(element: Element, selector: string): Element[] {

packages/playwright-core/src/utils/isomorphic/ariaSnapshot.ts

@@ -220,7 +220,8 @@ export function parseAriaSnapshot(yaml: YamlLibrary, text: string, options: yaml
 const emptyFragment: AriaTemplateRoleNode = { kind: 'role', role: 'fragment' };
 
 function normalizeWhitespace(text: string) {
-  return text.replace(/[\r\n\s\t]+/g, ' ').trim();
+  // TODO: why is this different from normalizeWhitespace in stringUtils.ts?
+  return text.replace(/[\u200b\u00ad]/g, '').replace(/[\r\n\s\t]+/g, ' ').trim();
 }
 
 export function valueOrRegex(value: string): string | AriaRegex {

packages/playwright-core/src/utils/isomorphic/stringUtils.ts

@@ -83,7 +83,7 @@ export function cacheNormalizedWhitespaces() {
 export function normalizeWhiteSpace(text: string): string {
   let result = normalizedWhitespaceCache?.get(text);
   if (result === undefined) {
-    result = text.replace(/\u200b/g, '').trim().replace(/\s+/g, ' ');
+    result = text.replace(/[\u200b\u00ad]/g, '').trim().replace(/\s+/g, ' ');
     normalizedWhitespaceCache?.set(text, result);
   }
   return result;

tests/library/role-utils.spec.ts

@@ -372,6 +372,13 @@ test('display:contents should be visible when contents are visible', async ({ pa
   await expect(page.getByRole('button')).toHaveCount(1);
 });
 
+test('should remove soft hyphens and zero-width spaces', async ({ page }) => {
+  await page.setContent(`
+    <button>1\u00ad2\u200b3</button>
+  `);
+  expect.soft(await getNameAndRole(page, 'button')).toEqual({ role: 'button', name: '123' });
+});
+
 test('label/labelled-by aria-hidden with descendants', async ({ page }) => {
   test.info().annotations.push({ type: 'issue', description: 'https://github.com/microsoft/playwright/issues/29796' });
 

tests/page/expect-to-have-text.spec.ts

@@ -71,6 +71,8 @@ test.describe('toHaveText with text', () => {
     await expect(locator).toHaveText('text CONTENT', { ignoreCase: true });
     // Should support falsy ignoreCase.
     await expect(locator).not.toHaveText('TEXT', { ignoreCase: false });
+    // Should normalize soft hyphens.
+    await expect(locator).toHaveText('T\u00ade\u00adxt content');
   });
 
   test('pass contain', async ({ page }) => {

tests/page/page-aria-snapshot.spec.ts

@@ -515,6 +515,7 @@ it('should normalize whitespace', async ({ page }) => {
       <summary> one  \n two <a href="#"> link &nbsp;\n  1 </a> </summary>
     </details>
     <input value='  hello   &nbsp; world '>
+    <button>hello\u00ad\u200bworld</button>
   `);
 
   await checkAndMatchSnapshot(page.locator('body'), `
@@ -522,6 +523,7 @@ it('should normalize whitespace', async ({ page }) => {
       - text: one two
       - link "link 1"
     - textbox: hello world
+    - button "helloworld"
   `);
 
   // Weird whitespace in the template should be normalized.
@@ -532,6 +534,7 @@ it('should normalize whitespace', async ({ page }) => {
           two
       - link "  link     1 "
     - textbox:        hello  world
+    - button "he\u00adlloworld\u200b"
   `);
 });