localization optimization to avoid regex

test/i18n.test.js

@@ -31,6 +31,39 @@ test('HTML generation escapes text and attributes, preserves i18n markers and up
   assert.equal(localizeHtml(result, translate), result);
 });
 
+test('HTML scanning respects quoted tag delimiters, comments, boolean attributes and nested content', () => {
+  const source = '<!-- <h1 data-i18n="ignored">do not translate</h1> -->'
+    + '<button disabled title=old data-i18n-title="label" data-note="a > b">?</button>'
+    + "<h1 data-i18n='heading'>old</h1>"
+    + '<p data-i18n="parent"><strong>Keep nested text</strong></p>';
+  const calls = [];
+  const result = localizeHtml(source, key => {
+    calls.push(key);
+    return { heading: 'New heading', label: 'Sam\'s "label"' }[key];
+  });
+  assert.deepEqual(calls, ['label', 'heading']);
+  assert.ok(result.includes('<!-- <h1 data-i18n="ignored">do not translate</h1> -->'));
+  assert.ok(result.includes('disabled title="Sam&#39;s &quot;label&quot;"'));
+  assert.ok(result.includes('data-note="a > b"'));
+  assert.ok(result.includes("<h1 data-i18n='heading'>New heading</h1>"));
+  assert.ok(result.includes('<p data-i18n="parent"><strong>Keep nested text</strong></p>'));
+});
+
+test('HTML scanning preserves incomplete tags, quoted attributes and comments', () => {
+  for (const tail of ['<h1 data-i18n="title"', '<button title="unfinished >', '<!-- unfinished <h1>']) {
+    const source = `Plain text${tail}`;
+    assert.equal(localizeHtml(source, () => assert.fail('Incomplete markup must not be translated')), source);
+  }
+});
+
+test('HTML scanning handles long malformed input without regex backtracking', { timeout: 2000 }, () => {
+  const source = '<'.repeat(200000) + ' unfinished';
+  assert.equal(localizeHtml(source, () => assert.fail('Malformed markup must not be translated')), source);
+  const attribute = 'x'.repeat(200000);
+  const html = `<h1 data-note="${attribute}" data-i18n="heading">old</h1>`;
+  assert.equal(localizeHtml(html, () => 'New heading'), `<h1 data-note="${attribute}" data-i18n="heading">New heading</h1>`);
+});
+
 test('translations support placeholders, plurals and German fallback without treating text as HTML', async () => {
   const context = {};
   const i18n = await loadI18n(context);