fix: harden parseProfilePage class and program extraction

2026-08-18 15:30:49 +08:00
parent 3c409ea048
commit 94d9b7a879
2 changed files with 24 additions and 2 deletions
@@ -34,6 +34,26 @@ test('parseProfilePage extracts header fields', () => {
  assert.strictEqual(p.fullName, 'ADRIAN ALFRED CHIONG BONPIN');
  assert.strictEqual(p.studentId, '22103625');
  assert.strictEqual(p.yearLevel, '4');
+  assert.strictEqual(p.studentType, 'CONTINUING');
  assert.strictEqual(p.program, 'BACHELOR OF SCIENCE IN INFORMATION TECHNOLOGY');
  assert.strictEqual(p.semester, '1ST Semester 2026');
 });
+
+test('parseProfilePage matches bold class token with single quotes and extra classes', () => {
+  const html = `
+    <h3><b>BACHELOR OF SCIENCE</b></h3>
+    <table>
+      <tr><td>Full Name</td><td class='bold text-right'>JOHN DOE</td></tr>
+      <tr><td>Student Type</td><td class="bold">CONTINUING</td></tr>
+    </table>
+  `;
+  const p = parseProfilePage(html);
+  assert.strictEqual(p.fullName, 'JOHN DOE');
+  assert.strictEqual(p.studentType, 'CONTINUING');
+  assert.strictEqual(p.program, 'BACHELOR OF SCIENCE');
+});
+
+test('parseProfilePage degrades gracefully on empty input', () => {
+  const p = parseProfilePage('');
+  assert.deepStrictEqual(p, {});
+});