feat(ingest): readability wrapper

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-01 03:34:51 +10:00
parent 8d2afcd040
commit c6e72e93d5
2 changed files with 41 additions and 0 deletions
--- a/tests/ingest/readability.test.js
+++ b/tests/ingest/readability.test.js
@@ -0,0 +1,25 @@
+import { describe, it, expect } from 'vitest';
+import { extract } from '../../lib/ingest/readability.js';
+
+const HTML = `
+  <html><head><title>Blackflame Notes</title>
+  <meta property="og:site_name" content="Hynesy"/>
+  </head><body><article>
+  <h1>Blackflame Notes</h1>
+  <p>An essay on the Cradle aesthetic and the blackflame motif. This is a longer paragraph that gives readability enough text to consider this the main content of the page.</p>
+  <p>A second paragraph also part of the article.</p>
+  </article></body></html>`;
+
+describe('readability.extract', () => {
+  it('pulls title and text', () => {
+    const out = extract(HTML, 'https://example.com/x');
+    expect(out.title).toMatch(/Blackflame/);
+    expect(out.textContent).toMatch(/Cradle/);
+    expect(out.siteName).toBe('Hynesy');
+  });
+
+  it('returns empty struct when nothing parseable', () => {
+    const out = extract('<html></html>', 'https://example.com');
+    expect(out.textContent).toBe('');
+  });
+});