fix: support structured DOM selectors

test/index.test.tsmodified+24-0

Changes to test/index.test.ts

		⋯116 linesShow the 116 lines above this hunk@@ -117,6 +117,30 @@ describe('Web Scraper - Static HTML Parsing', () => {
117	117	expect(doc.querySelector('ytd-display-ad-renderer')?.textContent).toContain('Ad')
118	118	expect(doc.querySelector('[data-ad-slot]')?.tagName).toBe('ytd-display-ad-renderer')
119	119	})
	120	+
	121	+ it('should follow descendant and child combinators for table rows', () => {
	122	+ const doc = parseHTML(`
	123	+ <table class="items">
	124	+ <tbody>
	125	+ <tr class="odd"><td><a href="/club/startseite/verein/31">Club</a></td></tr>
	126	+ <tr class="even"><td><a href="/club/startseite/verein/42">Other</a></td></tr>
	127	+ </tbody>
	128	+ </table>
	129	+ `)
	130	+
	131	+ expect(doc.querySelectorAll('table.items > tbody > tr')).toHaveLength(2)
	132	+ expect(doc.querySelectorAll('table.items a[href*="/verein/"]')).toHaveLength(2)
	133	+ expect(doc.querySelector('tr.odd > td > a')?.getAttribute('href')).toContain('/verein/31')
	134	+ })
	135	+
	136	+ it('should decode entities in text and attributes', () => {
	137	+ const doc = parseHTML('<a title="Brighton &amp; Hove Albion">Brighton &amp; Hove&nbsp;Albion &#x20AC;1.2bn</a>')
	138	+ const link = doc.querySelector('a')
	139	+
	140	+ expect(link?.getAttribute('title')).toBe('Brighton & Hove Albion')
	141	+ expect(link?.textContent).toBe('Brighton & Hove\u00A0Albion €1.2bn')
	142	+ expect(extractText('<span>one</span><span>two</span>')).toBe('one two')
	143	+ })
120	144	})
121	145
122	146	describe('Client-Side Scraper - Detection', () => {