fix: ignore hidden Amazon sponsored products

scripts/test-amazon-search-parser.ts

@@ -78,6 +78,19 @@ const html = `
           <a href="/dp/B000000010"><img src="https://example.com/hidden-aria.jpg" /></a>
         </article>
       </section>
+
+      <section class="featured-from-amazon-brands">
+        <span>Featured from Amazon brands</span>
+        <article data-asin="B000000011">
+          <h3
+            aria-label="Sponsored Ad - hidden brand-banner product"
+            style="position: absolute; width: 1px; height: 1px; overflow: hidden; clip: rect(0, 0, 0, 0)"
+          >Hidden brand-banner product</h3>
+          <a href="/sspa/click?url=%2Fdp%2FB000000011">
+            <img src="https://example.com/visible-brand-hero.jpg" />
+          </a>
+        </article>
+      </section>
     </div>
 
     <aside id="rightResultsATF" data-component-type="sponsored-display">
@@ -127,6 +140,7 @@ async function main() {
   assert.equal(items.filter((item) => item.asin === 'B000000008').length, 1);
   assert.equal(byAsin.has('B000000009'), false);
   assert.equal(byAsin.has('B000000010'), false);
+  assert.equal(byAsin.has('B000000011'), false);
   assert.deepEqual(
     items.map((item) => item.pageRank),
     [1, 2, 3, 4, 5, 6, 7, 8],

src/page-worker/web-injectors/amazon-search-parser.ts

@@ -42,7 +42,12 @@ export async function collectAmazonSearchPageData(_: SearchPageCollectorPayload)
           style.visibility === 'hidden' ||
           style.visibility === 'collapse' ||
           style.opacity === '0' ||
-          style.getPropertyValue('content-visibility') === 'hidden')
+          style.getPropertyValue('content-visibility') === 'hidden' ||
+          (style.clip !== 'auto' && style.clip !== '') ||
+          (style.clipPath !== 'none' && style.clipPath !== '') ||
+          (style.overflow === 'hidden' &&
+            Number.parseFloat(style.width) <= 1 &&
+            Number.parseFloat(style.height) <= 1))
       ) {
         return false;
       }
@@ -85,10 +90,11 @@ export async function collectAmazonSearchPageData(_: SearchPageCollectorPayload)
 
   const hasSponsoredDisclosure = (node: Element) => {
     const candidates = [
-      node,
+      ...(node.matches('[aria-label], [title], span') ? [node] : []),
-      ...Array.from(node.querySelectorAll('[aria-label], [title], span, a')),
+      ...Array.from(node.querySelectorAll('[aria-label], [title], span')),
-    ];
+    ].filter((candidate): candidate is HTMLElement => candidate instanceof HTMLElement);
     return candidates.some((candidate) => {
+      if (!isVisible(candidate)) return false;
       const accessibleLabels = [
         candidate.getAttribute('aria-label'),
         candidate.getAttribute('title'),
@@ -98,21 +104,10 @@ export async function collectAmazonSearchPageData(_: SearchPageCollectorPayload)
       ) {
         return true;
       }
-      return /^(Sponsored|赞助)(?:\b|\s|$)/i.test(text(candidate));
+      return candidate.matches('span') && /^(Sponsored|赞助)(?:\b|\s|$)/i.test(text(candidate));
     });
   };
 
-  const hasSponsoredLink = (node: Element) =>
-    Array.from(node.querySelectorAll<HTMLAnchorElement>('a[href]')).some((anchor) => {
-      const href = decodeRepeatedly(anchor.href).toLowerCase();
-      return (
-        href.includes('/sspa/click') ||
-        href.includes('/gp/slredirect') ||
-        href.includes('adid=') ||
-        href.includes('qualifier=')
-      );
-    });
-
   const metadata = (node: Element) =>
     [
       node.getAttribute('data-component-type'),
@@ -176,13 +171,7 @@ export async function collectAmazonSearchPageData(_: SearchPageCollectorPayload)
         'aside, #rightResultsATF, .s-right-column, [data-cel-widget*="right"], [data-component-type*="sidebar"]',
       ),
     );
-    const sponsored =
+    const sponsored = hasSponsoredDisclosure(card) || hasSponsoredDisclosure(module);
-      hasSponsoredDisclosure(card) ||
-      hasSponsoredDisclosure(module) ||
-      hasSponsoredLink(card) ||
-      hasSponsoredLink(module) ||
-      hasExplicitAdMetadata(card) ||
-      hasExplicitAdMetadata(module);
 
     if (!sponsored) {
       return {
@@ -205,7 +194,7 @@ export async function collectAmazonSearchPageData(_: SearchPageCollectorPayload)
         adFormat: module.querySelector('video') ? ('video' as const) : ('standard' as const),
       };
     }
-    if (hasSponsoredDisclosure(card) || hasSponsoredLink(card)) {
+    if (hasSponsoredDisclosure(card)) {
       return {
         adType: 'sponsoredProducts' as const,
         placement: 'middleOfSearch' as const,
@@ -249,6 +238,11 @@ export async function collectAmazonSearchPageData(_: SearchPageCollectorPayload)
   });
 
   const cards = Array.from(candidateSet)
+    .filter((card) => {
+      const titleNode = card.querySelector<HTMLElement>('h2[aria-label], h2, h3[aria-label], h3');
+      const imageNode = card.querySelector<HTMLImageElement>('img.s-image, img[src]');
+      return Boolean(titleNode && imageNode && isVisible(titleNode) && isVisible(imageNode));
+    })
     .filter(
       (card, index, all) =>
         !all.some(
@@ -284,6 +278,7 @@ export async function collectAmazonSearchPageData(_: SearchPageCollectorPayload)
 
     const titleNode = card.querySelector<HTMLElement>('h2[aria-label], h2, h3[aria-label], h3');
     const imageNode = card.querySelector<HTMLImageElement>('img.s-image, img[src]');
+    if (!titleNode || !imageNode || !isVisible(titleNode) || !isVisible(imageNode)) return items;
     const title =
       titleNode?.getAttribute('aria-label')?.trim() ||
       text(titleNode) ||