fix: preserve base Amazon product collection

scripts/test-amazon-search-parser.ts

@@ -78,19 +78,6 @@ 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">
@@ -140,12 +127,42 @@ 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],
   );
 
+  const hiddenDisclosureDom = new JSDOM(
+    `
+      <main id="search">
+        <div class="s-main-slot">
+          <article data-component-type="s-search-result" data-asin="B000000011">
+            <h2>Visible organic product</h2>
+            <span
+              aria-label="Sponsored Ad - hidden disclosure"
+              style="position: absolute; width: 1px; height: 1px; overflow: hidden; clip: rect(0, 0, 0, 0)"
+            >Sponsored</span>
+            <a href="/sspa/click?url=%2Fdp%2FB000000011">
+              <img class="s-image" src="https://example.com/visible-product.jpg" />
+            </a>
+          </article>
+        </div>
+      </main>
+    `,
+    { url: 'https://www.amazon.com/s?k=mirror' },
+  );
+  Object.assign(globalThis, {
+    document: hiddenDisclosureDom.window.document,
+    Node: hiddenDisclosureDom.window.Node,
+    Element: hiddenDisclosureDom.window.Element,
+    HTMLElement: hiddenDisclosureDom.window.HTMLElement,
+    HTMLAnchorElement: hiddenDisclosureDom.window.HTMLAnchorElement,
+    HTMLImageElement: hiddenDisclosureDom.window.HTMLImageElement,
+  });
+  const hiddenDisclosureItems = await collectAmazonSearchPageData({ pattern: 'pattern-2' });
+  assert.equal(hiddenDisclosureItems.length, 1);
+  assert.equal(hiddenDisclosureItems[0].adType, 'organic');
+
   console.log('Amazon search parser fixtures passed.');
 }
 

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

@@ -42,12 +42,7 @@ 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;
       }
@@ -62,6 +57,27 @@ export async function collectAmazonSearchPageData(_: SearchPageCollectorPayload)
     );
   };
 
+  const isVisuallyHiddenText = (node: HTMLElement) => {
+    if (!isVisible(node)) return true;
+    const style = node.ownerDocument.defaultView?.getComputedStyle(node);
+    if (!style) return false;
+
+    const width = Number.parseFloat(style.width);
+    const height = Number.parseFloat(style.height);
+    const screenReaderOnlyBox =
+      (style.position === 'absolute' || style.position === 'fixed') &&
+      style.overflow === 'hidden' &&
+      Number.isFinite(width) &&
+      Number.isFinite(height) &&
+      width <= 1 &&
+      height <= 1;
+    const zeroClip = /rect\(\s*0(?:px)?[, ]+\s*0(?:px)?[, ]+\s*0(?:px)?[, ]+\s*0(?:px)?\s*\)/i.test(
+      style.clip || '',
+    );
+    const insetClip = /inset\(\s*(?:50%|100%)/i.test(style.clipPath || '');
+    return screenReaderOnlyBox || zeroClip || insetClip;
+  };
+
   const decodeRepeatedly = (value: string) => {
     let decoded = value;
     for (let i = 0; i < 3; i++) {
@@ -94,7 +110,7 @@ export async function collectAmazonSearchPageData(_: SearchPageCollectorPayload)
       ...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;
+      if (isVisuallyHiddenText(candidate)) return false;
       const accessibleLabels = [
         candidate.getAttribute('aria-label'),
         candidate.getAttribute('title'),
@@ -238,11 +254,6 @@ 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(
@@ -278,7 +289,6 @@ 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) ||