From b3196354d1272e6440119ba170c9d8369149baa4 Mon Sep 17 00:00:00 2001
From: Randall Hand <Yeraze@users.noreply.github.com>
Date: Sun, 14 Dec 2025 06:15:53 -0500
Subject: fix: Fix Amazon product image extraction on amazon.com URLs (#2108)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The metascraper-amazon package extracts the first .a-dynamic-image
element, which on amazon.com is often the Prime logo instead of the
product image. This works fine on amazon.co.uk where the product
image appears first in the DOM.

Created a custom metascraper plugin that uses more specific selectors
(#landingImage, #imgTagWrapperId, #imageBlock) to target the actual
product image. By placing this plugin before metascraperAmazon() in
the chain, we fix image extraction while preserving all other Amazon
metadata (title, brand, description).

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-authored-by: Claude <noreply@anthropic.com>
---
 .../metascraper-amazon-improved.ts                 | 77 ++++++++++++++++++++++
 1 file changed, 77 insertions(+)
 create mode 100644 apps/workers/metascraper-plugins/metascraper-amazon-improved.ts

(limited to 'apps/workers/metascraper-plugins')

diff --git a/apps/workers/metascraper-plugins/metascraper-amazon-improved.ts b/apps/workers/metascraper-plugins/metascraper-amazon-improved.ts
new file mode 100644
index 00000000..ea9bf2e9
--- /dev/null
+++ b/apps/workers/metascraper-plugins/metascraper-amazon-improved.ts
@@ -0,0 +1,77 @@
+import type { Rules } from "metascraper";
+
+/**
+ * Improved Amazon metascraper plugin that fixes image extraction.
+ *
+ * The default metascraper-amazon package uses `.a-dynamic-image` selector
+ * which matches the FIRST element with that class. On amazon.com pages,
+ * this is often the Prime logo instead of the product image.
+ *
+ * This plugin uses more specific selectors to target the actual product
+ * image:
+ * - #landingImage: The main product image ID
+ * - #imgTagWrapperId img: Fallback container for product images
+ * - #imageBlock img: Additional fallback for newer Amazon layouts
+ *
+ * By placing this plugin BEFORE metascraperAmazon() in the plugin chain,
+ * we ensure the correct image is extracted while keeping all other Amazon
+ * metadata (title, brand, description) from the original plugin.
+ */
+
+const REGEX_AMAZON_URL =
+  /https?:\/\/(.*amazon\..*\/.*|.*amzn\..*\/.*|.*a\.co\/.*)/i;
+
+const test = ({ url }: { url: string }): boolean => REGEX_AMAZON_URL.test(url);
+
+const metascraperAmazonImproved = () => {
+  const rules: Rules = {
+    pkgName: "metascraper-amazon-improved",
+    test,
+    image: ({ htmlDom }) => {
+      // Try the main product image ID first (most reliable)
+      // Prefer data-old-hires attribute for high-resolution images
+      const landingImageHires = htmlDom("#landingImage").attr("data-old-hires");
+      if (landingImageHires) {
+        return landingImageHires;
+      }
+
+      const landingImageSrc = htmlDom("#landingImage").attr("src");
+      if (landingImageSrc) {
+        return landingImageSrc;
+      }
+
+      // Fallback to image block container
+      const imgTagHires = htmlDom("#imgTagWrapperId img").attr(
+        "data-old-hires",
+      );
+      if (imgTagHires) {
+        return imgTagHires;
+      }
+
+      const imgTagSrc = htmlDom("#imgTagWrapperId img").attr("src");
+      if (imgTagSrc) {
+        return imgTagSrc;
+      }
+
+      // Additional fallback for newer Amazon layouts
+      const imageBlockHires = htmlDom("#imageBlock img")
+        .first()
+        .attr("data-old-hires");
+      if (imageBlockHires) {
+        return imageBlockHires;
+      }
+
+      const imageBlockSrc = htmlDom("#imageBlock img").first().attr("src");
+      if (imageBlockSrc) {
+        return imageBlockSrc;
+      }
+
+      // Return undefined to allow next plugin to try
+      return undefined;
+    },
+  };
+
+  return rules;
+};
+
+export default metascraperAmazonImproved;
-- 
cgit v1.2.3-70-g09d2