Fix Amazon product image provenance

scripts/bsr/creators-client.mjs

@@ -275,6 +275,7 @@ export function mapCreatorsItem(item, fallback = {}) {
     asin: item.asin,
     title: displayValue(itemInfo.title) || fallback.title,
     images: [...new Set(imageCandidates)].slice(0, 6),
+    imageGalleryVerified: true,
     price: listing?.price?.money?.amount,
     listPrice: listing?.price?.savingBasis?.money?.amount,
     rating:

scripts/bsr/fill-quota.mjs

@@ -310,15 +310,17 @@ const products = structuredClone(snap.products)
         !String(u).startsWith('/brand/landing-forest') &&
         !String(u).startsWith('/brand/hero'),
     )
-    // If no Amazon photo left but we have an ASIN, keep ASIN image attempts for the client
-    if (cleaned.length === 0 && p.asin) {
+    const externallySourced =
+      p.source === 'amazon-bsr' || p.source === 'amazon-search'
+    const safeImages = externallySourced && !p.imageGalleryVerified
+      ? cleaned.slice(0, 1)
+      : cleaned
+    // If no Amazon photo left but we have an ASIN, keep an ASIN image attempt for the client
+    if (safeImages.length === 0 && p.asin) {
       const a = String(p.asin).toUpperCase()
-      cleaned.push(
-        `https://m.media-amazon.com/images/P/${a}.01._SCLZZZZZZZ_SX500_.jpg`,
-        `https://images-na.ssl-images-amazon.com/images/P/${a}.01.LZZZZZZZ.jpg`,
-      )
+      safeImages.push(`https://m.media-amazon.com/images/P/${a}.01._SCLZZZZZZZ_SX500_.jpg`)
     }
-    return { ...p, images: cleaned }
+    return { ...p, images: safeImages }
   })
 const weekOf = snap.weekOf
 const expiresAt = snap.expiresAt

scripts/bsr/import-bsr.mjs

@@ -133,7 +133,7 @@ function upgradeAmazonImageUrl(url) {
 }
 
 /** Extract product photo URLs from a list-card HTML chunk. */
-function extractListImages(chunk, { max = 4 } = {}) {
+function extractListImages(chunk, { max = 1 } = {}) {
   const found = []
   const patterns = [
     /src="(https:\/\/[^"]+(?:media-amazon|ssl-images-amazon)\.com\/images\/I\/[^"]+)"/gi,
@@ -385,7 +385,7 @@ async function enrichAsin(asin) {
     return {
       asin,
       title,
-      images: images.slice(0, 6),
+      images: images.slice(0, 1),
       price,
       rating,
       reviewCount,
@@ -587,6 +587,15 @@ function toProduct(enriched, meta, weekOf, expiresAt) {
     outdoor: 120,
     baby: 200,
   }
+  const resolvedImages = resolveProductImages({
+    listImages: meta.images || [],
+    enrichedImages: enriched.images || [],
+    category: meta.ibambooCategory,
+    asin: enriched.asin,
+  })
+  const images = enriched.imageGalleryVerified
+    ? resolvedImages
+    : resolvedImages.slice(0, 1)
 
   return {
     id: `bsr-${enriched.asin}`,
@@ -628,12 +637,8 @@ function toProduct(enriched, meta, weekOf, expiresAt) {
     asin: enriched.asin,
     searchKeywords: enriched.title,
     badge,
-    images: resolveProductImages({
-      listImages: meta.images || [],
-      enrichedImages: enriched.images || [],
-      category: meta.ibambooCategory,
-      asin: enriched.asin,
-    }),
+    images,
+    ...(enriched.imageGalleryVerified ? { imageGalleryVerified: true } : {}),
     ...(enriched.rating != null ? { rating: enriched.rating } : {}),
     ...(enriched.reviewCount != null
       ? { reviewCount: enriched.reviewCount }
@@ -873,13 +878,11 @@ async function main() {
       continue
     }
 
-    // Prefer Creators images; fill gaps from list-page photos
+    // Creators galleries are ASIN-scoped. Never mix them with scraped page
+    // images; use the single list-card primary only when the API has no image.
     if ((!enriched.images || enriched.images.length === 0) && meta.images?.length) {
       enriched.images = meta.images
-    } else if (meta.images?.length) {
-      for (const img of meta.images) {
-        if (!enriched.images.includes(img)) enriched.images.push(img)
-      }
+      enriched.imageGalleryVerified = false
     }
     console.log(
       `OK ${enriched.title.slice(0, 50)} imgs=${(enriched.images || meta.images || []).length}`,

src/data/types.ts

@@ -48,6 +48,8 @@ export interface Product {
   badge?: string
   /** Primary + gallery images (Amazon CDN, brand, or photography) */
   images: string[]
+  /** True only when every secondary image came from an ASIN-scoped trusted API. */
+  imageGalleryVerified?: boolean
   rating?: number
   reviewCount?: number
   hue: number

src/lib/amazon.ts

@@ -4,7 +4,7 @@
  */
 
 export const AMAZON_ASSOCIATE_TAG =
-  import.meta.env.VITE_AMAZON_ASSOCIATE_TAG || 'iu0e3-20'
+  import.meta.env?.VITE_AMAZON_ASSOCIATE_TAG || 'iu0e3-20'
 
 /** Product detail page with Associates tag. */
 export function amazonProductUrl(asin: string, extras?: Record<string, string>) {

src/lib/productImages.ts

@@ -3,8 +3,8 @@
  * busy brand lifestyle fillers as the primary merchandising image.
  */
 
-import { amazonAsinImage } from './amazon'
-import type { Category, Product } from '../data/types'
+import { amazonAsinImage } from './amazon.ts'
+import type { Category, Product } from '../data/types.ts'
 
 const BUSY_BRAND_PATHS = [
   '/brand/products-flatlay.webp',
@@ -219,6 +219,20 @@ function upgradeAmazonThumb(url: string, size: 500 | 1000 = 500): string {
     .replace(/\._SS\d+_\./i, `._SL${size}_.`)
 }
 
+/**
+ * Scraped Amazon pages can contain recommendation images outside the active
+ * ASIN's gallery. Until an ASIN-scoped API verifies the gallery, trust only
+ * the primary catalog image and use ASIN-derived fallbacks after it.
+ */
+function catalogImagesForProduct(product: Product): string[] {
+  const images = product.images || []
+  const externallySourced =
+    product.source === 'amazon-bsr' || product.source === 'amazon-search'
+  return externallySourced && !product.imageGalleryVerified
+    ? images.slice(0, 1)
+    : images
+}
+
 /**
  * Ordered gallery for display / onError chain:
  * 1) Real Amazon CDN images from catalog
@@ -245,12 +259,12 @@ export function resolveProductImages(
     out.push(u)
   }
 
-  for (const img of product.images || []) {
+  for (const img of catalogImagesForProduct(product)) {
     if (isAmazonCdnImage(img)) push(img)
   }
 
   // Non-Amazon, non-busy extras (e.g. local brand product photography later)
-  for (const img of product.images || []) {
+  for (const img of catalogImagesForProduct(product)) {
     if (
       !isAmazonCdnImage(img) &&
       !isBusyBrandFallback(img) &&
@@ -284,7 +298,7 @@ export function galleryThumbImages(
 ): string[] {
   const out: string[] = []
   const seen = new Set<string>()
-  for (const img of product.images || []) {
+  for (const img of catalogImagesForProduct(product)) {
     if (!isReliableAmazonImage(img)) continue
     const u = upgradeAmazonThumb(img.trim(), size)
     if (!u || seen.has(u)) continue

tests/e2e/content-balloon-layout.spec.ts

@@ -4,6 +4,7 @@ import { bsrProducts } from '../../src/data/products.bsr.generated'
 
 const PRODUCT_PATH = '/product/riveira-dark-bamboo-wooden-spoons-for-cooking-6-piece-apartment-essentials-wood-'
 const NIAGARA_PATH = '/product/niagara-sleep-solution-ultra-soft-queen-size-mattress-topper-rayon-derived-from-'
+const SMIRLY_PATH = '/product/smirly-bamboo-cutting-boards-for-kitchen-wood-cutting-board-for-meal-prep-servin'
 const WIDTHS = [390, 768, 1024, 1440, 2560]
 const products = (() => {
   const seenAsins = new Set<string>()
@@ -143,6 +144,17 @@ test('a standard PDP has three useful, separated placements', async ({ page }) =
   )).size).toBe(3)
 })
 
+test('an unverified Amazon gallery never exposes recommendation images', async ({ page }) => {
+  await page.goto(SMIRLY_PATH)
+  const media = page.locator('[data-product-surface="media"]')
+  await expect(media.locator('img')).toHaveAttribute('src', /81FoZNCStHL/)
+  await expect(media.locator('[data-has-thumbnail-rail]')).toHaveAttribute(
+    'data-has-thumbnail-rail',
+    'false',
+  )
+  await expect(media.locator('button')).toHaveCount(0)
+})
+
 test('mobile reads image, purchase decision, then specifications', async ({ page }) => {
   await page.setViewportSize({ width: 390, height: 844 })
   await loadWithFacts(page, NIAGARA_PATH)

tests/product-images.test.mjs

@@ -0,0 +1,58 @@
+import assert from 'node:assert/strict'
+import { readFileSync } from 'node:fs'
+import test from 'node:test'
+
+import { galleryThumbImages, resolveProductImages } from '../src/lib/productImages.ts'
+
+const primary = 'https://m.media-amazon.com/images/I/primary._AC_SL1000_.jpg'
+const unrelated = [
+  'https://m.media-amazon.com/images/I/recommendation-one.jpg',
+  'https://m.media-amazon.com/images/I/recommendation-two.jpg',
+  'https://m.media-amazon.com/images/I/recommendation-three.jpg',
+]
+const product = {
+  id: 'bsr-B0829DCVN6',
+  slug: 'smirly-cutting-boards',
+  name: 'SMIRLY Bamboo Cutting Boards',
+  tagline: 'Weekly list',
+  description: 'A cutting board set.',
+  category: 'cutting-boards',
+  collection: 'Boards',
+  material: 'Bamboo',
+  features: [],
+  specs: [],
+  priceHint: 0,
+  asin: 'B0829DCVN6',
+  searchKeywords: 'SMIRLY cutting boards',
+  images: [primary, ...unrelated],
+  hue: 70,
+  source: 'amazon-bsr',
+}
+
+test('unverified Amazon catalog entries expose only their primary image', () => {
+  assert.deepEqual(galleryThumbImages(product, 1000), [primary])
+  const resolved = resolveProductImages(product, 1000)
+  assert.equal(resolved[0], primary)
+  assert.ok(unrelated.every((url) => !resolved.includes(url)))
+})
+
+test('ASIN-scoped verified galleries retain their secondary images', () => {
+  const verified = { ...product, imageGalleryVerified: true }
+  assert.deepEqual(galleryThumbImages(verified, 1000), [primary, ...unrelated])
+})
+
+test('generated Amazon catalog persists no unverified secondary images', () => {
+  const snapshot = JSON.parse(
+    readFileSync(new URL('../src/data/bsr-snapshot.json', import.meta.url), 'utf8'),
+  )
+  const unsafe = snapshot.products.filter(
+    (entry) =>
+      ['amazon-bsr', 'amazon-search'].includes(entry.source) &&
+      !entry.imageGalleryVerified &&
+      entry.images.length > 1,
+  )
+  assert.deepEqual(
+    unsafe.map((entry) => entry.asin),
+    [],
+  )
+})