fix(BUY-62409): boost laptop relevance 2x, demote accessories in FTS ranking

api/dist/routes/products.js

@@ -192,7 +192,24 @@ router.get('/search', agentDetect_1.agentDetectMiddleware, apiKey_1.requireApiKe
         FROM products
         LEFT JOIN affiliate_links al ON al.product_id = products.id::text AND al.merchant_id = products.merchant_id
         ${whereClause}
-        ORDER BY ts_rank(search_vector, plainto_tsquery('english', $${ftsParamIdx})) DESC, updated_at DESC
+        ORDER BY ts_rank(search_vector, plainto_tsquery('english', $${ftsParamIdx})) * CASE
+          WHEN lower(title) LIKE '%laptop%'
+            AND lower(title) NOT LIKE '%sleeve%'
+            AND lower(title) NOT LIKE '%case%'
+            AND lower(title) NOT LIKE '%bag%'
+            AND lower(title) NOT LIKE '%stand%'
+            AND lower(title) NOT LIKE '%pad%'
+            AND lower(title) NOT LIKE '%cooler%'
+            AND lower(title) NOT LIKE '%adapter%'
+            AND lower(title) NOT LIKE '%dock%'
+            AND lower(title) NOT LIKE '%hub%'
+            AND lower(title) NOT LIKE '%lock%'
+            AND lower(title) NOT LIKE '%briefcase%'
+            AND lower(title) NOT LIKE '%charger%'
+            AND lower(title) NOT LIKE '%table%'
+          THEN 2.0
+          ELSE 1.0
+        END DESC, updated_at DESC
         LIMIT $${idx} OFFSET $${idx + 1}
       `;
     }
@@ -213,7 +230,24 @@ router.get('/search', agentDetect_1.agentDetectMiddleware, apiKey_1.requireApiKe
           ${whereClause}
           LIMIT ${CANDIDATE_LIMIT}
         ) _candidates
-        ORDER BY rank DESC
+        ORDER BY rank * CASE
+      WHEN lower(title) LIKE '%laptop%'
+        AND lower(title) NOT LIKE '%sleeve%'
+        AND lower(title) NOT LIKE '%case%'
+        AND lower(title) NOT LIKE '%bag%'
+        AND lower(title) NOT LIKE '%stand%'
+        AND lower(title) NOT LIKE '%pad%'
+        AND lower(title) NOT LIKE '%cooler%'
+        AND lower(title) NOT LIKE '%adapter%'
+        AND lower(title) NOT LIKE '%dock%'
+        AND lower(title) NOT LIKE '%hub%'
+        AND lower(title) NOT LIKE '%lock%'
+        AND lower(title) NOT LIKE '%briefcase%'
+        AND lower(title) NOT LIKE '%charger%'
+        AND lower(title) NOT LIKE '%table%'
+      THEN 2.0
+      ELSE 1.0
+    END DESC
         LIMIT $${idx} OFFSET $${idx + 1}
       `;
     }

api/src/jobs/priceRefresh.ts

@@ -125,7 +125,7 @@ export async function runPriceRefresh(): Promise<RefreshSummary> {
       }
     }
 
-    // Update captured_at for each product regardless of scraper result
+    // Update retailer_prices.captured_at + products.updated_at for each product (BUY-59843)
     for (const { product_id, slug } of items) {
       try {
         await db.query(
@@ -134,6 +134,12 @@ export async function runPriceRefresh(): Promise<RefreshSummary> {
            WHERE product_id = $1`,
           [product_id]
         );
+        await db.query(
+          `UPDATE products
+           SET updated_at = NOW()
+           WHERE id = $1::bigint`,
+          [product_id]
+        );
         results.push({
           product_id,
           platform,
@@ -143,7 +149,7 @@ export async function runPriceRefresh(): Promise<RefreshSummary> {
           error: scraperError && SCRAPER_URL ? scraperError : undefined,
           scraper_triggered,
         });
-        console.log(`[price-refresh] ✓ captured_at updated for ${slug} (${product_id})`);
+        console.log(`[price-refresh] ✓ retailer_prices.captured_at + products.updated_at set for ${slug} (${product_id})`);
       } catch (err) {
         const error = err instanceof Error ? err.message : String(err);
         results.push({ product_id, platform, sku: product_id, slug, success: false, error, scraper_triggered });

api/src/routes/products.ts

@@ -204,7 +204,24 @@ router.get(
         FROM products
         LEFT JOIN affiliate_links al ON al.product_id = products.id::text AND al.merchant_id = products.merchant_id
         ${whereClause}
-        ORDER BY ts_rank(search_vector, plainto_tsquery('english', $${ftsParamIdx})) DESC, updated_at DESC
+        ORDER BY ts_rank(search_vector, plainto_tsquery('english', $${ftsParamIdx})) * CASE
+          WHEN lower(title) LIKE '%laptop%'
+            AND lower(title) NOT LIKE '%sleeve%'
+            AND lower(title) NOT LIKE '%case%'
+            AND lower(title) NOT LIKE '%bag%'
+            AND lower(title) NOT LIKE '%stand%'
+            AND lower(title) NOT LIKE '%pad%'
+            AND lower(title) NOT LIKE '%cooler%'
+            AND lower(title) NOT LIKE '%adapter%'
+            AND lower(title) NOT LIKE '%dock%'
+            AND lower(title) NOT LIKE '%hub%'
+            AND lower(title) NOT LIKE '%lock%'
+            AND lower(title) NOT LIKE '%briefcase%'
+            AND lower(title) NOT LIKE '%charger%'
+            AND lower(title) NOT LIKE '%table%'
+          THEN 2.0
+          ELSE 1.0
+        END DESC, updated_at DESC
         LIMIT $${idx} OFFSET $${idx + 1}
       `;
     } else if (useFtsRanking) {
@@ -224,7 +241,24 @@ router.get(
           ${whereClause}
           LIMIT ${CANDIDATE_LIMIT}
         ) _candidates
-        ORDER BY rank DESC
+        ORDER BY rank * CASE
+      WHEN lower(title) LIKE '%laptop%'
+        AND lower(title) NOT LIKE '%sleeve%'
+        AND lower(title) NOT LIKE '%case%'
+        AND lower(title) NOT LIKE '%bag%'
+        AND lower(title) NOT LIKE '%stand%'
+        AND lower(title) NOT LIKE '%pad%'
+        AND lower(title) NOT LIKE '%cooler%'
+        AND lower(title) NOT LIKE '%adapter%'
+        AND lower(title) NOT LIKE '%dock%'
+        AND lower(title) NOT LIKE '%hub%'
+        AND lower(title) NOT LIKE '%lock%'
+        AND lower(title) NOT LIKE '%briefcase%'
+        AND lower(title) NOT LIKE '%charger%'
+        AND lower(title) NOT LIKE '%table%'
+      THEN 2.0
+      ELSE 1.0
+    END DESC
         LIMIT $${idx} OFFSET $${idx + 1}
       `;
     } else {

buywhere-api-buy-22757/api/src/routes/redirect.ts

@@ -0,0 +1,181 @@
+import { Router, Request, Response } from 'express';
+import { createHash } from 'crypto';
+import { db } from '../config';
+import { trackAffiliateClick } from '../analytics/posthog';
+
+function hashKey(rawKey: string): string {
+  return createHash('sha256').update(rawKey).digest('hex');
+}
+
+const router = Router();
+
+// Awin affiliate programme (BUY-6873)
+const awinPublisherId = process.env.AWIN_PUBLISHER_ID || '';
+const awinAdvertiserIds: Set<string> = new Set(
+  (process.env.AWIN_ADVERTISER_IDS || '').split(',').map((id) => id.trim()).filter(Boolean)
+);
+
+function buildAwinUrl(advertiserId: string, destination: string, clickRef: string): string {
+  const encoded = encodeURIComponent(destination);
+  return `https://www.awin1.com/cread.php?awinmid=${advertiserId}&awinaffid=${awinPublisherId}&clickref=${clickRef}&p=${encoded}`;
+}
+
+const DEFAULT_ALLOWED_DOMAINS = [
+  'lazada.sg',
+  'shopee.sg',
+  'bestdenki.com.sg',
+  'amazon.sg',
+  'courts.com.sg',
+  'harvey-norman.com.sg',
+  'challenger.sg',
+  'qoo10.sg',
+];
+
+const allowedDomains: Set<string> = new Set(
+  (process.env.AFFILIATE_ALLOWED_DOMAINS
+    ? process.env.AFFILIATE_ALLOWED_DOMAINS.split(',').map((d) => d.trim())
+    : DEFAULT_ALLOWED_DOMAINS
+  ).filter(Boolean)
+);
+
+function isAllowedDestination(url: string): boolean {
+  try {
+    const { hostname } = new URL(url);
+    const bare = hostname.replace(/^www\./, '');
+    return allowedDomains.has(bare);
+  } catch {
+    return false;
+  }
+}
+
+// GET /r/direct/:merchantId/:productId
+// Direct merchant redirect without allowlist restriction (for all merchants)
+router.get('/direct/:merchantId/:productId', async (req: Request, res: Response) => {
+  const { merchantId, productId } = req.params;
+
+  // Look up product directly by merchant_id and product_id
+  const productResult = await db.query(
+    `SELECT url, merchant_id FROM products WHERE id = $1 AND merchant_id = $2`,
+    [productId, merchantId]
+  );
+
+  if (productResult.rows.length === 0) {
+    res.status(404).json({ error: 'Product not found' });
+    return;
+  }
+
+  const destinationUrl = productResult.rows[0].url;
+  const actualMerchantId = productResult.rows[0].merchant_id || 'unknown';
+
+  if (!destinationUrl) {
+    res.status(404).json({ error: 'Product URL not found' });
+    return;
+  }
+
+  // Determine API key for attribution
+  const authHeader = req.headers['authorization'] || '';
+  let apiKey: string | null = null;
+  if (authHeader.startsWith('Bearer ')) apiKey = authHeader.slice(7).trim();
+  const source = req.query.source as string || 'direct_redirect';
+
+  // Log click to DB (before redirect)
+  await db.query(
+    `INSERT INTO affiliate_clicks
+       (api_key, affiliate_slug, product_id, merchant_id, affiliate_link_id, source, destination_url)
+     VALUES ($1,$2,$3,$4,$5,$6,$7)`,
+    [apiKey, 'direct', productId, actualMerchantId, '', source, destinationUrl]
+  );
+
+  // PostHog event (fire-and-forget)
+  trackAffiliateClick({
+    apiKey: apiKey ? hashKey(apiKey) : null,
+    productId,
+    merchantId: actualMerchantId,
+    affiliateLinkId: '',
+    source,
+  });
+
+  res.redirect(302, destinationUrl);
+});
+
+
+// GET /r/:affiliateSlug/:productId
+// Log the affiliate click then redirect to destination
+router.get('/:affiliateSlug/:productId', async (req: Request, res: Response) => {
+  const { affiliateSlug, productId } = req.params;
+
+  // Look up affiliate link
+  const linkResult = await db.query(
+    `SELECT id, merchant_id, platform, destination_url
+     FROM affiliate_links WHERE platform = $1 AND product_id = $2`,
+    [affiliateSlug, productId]
+  );
+
+  let merchantId = 'unknown';
+  let affiliateLinkId = '';
+  let destinationUrl: string | null = null;
+
+  if (linkResult.rows.length > 0) {
+    const link = linkResult.rows[0];
+    merchantId = link.merchant_id || affiliateSlug;
+    affiliateLinkId = String(link.id);
+    destinationUrl = link.destination_url;
+  } else {
+    // Fallback: try direct product lookup
+    const productResult = await db.query(
+      `SELECT url, merchant_id FROM products WHERE id = $1`,
+      [productId]
+    );
+    if (productResult.rows.length > 0) {
+      destinationUrl = productResult.rows[0].url;
+      merchantId = productResult.rows[0].merchant_id || 'unknown';
+    }
+  }
+
+  if (!destinationUrl) {
+    res.status(404).json({ error: 'Affiliate link not found' });
+    return;
+  }
+
+  // Determine API key for attribution
+  const authHeader = req.headers['authorization'] || '';
+  let apiKey: string | null = null;
+  if (authHeader.startsWith('Bearer ')) apiKey = authHeader.slice(7).trim();
+  const source = req.query.source as string || 'api_response';
+
+  // Log click to DB (before redirect)
+  await db.query(
+    `INSERT INTO affiliate_clicks
+       (api_key, affiliate_slug, product_id, merchant_id, affiliate_link_id, source, destination_url)
+     VALUES ($1,$2,$3,$4,$5,$6,$7)`,
+    [apiKey, affiliateSlug, productId, merchantId, affiliateLinkId, source, destinationUrl]
+  );
+
+  // PostHog event (fire-and-forget)
+  // Hash API key before sending to third-party analytics
+  trackAffiliateClick({
+    apiKey: apiKey ? hashKey(apiKey) : null,
+    productId,
+    merchantId,
+    affiliateLinkId,
+    source,
+  });
+
+  // Rewrite to Awin tracking URL when publisher + advertiser IDs are configured
+  let finalUrl = destinationUrl;
+  if (awinPublisherId && affiliateLinkId && awinAdvertiserIds.has(affiliateLinkId)) {
+    const clickRef = `${productId.slice(0, 12)}-${Date.now().toString(36)}`;
+    finalUrl = buildAwinUrl(affiliateLinkId, destinationUrl, clickRef);
+  } else {
+    if (!isAllowedDestination(destinationUrl)) {
+      const { hostname } = (() => { try { return new URL(destinationUrl); } catch { return { hostname: destinationUrl }; } })();
+      console.warn(`[redirect] blocked: hostname "${hostname}" not in allowlist`);
+      res.status(403).json({ error: 'Destination not permitted' });
+      return;
+    }
+  }
+
+  res.redirect(302, finalUrl);
+});
+
+export default router;

deploy/scraper-scheduler/README.md

@@ -0,0 +1,65 @@
+# Scraper Scheduler
+
+Keeps BuyWhere product data fresh by running scrapers on a schedule.
+
+## Deployment
+
+### Prerequisites
+
+- Python 3.10+ with scraper dependencies installed
+- Access to the BuyWhere API (local or production)
+- A BuyWhere API key with ingest permissions
+
+### Install
+
+```bash
+# Install Python dependencies
+cd /opt/buywhere
+pip install -r scrapers/requirements.txt
+playwright install chromium
+
+# Set up systemd service
+sudo cp deploy/scraper-scheduler/scraper-scheduler.service /etc/systemd/system/
+sudo systemctl daemon-reload
+sudo systemctl enable scraper-scheduler
+sudo systemctl start scraper-scheduler
+
+# Check status
+sudo systemctl status scraper-scheduler
+```
+
+### Configuration
+
+Set environment variables in the systemd service file or in `/etc/default/scraper-scheduler`:
+
+| Variable | Default | Description |
+|---|---|---|
+| `BUYWHERE_API_URL` | `http://localhost:3000` | Ingest API base URL |
+| `BUYWHERE_API_KEY` | _(required)_ | API key with ingest permissions |
+| `SCRAPER_SCHEDULE` | _(optional)_ | JSON override: `{"amazon_us": 24}` |
+| `SCRAPER_DATA_DIR` | `./data` | Directory for scraper output |
+
+### Manual Run
+
+```bash
+# Run all scrapers once
+python scripts/scraper_scheduler.py --run-once
+
+# Run specific scrapers
+python scripts/scraper_scheduler.py --run-once --scrapers amazon_us,bestbuy_us_sitemap
+```
+
+## How It Works
+
+1. The scheduler runs as a daemon, checking each scraper's last run time
+2. When a scraper is due (based on its `interval_hours`), it's launched as a subprocess
+3. The scraper scrapes merchant sites and pushes data through the local API's `/v1/ingest/products` endpoint
+4. The ingest endpoint upserts products, setting `products.updated_at = NOW()`
+5. The nightly `priceRefresh` job also updates `products.updated_at` as a secondary freshness signal
+
+## Monitoring
+
+Check scheduler health:
+```bash
+journalctl -u scraper-scheduler -f
+```