From ea6612125cd5ebb3eadd547e702ae28e2c01f8c2 Mon Sep 17 00:00:00 2001
From: jason <stedwell@gmail.com>
Date: Sat, 6 Jun 2026 14:14:15 -0500
Subject: [PATCH] scaper wait time and debug

---
 backend/src/scraper.ts | 76 ++++++++++++++++++++++++++++++++++--------
 1 file changed, 63 insertions(+), 13 deletions(-)

diff --git a/backend/src/scraper.ts b/backend/src/scraper.ts
index 5da9bf8..a8c76a4 100644
--- a/backend/src/scraper.ts
+++ b/backend/src/scraper.ts
@@ -70,24 +70,55 @@ export async function checkStockStatus(url: string): Promise<ScrapeResult> {
 
     await page.goto(url, { waitUntil: 'networkidle2', timeout: 45000 });
 
-    // Wait for React hydration
+    // Wait for React hydration, then actively wait for a <button> to appear
+    // (hydration on store.ui.com sometimes takes 4-6s in the Unraid container).
     await new Promise(r => setTimeout(r, 2500));
+    try {
+      await page.waitForFunction(
+        () => document.querySelectorAll('button').length > 0,
+        { timeout: 8000 }
+      );
+    } catch {
+      // Fall through — the _debug payload below will record buttonCount: 0.
+    }
 
     const result = await page.evaluate(() => {
       let status: 'in_stock' | 'sold_out' | 'unknown' = 'unknown';
 
-      // In stock: primary button carries label="Add to Cart" as an attribute.
-      // Why: styled-components class hashes rebuild every deploy, and "Add to Cart"
-      // text alone can appear in related-product cards on the page.
-      if (document.querySelector('button[label="Add to Cart"]')) {
+      // Normalize text: collapse whitespace (incl. NBSP), lowercase. Necessary
+      // because the page text may include NBSPs, leading icons that contribute
+      // whitespace to textContent, or casing changes between deploys.
+      const norm = (s: string | null | undefined) =>
+        (s ?? '').replace(/ /g, ' ').replace(/\s+/g, ' ').trim().toLowerCase();
+
+      const buttons = Array.from(document.querySelectorAll('button'));
+
+      // In stock detection — try most specific signal first, fall back to text.
+      // 1. button with label="Add to Cart" attribute (styled-components prop pass-through)
+      // 2. button whose normalized text contains "add to cart"
+      // 3. any span text exactly "Add to Cart" inside a button (legacy DOM fallback)
+      const hasLabelAttr = buttons.some(
+        b => norm(b.getAttribute('label')) === 'add to cart'
+      );
+      const hasCartText = buttons.some(b => norm(b.textContent).includes('add to cart'));
+
+      if (hasLabelAttr || hasCartText) {
         status = 'in_stock';
       } else {
-        const soldOutPhrases = ['Notify me when available', 'Sold Out', 'Out of Stock'];
-        const buttons = document.querySelectorAll('button');
-        for (const btn of buttons) {
-          const text = btn.textContent?.trim();
-          if (text && soldOutPhrases.includes(text)) { status = 'sold_out'; break; }
-        }
+        // Sold-out / unavailable phrases — substring match on normalized button text.
+        const soldOutPhrases = [
+          'notify me when available',
+          'notify me',
+          'sold out',
+          'out of stock',
+          'currently unavailable',
+          'coming soon',
+        ];
+        const hit = buttons.some(b => {
+          const t = norm(b.textContent);
+          return t && soldOutPhrases.some(p => t.includes(p));
+        });
+        if (hit) status = 'sold_out';
       }
 
       // Product name: og:title is most reliable for single-product pages
@@ -108,10 +139,29 @@ export async function checkStockStatus(url: string): Promise<ScrapeResult> {
         thumbnail = ogImage.content;
       }
 
-      return { status, name, thumbnail };
+      // Debug payload — populated only when we can't determine status, so we
+      // can see what the page actually looked like to the scraper.
+      let _debug: Record<string, unknown> | undefined;
+      if (status === 'unknown') {
+        _debug = {
+          title: document.title,
+          url: location.href,
+          bodyLen: document.body?.innerHTML.length ?? 0,
+          buttonCount: buttons.length,
+          buttonLabels: buttons.map(b => b.getAttribute('label')).filter(Boolean).slice(0, 20),
+          buttonTexts: buttons.map(b => norm(b.textContent)).filter(Boolean).slice(0, 30),
+          h1: document.querySelector('h1')?.textContent?.trim().slice(0, 120),
+        };
+      }
+
+      return { status, name, thumbnail, _debug };
     });
 
-    return result as ScrapeResult;
+    if ((result as { _debug?: unknown })._debug) {
+      console.log(`[Scraper] UNKNOWN debug for ${url}:`, JSON.stringify((result as { _debug: unknown })._debug, null, 2));
+    }
+
+    return { status: result.status, name: result.name, thumbnail: result.thumbnail } as ScrapeResult;
   } finally {
     await browser.close();
     scrapeSemaphore.release();