Fix triệt để: Excel đơn giá (đ→d), OCR AI fallback + comma-VND, URL ảnh chain fallback, redirect to detail

Dockerfile

@@ -11,4 +11,4 @@ EXPOSE 7860
 USER bun
 CMD ["bun", "index.ts"]
 
-# Rebuild trigger: 1786976466 cache-bust-v9
+# Rebuild trigger: 1786990001 fix-excel-vnd-ocr-trien-de

bulk-import.ts

@@ -48,6 +48,8 @@ function _normHeader(s: string): string {
   return String(s || "")
     .normalize("NFD")
     .replace(/[\u0300-\u036f]/g, "")
+    .replace(/đ/g, "d")
+    .replace(/Đ/g, "d")
     .toLowerCase()
     .replace(/\s+/g, " ")
     .trim();
@@ -415,13 +417,16 @@ function parsePdfTextLines(lines: string[], opts: any, warnings: string[]): any
 // null when the line has no parseable price (then callers keep other strategies).
 function parseRaggedRow(line: string, opts: any): any | null {
   const L = String(line || "");
-  // (a) dot-grouped thousands: 1.290.000, 5.500.000 (not part of a code like GL-400)
-  let mt = L.match(/(?<![\d.])(\d{1,3}(?:\.\d{3})+)(?!\d)(\s*(?:đ|₫|đồng|dong|vnd))?/i);
+  // (a) grouped thousands — accept BOTH dot (1.290.000) and comma (6,290,000)
+  //     VND separators, plus optional currency suffix.
+  let mt = L.match(/(?<![\d.,])(\d{1,3}(?:[.,]\d{3})+)(?!\d)(\s*(?:đ|₫|đồng|dong|vnd))?/i);
   if (!mt) {
     // (b) standalone integer token of 4+ digits separated from letters
     mt = L.match(/(?<![\wđ₫])(\d{4,})(?![\wđ₫])(\s*(?:đ|₫|đồng|dong|vnd))?/i);
   }
   if (!mt) return null;
+  // Normalize the captured thousands to dots (so parseVnd parses them correctly);
+  // a pure 4-digit "6,290" without three trailing digits is NOT thousands.
   const priceText = mt[1] + (mt[2] ? mt[2].trim() : "");
   const priceStart = mt.index as number;
   const full = mt[0];
@@ -429,8 +434,10 @@ function parseRaggedRow(line: string, opts: any): any | null {
   let after = L.slice(priceStart + full.length).trim();
   const out: any = { price: priceText };
   // numeric qty token immediately after the price → drop (not product text)
-  const aftk = after.split(/\s+/);
+  let aftk = after.split(/\s+/);
   if (aftk.length && /^\d{1,3}$/.test(aftk[0])) after = aftk.slice(1).join(" ").trim();
+  // A lone currency token (đ, d, ₫, vnd, đồng) right after the price is NOT a brand.
+  if (/^[đd₫vnd]{1,4}$/i.test(after)) after = "";
   if (after && /[a-zA-ZÀ-ỹ]/.test(after)) out.brand = after;
   // trailing code-like SKU at the end of the "before" segment: "EH-90", "LO-45",
   // "MQ-8", "MM-6030" (letters(+dash/space)+digits, short, contains a digit).
@@ -441,6 +448,12 @@ function parseRaggedRow(line: string, opts: any): any | null {
   } else {
     out.name = before;
   }
+  // Strip a leading OCR row ordinal ("1 May giat..." -> "May giat...") when the
+  // rest looks like a product name (starts with a letter). Never touch names
+  // that genuinely begin with digits.
+  if (out.name && /^\d{1,3}[.\s]\s*[A-Za-zÀ-ỹ]/.test(out.name)) {
+    out.name = out.name.replace(/^\d{1,3}[.\s]+\s*/, "").trim();
+  }
   if (!out.name && out.sku) out.name = "";
   return out;
 }
@@ -657,6 +670,9 @@ function finalizeProducts(products: any[], opts: any, warnings: string[]): any {
     if (!nm && !skuV) return;
     if (nm && _isNoiseName(nm)) return;                    // "TỔNG CỘNG", "Cộng tiền", "STT"
     if (/^(trang|page)\s*\d+$/i.test(nm)) return;          // "Trang 2", "page 5" page footers
+    // Non-data rows: unit-of-measure, note/label lines, section markers.
+    if (/^(don vi(tinh)?\s*[:：]?|đơn vị\s*[:：]?|ghi chu\s*[:：]?|ky hieu\s*[:：]?|loai\s*[:：]?|hang\s*[:：]?)\s*.{0,40}$/i.test(nm)) return;
+    if (/^(nhom|phan nhom|danh muc|muc luc|bang gia|toan bo|san pham)$/i.test(nm)) return;
     if (nm && /^[\d.,\sđ₫%]+$/.test(nm) && !skuV) return;  // numeric-only name (a total amount)
     if (nm && /[a-zA-ZÀ-ỹ]/.test(nm) === false && !skuV) return;
     if (!p.name && !p.sku) return;

index.html

@@ -830,7 +830,7 @@
          first paint — the welcome/avatar-picker modal renders immediately. -->
     <script defer src="./src/cart-quote.js?gc=34"></script>    <!-- Cart + Quote module (Giỏ hàng / Báo giá) -->
     <script defer src="./src/customers.js?gc=18"></script>    <!-- Customer list (Danh sách khách hàng) -->
-<script defer src="./src/vaix-rag.js?gc=61"></script>    <!-- V.AI STUDIO RAG Module - loads FASTER via light index; lazy galleries; spec-aware search -->
+<script defer src="./src/vaix-rag.js?gc=62"></script>    <!-- V.AI STUDIO RAG Module - loads FASTER via light index; lazy galleries; spec-aware search -->
 	    <script defer src="./src/greeting-news.js?gc=41"></script>  <!-- Greeting HOT news cards + source links -->
 	    <script defer src="./src/order-sync.js?gc=17"></script>    <!-- Order sync with V.AISTUDIO backend -->
 	    <script type="module" src="./src/app.js?gc=61"></script>
@@ -862,7 +862,7 @@
 
     <!-- ── Thêm SP (Add Product) module: 100% port from V.AISTUDIO, gated by the
          homepage V.AISTUDIO access code (window.isVaAdmin()). ── -->
-    <script defer src="./src/vaix-add-product.js?v=9"></script>
-    <script defer src="./src/vaix-bulk-import.js?v=2"></script>
+    <script defer src="./src/vaix-add-product.js?v=10"></script>
+    <script defer src="./src/vaix-bulk-import.js?v=3"></script>
   </body>
 </html>

index.ts

@@ -423,6 +423,78 @@ async function chatOnce(model: string, messages: any[], token: string) {
   throw err;
 }
 
+// ── AI-assisted extraction from OCR text (PDF/ảnh đã OCR) ──────────────────
+// The deterministic line parser (`lineRowsToProducts`) struggles with ragged
+// tesseract OCR output. When the user uploads a PDF/ảnh and the OCR text is
+// noisy, send the raw OCR text to Gemma (same router as chat) and ask it to
+// return CLEAN structured products as JSON. Deterministic parse stays the
+// cheap/fast path for well-formed input; AI is the accuracy fallback.
+async function aiExtractProductsFromText(ocrText: string, opts: any = {}): Promise<any> {
+  const token = (process.env.HF_INFERENCE_TOKEN || process.env.HF_TOKEN || "").trim();
+  if (!token) return { ok: false, reason: "no HF token" };
+  const text = String(ocrText || "").slice(0, 24000);
+  if (!text.trim()) return { ok: false, reason: "empty text" };
+  const model = opts.aiTextModel || "google/gemma-3-12b-it";
+  const sys = [
+    "Bạn là bộ phận trích xuất sản phẩm từ văn bản OCR (quét catalogue giá thị trường hoặc PDF thiết bị nhà bếp bằng tiếng Việt).",
+    "OCR thường bị lẫn dòng, sai chỗ ngắt cột, dính chữ. Dựa vào NGỮ NGHĨA để bóc tách đúng từng sản phẩm.",
+    "Mỗi dòng/sụm dòng là MỘT sản phẩm. Tách: name (tên SP đầy đủ), sku/mã (nếu có, VD 'EH-90','GL-400','357869'), brand (thương hiệu: Heatlock, Hafele, Eurogold, Malloca, Bosch, Malloca...), price (giá số nguyên, bỏ dấu chấm phẩy, VD 6290000 cho '6.290.000đ'; nếu '5tr5' -> 5500000), description (mô tả ngắn).",
+    "BỎ các dòng không phải sản phẩm: 'TỔNG CỘNG', 'Trang 2', 'STT', 'Cộng tiền', 'Đơn vị tính', tiêu đề cột, dòng chỉ chứa số thứ tự.",
+    "Trả về DUY NHẤT một JSON array, không markdown, không text thừa. Mỗi phần tử object chỉ dùng các key: name, sku, brand, price, description.",
+    "Nếu không trích được sản phẩm nào, trả về [].",
+  ];
+  const body = {
+    model,
+    messages: [
+      { role: "system", content: sys.join("\n") },
+      { role: "user", content: "Văn bản OCR catalogue:\n---\n" + text + "\n---\nHãy trả về JSON array các sản phẩm." },
+    ],
+    max_tokens: 2048,
+    temperature: 0,
+    top_p: 0.95,
+    stream: false,
+  };
+  let resp: Response;
+  try {
+    resp = await fetch(ROUTER, {
+      method: "POST",
+      headers: { "Content-Type": "application/json", "Authorization": "Bearer " + token, "User-Agent": "vai-avatar2-extract" },
+      body: JSON.stringify(body),
+      signal: AbortSignal.timeout(60000),
+    });
+  } catch (e: any) { return { ok: false, reason: String(e?.message || e) }; }
+  if (!resp.ok) {
+    const t = await resp.text().catch(() => "");
+    return { ok: false, reason: "HTTP " + resp.status + " " + t.slice(0, 120), status: resp.status };
+  }
+  let content = "";
+  try { const d = await resp.json(); content = String(d?.choices?.[0]?.message?.content || "").trim(); }
+  catch (e: any) { return { ok: false, reason: "Bad AI response JSON" }; }
+  let arr: any = null;
+  try { arr = JSON.parse(content); }
+  catch (e1: any) {
+    const m = content.match(/\[[\s\S]*\]/);
+    if (m) { try { arr = JSON.parse(m[0]); } catch (e2: any) {} }
+  }
+  if (!Array.isArray(arr)) return { ok: false, reason: "AI returned non-array JSON" };
+  const cleaned = arr
+    .filter((p: any) => p && typeof p === "object" && (p.name || p.sku))
+    .map((p: any) => {
+      const price = Number(String(p.price ?? "").replace(/[^\d]/g, ""));
+      return {
+        name: String(p.name || "").trim(),
+        sku: String(p.sku || p.model || p.ma || "").trim(),
+        brand: String(p.brand || "").trim(),
+        price: price || undefined,
+        description: String(p.description || p.desc || "").trim(),
+        features: (Array.isArray(p.features) ? p.features : (p.features ? [p.features] : [])).map(String).map(s => s.trim()).filter(Boolean),
+      };
+    })
+    .filter((p: any) => p.name || p.sku);
+  if (!cleaned.length) return { ok: false, reason: "AI extracted 0 products" };
+  return { ok: true, products: cleaned, model };
+}
+
 // ── Server-side tool execution (restores news/web/product search in /api/chat) ──
 const CHAT_TOOLS = [
   { type: "function", function: {
@@ -1669,7 +1741,25 @@ const server = Bun.serve({
           const body: any = await req.json().catch(() => ({}));
           const text = String(body?.text || "");
           const opts: any = { defaultCategory: String(body?.defaultCategory || "") || undefined };
+          // First try the fast deterministic parser.
           const res: any = await parseTextContent(text, opts);
+          const detProds = (Array.isArray(res?.products) ? res.products : []);
+          const detTrustworthy = detProds.length >= 1 && (!res.reason || /product rows/i.test(String(res.reason)));
+          // If deterministic gave nothing (ragged OCR, merged columns), fall back
+          // to the Gemma structured extractor for accuracy — exactly the "PDF/ảnh
+          // trích xuất tầm bậy" case the admin reported.
+          if (!detTrustworthy || body?.forceAI) {
+            try {
+              const ai = await aiExtractProductsFromText(text, opts);
+              if (ai.ok && Array.isArray(ai.products) && ai.products.length) {
+                const aiProds = ai.products.map((p: any) => normalizeCatalogProduct(p, opts)).filter((p: any) => p && (p.name || p.sku));
+                const merged = detProds.length && aiProds.length ? aiProds : (aiProds.length ? aiProds : detProds);
+                const finalRes: any = { ok: true, products: merged, ready: true, count: merged.length, aiExtracted: aiProds.length > 0, model: ai.model };
+                if (detProds.length && !aiProds.length) finalRes.warnings = res.warnings || [];
+                return Response.json(finalRes);
+              }
+            } catch (_e) { /* fall through to deterministic result */ }
+          }
           return Response.json(res);
         } catch (e: any) {
           return Response.json({ ok: false, products: [], ready: false, reason: String(e?.message || e) }, { status: 200 });

src/vaix-add-product.js

@@ -1214,7 +1214,16 @@
       _pendingAdd = null;
       var moreBtn = $('addMoreBtn');
       if (moreBtn) moreBtn.style.display = 'inline-flex';
-      setTimeout(function () { window.closeAddProductModal(); var sb = $('addProductSubmit'); if (sb) { sb.textContent = 'Thêm SP'; sb.removeAttribute('data-confirm'); } }, 2600);
+      // ✅ User request: after adding, jump straight to THAT product's detail
+      // page so they can review the new item (image + price) immediately.
+      // `?product=<slug>` is handled by handleProductUrlParam in vaix-rag.js.
+      var detailSlug = slugify((sku ? sku + ' ' : '') + name);
+      setTimeout(function () {
+        try {
+          var toDetail = (window.location.origin + window.location.pathname).replace(/\/+$/, '') + '/?product=' + encodeURIComponent(detailSlug);
+          window.location.href = toDetail;
+        } catch (e2) { }
+      }, 1400);
     } catch (err) {
       status('❌ Lỗi: ' + (err && err.message || err), 'err');
       if (submitBtn) { submitBtn.disabled = false; submitBtn.textContent = '✅ Xác nhận thêm SP'; }

src/vaix-rag.js

@@ -1821,9 +1821,38 @@ function renderPanelResults(products) {
     // Always store the SKU so reapplyCk() can resolve the product even when the
     // discount map is applied after render (fixes giá CK missing on other brands).
     if (p.sku) card.setAttribute("data-sku", p.sku);
-    if (p.image) { const ie = document.createElement("img"); ie.className = "product-card-img"; ie.setAttribute("data-raw", p.image); ie.src = primaryImgSrc(p.image); ie.alt = p.title_clean || ""; ie.loading = "lazy"; ie.onerror = function(){ this.style.display = "none"; const w = this.closest && this.closest(".product-card-img-wrap"); if (w) w.style.display = "none"; }; card.appendChild(ie); }
-    else if (p.image) { const ie = document.createElement("img"); ie.className = "product-card-img"; ie.src = p.image; ie.alt = p.title_clean || ""; ie.loading = "lazy"; ie.onerror = function(){ this.style.display = "none"; }; card.appendChild(ie); }
-    else { const pl = document.createElement("div"); pl.className = "product-card-img"; pl.style.cssText = "flex-shrink:0;background:#e2e8f0;display:flex;align-items:center;justify-content:center;font-size:1.5rem"; pl.textContent = "📦"; card.appendChild(pl); }
+// Image chain: try p.image, then each p.images[] (in case the primary is a
+	    // broken/no-extension URL that the CDN 403s), then a plain (un-proxied)
+	    // variant. Only hide the card image once EVERY candidate in the chain fails.
+	    var _imgCandidates = (function () {
+	      var arr = [];
+	      if (p.image) arr.push(p.image);
+	      if (Array.isArray(p.images)) { for (var ii = 0; ii < p.images.length; ii++) { if (p.images[ii] && arr.indexOf(p.images[ii]) < 0) arr.push(p.images[ii]); } }
+	      return arr;
+	    })();
+	    if (_imgCandidates.length) {
+	      var ie = document.createElement("img"); ie.className = "product-card-img"; ie.alt = p.title_clean || ""; ie.loading = "lazy";
+	      var _imgIdx = 0;
+	      ie.setAttribute("data-raw", _imgCandidates[0]);
+	      ie.src = primaryImgSrc(_imgCandidates[0]);
+	      ie.onerror = function () {
+	        _imgIdx++;
+	        if (_imgIdx < _imgCandidates.length) {
+	          ie.setAttribute("data-raw", _imgCandidates[_imgIdx]);
+	          ie.src = primaryImgSrc(_imgCandidates[_imgIdx]);
+	          return;
+	        }
+	        var raw = _imgCandidates[_imgCandidates.length - 1];
+	        if (raw && ie.getAttribute("data-raw") !== "__raw") {
+	          ie.setAttribute("data-raw", "__raw");
+	          ie.src = raw;
+	          return;
+	        }
+	        this.style.display = "none";
+	        var wrp = this.closest && this.closest(".product-card-img-wrap"); if (wrp) wrp.style.display = "none";
+	      };
+	      card.appendChild(ie);
+	    } else { const pl = document.createElement("div"); pl.className = "product-card-img"; pl.style.cssText = "flex-shrink:0;background:#e2e8f0;display:flex;align-items:center;justify-content:center;font-size:1.5rem"; pl.textContent = "📦"; card.appendChild(pl); }
     const info = document.createElement("div"); info.className = "product-card-info";
     const t = document.createElement("p"); t.className = "product-card-title"; t.textContent = p.title_clean || p.name;
     const b = document.createElement("p"); b.className = "product-card-brand"; b.textContent = p.brand || "";
@@ -2027,17 +2056,38 @@ function renderDetailImages(p) {
   const zoomBadge = document.getElementById("detail-zoom-badge");
   if (mainImg) {
     p = p || {};
-    let src = p.image || (Array.isArray(p.images) && p.images[0] ? p.images[0] : '') || '';
-    if (src) {
-      mainImg.setAttribute("data-raw", src);
-      mainImg.src = primaryImgSrc(src);
-      mainImg.style.display = 'block';
-      mainImg.onerror = function(){ this.style.display = 'none'; };
-      if (zoomBadge) zoomBadge.style.display = 'flex';
-    } else {
-      mainImg.style.display = 'none';
-      if (zoomBadge) zoomBadge.style.display = 'none';
-    }
+let src = p.image || (Array.isArray(p.images) && p.images[0] ? p.images[0] : '') || '';
+	    if (src) {
+	      // Image chain fallback (like card): try each candidate in p.image +
+	      // p.images[], then plain raw. Fixes the case where the primary is a
+	      // broken no-extension CDN URL that 403s even via the proxy.
+	      const _chain = [];
+	      if (p.image) _chain.push(p.image);
+	      if (Array.isArray(p.images)) { for (let _ci = 0; _ci < p.images.length; _ci++) { const _u = p.images[_ci]; if (_u && _chain.indexOf(_u) < 0) _chain.push(_u); } }
+	      var _mi = 0;
+	      mainImg.setAttribute("data-raw", _chain[0]);
+	      mainImg.src = primaryImgSrc(_chain[0]);
+	      mainImg.style.display = 'block';
+	      mainImg.onerror = function () {
+	        _mi++;
+	        if (_mi < _chain.length) {
+	          mainImg.setAttribute("data-raw", _chain[_mi]);
+	          mainImg.src = primaryImgSrc(_chain[_mi]);
+	          return;
+	        }
+	        var _raw = _chain[_chain.length - 1];
+	        if (_raw && mainImg.getAttribute("data-raw") !== "__raw") {
+	          mainImg.setAttribute("data-raw", "__raw");
+	          mainImg.src = _raw;
+	          return;
+	        }
+	        this.style.display = 'none';
+	      };
+	      if (zoomBadge) zoomBadge.style.display = 'flex';
+	    } else {
+	      mainImg.style.display = 'none';
+	      if (zoomBadge) zoomBadge.style.display = 'none';
+	    }
   }
   if (!ic) return;
   let imgs = [];