add /api/url-product: server-side malloca/JSON-LD product parser (name, sku, price, images, description, specs, features)

index.ts

@@ -1228,6 +1228,161 @@ const server = Bun.serve({
         return new Response(buf,{headers:{"Content-Type":"image/jpeg","Cache-Control":"public, max-age=3600","Access-Control-Allow-Origin":"*"}});
       } catch(e:any){ return new Response("proxy error: "+String(e?.message||e),{status:502}); }
     }},
+    "/api/url-product": { GET: async (req: Request) => {
+      // Server-side product-URL parser: fetch the merchant page directly from Bun
+      // (no CORS, no external proxies) and extract full product info. Supports
+      // malloca.com (primary) + generic JSON-LD/meta fallback for any e-commerce URL.
+      // Returns: { name, sku, model, brand, price, priceText, currency, image, images,
+      //            description, summary, specs, features, category, link }
+      try {
+        const raw = new URL(req.url).searchParams.get("url") || "";
+        if (!raw || !/^https?:\/\//i.test(raw)) return new Response("missing url", { status: 400 });
+        const u = new URL(raw);
+        if (["http:", "https:"].indexOf(u.protocol) < 0) return new Response("bad proto", { status: 400 });
+        const resp = await fetch(u.href, {
+          headers: { "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 Chrome/120 Safari/537.36", "Accept-Language": "vi-VN,vi;q=0.9" },
+          signal: AbortSignal.timeout(40000),
+          redirect: "follow",
+        });
+        if (!resp.ok) return new Response("upstream " + resp.status, { status: 502 });
+        const html = await resp.text();
+        const out: Record<string, any> = { link: u.href };
+
+        // ── JSON-LD (Product blocks) ──
+        let ldProduct: any = null, ldDesc = "", ldPrice = 0;
+        const ldRe = /<script[^>]*type=["']application\/ld\+json["'][^>]*>([\s\S]*?)<\/script>/g;
+        let lm: RegExpExecArray | null;
+        while ((lm = ldRe.exec(html))) {
+          try {
+            const data = JSON.parse(lm[1]);
+            const walk = (n: any) => {
+              if (!n) return;
+              if (Array.isArray(n)) { n.forEach(walk); return; }
+              if (typeof n === "object") {
+                if (n["@type"] && String(n["@type"]).toLowerCase() === "product" && !ldProduct) ldProduct = n;
+                if (Array.isArray(n["@graph"])) n["@graph"].forEach(walk);
+                walk(n["@type"] === "Product" ? n : n["itemListElement"]);
+              }
+            };
+            walk(data);
+          } catch (e) { /* skip malformed */ }
+        }
+        if (ldProduct) {
+          if (!ldDesc && ldProduct.description) ldDesc = String(ldProduct.description).replace(/&[a-z]+;/g, " ").trim();
+          if (!ldPrice && ldProduct.offers && ldProduct.offers.price) ldPrice = Number(ldProduct.offers.price) || 0;
+        }
+
+        // ── <title> / og / meta ──
+        const title = (html.match(/<title[^>]*>([\s\S]*?)<\/title>/i) || [])[1]?.replace(/\s+/g, " ").trim() || "";
+        const meta = (name: string) => {
+          const m = html.match(new RegExp(`<meta[^>]+(?:name|property)=["']${name}["'][^>]+content=["']([^"']*)["']`, "i")) ||
+                    html.match(new RegExp(`<meta[^>]+content=["']([^"']*)["'][^>]+(?:name|property)=["']${name}["']`, "i"));
+          return m ? m[1] : "";
+        };
+        const ogTitle = meta("og:title") || "", ogDesc = meta("og:description") || "", ogImage = meta("og:image") || meta("twitter:image");
+
+        // ── name ──
+        let name = "";
+        if (ldProduct && ldProduct.name) name = String(ldProduct.name).trim();
+        if (!name && ogTitle) name = ogTitle.split(/[|–—-]/)[0].trim();
+        if (!name && title) name = title.split(/[|–—-]/)[0].trim();
+
+        // ── sku / model: prefer URL slug token (MS7743SI) then HTML "Mã sản phẩm" then JSON-LD sku ──
+        const slugToken = (u.pathname.split("/").filter(Boolean).pop() || "").match(/[A-Z]{2,}\d{3,}[A-Z]*/i);
+        let sku = slugToken ? slugToken[0].toUpperCase() : "";
+        const msku = html.match(/Mã (?:sản phẩm|SP)[^>]*>?\s*([^<]{2,30})</i) || html.match(/(?:data-sku|itemprop="sku")[^>]*["']([A-Z0-9-]{3,30})["']/i);
+        if (!sku && msku) sku = msku[1].trim();
+        if (!sku && ldProduct && ldProduct.sku) sku = String(ldProduct.sku).trim().slice(0, 20);
+
+        // ── brand ──
+        let brand = "";
+        if (ldProduct && ldProduct.brand) brand = (typeof ldProduct.brand === "string" ? ldProduct.brand : (ldProduct.brand.name || "")).trim();
+        if (!brand) { const mb = html.match(/Malloca/i); if (mb) brand = "Malloca"; }
+
+        // ── price ──
+        let price = 0, priceText = "";
+        const ogPrice = meta("og:price:amount");
+        if (ogPrice) price = Number(String(ogPrice).replace(/[^\d]/g, "")) || 0;
+        if (!price && ldPrice) price = ldPrice;
+        if (!price) {
+          const m = html.match(/(\d{1,3}(?:[.,]\d{3}){1,3})\s*(?:đ|₫|vnd)/i);
+          if (m) price = Number(m[1].replace(/[.,](?=\d{3}\b)/g, "").replace(/\./g, "")) || 0;
+        }
+        if (price) priceText = price.toLocaleString("vi-VN") + "đ";
+
+        // ── images: only real product images (path contains /products/) ──
+        const imgs: string[] = [];
+        const allImgs = html.match(/https?:\/\/[^"'\\\s]+\.(?:jpg|jpeg|png|webp)/gi) || [];
+        for (const im of allImgs) {
+          let c = im.replace(/\\\//g, "/");
+          if (!/\/products\//i.test(c)) continue;
+          c = c.replace(/\/thumb\/(?:grande|large|medium|small|compact)\//, "/");
+          c = c.split("?")[0];
+          if (imgs.indexOf(c) < 0) imgs.push(c);
+        }
+        if (!imgs.length && ogImage) imgs.push(ogImage);
+        const image = imgs[0] || ogImage || "";
+
+        // ── description / summary ──
+        let description = ldDesc || ogDesc || meta("description") || "";
+        description = description.replace(/\s+/g, " ").trim();
+
+        // ── specs: scan "Thông số kỹ thuật" / "Thông số" / "Thông tin" section for label:value rows ──
+        const specs: Record<string, string> = {};
+        const specSection = html.match(/(?:Thông số kỹ thuật|Thông số|Thông tin sản phẩm|Chi tiết sản phẩm|Đặc điểm nổi bật)[\s\S]{0,6000}/i);
+        if (specSection) {
+          const seg = specSection[0];
+          const rows = seg.match(/<[^>]{0,80}?>(?:<[^>]+>)*?\s*([A-ZÀ-Ỹ][^<>]{2,60}?)\s*(?:<\/[^>]+>)?\s*:?\s*<\/[^>]+>\s*(?:<[^>]+>)*?\s*([^<>]{1,200}?)\s*<\/[^>]+>/g) || [];
+          for (const rw of rows) {
+            const clean = rw.replace(/<[^>]+>/g, " ").replace(/\s+/g, " ").trim();
+            const mm = clean.match(/^(.{2,60}?)\s*[:：]\s*(.{1,180})$/);
+            if (mm && mm[1].length <= 60 && specs[mm[1].trim()] === undefined) specs[mm[1].trim()] = mm[2].trim();
+          }
+        }
+        // also try JSON-LD or property rows (dt/dd)
+        const ddRe = html.match(/<dt[^>]*>([\s\S]*?)<\/dt>\s*<dd[^>]*>([\s\S]*?)<\/dd>/gi);
+        if (ddRe && !Object.keys(specs).length) {
+          for (const blk of ddRe.slice(0, 40)) {
+            const c = blk.replace(/<[^>]+>/g, " ").replace(/\s+/g, " ").trim();
+            const mm = c.match(/^(.{2,60}?)\s*[:：]?\s*(.{1,180})$/);
+            if (mm && specs[mm[1].trim()] === undefined) specs[mm[1].trim()] = mm[2].trim();
+          }
+        }
+
+        // ── features: bullets under "Tính năng" / "Đặc điểm" ──
+        const features: string[] = [];
+        const featSec = html.match(/(?:Tính năng|Đặc điểm nổi bật|Điểm nổi bật)[\s\S]{0,4000}/i);
+        if (featSec) {
+          const bullets = featSec[0].match(/<li[^>]*>([\s\S]*?)<\/li>/gi) || featSec[0].match(/<p[^>]*>([\s\S]*?)<\/p>/gi) || [];
+          for (const b of bullets.slice(0, 12)) {
+            const t = b.replace(/<[^>]+>/g, " ").replace(/\s+/g, " ").trim();
+            if (t.length > 6 && t.length < 220 && features.indexOf(t) < 0) features.push(t);
+          }
+        }
+
+        // ── category ──
+        let category = "";
+        const bc = html.match(/<nav[^>]*aria-label=["']breadcrumb["'][\s\S]*?<\/nav>/i) || html.match(/class=["'][^"']*breadcrumb[^"']*["'][\s\S]{0,3000}?<\/nav>/i);
+        if (bc) {
+          const links = bc[0].match(/<a[^>]*>([\s\S]*?)<\/a>/gi) || [];
+          for (const lk of links) {
+            const t = lk.replace(/<[^>]+>/g, "").replace(/\s+/g, " ").trim();
+            if (t && t.length <= 60 && t.toLowerCase().indexOf("trang chủ") < 0 && t.toLowerCase().indexOf("malloca") < 0) category = t;
+          }
+        }
+
+        out.name = name || "Sản phẩm";
+        out.sku = sku; out.model = sku; out.brand = brand;
+        out.price = price; out.priceText = priceText; out.currency = "VND";
+        out.image = image; out.images = imgs;
+        out.description = description; out.summary = description.slice(0, 300);
+        out.specs = specs; out.features = features; out.category = category;
+        const body = JSON.stringify(out);
+        return new Response(body, { headers: { "Content-Type": "application/json; charset=utf-8", "Cache-Control": "public, max-age=600", "Access-Control-Allow-Origin": "*" } });
+      } catch (e: any) {
+        return new Response("proxy error: " + String(e?.message || e), { status: 502 });
+      }
+    }},
     "/api/img": { GET: async (req: Request) => {