Web search: fallback chain DDG → Bing RSS → Google HTML (HF egress hay bị block bởi DDG/Google HTML)

index.ts

@@ -1458,7 +1458,29 @@ const server = Bun.serve({
         }
         return res;
       };
-      // 2) Google HTML fallback (used when DDG returns nothing / is unreachable)
+      // 2) Bing RSS fallback (used when DDG returns nothing / is unreachable).
+      //    RSS endpoints are far more permissive than HTML scraping from HF egress.
+      const bingRss = async (): Promise<Array<any>> => {
+        const r = await fetch("https://www.bing.com/search?format=rss&q=" + encodeURIComponent(q), { headers: { "User-Agent": UA, "Accept-Language": "vi-VN,vi;q=0.9" }, signal: AbortSignal.timeout(12000) });
+        const h = await r.text();
+        const res: Array<any> = [];
+        const items = [...h.matchAll(/<item>([\s\S]*?)<\/item>/gi)];
+        for (const it of items) {
+          const body = it[1];
+          const tm = body.match(/<title>(?:<!\[CDATA\[)?([\s\S]*?)(?:\]\]>)?<\/title>/i);
+          const lm2 = body.match(/<link>(?:<!\[CDATA\[)?([\s\S]*?)(?:\]\]>)?<\/link>/i);
+          const dm = body.match(/<description>(?:<!\[CDATA\[)?([\s\S]*?)(?:\]\]>)?<\/description>/i);
+          const title = tm ? dec(tm[1]).trim() : '';
+          let href = lm2 ? lm2[1].trim() : '';
+          const snippet = dm ? dec(dm[1]).replace(/<[^>]+>/g, " ").replace(/\s+/g, " ").trim().slice(0, 160) : '';
+          if (!title || !href || !/^https?:\/\//i.test(href)) continue;
+          let source = '';
+          try { source = new URL(href).hostname.replace(/^www\./, ''); } catch (_) {}
+          res.push({ title, snippet, url: href, source });
+        }
+        return res;
+      };
+      // 3) Google HTML fallback (last resort — often blocked from HF egress)
       const ggl = async (): Promise<Array<any>> => {
         const r = await fetch("https://www.google.com/search?q=" + encodeURIComponent(q) + "&num=10&hl=vi", { headers: { "User-Agent": UA, "Accept-Language": "vi-VN,vi;q=0.9" }, signal: AbortSignal.timeout(12000) });
         const h = await r.text();
@@ -1507,11 +1529,14 @@ const server = Bun.serve({
       try {
         let results: Array<any> = [];
         try { results = await ddg(); } catch (_e) { results = []; }
+        if (!results.length) {
+          try { results = await bingRss(); } catch (_e) { results = []; }
+        }
         if (!results.length) {
           try { results = await ggl(); } catch (_e) { results = []; }
         }
         if (!results.length) return Response.json({ error: "Web search unreachable." }, { status: 502 });
-        return Response.json({ results: dedupeFn(results), engine: "ddg+google" });
+        return Response.json({ results: dedupeFn(results), engine: "ddg+google+bing" });
       } catch { return Response.json({ error: "Web search unreachable." }, { status: 502 }); }
     }},
     "/api/web/content": { GET: async (req: Request) => {