refactor: handle facebook route-aware failure states

packages/core/src/scrapers/facebook.ts

@@ -283,7 +283,7 @@ async function fetchHtml(
     onRateInfo?: (remaining: string | null, reset: string | null) => void;
     cookies?: string;
   },
-): Promise<HTMLString> {
+): Promise<{ html: HTMLString; responseUrl: string }> {
   const maxRetries = opts?.maxRetries ?? 3;
   const retryBaseMs = opts?.retryBaseMs ?? 500;
 
@@ -354,7 +354,7 @@ async function fetchHtml(
       const html = await res.text();
       // Respect per-request delay to keep at or under REQUESTS_PER_SECOND
       await delay(DELAY_MS);
-      return html;
+      return { html, responseUrl: res.url || url };
     } catch (err) {
       if (attempt >= maxRetries) throw err;
       await delay((attempt + 1) * retryBaseMs);
@@ -394,6 +394,10 @@ export function classifyFacebookResponse(
     return { kind: "unavailable" as const, authGated: false, unavailable: true };
   }
 
+  if (responseUrl.includes("/marketplace/item/")) {
+    return { kind: "item" as const, authGated: false, unavailable: false };
+  }
+
   if (htmlString.includes("XCometMarketplaceSearchController")) {
     return { kind: "search" as const, authGated: false, unavailable: false };
   }
@@ -1085,8 +1089,9 @@ export default async function fetchFacebookItems(
   console.log(`Using ${cookies.length} cookies for authentication`);
 
   let searchHtml: string;
+  let searchResponseUrl = searchUrl;
   try {
-    searchHtml = await fetchHtml(searchUrl, DELAY_MS, {
+    const response = await fetchHtml(searchUrl, DELAY_MS, {
       maxRetries: 3,
       onRateInfo: (remaining, reset) => {
         if (remaining && reset) {
@@ -1097,6 +1102,8 @@ export default async function fetchFacebookItems(
       },
       cookies: cookiesHeader,
     });
+    searchHtml = response.html;
+    searchResponseUrl = response.responseUrl;
   } catch (err) {
     if (err instanceof HttpError) {
       console.warn(
@@ -1112,6 +1119,24 @@ export default async function fetchFacebookItems(
     throw err;
   }
 
+  const classification = classifyFacebookResponse(searchHtml, searchResponseUrl);
+  if (classification.authGated) {
+    console.warn("Facebook marketplace search redirected to login. Cookies may be expired.");
+    return [];
+  }
+
+  if (classification.unavailable) {
+    console.warn("Facebook marketplace search returned an unavailable route.");
+    return [];
+  }
+
+  if (classification.kind !== "search") {
+    console.warn(
+      `Facebook marketplace search returned unexpected route kind: ${classification.kind}.`,
+    );
+    return [];
+  }
+
   const ads = extractFacebookMarketplaceData(searchHtml);
   if (!ads || ads.length === 0) {
     console.warn("No ads parsed from Facebook marketplace page.");
@@ -1163,8 +1188,9 @@ export async function fetchFacebookItem(
   console.log(`Fetching Facebook marketplace item: ${itemUrl}`);
 
   let itemHtml: string;
+  let itemResponseUrl = itemUrl;
   try {
-    itemHtml = await fetchHtml(itemUrl, 1000, {
+    const response = await fetchHtml(itemUrl, 1000, {
       onRateInfo: (remaining, reset) => {
         if (remaining && reset) {
           console.log(
@@ -1174,6 +1200,8 @@ export async function fetchFacebookItem(
       },
       cookies: cookiesHeader,
     });
+    itemHtml = response.html;
+    itemResponseUrl = response.responseUrl;
   } catch (err) {
     if (err instanceof HttpError) {
       console.warn(
@@ -1214,26 +1242,32 @@ export async function fetchFacebookItem(
     throw err;
   }
 
+  const classification = classifyFacebookResponse(itemHtml, itemResponseUrl);
+
+  if (classification.authGated) {
+    logExtractionMetrics(false, itemId);
+    console.warn(`Authentication failed for item ${itemId}. Cookies may be expired.`);
+    return null;
+  }
+
+  if (classification.unavailable || itemHtml.includes("This item has been sold")) {
+    logExtractionMetrics(false, itemId);
+    console.warn(`Item ${itemId} appears to be sold or removed from marketplace.`);
+    return null;
+  }
+
+  if (classification.kind !== "item") {
+    logExtractionMetrics(false, itemId);
+    console.warn(
+      `Item ${itemId} returned unexpected route kind: ${classification.kind}.`,
+    );
+    return null;
+  }
+
   const itemData = extractFacebookItemData(itemHtml);
   if (!itemData) {
     logExtractionMetrics(false, itemId);
 
-    const classification = classifyFacebookResponse(itemHtml, itemUrl);
-
-    if (classification.authGated) {
-      console.warn(
-        `Authentication failed for item ${itemId}. Cookies may be expired.`,
-      );
-      return null;
-    }
-
-    if (classification.unavailable || itemHtml.includes("This item has been sold")) {
-      console.warn(
-        `Item ${itemId} appears to be sold or removed from marketplace.`,
-      );
-      return null;
-    }
-
     console.warn(
       `No item data found in Facebook marketplace page for item ${itemId}. This may indicate:`,
     );