fix: harden scraper price parsing

2026-04-23 10:31:08 -04:00
parent 807849e257
commit 244a88e63c
4 changed files with 77 additions and 33 deletions
--- a/packages/core/src/scrapers/ebay.ts
+++ b/packages/core/src/scrapers/ebay.ts
@@ -240,7 +240,6 @@ function parseEbayListings(
            !text.includes("core") &&
            !text.includes("ram") &&
            !text.includes("ssd") &&
-            !/\d{4}/.test(text) && // Avoid years like "2024"
            !text.includes('"') // Avoid measurements
          ) {
            priceElement = el;
--- a/packages/core/src/scrapers/facebook.ts
+++ b/packages/core/src/scrapers/facebook.ts
@@ -890,30 +890,15 @@ export function parseFacebookAds(
            : priceObj.amount;
        cents = Math.round(dollars * 100);
      } else if (priceObj.amount_with_offset_in_currency != null) {
-        // Fallback: try to extract cents from amount_with_offset_in_currency
-        // This appears to use some exchange rate/multiplier format
-        const encodedAmount = Number(priceObj.amount_with_offset_in_currency);
-        if (!Number.isNaN(encodedAmount) && encodedAmount > 0) {
-          // Estimate roughly - this field doesn't contain real cents
-          // Use formatted_amount to get the actual dollar amount
-          if (priceObj.formatted_amount) {
-            const match = priceObj.formatted_amount.match(/[\d,]+\.?\d*/);
-            if (match) {
-              const dollars = Number.parseFloat(match[0].replace(/,/g, ""));
-              if (!Number.isNaN(dollars)) {
-                cents = Math.round(dollars * 100);
-              } else {
-                cents = encodedAmount; // fallback
-              }
-            } else {
-              cents = encodedAmount; // fallback
-            }
-          } else {
-            cents = encodedAmount; // fallback
-          }
-        } else {
-          continue; // Invalid price
-        }
+        if (!priceObj.formatted_amount) continue;
+
+        const match = priceObj.formatted_amount.match(/[\d,]+\.?\d*/);
+        if (!match) continue;
+
+        const dollars = Number.parseFloat(match[0].replace(/,/g, ""));
+        if (Number.isNaN(dollars)) continue;
+
+        cents = Math.round(dollars * 100);
      } else {
        continue; // No price available
      }
@@ -977,7 +962,9 @@ export function parseFacebookAds(
      };

      results.push(listingDetails);
-    } catch {}
+    } catch (error) {
+      console.warn("Failed to parse Facebook ad:", error);
+    }
  }

  return results;
--- a/packages/core/test/ebay-core.test.ts
+++ b/packages/core/test/ebay-core.test.ts
@@ -12,18 +12,14 @@ type IsExact<T, U> =
    : false;

 const getDefaultEbayItems = async () => fetchEbayItems("laptop");
-const unstableEbayItemsPromise = fetchEbayItems(
-  "laptop",
-  1000,
-  {},
-  { hideUnstableResults: true },
-);
+const getUnstableEbayItems = async () =>
+  fetchEbayItems("laptop", 1000, {}, { hideUnstableResults: true });
 type _EbayDefaultReturn = Assert<
  IsExact<Awaited<ReturnType<typeof getDefaultEbayItems>>, EbayListingDetails[]>
 >;
 type _EbayUnstableReturn = Assert<
  IsExact<
-    Awaited<typeof unstableEbayItemsPromise>,
+    Awaited<ReturnType<typeof getUnstableEbayItems>>,
    UnstableListingBuckets<EbayListingDetails>
  >
 >;
@@ -302,6 +298,38 @@ describe("eBay Scraper Cookie Handling", () => {
    ]);
  });

+  test("accepts higher fallback prices without price classes", async () => {
+    global.fetch = mock(() =>
+      Promise.resolve({
+        ok: true,
+        text: () =>
+          Promise.resolve(`
+            <html><body>
+              <li class="s-item">
+                <a href="/itm/123"></a>
+                <h3>Studio Microphone Bundle</h3>
+                <div>CA $2500.00</div>
+              </li>
+            </body></html>
+          `),
+      }),
+    ) as typeof fetch;
+
+    const results = await fetchEbayItems("microphone", 1000, {
+      keywords: ["microphone"],
+    });
+
+    expect(results).toEqual([
+      expect.objectContaining({
+        title: "Studio Microphone Bundle",
+        listingPrice: expect.objectContaining({
+          amountFormatted: "CA $2500.00",
+          cents: 250000,
+        }),
+      }),
+    ]);
+  });
+
  test("retains free items when the requested price range includes zero", async () => {
    global.fetch = mock(() =>
      Promise.resolve({
--- a/packages/core/test/facebook-core.test.ts
+++ b/packages/core/test/facebook-core.test.ts
@@ -1613,6 +1613,10 @@ describe("Facebook Marketplace Scraper Core Tests", () => {
      });

      test("should handle malformed ads gracefully", () => {
+        const originalWarn = console.warn;
+        const warnMock = mock(() => {});
+        console.warn = warnMock;
+
        const ads = [
          {
            node: {
@@ -1638,6 +1642,9 @@ describe("Facebook Marketplace Scraper Core Tests", () => {
        const results = parseFacebookAds(ads);
        expect(results).toHaveLength(1);
        expect(results[0].title).toBe("Valid Ad");
+        expect(warnMock).toHaveBeenCalledTimes(1);
+
+        console.warn = originalWarn;
      });

      test("parses formatted fallback prices with multiple commas", () => {
@@ -1667,6 +1674,29 @@ describe("Facebook Marketplace Scraper Core Tests", () => {
        ]);
      });

+      test("does not trust amount_with_offset_in_currency without a parseable formatted price", () => {
+        const ads = [
+          {
+            node: {
+              listing: {
+                id: "bad-offset",
+                marketplace_listing_title: "Broken Price Listing",
+                listing_price: {
+                  amount_with_offset_in_currency: "123456789",
+                  formatted_amount: "price unavailable",
+                  currency: "CAD",
+                },
+                is_live: true,
+              },
+            },
+          },
+        ];
+
+        const results = parseFacebookAds(ads);
+
+        expect(results).toEqual([]);
+      });
+
      test("keeps valid free search listings", () => {
        const ads = [
          {