old4ever/ca-marketplace-scraper
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: tighten ebay result parsing

Browse Source
This commit is contained in:
Dmytro Stanchiev
2026-04-23 05:13:40 -04:00
parent 0a0723a560
commit 08d59ab497
2 changed files with 107 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
packages/core/src/scrapers/ebay.ts
View File

@@ -43,7 +43,7 @@ const EBAY_PRICE_TEXT_RE = /^(?:\s*(?:CA|C)\s*\$|\s*[$£€¥])/u;
function canonicalizeEbayItemUrl(url: string): string { function canonicalizeEbayItemUrl(url: string): string {
try { try {
const parsed = new URL(url, "https://www.ebay.ca"); const parsed = new URL(url, "https://www.ebay.ca");
const match = parsed.pathname.match(/\/itm\/[^/?#]+/); const match = parsed.pathname.match(/\/itm\/(?:[^/?#]+\/)?\d+/);
return match ? `${parsed.origin}${match[0]}` : `${parsed.origin}${parsed.pathname}`; return match ? `${parsed.origin}${match[0]}` : `${parsed.origin}${parsed.pathname}`;
} catch { } catch {
return url; return url;
@@ -267,8 +267,7 @@ function parseEbayListings(
if ( if (
text && text &&
EBAY_PRICE_TEXT_RE.test(text) && EBAY_PRICE_TEXT_RE.test(text) &&
text.length < 50 && text.length < 50
!/\d{4}/.test(text)
) { ) {
actualPrices.push(el); actualPrices.push(el);
} }
@@ -512,7 +511,7 @@ export default async function fetchEbayItems(
// Filter by price range (additional safety check) // Filter by price range (additional safety check)
const filteredListings = listings.filter((listing) => { const filteredListings = listings.filter((listing) => {
const cents = listing.listingPrice?.cents; const cents = listing.listingPrice?.cents;
return cents && cents >= minPrice && cents <= maxPrice; return typeof cents === "number" && cents >= minPrice && cents <= maxPrice;
}); });
console.log(`Parsed ${filteredListings.length} eBay listings.`); console.log(`Parsed ${filteredListings.length} eBay listings.`);

104
packages/core/test/ebay-core.test.ts
View File

@@ -131,6 +131,48 @@ describe("eBay Scraper Cookie Handling", () => {
); );
}); });
test("deduplicates tracking variants of SEO-style item URLs", async () => {
global.fetch = mock(() =>
Promise.resolve({
ok: true,
text: () =>
Promise.resolve(`
<html><body>
<li class="s-item">
<a href="/itm/title-slug/1234567890?_trkparms=foo"></a>
<h3>Stable Laptop Bundle</h3>
<span class="s-item__price">CA $100.00</span>
</li>
<li class="s-item">
<a href="https://www.ebay.ca/itm/title-slug/1234567890?hash=item123"></a>
<h3>Stable Laptop Bundle</h3>
<span class="s-item__price">CA $100.00</span>
</li>
<li class="s-item">
<a href="https://www.ebay.ca/itm/title-slug/9999999999?hash=item999"></a>
<h3>Another Laptop Bundle</h3>
<span class="s-item__price">CA $110.00</span>
</li>
</body></html>
`),
}),
) as typeof fetch;
const results = await fetchEbayItems("laptop", 1000);
expect(results).toHaveLength(2);
expect(results[0]).toEqual(
expect.objectContaining({
url: "https://www.ebay.ca/itm/title-slug/1234567890?_trkparms=foo",
}),
);
expect(results[1]).toEqual(
expect.objectContaining({
url: "https://www.ebay.ca/itm/title-slug/9999999999?hash=item999",
}),
);
});
test("treats bare dollar prices as CAD on ebay.ca", async () => { test("treats bare dollar prices as CAD on ebay.ca", async () => {
global.fetch = mock(() => global.fetch = mock(() =>
Promise.resolve({ Promise.resolve({
@@ -189,6 +231,68 @@ describe("eBay Scraper Cookie Handling", () => {
]); ]);
}); });
test("prefers discounted Canadian prices that contain four consecutive digits", async () => {
global.fetch = mock(() =>
Promise.resolve({
ok: true,
text: () =>
Promise.resolve(`
<html><body>
<li class="s-item">
<a href="/itm/123"></a>
<h3>Stable Laptop Bundle</h3>
<span class="s-item__price">
<s>CA $1500.00</s>
<span>CA $1000.00</span>
</span>
</li>
</body></html>
`),
}),
) as typeof fetch;
const results = await fetchEbayItems("laptop", 1000);
expect(results).toEqual([
expect.objectContaining({
listingPrice: expect.objectContaining({
amountFormatted: "CA $1000.00",
cents: 100000,
}),
}),
]);
});
test("retains free items when the requested price range includes zero", async () => {
global.fetch = mock(() =>
Promise.resolve({
ok: true,
text: () =>
Promise.resolve(`
<html><body>
<li class="s-item">
<a href="/itm/123"></a>
<h3>Free Laptop Bundle</h3>
<span class="s-item__price">$0.00</span>
</li>
</body></html>
`),
}),
) as typeof fetch;
const results = await fetchEbayItems("laptop", 1000, {
minPrice: 0,
maxPrice: 0,
});
expect(results).toEqual([
expect.objectContaining({
title: "Free Laptop Bundle",
listingPrice: expect.objectContaining({ cents: 0 }),
}),
]);
});
test("returns results and unstableResults when unstable mode is enabled", async () => { test("returns results and unstableResults when unstable mode is enabled", async () => {
global.fetch = mock(() => global.fetch = mock(() =>
Promise.resolve({ Promise.resolve({