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

fix: tighten item price and pacing behavior

Browse Source
This commit is contained in:
Dmytro Stanchiev
2026-04-23 10:59:33 -04:00
parent 9c8643086a
commit 10c2856bf6
4 changed files with 135 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
packages/core/src/scrapers/facebook.ts
View File

@@ -984,13 +984,13 @@ export function parseFacebookItem(
const url = `https://www.facebook.com/marketplace/item/${item.id}`;
// Extract price information
let cents = 0;
let cents: number | undefined;
let currency = "CAD"; // Default
let amountFormatted = item.formatted_price?.text || "FREE";
let amountFormatted = item.formatted_price?.text;
if (item.listing_price) {
currency = item.listing_price.currency || "CAD";
if (item.listing_price.amount && item.listing_price.amount !== "0.00") {
if (item.listing_price.amount != null) {
const amount = Number.parseFloat(item.listing_price.amount);
if (!Number.isNaN(amount)) {
cents = Math.round(amount * 100);
@@ -1037,6 +1037,13 @@ export function parseFacebookItem(
listingType = "vehicle";
}
if (cents == null || !amountFormatted) {
if (!listingStatus || listingStatus === "ACTIVE") return null;
cents = 0;
amountFormatted = item.formatted_price?.text || "PRICE_UNAVAILABLE";
}
const listingDetails: FacebookListingDetails = {
url,
title,

16
packages/core/src/scrapers/kijiji.ts
View File

@@ -889,15 +889,19 @@ export default async function fetchKijijiItems(
progressBar?.start(totalProgress, currentProgress);
// Process in batches for controlled concurrency
const CONCURRENT_REQUESTS = 1;
const CONCURRENT_REQUESTS = Math.max(1, Math.floor(requestsPerSecond));
const results: (DetailedListing | null)[] = [];
for (let i = 0; i < newListingLinks.length; i += CONCURRENT_REQUESTS) {
const batch = newListingLinks.slice(i, i + CONCURRENT_REQUESTS);
const batchPromises = batch.map(async (link) => {
const batchPromises = batch.map(async (link, batchIndex) => {
try {
const html = await fetchHtml(link, DELAY_MS, {
// Per-request delay keeps detail fetches within REQUESTS_PER_SECOND.
if (batchIndex > 0) {
await new Promise((resolve) => setTimeout(resolve, DELAY_MS * batchIndex));
}
const html = await fetchHtml(link, 0, {
// Staggered starts keep request pacing within REQUESTS_PER_SECOND.
onRateInfo: (remaining, reset) => {
if (remaining && reset) {
console.log(
@@ -936,6 +940,10 @@ export default async function fetchKijijiItems(
const batchResults = await Promise.all(batchPromises);
results.push(...batchResults);
if (i + CONCURRENT_REQUESTS < newListingLinks.length) {
await new Promise((resolve) => setTimeout(resolve, DELAY_MS));
}
}
allListings.push(

19
packages/core/test/facebook-core.test.ts
View File

@@ -1532,10 +1532,21 @@ describe("Facebook Marketplace Scraper Core Tests", () => {
};
const result = parseFacebookItem(item);
expect(result).not.toBeNull();
expect(result?.title).toBe("Minimal Item");
expect(result?.description).toBeUndefined();
expect(result?.seller).toBeUndefined();
expect(result).toBeNull();
});
test("returns null when item price data is present but unparseable", () => {
const item = {
id: "456b",
__typename: "GroupCommerceProductItem" as const,
marketplace_listing_title: "Broken Price Item",
formatted_price: { text: "price unavailable" },
listing_price: { amount: "not-a-number", currency: "CAD" },
};
const result = parseFacebookItem(item);
expect(result).toBeNull();
});
test("should identify vehicle listings", () => {

98
packages/core/test/kijiji-core.test.ts
View File

@@ -428,6 +428,104 @@ describe("fetchKijijiItems", () => {
expect(maxActiveDetailRequests).toBe(1);
});
test("allows bounded concurrency to scale with REQUESTS_PER_SECOND", async () => {
const searchHtml = `
<html>
<script id="__NEXT_DATA__" type="application/json">
${JSON.stringify({
props: {
pageProps: {
__APOLLO_STATE__: {
"Listing:1": { url: "/v-one/k0l0", title: "One" },
"Listing:2": { url: "/v-two/k0l0", title: "Two" },
},
},
},
})}
</script>
</html>
`;
const listingHtml = (title: string, slug: string) => `
<html>
<script id="__NEXT_DATA__" type="application/json">
${JSON.stringify({
props: {
pageProps: {
__APOLLO_STATE__: {
"Listing:detail": {
url: `/${slug}`,
title,
price: { amount: 10000, currency: "CAD", type: "FIXED" },
type: "OFFER",
status: "ACTIVE",
},
},
},
},
})}
</script>
</html>
`;
let activeDetailRequests = 0;
let maxActiveDetailRequests = 0;
global.fetch = mock(async (input: string | URL | Request) => {
const url = typeof input === "string" ? input : input.toString();
if (url.includes("/k0c0l1700272")) {
return {
ok: true,
text: () => Promise.resolve(searchHtml),
headers: { get: () => null },
url,
};
}
activeDetailRequests++;
maxActiveDetailRequests = Math.max(
maxActiveDetailRequests,
activeDetailRequests,
);
await new Promise((resolve) => setTimeout(resolve, 300));
activeDetailRequests--;
if (url.endsWith("/v-one/k0l0")) {
return {
ok: true,
text: () => Promise.resolve(listingHtml("One", "v-one/k0l0")),
headers: { get: () => null },
url,
};
}
if (url.endsWith("/v-two/k0l0")) {
return {
ok: true,
text: () => Promise.resolve(listingHtml("Two", "v-two/k0l0")),
headers: { get: () => null },
url,
};
}
throw new Error(`Unexpected URL: ${url}`);
}) as typeof fetch;
const results = await fetchKijijiItems(
"phone",
4,
"https://www.kijiji.ca",
{ maxPages: 1 },
);
expect(results).toHaveLength(2);
expect(maxActiveDetailRequests).toBeGreaterThan(1);
expect(maxActiveDetailRequests).toBeLessThanOrEqual(4);
});
test("classifies the filtered Kijiji result set in unstable mode", async () => {
const searchHtml = `
<html>