fix: align scraper unstable mode behavior

This commit is contained in:
2026-04-22 23:36:00 -04:00
parent c7fc8352ac
commit 08edfa8097
5 changed files with 54 additions and 18 deletions

View File

@@ -1,5 +1,6 @@
import { parseHTML } from "linkedom";
import type {
HTMLString,
UnstableListingBuckets,
UnstableListingModeOptions,
} from "../types/common";
@@ -114,7 +115,7 @@ function parseEbayListings(
if (!href.startsWith("http")) {
href = href.startsWith("//")
? `https:${href}`
: `https://www.ebay.com${href}`;
: `https://www.ebay.ca${href}`;
}
// Find the container - go up several levels to find the item container
@@ -397,6 +398,8 @@ export default async function fetchEbayItems(
} = {},
unstableMode: UnstableListingModeOptions = {},
) {
const requestsPerSecond = REQUESTS_PER_SECOND > 0 ? REQUESTS_PER_SECOND : 1;
const finalizeResults = (
listings: EbayListingDetails[],
): EbayListingDetails[] | UnstableListingBuckets<EbayListingDetails> => {
@@ -436,7 +439,7 @@ export default async function fetchEbayItems(
const searchUrl = `https://www.ebay.ca/sch/i.html?${urlParams.toString()}`;
const DELAY_MS = Math.max(1, Math.floor(1000 / REQUESTS_PER_SECOND));
const DELAY_MS = Math.max(1, Math.floor(1000 / requestsPerSecond));
console.log(`Fetching eBay search: ${searchUrl}`);

View File

@@ -1086,6 +1086,8 @@ export default async function fetchFacebookItems(
MAX_ITEMS = 25,
unstableMode: UnstableListingModeOptions = {},
) {
const requestsPerSecond = REQUESTS_PER_SECOND > 0 ? REQUESTS_PER_SECOND : 1;
const finalizeResults = (
listings: FacebookListingDetails[],
): FacebookListingDetails[] | UnstableListingBuckets<FacebookListingDetails> => {
@@ -1093,7 +1095,11 @@ export default async function fetchFacebookItems(
return listings.slice(0, MAX_ITEMS);
}
return classifyUnstableListings(listings.slice(0, MAX_ITEMS));
const classified = classifyUnstableListings(listings);
return {
results: classified.results.slice(0, MAX_ITEMS),
unstableResults: classified.unstableResults,
};
};
const cookies = await ensureFacebookCookies();
@@ -1107,7 +1113,7 @@ export default async function fetchFacebookItems(
);
}
const DELAY_MS = Math.max(1, Math.floor(1000 / REQUESTS_PER_SECOND));
const DELAY_MS = Math.max(1, Math.floor(1000 / requestsPerSecond));
// Encode search query for URL
const encodedQuery = encodeURIComponent(SEARCH_QUERY);

View File

@@ -725,6 +725,8 @@ export default async function fetchKijijiItems(
listingOptions: ListingFetchOptions = {},
unstableMode: UnstableListingModeOptions = {},
) {
const requestsPerSecond = REQUESTS_PER_SECOND > 0 ? REQUESTS_PER_SECOND : 1;
const finalizeResults = (
listings: DetailedListing[],
): DetailedListing[] | UnstableListingBuckets<DetailedListing> => {
@@ -735,7 +737,7 @@ export default async function fetchKijijiItems(
return classifyUnstableListings(listings);
};
const DELAY_MS = Math.max(1, Math.floor(1000 / REQUESTS_PER_SECOND));
const DELAY_MS = Math.max(1, Math.floor(1000 / requestsPerSecond));
// Load Kijiji cookies (optional - helps bypass bot detection)
const cookies = await loadCookiesOptional(
@@ -824,7 +826,7 @@ export default async function fetchKijijiItems(
progressBar?.start(totalProgress, currentProgress);
// Process in batches for controlled concurrency
const CONCURRENT_REQUESTS = REQUESTS_PER_SECOND * 2; // 2x rate for faster processing
const CONCURRENT_REQUESTS = Math.max(1, Math.floor(requestsPerSecond * 2)); // 2x rate for faster processing
const results: (DetailedListing | null)[] = [];
for (let i = 0; i < newListingLinks.length; i += CONCURRENT_REQUESTS) {