refactor: improve Kijiji scraping architecture

Introduces types, utilities, and a dedicated HTTP client for improved structure and maintainability.  Includes rate limit handling, retry logic, and error handling.
This commit is contained in:
2025-09-17 20:45:57 -04:00
parent 23dc9cae4c
commit b7a61423c3

View File

@@ -1,159 +1,234 @@
/* eslint-disable @typescript-eslint/no-explicit-any */
import { parseHTML } from "linkedom";
// ----------------------------- Types -----------------------------
type HTMLString = string;
type SearchListing = {
name: string;
listingLink: string;
};
interface ApolloSearchState {
[key: string]: {
[key: string]: unknown;
type ApolloRecord = Record<string, unknown>;
interface ApolloSearchItem {
url?: string;
title?: string;
[k: string]: unknown;
}
interface ApolloListingRoot {
url?: string;
title?: string;
description?: string;
price?: { amount?: number | string; currency?: string };
type?: string;
status?: string;
activationDate?: string;
endDate?: string;
metrics?: { views?: number | string };
location?: { address?: string | null };
[k: string]: unknown;
}
type ListingDetails = {
url: string;
title: string;
description?: string;
listingPrice?: {
amountFormatted: string;
cents?: number;
currency?: string;
};
listingType?: string;
listingStatus?: string;
creationDate?: string;
endDate?: string;
numberOfViews?: number;
address?: string | null;
};
}
interface ApolloListingState {
[key: string]: any;
}
// ----------------------------- Config -----------------------------
function formatCentsToCurrency(num: number | string, locale = "en-US") {
if (typeof num === "string") num = parseInt(num);
const numberInDollars = num / 100;
const REQUESTS_PER_SECOND = 1;
const DELAY_MS = Math.max(1, Math.floor(1000 / REQUESTS_PER_SECOND));
const BASE_URL = "https://www.kijiji.ca";
const SEARCH_QUERY = "playstation 5";
// ----------------------------- Utilities -----------------------------
/**
* Turns cents to localized currency string.
*/
function formatCentsToCurrency(
num: number | string | undefined,
locale = "en-US",
): string {
if (num == null) return "";
const cents = typeof num === "string" ? Number.parseInt(num, 10) : num;
if (Number.isNaN(cents)) return "";
const dollars = cents / 100;
const formatter = new Intl.NumberFormat(locale, {
minimumFractionDigits: 2,
maximumFractionDigits: 2,
useGrouping: true,
});
return formatter.format(numberInDollars);
return formatter.format(dollars);
}
const searchQuery = "playstation 5";
const REQUESTS_PER_SECOND = 1;
const DELAY_MS = 1000 / REQUESTS_PER_SECOND;
// const exampleSearchHTML = Bun.file("./example-kijiji-search.html");
// const exampleSearchHTMLData = await exampleSearchHTML.text();
function parseSearch(htmlString: string) {
const { document } = parseHTML(htmlString);
const nextData = document.getElementById("__NEXT_DATA__");
if (!nextData) {
console.error("Could not find __NEXT_DATA__ script element.");
return [];
function isRecord(value: unknown): value is Record<string, unknown> {
return typeof value === "object" && value !== null;
}
async function delay(ms: number): Promise<void> {
await new Promise((resolve) => setTimeout(resolve, ms));
}
class HttpError extends Error {
constructor(
message: string,
public readonly status: number,
public readonly url: string,
) {
super(message);
this.name = "HttpError";
}
}
// ----------------------------- HTTP Client -----------------------------
/**
Fetch HTML with a basic retry strategy and simple rate-limit delay between calls.
- Retries on 429 and 5xx
- Respects X-RateLimit-Reset when present (seconds)
*/
async function fetchHtml(
url: string,
opts?: {
maxRetries?: number;
retryBaseMs?: number;
onRateInfo?: (remaining: string | null, reset: string | null) => void;
},
): Promise<HTMLString> {
const maxRetries = opts?.maxRetries ?? 3;
const retryBaseMs = opts?.retryBaseMs ?? 500;
for (let attempt = 0; attempt <= maxRetries; attempt++) {
try {
if (!nextData.textContent) {
console.error("__NEXT_DATA__ element is empty!");
return [];
}
const jsonData = JSON.parse(nextData.textContent);
const apolloState: ApolloSearchState =
jsonData.props.pageProps.__APOLLO_STATE__;
const listingsKeys: string[] = [];
for (const key in apolloState) {
if (key.includes("Listing")) {
listingsKeys.push(key);
}
}
const searchListings: SearchListing[] = listingsKeys.map((key) => {
const listing = apolloState[key];
return {
listingLink: listing!.url,
name: listing!.title,
};
});
// console.log(searchListings);
return searchListings;
} catch (error) {
console.error("Error parsing __NEXT_DATA__:", error);
return [];
}
}
const makeKijijiRequest = async <T>(url: string): Promise<T> => {
console.log(`Making a request at ${new Date()}`);
const response = await fetch(url, {
const res = await fetch(url, {
method: "GET",
headers: {
accept:
"text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.7",
"accept-language": "en-GB,en-US;q=0.9,en;q=0.8",
"cache-control": "max-age=0",
priority: "u=0, i",
"sec-ch-ua": '"Not)A;Brand";v="8", "Chromium";v="138"',
"sec-ch-ua-mobile": "?0",
"sec-ch-ua-platform": '"Linux"',
"sec-fetch-dest": "document",
"sec-fetch-mode": "navigate",
"sec-fetch-site": "same-origin",
"sec-fetch-user": "?1",
"cache-control": "no-cache",
"upgrade-insecure-requests": "1",
"user-agent":
"Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120 Safari/537.36",
},
body: null,
method: "GET",
});
const rateLimitRemaining = response.headers.get("X-RateLimit-Remaining");
const rateLimitReset = response.headers.get("X-RateLimit-Reset");
const rateLimitRemaining = res.headers.get("X-RateLimit-Remaining");
const rateLimitReset = res.headers.get("X-RateLimit-Reset");
opts?.onRateInfo?.(rateLimitRemaining, rateLimitReset);
if (rateLimitRemaining !== null && rateLimitReset !== null) {
console.log(
`Rate limit remaining: ${rateLimitRemaining}, Reset in: ${rateLimitReset} seconds`,
if (!res.ok) {
// Respect 429 reset if provided
if (res.status === 429) {
const resetSeconds = rateLimitReset ? Number(rateLimitReset) : NaN;
const waitMs = Number.isFinite(resetSeconds)
? Math.max(0, resetSeconds * 1000)
: (attempt + 1) * retryBaseMs;
await delay(waitMs);
continue;
}
// Retry on 5xx
if (res.status >= 500 && res.status < 600 && attempt < maxRetries) {
await delay((attempt + 1) * retryBaseMs);
continue;
}
throw new HttpError(
`Request failed with status ${res.status}`,
res.status,
url,
);
}
const data: T = (await response.text()) as T;
return data;
};
async function delay(ms: number): Promise<void> {
return new Promise((resolve) => setTimeout(resolve, ms));
const html = await res.text();
// Respect per-request delay to keep at or under REQUESTS_PER_SECOND
await delay(DELAY_MS);
return html;
} catch (err) {
if (attempt >= maxRetries) throw err;
await delay((attempt + 1) * retryBaseMs);
}
}
// const exampleListing = await Bun.file("./examples/apollo_listing.json").json();
// const exampleListingApolloState =
// exampleListing.props.pageProps.__APOLLO_STATE__;
throw new Error("Exhausted retries without response");
}
const parseListing = (htmlString: string) => {
// ----------------------------- Parsing -----------------------------
/**
Extracts json.props.pageProps.__APOLLO_STATE__ safely from a Kijiji page HTML.
*/
function extractApolloState(htmlString: HTMLString): ApolloRecord | null {
const { document } = parseHTML(htmlString);
const nextData = document.getElementById("__NEXT_DATA__");
if (!nextData || !nextData.textContent) return null;
if (!nextData) {
console.error("Could not find __NEXT_DATA__ script element.");
return [];
}
if (!nextData.textContent) {
console.error("__NEXT_DATA__ element is empty!");
return [];
}
try {
const jsonData = JSON.parse(nextData.textContent);
const apolloState: ApolloListingState =
jsonData.props.pageProps.__APOLLO_STATE__;
const getListingId = (apolloState: { [key: string]: any }):
| string
| undefined => {
const apolloStateKeys = Object.keys(apolloState);
const key = apolloStateKeys.find((key) => key.includes("Listing"));
if (!key) return undefined;
return key;
};
const listingKey = getListingId(apolloState);
if (!listingKey) {
throw new Error("No listing key found in listing apolloState!");
const apollo = jsonData?.props?.pageProps?.__APOLLO_STATE__;
return isRecord(apollo) ? apollo : null;
} catch {
return null;
}
}
/**
Parse search page apollo state into SearchListing[].
Filters keys likely to be listing entities and ensures url/title exist.
*/
function parseSearch(htmlString: HTMLString): SearchListing[] {
const apolloState = extractApolloState(htmlString);
if (!apolloState) return [];
const results: SearchListing[] = [];
for (const [key, value] of Object.entries(apolloState)) {
// Heuristic: Kijiji listing keys usually contain "Listing"
if (!key.includes("Listing")) continue;
if (!isRecord(value)) continue;
const item = value as ApolloSearchItem;
if (typeof item.url === "string" && typeof item.title === "string") {
results.push({
listingLink: item.url.startsWith("http")
? item.url
: `${BASE_URL}${item.url}`,
name: item.title,
});
}
}
return results;
}
/**
Parse a listing page into a typed object.
*/
function parseListing(htmlString: HTMLString): ListingDetails | null {
const apolloState = extractApolloState(htmlString);
if (!apolloState) return null;
// Find the listing root key
const listingKey = Object.keys(apolloState).find((k) =>
k.includes("Listing"),
);
if (!listingKey) return null;
const root = apolloState[listingKey];
if (!isRecord(root)) return null;
const {
url,
@@ -165,66 +240,105 @@ const parseListing = (htmlString: string) => {
activationDate,
endDate,
metrics,
// attributes,
location,
} = apolloState[listingKey];
} = root as ApolloListingRoot;
const listingObject = {
url,
const cents = price?.amount != null ? Number(price.amount) : undefined;
const amountFormatted = formatCentsToCurrency(cents);
const numberOfViews =
metrics?.views != null ? Number(metrics.views) : undefined;
const listingUrl =
typeof url === "string"
? url.startsWith("http")
? url
: `${BASE_URL}${url}`
: "";
if (!listingUrl || !title) return null;
return {
url: listingUrl,
title,
description,
listingPrice: {
amount: formatCentsToCurrency(price.amount),
currency: price.currency,
},
listingPrice: amountFormatted
? {
amountFormatted,
cents: Number.isFinite(cents!) ? cents : undefined,
currency: price?.currency,
}
: undefined,
listingType: type,
listingStatus: status,
creationDate: activationDate,
endDate,
numberOfViews: metrics.views,
// condition: attributes.all.find(
// (attr: { [key: string]: unknown }) => attr.canonicalName === "condition",
// ).canonicalValues[0],
address: location.address,
numberOfViews: Number.isFinite(numberOfViews!) ? numberOfViews : undefined,
address: location?.address ?? null,
};
}
return listingObject;
};
// ----------------------------- Main -----------------------------
const searchHtml: string = await makeKijijiRequest(
`https://www.kijiji.ca/b-canada/${searchQuery}/k0l0?dc=true&view=list`,
async function main() {
const searchUrl = `${BASE_URL}/b-canada/${encodeURIComponent(SEARCH_QUERY)}/k0l0?dc=true&view=list`;
console.log(`Fetching search: ${searchUrl}`);
const searchHtml = await fetchHtml(searchUrl, {
onRateInfo: (remaining, reset) => {
if (remaining && reset) {
console.log(
`Search - Rate limit remaining: ${remaining}, reset in: ${reset}s`,
);
const searchResults = parseSearch(searchHtml);
// if (searchResults.length === 0) {
// throw new Error("Search didn't return an HTML!")
// }
// console.log(searchResults);
const fetchAllWithRateLimit = async (links: string[]) => {
const results: string[] = [];
for (const link of links) {
try {
const data: string = await makeKijijiRequest(link);
// console.log(data);
results.push(data);
} catch (error) {
console.error(`Failed to fetch data from ${link}:`, error);
}
await delay(DELAY_MS);
}
return results;
};
const listingsLinks: string[] = searchResults.map((item) => {
return item.listingLink;
},
});
// console.log(listingsLinks);
const searchResults = parseSearch(searchHtml);
if (searchResults.length === 0) {
console.warn("No search results parsed from page.");
return;
}
const fetchResults = await fetchAllWithRateLimit(listingsLinks);
// Deduplicate links
const listingLinks = Array.from(
new Set(searchResults.map((r) => r.listingLink)),
);
const itemsData = fetchResults.map((itemHtml) => parseListing(itemHtml));
console.log(
`Found ${listingLinks.length} listing links. Fetching details...`,
);
console.log(itemsData);
const items: ListingDetails[] = [];
for (const link of listingLinks) {
try {
const html = await fetchHtml(link, {
onRateInfo: (remaining, reset) => {
if (remaining && reset) {
console.log(
`Item - Rate limit remaining: ${remaining}, reset in: ${reset}s`,
);
}
},
});
const parsed = parseListing(html);
if (parsed) items.push(parsed);
} catch (err) {
if (err instanceof HttpError) {
console.error(`Failed to fetch ${link} - ${err.status} ${err.message}`);
} else {
console.error(
`Failed to fetch ${link} - ${String((err as Error)?.message || err)}`,
);
}
}
}
console.log(`Parsed ${items.length} listings.`);
console.log(items);
}
void main().catch((err) => {
console.error("Fatal error:", err);
process.exitCode = 1;
});