jobs-monorepo/packages/glassdoor-scraper/index.ts
2025-11-01 10:59:58 +05:00

287 lines
8 KiB
TypeScript

import puppeteer, { Browser, Page } from "puppeteer-core";
import { ChildProcess } from "child_process";
import * as dotenv from "dotenv";
import {
ChromeVersionResponse,
launchChromeWithDebugging,
randomDelay,
waitForChromeReady,
} from "./utils/browser";
import {
RabbitMQClient,
SearchQuery,
createRabbitMQClient,
} from "./utils/rabbitmq";
import { BrowserService } from "./services/browserService";
dotenv.config();
let globalBrowser: Browser | null = null;
let chrome: ChildProcess;
async function cleanup() {
console.log("Starting cleanup process...");
if (rabbitMQClient) {
console.log("Closing RabbitMQ connection");
await rabbitMQClient.close();
rabbitMQClient = null;
}
if (globalBrowser) {
console.log("Closing browser on app shutdown");
await globalBrowser.close();
globalBrowser = null;
}
if (chrome) {
chrome.kill("SIGINT");
}
console.log("Cleanup completed");
}
async function initializeBrowser(): Promise<{ browser: Browser; page: Page }> {
chrome = launchChromeWithDebugging();
await waitForChromeReady();
const response = await fetch("http://localhost:9222/json/version", {
headers: {
Origin: "",
},
});
if (!response.ok) {
throw new Error("Failed to connect to Chrome debugging port");
}
const chromeInstance = (await response.json()) as ChromeVersionResponse;
const { webSocketDebuggerUrl } = chromeInstance;
const browser = await puppeteer.connect({
browserWSEndpoint: webSocketDebuggerUrl,
defaultViewport: null,
});
globalBrowser = browser;
console.log("Connected to browser successfully");
let pages = await browser.pages();
console.log(`Found ${pages.length} pages`);
let existingcookies = await browser.cookies();
await browser.deleteCookie(...existingcookies);
let page: Page;
if (pages.length === 0) {
console.log("No pages found, creating a new page");
page = await browser.newPage();
} else {
page = pages[0];
}
await page.setUserAgent(
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36"
);
return { browser, page };
}
async function scrapeGlassdoorJobs(searchQuery: SearchQuery): Promise<void> {
console.log(
`Starting Glassdoor scrape for: ${searchQuery.keywords} in ${searchQuery.location}`
);
try {
const { page } = await initializeBrowser();
const browserService = new BrowserService(page);
// Build Glassdoor search URL
const searchUrl = buildGlassdoorSearchURL(searchQuery);
console.log(`Navigating to: ${searchUrl}`);
await page.goto(searchUrl, {
waitUntil: ["networkidle2", "domcontentloaded"],
timeout: 60000,
});
await randomDelay(1500, 2500);
await browserService.searchJobs(searchQuery.keywords, searchQuery.location);
const jobs = await browserService.extractJobListings();
console.log(`\n=== SCRAPING RESULTS ===`);
console.log(`Successfully extracted ${jobs.length} jobs from Glassdoor`);
if (jobs.length > 0) {
console.log(`\n=== JOB LISTINGS ===`);
jobs.forEach((job, index) => {
console.log(`\n${index + 1}. ${job.title}`);
console.log(` Company: ${job.company}`);
console.log(` Location: ${job.location}`);
console.log(` Age: ${job.age}`);
console.log(` Job ID: ${job.jobId}`);
if (job.skills.length > 0) {
console.log(` Skills: ${job.skills.join(", ")}`);
}
if (job.jobLink) {
console.log(` Job Link: ${job.jobLink}`);
}
if (job.description) {
console.log(
` Description: ${job.description.substring(0, 200)}...`
);
}
});
// Here you can process the jobs further:
// - Save to database
// - Send to another queue
// - Perform additional analysis
console.log(`\n=== END RESULTS ===`);
} else {
console.log("No jobs found for this search query");
}
await randomDelay(1500, 2500);
} catch (error) {
console.error("Error scraping Glassdoor jobs:", error);
throw error;
}
}
function getGlassdoorDomain(location: string): string {
const locationLower = location.toLowerCase();
// Country-specific Glassdoor domains
const countryDomains: { [key: string]: string } = {
// Asia Pacific
japan: "https://www.glassdoor.com/Job", // Glassdoor Japan uses .com
singapore: "https://www.glassdoor.sg/Job",
australia: "https://www.glassdoor.com.au/Job",
"hong kong": "https://www.glassdoor.com/Job", // Uses .com with HK location
india: "https://www.glassdoor.co.in/Job",
// Europe
"united kingdom": "https://www.glassdoor.co.uk/Job",
uk: "https://www.glassdoor.co.uk/Job",
germany: "https://www.glassdoor.de/Job",
france: "https://www.glassdoor.fr/Job",
netherlands: "https://www.glassdoor.nl/Job",
belgium: "https://www.glassdoor.be/Job",
austria: "https://www.glassdoor.at/Job",
switzerland: "https://www.glassdoor.ch/Job",
ireland: "https://www.glassdoor.ie/Job",
spain: "https://www.glassdoor.es/Job",
italy: "https://www.glassdoor.it/Job",
// North America
canada: "https://www.glassdoor.ca/Job",
mexico: "https://www.glassdoor.com.mx/Job",
// South America
brazil: "https://www.glassdoor.com.br/Job",
argentina: "https://www.glassdoor.com.ar/Job",
};
// Check for country matches
for (const [country, domain] of Object.entries(countryDomains)) {
if (locationLower.includes(country)) {
return domain;
}
}
// Check for city-specific mappings
const cityMappings: { [key: string]: string } = {
tokyo: "https://www.glassdoor.com/Job",
osaka: "https://www.glassdoor.com/Job",
kyoto: "https://www.glassdoor.com/Job",
london: "https://www.glassdoor.co.uk/Job",
manchester: "https://www.glassdoor.co.uk/Job",
berlin: "https://www.glassdoor.de/Job",
munich: "https://www.glassdoor.de/Job",
paris: "https://www.glassdoor.fr/Job",
amsterdam: "https://www.glassdoor.nl/Job",
toronto: "https://www.glassdoor.ca/Job",
vancouver: "https://www.glassdoor.ca/Job",
sydney: "https://www.glassdoor.com.au/Job",
melbourne: "https://www.glassdoor.com.au/Job",
singapore: "https://www.glassdoor.sg/Job",
mumbai: "https://www.glassdoor.co.in/Job",
bangalore: "https://www.glassdoor.co.in/Job",
delhi: "https://www.glassdoor.co.in/Job",
};
for (const [city, domain] of Object.entries(cityMappings)) {
if (locationLower.includes(city)) {
return domain;
}
}
// Default to US site
return "https://www.glassdoor.com/Job";
}
function buildGlassdoorSearchURL(query: SearchQuery): string {
const baseDomain = getGlassdoorDomain(query.location);
const baseUrl = `${baseDomain}/jobs.htm`;
console.log(
`Selected Glassdoor domain for location "${query.location}": ${baseDomain}`
);
const params = new URLSearchParams();
// if (query.Keywords) {
// params.set("sc.keyword", query.Keywords);
// }
// if (query.Location) {
// params.set("locT", "C");
// params.set("locId", query.Location);
// }
return `${baseUrl}?${params.toString()}`;
}
let rabbitMQClient: RabbitMQClient | null = null;
async function subscribeToGlassdoorQueue(): Promise<void> {
try {
rabbitMQClient = await createRabbitMQClient();
await rabbitMQClient.subscribe(async (searchQuery: SearchQuery) => {
await scrapeGlassdoorJobs(searchQuery);
});
} catch (error) {
console.error("Failed to set up RabbitMQ subscription:", error);
process.exit(1);
}
}
async function main() {
console.log("Starting Glassdoor scraper service...");
await subscribeToGlassdoorQueue();
}
process.on("SIGINT", async () => {
console.log("Received SIGINT signal");
await cleanup();
if (chrome) {
chrome.kill("SIGINT");
}
process.exit(0);
});
process.on("SIGTERM", async () => {
console.log("Received SIGTERM signal");
await cleanup();
if (chrome) {
chrome.kill("SIGTERM");
}
process.exit(0);
});
main().catch((error) => {
console.error("Fatal error:", error);
process.exit(1);
});