Merge pull request #22 from austin1237/local-jobApi

local script now uses job api instead of writing to a file.

local/package.json

@@ -5,11 +5,13 @@
     },
     "devDependencies": {
         "@types/jest": "^29.5.12",
+        "axios-mock-adapter": "^1.22.0",
         "jest": "^29.7.0",
         "ts-jest": "^29.1.2",
         "typescript": "^5.3.3"
     },
     "dependencies": {
+        "axios": "^1.6.8",
         "puppeteer": "^22.4.0"
     }
 }

local/src/index.ts

@@ -1,15 +1,22 @@
 import puppeteer from 'puppeteer';
 import { scrap } from './scraper/sitea';
+import { JobService } from './jobService';
 
 const main = async () =>{
     let jobCount = 0;
     const link = process.env.LOCAL_SCRAPER_SITEA;
+    const jobApiEndpoint = process.env.JOB_API_ENDPOINT;
     if (!link) {
         console.error('LOCAL_SCRAPER_SITEA environment variable isnt set');
         return;
     }
+    if (!jobApiEndpoint) {
+        console.error('JOB_API_ENDPOINT environment variable isnt set');
+        return;
+    }
     const browser = await puppeteer.launch({ headless: false });
-    await scrap(browser, link, jobCount);
+    const jobService = new JobService(jobApiEndpoint);
+    await scrap(browser, link, jobCount, jobService);
     console.log('Scraping complete');
     await browser.close();
 }

local/src/jobService.ts

@@ -0,0 +1,27 @@
+import axios from 'axios';
+
+export interface Job {
+    title: string;
+    company: string;
+    keyword: string;
+    link: string;
+}
+
+export interface Response {
+    total: number;
+    uncached: number;
+    duplicates: number;
+}
+
+export class JobService {
+    private endpoint: string;
+
+    constructor(endpoint: string) {
+        this.endpoint = endpoint;
+    }
+
+    async sendJobs(jobs: Job[]): Promise<Response> {
+        const response = await axios.post(this.endpoint, { jobs });
+        return response.data as Response;
+    }
+}

local/src/scraper/sitea.ts

@@ -1,6 +1,6 @@
 import { Browser, Page } from 'puppeteer';
 import { evaluateJobInterest } from '../interest';
-import { writeToFile } from '../fileHandler';
+import { JobService, Job } from '../jobService';
 const urls: string[] = [];
 
 const didUrlChange = async (page: Page): Promise<boolean> => {
@@ -18,7 +18,8 @@ const didUrlChange = async (page: Page): Promise<boolean> => {
 };
 
 
-export const scrap = async(browser : Browser, link : string, jobCount : number) => {
+export const scrap = async(browser : Browser, link : string, jobCount : number, jobService: JobService) => {
+    let interestingJobs: Job[] = [];
     try{    
         const page = await browser.newPage();
         await page.goto(link);
@@ -54,16 +55,22 @@ export const scrap = async(browser : Browser, link : string, jobCount : number)
                         const pageUrl = urls[urls.length - 1]; // Get the current page URL
                         const jobCategory = evaluateJobInterest(jobTitle, companyName, jobDescriptionText);
 
-                        if(jobCategory){
-                            writeToFile(pageUrl, jobTitle, companyName, jobCategory);
-                            jobCount++;
-                            console.log(`Job found ${jobCount}`);
+                        if(jobCategory && jobTitle && companyName && pageUrl){
+                            interestingJobs.push({ title: jobTitle, company: companyName, keyword: jobCategory, link: pageUrl });
                         }
 
                         await new Promise(resolve => setTimeout(resolve, 5000)); // Sleep for 5 seconds to avoid bot detection
                     }
                 }
             }
+
+            // Send the jobs to the API
+            if (interestingJobs.length) {
+                const response = await jobService.sendJobs(interestingJobs);
+                jobCount += response.uncached;
+                console.log(`Total jobs: ${jobCount}`);
+                interestingJobs = []
+            }
 
             // Click on the "Next" button and wait for the next page to load
             const nextButton = await page.$('[data-testid="pagination-page-next"]');

local/tests/jobService.test.ts

@@ -0,0 +1,43 @@
+import axios from 'axios';
+import MockAdapter from 'axios-mock-adapter';
+import { JobService, Job, Response } from '../src/jobService';
+
+describe('JobService', () => {
+    let mockAxios: MockAdapter;
+    let jobService: JobService;
+
+    beforeEach(() => {
+        mockAxios = new MockAdapter(axios);
+        jobService = new JobService('http://mock-endpoint.com');
+    });
+
+    afterEach(() => {
+        mockAxios.reset();
+    });
+
+    it('should send jobs and return the response', async () => {
+        const jobs: Job[] = [
+            { title: 'Software Engineer', company: 'Company1', keyword: 'Go', link: 'http://example.com/job1' },
+            { title: 'Data Analyst', company: 'Company2', keyword: 'Python', link: 'http://example.com/job2' },
+            { title: 'Financial Advisor', company: 'Company3', keyword: 'Finance', link: 'http://example.com/job3' },
+        ];
+
+        const expectedResponse: Response = { total: 3, uncached: 1, duplicates: 2 };
+
+        mockAxios.onPost('http://mock-endpoint.com').reply(200, expectedResponse);
+
+        const response = await jobService.sendJobs(jobs);
+
+        expect(response).toEqual(expectedResponse);
+    });
+
+    it('should throw an error if the request fails', async () => {
+        const jobs: Job[] = [
+            { title: 'Software Engineer', company: 'Company1', keyword: 'Go', link: 'http://example.com/job1' },
+        ];
+
+        mockAxios.onPost('http://mock-endpoint.com').networkError();
+
+        await expect(jobService.sendJobs(jobs)).rejects.toThrow('Network Error');
+    });
+});