Merge pull request #1 from NarwhalChen/main

feat: adding feat of get ng job

Author: Sma1lboy

src/index.ts

@@ -1,13 +1,13 @@
 import { WechatyBuilder, Contact, Room, Message } from 'wechaty';
 import { ContactImpl } from 'wechaty/impls';
 import qrcodeTerminal from 'qrcode-terminal';
-import { InternJobProvider, Job } from './intern-job-provider';
+import { InternJobProvider, NGJobProvider, Job } from './intern-job-provider';
 import { jobWxBotConfig } from '../package.json';
 
 const wechaty = WechatyBuilder.build();
 let targetRooms: Room[] = [];
-const jobProvider = new InternJobProvider();
-
+const InternJob = new InternJobProvider();
+const NGJob = new NGJobProvider();
 function displayStartupBanner() {
   const banner = `
   ____    __  __    _    _     _     ____   ___ _____
@@ -32,14 +32,35 @@ function displayStartupBanner() {
   console.log('\x1b[35m%s\x1b[0m', '🔍 Scanning QR Code to log in...\n'); // Magenta color
 }
 
-async function sendJobUpdates() {
+async function sendInternJobUpdates() {
+  if (targetRooms.length === 0) {
+    console.log('No target rooms set');
+    return;
+  }
+  const newInternJobs = await InternJob.getNewJobs();
+  if (newInternJobs.length > 0) {
+    const messages = InternJob.formatJobMessages(newInternJobs);
+    for (const room of targetRooms) {
+      for (const message of messages) {
+        await room.say(message);
+        await new Promise((resolve) => setTimeout(resolve, 1000));
+      }
+    }
+    return true;
+  } else {
+    console.log('No new jobs found');
+    return false;
+  }
+}
+
+async function sendNGJobUpdates() {
   if (targetRooms.length === 0) {
     console.log('No target rooms set');
     return;
   }
-  const newJobs = await jobProvider.getNewJobs();
-  if (newJobs.length > 0) {
-    const messages = jobProvider.formatJobMessages(newJobs);
+  const newNGJobs = await NGJob.getNewJobs();
+  if (newNGJobs.length > 0) {
+    const messages = NGJob.formatJobMessages(newNGJobs);
     for (const room of targetRooms) {
       for (const message of messages) {
         await room.say(message);
@@ -76,8 +97,10 @@ wechaty
         '\x1b[36m%s\x1b[0m',
         `🚀 ${targetRooms.length} target room(s) found. Bot is ready!`,
       ); // Cyan color
-      setInterval(sendJobUpdates, jobWxBotConfig.minsCheckInterval * 60 * 1000);
-      sendJobUpdates();
+      setInterval(sendInternJobUpdates, jobWxBotConfig.minsCheckInterval * 60 * 1000);
+      setInterval(sendNGJobUpdates, jobWxBotConfig.minsCheckInterval * 60 * 1000);
+      sendInternJobUpdates();
+      sendNGJobUpdates();
     } else {
       console.log('\x1b[31m%s\x1b[0m', '❌ No target rooms found. Bot cannot operate.'); // Red color
     }
@@ -86,9 +109,14 @@ wechaty
     //TODO: need commands module
     //TOOD: need debug mode
     console.log(`Message received: ${message.text()}`);
-    if (message.text().toLowerCase() === 'jobs') {
-      if (!(await sendJobUpdates())) {
-        message.say('No new jobs found for Intern/NG roles');
+    if (message.text().toLowerCase() === '/internjobs') {
+      if (!(await sendInternJobUpdates())) {
+        message.say('No new jobs found for Intern roles');
+      }
+    }
+    if (message.text().toLowerCase() === '/ngjobs') {
+      if (!(await sendNGJobUpdates())) {
+        message.say('No new jobs found for ng roles');
       }
     }
   });

src/intern-job-provider.ts

@@ -30,7 +30,7 @@ export class InternJobProvider {
     if (!fs.existsSync(cacheDir)) {
       fs.mkdirSync(cacheDir, { recursive: true });
     }
-    this.sentJobsPath = path.join(cacheDir, 'sent_jobs.json');
+    this.sentJobsPath = path.join(cacheDir, 'sent_Intern_jobs.json');
     this.config = { ...jobWxBotConfig, jobsPerMessage: jobWxBotConfig.jobsPerMessage || 3 };
   }
 
@@ -188,7 +188,7 @@ export class InternJobProvider {
     const messages: string[] = [];
     for (let i = 0; i < jobs.length; i += this.config.jobsPerMessage) {
       const jobGroup = jobs.slice(i, i + this.config.jobsPerMessage);
-      let message = '📢 New Job Opportunities 📢\n\n';
+      let message = '📢 New Job Opportunities for INTERN📢\n\n';
       jobGroup.forEach((job) => {
         message += this.formatJobMessage(job);
       });
@@ -204,6 +204,204 @@ export class InternJobProvider {
 📍 Location: ${job.location}
 🔗 Apply: ${job.applicationLink}
 📅 Posted: ${job.datePosted}
+🧢 Type: INTERN
+`;
+    if (job.annotations.length > 0) {
+      message += `⚠️ Note: ${job.annotations.join(', ')}\n`;
+    }
+    message += '----------------------------\n';
+    return message;
+  }
+}
+
+
+export class NGJobProvider {
+  private sentJobsPath: string;
+  private config: Config;
+  private githubUrl: string =
+    'https://raw.githubusercontent.com/SimplifyJobs/New-Grad-Positions/dev/README.md';
+
+  constructor() {
+    const homeDir = os.homedir();
+    const cacheDir = path.join(homeDir, '.job-wx-bot');
+    if (!fs.existsSync(cacheDir)) {
+      fs.mkdirSync(cacheDir, { recursive: true });
+    }
+    this.sentJobsPath = path.join(cacheDir, 'sent_NG_jobs.json');
+    this.config = { ...jobWxBotConfig, jobsPerMessage: jobWxBotConfig.jobsPerMessage || 3 };
+  }
+
+  private async fetchJobsFromGithub(): Promise<string> {
+    try {
+      const response = await axios.get(this.githubUrl);
+      return response.data;
+    } catch (error) {
+      console.error('Error fetching data from GitHub:', error);
+      throw error;
+    }
+  }
+
+  private extractTableFromMarkdown(markdownContent: string): Job[] {
+    const tablePattern = /\| Company.*?\n([\s\S]*?)\n\n/;
+    const tableMatch = markdownContent.match(tablePattern);
+    if (!tableMatch) return [];
+
+    const tableContent = tableMatch[1];
+    const rows = tableContent.trim().split('\n');
+
+    let lastValidCompany = '';
+    return rows
+      .map((row) => {
+        const columns = row.split('|');
+        if (columns.length >= 6) {
+          const company = this.cleanCompanyName(columns[1].trim());
+          if (company !== '↳') {
+            lastValidCompany = company;
+          }
+          const role = columns[2].trim();
+          const annotations = this.getJobAnnotations(role);
+
+          // 如果职位已关闭，则跳过
+          if (annotations.includes('Closed')) {
+            return null;
+          }
+
+          return {
+            company: company === '↳' ? lastValidCompany : company,
+            role: this.cleanRole(role),
+            location: this.cleanLocation(columns[3].trim()),
+            applicationLink: this.extractApplicationLink(columns[4].trim()),
+            datePosted: columns[5].trim(),
+            annotations: annotations,
+          };
+        }
+        return null;
+      })
+      .filter((job): job is Job => job !== null);
+  }
+
+  private getJobAnnotations(role: string): string[] {
+    const annotations: string[] = [];
+    if (role.includes('🛂')) annotations.push('No Sponsorship');
+    if (role.includes('🇺🇸')) annotations.push('U.S. Citizenship Required');
+    if (role.includes('🔒')) annotations.push('Closed');
+    return annotations;
+  }
+
+  private cleanRole(role: string): string {
+    return role.replace(/[🛂🇺🇸🔒]/g, '').trim();
+  }
+
+  private cleanCompanyName(company: string): string {
+    return company.replace(/\*\*\[(.*?)\].*?\*\*/, '$1');
+  }
+
+  private cleanLocation(location: string): string {
+    return location
+      .replace(/<br\s*\/?>/gi, ', ')
+      .replace(/<[^>]*>/g, '')
+      .replace(/\s+/g, ' ')
+      .trim();
+  }
+
+  private extractApplicationLink(htmlString: string): string {
+    const linkPattern = /href="([^"]*)/;
+    const match = htmlString.match(linkPattern);
+    if (match) {
+      let link = match[1];
+
+      link = link.replace(/([?&]utm_source=Simplify)(&ref=Simplify)?($|&)/, '');
+      link = link.replace(/([?&])utm_source=Simplify(&ref=Simplify)?&/, '$1');
+      link = link.replace(/[?&]$/, '');
+
+      return link;
+    }
+    return 'No link available';
+  }
+  private filterJobsByDate(jobs: Job[], days: number): Job[] {
+    const today = new Date();
+    today.setHours(0, 0, 0, 0);
+    const cutoffDate = new Date(today);
+    cutoffDate.setDate(today.getDate() - days);
+
+    const months = [
+      'Jan',
+      'Feb',
+      'Mar',
+      'Apr',
+      'May',
+      'Jun',
+      'Jul',
+      'Aug',
+      'Sep',
+      'Oct',
+      'Nov',
+      'Dec',
+    ];
+
+    return jobs.filter((job) => {
+      const [month, day] = job.datePosted.split(' ');
+      const jobDate = new Date(today.getFullYear(), months.indexOf(month), parseInt(day));
+
+      if (jobDate > today) {
+        jobDate.setFullYear(jobDate.getFullYear() - 1);
+      }
+
+      return jobDate >= cutoffDate;
+    });
+  }
+
+  public async getNewJobs(): Promise<Job[]> {
+    const markdownContent = await this.fetchJobsFromGithub();
+    const allJobs = this.extractTableFromMarkdown(markdownContent);
+    const filteredJobs = this.filterJobsByDate(allJobs, this.config.maxDays);
+
+    let sentJobs: Job[] = [];
+    try {
+      sentJobs = JSON.parse(fs.readFileSync(this.sentJobsPath, 'utf8'));
+    } catch (error) {
+      console.log('No previous sent jobs found');
+    }
+
+    const newJobs = filteredJobs.filter(
+      (job) =>
+        !sentJobs.some(
+          (sentJob) =>
+            sentJob.company === job.company &&
+            sentJob.role === job.role &&
+            sentJob.datePosted === job.datePosted,
+        ),
+    );
+
+    if (newJobs.length > 0) {
+      sentJobs = [...sentJobs, ...newJobs];
+      fs.writeFileSync(this.sentJobsPath, JSON.stringify(sentJobs));
+    }
+
+    return newJobs;
+  }
+
+  public formatJobMessages(jobs: Job[]): string[] {
+    const messages: string[] = [];
+    for (let i = 0; i < jobs.length; i += this.config.jobsPerMessage) {
+      const jobGroup = jobs.slice(i, i + this.config.jobsPerMessage);
+      let message = '📢 New Job Opportunities for NG 📢\n\n';
+      jobGroup.forEach((job) => {
+        message += this.formatJobMessage(job);
+      });
+      messages.push(message);
+    }
+    return messages;
+  }
+
+  public formatJobMessage(job: Job): string {
+    let message = `
+🏢 Company: ${job.company}
+💼 Role: ${job.role}
+📍 Location: ${job.location}
+🔗 Apply: ${job.applicationLink}
+📅 Posted: ${job.datePosted}
+🧢 Type: NG
 `;
     if (job.annotations.length > 0) {
       message += `⚠️ Note: ${job.annotations.join(', ')}\n`;