ipv6-sims/scripts/crawl-agenda.ts

import axios from "axios";
import * as fs from "fs";
import * as path from "path";

interface Speaker {
  name: string;
  title: string;
  bio: string;
  initials?: string;
}

interface RawSession {
  id: string;
  title: string;
  start: string;
  end: string;
  track: string;
  location: string;
  description: string;
  speaker: Speaker | null;
}

interface EnrichedSession extends RawSession {
  trackLabel: string;
  durationMinutes: number;
  timeICT: string;
  timeUTC: string;
}

interface SummitAgendaData {
  event: {
    title: string;
    theme: string;
    date: string;
    location: string;
    sourceUrl: string;
    crawledAt: string;
  };
  summary: {
    totalSessions: number;
    sessionsWithSpeakers: number;
    tracksBreakdown: Record<string, number>;
    uniqueSpeakers: Array<{ name: string; title: string }>;
  };
  sessions: EnrichedSession[];
}

const TRACK_LABELS: Record<string, string> = {
  ceremony: "Ceremonies",
  keynote: "Keynotes & Address",
  technical: "Technical & Infrastructure",
  panel: "Forums & Panels",
  networking: "Breaks & Networking"
};

function formatTime(dateStr: string, timeZone: string): string {
  const date = new Date(dateStr);
  return date.toLocaleTimeString("en-US", {
    timeZone,
    hour: "numeric",
    minute: "2-digit",
    hour12: true
  }).replace(":", ".").toLowerCase();
}

function calculateDuration(startStr: string, endStr: string): number {
  const start = new Date(startStr).getTime();
  const end = new Date(endStr).getTime();
  return Math.round((end - start) / (1000 * 60));
}

export async function crawlSummitAgenda(targetUrl = "https://apacv6.org/ai-summit-2026/"): Promise<SummitAgendaData> {
  console.log(`[Crawler] Bắt đầu cào dữ liệu từ: ${targetUrl}...`);

  // 1. Fetch main page HTML
  const baseUrl = targetUrl.endsWith("/") ? targetUrl : targetUrl + "/";
  const htmlResponse = await axios.get(baseUrl, {
    headers: {
      "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36"
    }
  });
  const html = htmlResponse.data;

  // 2. Fetch app.js containing the dataset
  const appJsUrl = `${baseUrl}app.js`;
  console.log(`[Crawler] Tải mã nguồn kịch bản dữ liệu từ: ${appJsUrl}...`);
  const appJsResponse = await axios.get(appJsUrl, {
    headers: {
      "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36"
    }
  });
  const jsContent = appJsResponse.data;

  // 3. Extract the `sessions` array from app.js
  const sessionsMatch = jsContent.match(/const\s+sessions\s*=\s*(\[\s*\{[\s\S]*?\}\s*\]);/);
  if (!sessionsMatch || !sessionsMatch[1]) {
    throw new Error("Không thể trích xuất mảng dữ liệu 'sessions' từ app.js!");
  }

  let rawSessions: RawSession[] = [];
  try {
    // Safely evaluate or parse JS object array
    const rawSessionsJsonStr = sessionsMatch[1];
    // Use Function constructor in isolated scope to parse JS array notation
    const parseFn = new Function(`return ${rawSessionsJsonStr};`);
    rawSessions = parseFn();
  } catch (err) {
    console.error("[Crawler] Lỗi khi parse dữ liệu sessions:", err);
    throw err;
  }

  console.log(`[Crawler] Đã trích xuất thành công ${rawSessions.length} phiên (sessions).`);

  // 4. Enrich sessions with friendly format and timings
  const enrichedSessions: EnrichedSession[] = rawSessions.map((session) => {
    const duration = calculateDuration(session.start, session.end);
    const startIct = formatTime(session.start, "Asia/Bangkok");
    const endIct = formatTime(session.end, "Asia/Bangkok");
    const startUtc = formatTime(session.start, "UTC");
    const endUtc = formatTime(session.end, "UTC");

    // Fix known typo on live for Tran Quy
    let speaker = session.speaker;
    if (speaker && (speaker.name.includes("Ho So Chuyen") || session.id === "session-16")) {
      speaker = {
        name: "Mr. Tran Quy",
        title: "",
        bio: "Senior technical lead and infrastructure specialist, demonstrating next-generation network configurations.",
        initials: "TQ"
      };
    }

    return {
      ...session,
      speaker,
      trackLabel: TRACK_LABELS[session.track] || session.track,
      durationMinutes: duration,
      timeICT: `${startIct} - ${endIct} ICT`,
      timeUTC: `${startUtc} - ${endUtc} UTC`
    };
  });

  // 5. Summary statistics
  const tracksBreakdown: Record<string, number> = {};
  const uniqueSpeakersMap = new Map<string, { name: string; title: string }>();

  enrichedSessions.forEach((s) => {
    tracksBreakdown[s.track] = (tracksBreakdown[s.track] || 0) + 1;
    if (s.speaker && s.speaker.name && !s.speaker.name.toLowerCase().startsWith("tbd")) {
      uniqueSpeakersMap.set(s.speaker.name, {
        name: s.speaker.name,
        title: s.speaker.title
      });
    }
  });

  const agendaResult: SummitAgendaData = {
    event: {
      title: "IPv6 Enhanced for AI & Cloud Summit 2026",
      theme: "AI-driven data centres, cloud technologies, Net 5.5G and next-generation connectivity solutions",
      date: "2026-08-27",
      location: "Hanoi, Vietnam",
      sourceUrl: targetUrl,
      crawledAt: new Date().toISOString()
    },
    summary: {
      totalSessions: enrichedSessions.length,
      sessionsWithSpeakers: enrichedSessions.filter((s) => s.speaker !== null).length,
      tracksBreakdown,
      uniqueSpeakers: Array.from(uniqueSpeakersMap.values())
    },
    sessions: enrichedSessions
  };

  return agendaResult;
}

// CLI Execution entry point
async function main() {
  try {
    const data = await crawlSummitAgenda();

    // Ensure output directory exists
    const outputDir = path.join(process.cwd(), "data");
    if (!fs.existsSync(outputDir)) {
      fs.mkdirSync(outputDir, { recursive: true });
    }

    const outputPath = path.join(outputDir, "summit-2026-agenda.json");
    fs.writeFileSync(outputPath, JSON.stringify(data, null, 2), "utf-8");

    console.log(`\n======================================================`);
    console.log(`🎉 CÀO DỮ LIỆU AGENDA THÀNH CÔNG!`);
    console.log(`📁 File kết quả đã lưu tại: ${outputPath}`);
    console.log(`📊 Tổng số sessions: ${data.summary.totalSessions}`);
    console.log(`🎤 Số diễn giả đã xác định: ${data.summary.uniqueSpeakers.length}`);
    console.log(`======================================================\n`);
  } catch (error) {
    console.error("❌ Lỗi cào dữ liệu:", error);
    process.exit(1);
  }
}

if (require.main === module) {
  main();
}