report.js

'use strict';
/**
 * report.js — 日报 / 周报生成（重构版）
 *
 * 核心改进：
 *   1. 周报生成前先批量补充未分类条目（batchClassify）
 *   2. 双层噪声过滤（filter.js isReportNoise）
 *   3. 周报 AI 先精选 top-30 再生成，输出接近手工整理水平
 *   4. 统一从 config.js 读配置
 */

const { db }                    = require('./db');
const { isReportNoise } = require('./filter');
const { batchClassify, generateDailySummary, generateWeeklySummary } = require('./ai');
const { callAI } = require('./ai-provider');
const { insightDAO } = require('./dao');
const { sendReportToWeCom }     = require('./wecom');
const { fetchMacroPanel, fetchMacroContext } = require('./macro-market');
const { sendWeeklyReportEmail } = require('./email-report');
const { getWikiContext }        = require('./wiki-context');
const { createClient }          = require('@supabase/supabase-js');
const { REPORT }                = require('./config');
require('dotenv').config();

const USE_SUPABASE = process.env.USE_SUPABASE === 'true';
let supabase = null;
if (USE_SUPABASE && process.env.SUPABASE_URL && process.env.SUPABASE_KEY) {
  supabase = createClient(process.env.SUPABASE_URL, process.env.SUPABASE_KEY);
}

// ── 时间工具 ──────────────────────────────────────────────────────────────────
function getTodayMidnightBJ() {
  const bj = new Date(Date.now() + 8 * 3600000);
  return Date.UTC(bj.getUTCFullYear(), bj.getUTCMonth(), bj.getUTCDate(), -8, 0, 0, 0);
}

function getWeekStartBJ() {
  const bj  = new Date(Date.now() + 8 * 3600000);
  const day = bj.getUTCDay();
  const diff = day === 0 ? 6 : day - 1;
  return Date.UTC(bj.getUTCFullYear(), bj.getUTCMonth(), bj.getUTCDate() - diff, -8, 0, 0, 0);
}

function formatDateBJ(ts) {
  return new Date(ts + 8 * 3600000).toISOString().split('T')[0];
}

// ── 数据获取 ──────────────────────────────────────────────────────────────────
async function fetchNewsForReport(since, limit = 500) {
  if (USE_SUPABASE && supabase) {
    try {
      // 先尝试按 alpha_score 排序（如果列存在）
      const query = supabase
        .from('news')
        .select('*')
        .gte('timestamp', since)
        .order('timestamp', { ascending: false })
        .limit(limit);

      const { data, error } = await query;

      if (error) {
        // 如果失败，可能是 alpha_score 列不存在，尝试只用 timestamp 排序
        console.log('[Report] Retrying without alpha_score sort...');
        const { data: data2, error: error2 } = await supabase
          .from('news')
          .select('*')
          .gte('timestamp', since)
          .order('timestamp', { ascending: false })
          .limit(limit);
        if (error2) { console.error('[Report] Supabase error:', error2.message); return []; }
        return data2 || [];
      }
      return data || [];
    } catch (e) {
      console.error('[Report] Supabase fetch error:', e.message);
      return [];
    }
  }
  if (!db) return [];
  // SQLite 使用 alpha_score（本地数据库有该列）
  return db.prepare(
    'SELECT * FROM news WHERE timestamp > ? ORDER BY alpha_score DESC, timestamp DESC LIMIT ?',
  ).all(since, limit);
}

// ── 批量补充分类（报告前预处理）───────────────────────────────────────────────
/**
 * 找出未分类的条目（business_category 为空/其他），批量调用 AI 填充。
 * 限制处理数量（MAX_ITEMS_FOR_AI），避免 token 超支。
 */
async function fillMissingCategories(rows) {
  const unclassified = rows.filter(r =>
    !r.business_category || r.business_category === '' || r.business_category === '其他',
  ).slice(0, REPORT.MAX_ITEMS_FOR_AI);

  if (unclassified.length === 0) return rows;

  console.log(`[Report] Batch classifying ${unclassified.length} unclassified items…`);
  const resultMap = await batchClassify(unclassified);

  // 把 AI 结果写回 rows（仅内存，不再写 DB 以减少干扰）
  const unclassifiedMap = new Map(unclassified.map((r, i) => [i, r]));
  resultMap.forEach((aiResult, idx) => {
    const item = unclassifiedMap.get(idx);
    if (item && aiResult) {
      Object.assign(item, {
        business_category:   aiResult.business_category   || item.business_category,
        competitor_category: aiResult.competitor_category || item.competitor_category,
        detail:              aiResult.detail              || item.detail,
        alpha_score:         aiResult.alpha_score         || item.alpha_score,
        impact:              aiResult.impact              || item.impact,
        bitv_action:         aiResult.bitv_action         || item.bitv_action,
        is_important:        aiResult.is_important        ?? item.is_important,
      });
    }
  });

  return rows;
}

// ── 竞品动态矩阵 ───────────────────────────────────────────────────────────────
/**
 * 按数据源（交易所/香港合规所）分组，生成本周各竞品主要动作的矩阵视图。
 * 只展示有实质动态（alpha_score >= 65 或 is_important）的来源。
 */
function buildCompetitorMatrix(rows) {
  // 香港合规所（优先展示）
  const HK_SOURCES = ['HashKeyExchange', 'HashKeyGroup', 'OSL', 'Exio', 'TechubNews'];
  // 头部离岸所
  const OFFSHORE_SOURCES = ['Binance', 'OKX', 'Bybit', 'Gate', 'MEXC', 'Bitget', 'HTX', 'KuCoin'];
  const ALL_COMPETITOR_SOURCES = [...HK_SOURCES, ...OFFSHORE_SOURCES];

  const SOURCE_LABEL = {
    'HashKeyExchange': 'HashKey Exchange',
    'HashKeyGroup':    'HashKey Group',
    'TechubNews':      'TechubNews（HashKey媒体）',
  };

  // 按来源分桶，只保留有价值的条目
  const bySource = {};
  rows
    .filter(r =>
      ALL_COMPETITOR_SOURCES.includes(r.source) &&
      (r.alpha_score >= 65 || r.is_important === 1) &&
      (r.detail?.length > 5 || r.title?.length > 8),
    )
    .forEach(r => {
      if (!bySource[r.source]) bySource[r.source] = [];
      bySource[r.source].push(r);
    });

  // 每个来源内部按 alpha_score 降序
  Object.values(bySource).forEach(items =>
    items.sort((a, b) => (b.alpha_score || 0) - (a.alpha_score || 0)),
  );

  // 排序：有动态的 HK 来源在前，离岸所在后；各组内按条目数降序
  const sortedSources = [
    ...HK_SOURCES.filter(s => bySource[s]).sort((a, b) => bySource[b].length - bySource[a].length),
    ...OFFSHORE_SOURCES.filter(s => bySource[s]).sort((a, b) => bySource[b].length - bySource[a].length),
  ];

  if (sortedSources.length === 0) return '';

  let matrix = '';
  sortedSources.forEach(source => {
    const items = bySource[source];
    const name  = SOURCE_LABEL[source] || source;
    const tag   = HK_SOURCES.includes(source) ? '🏛️' : '🌐';

    matrix += `\n${tag} **${name}** · ${items.length}条重要动态\n`;
    items.slice(0, 3).forEach(item => {
      const emoji = item.alpha_score >= 90 ? '🔥' : item.alpha_score >= 70 ? '⭐️' : '📡';
      const cat   = item.business_category ? ` \`${item.business_category}\`` : '';
      const score = item.alpha_score ? ` \`${item.alpha_score}\`` : '';
      matrix += `${emoji} ${item.title}${cat}${score}\n`;
      if (item.detail) matrix += `   > ${item.detail}\n`;
      if (item.bitv_action) matrix += `   > 💡 ${item.bitv_action}\n`;
    });
  });

  return matrix;
}

// ── 统计面板 ──────────────────────────────────────────────────────────────────
function buildStatsPanel(items, period = '今日') {
  const total      = items.length;
  const important  = items.filter(i => i.is_important === 1).length;
  const sources    = new Set(items.map(i => i.source)).size;
  const withDetail = items.filter(i => i.detail?.length > 5).length;
  return `📊 **${period}数据概览**\n> 抓取 **${total}** 条 | 重要 **${important}** 条 | AI 摘要 **${withDetail}** 条 | 来源 **${sources}** 个`;
}

// ═══════════════════════════════════════════════════════════════════════
//  日报
// ═══════════════════════════════════════════════════════════════════════
async function runDailyReport(dryRun = false) {
  console.log('[DailyReport] Start…');
  console.log(`[DailyReport] dryRun=${dryRun}, Time: ${new Date().toLocaleString('zh-CN', { timeZone: 'Asia/Shanghai' })}`);

  const since   = getTodayMidnightBJ();
  const dateStr = formatDateBJ(Date.now());
  console.log(`[DailyReport] Fetching news since: ${new Date(since).toISOString()}`);

  let rawRows = [];
  try {
    rawRows = await fetchNewsForReport(since);
    console.log(`[DailyReport] Fetched ${rawRows.length} raw rows`);
  } catch (err) {
    console.error('[DailyReport] Error fetching news:', err.message);
    return null;
  }

  // 噪声过滤：仅去除报告噪声源（不重新跑完整的 filterNewsItems，
  // 因为 DB 中的数据已经过滤过，重跑 validateTimestamp 会误删有效条目）
  let rows = rawRows.filter(r => !isReportNoise(r));
  console.log(`[DailyReport] After filtering: ${rows.length} rows`);

  if (rows.length === 0) {
    console.log('[DailyReport] No qualifying news today.');
    // 即使没有新闻，也发送一个空报告通知，避免用户以为系统故障
    if (!dryRun) {
      const { sendReportToWeCom } = require('./wecom');
      await sendReportToWeCom(`📋 **Web3Watch HK 行业日报 | ${dateStr}**\n\n今日暂无符合条件的行业动态。\n\n---\n*Web3Watch HK 战略分析引擎*`, '日报');
    }
    return null;
  }

  // 并行：补充分类 + 拉取宏观数据 + 读取历史趋势记忆
  [rows] = await Promise.all([
    fillMissingCategories(rows),
  ]);
  const [macroPanel, macroContext, recentTrends] = await Promise.all([
    fetchMacroPanel(),
    fetchMacroContext(),
    insightDAO.getRecent(4).catch(() => []),
  ]);

  // Pattern A: 注入行研知识库战略上下文
  const wikiContext = getWikiContext(rows);
  if (wikiContext) console.log('[DailyReport] Wiki context injected:', wikiContext.length, 'chars');

  // AI 总结（注入宏观上下文 + 历史趋势记忆 + 行研知识库）
  const aiInput   = rows.filter(r => r.detail || r.alpha_score >= 70);
  const aiSummary = await generateDailySummary(
    aiInput.length ? aiInput : rows.slice(0, 30),
    macroContext || '',
    recentTrends,
    wikiContext,
  );

  // 组装报告：头部 → 宏观背景 → AI总结
  let report = `📋 **Web3Watch HK 行业日报 | ${dateStr}**\n\n`;
  if (macroPanel) report += `${macroPanel}\n\n`;
  if (aiSummary) report += `---\n\n${aiSummary}\n\n`;
  report    += `---\n*Web3Watch HK | ${new Date().toLocaleString('zh-CN', { timeZone: 'Asia/Shanghai' })}*`;

  if (dryRun) {
    console.log('\n=== DAILY REPORT (DRY RUN) ===\n', report, '\n=== END ===\n');
    return report;
  }

  await sendReportToWeCom(report, '日报');
  console.log('[DailyReport] Done.');
  return report;
}

// ── 趋势提炼（记忆系统核心） ───────────────────────────────────────────────
/**
 * 从周报数据中提炼 1-3 条长期趋势/共识，存入记忆系统
 */
async function extractAndSaveTrends(rows) {
  if (!rows || rows.length < 5) return;

  const topItems = rows
    .filter(r => r.alpha_score >= 75)
    .slice(0, 30)
    .map(r => `- ${r.title}: ${r.detail || ''}`)
    .join('\n');

  const prompt = `你是一个行业研究主管。请根据以下本周的重点情报，提炼出 1-3 条“行业共识”或“重要长期趋势”。
这些趋势将作为系统的“长期记忆”，用于指导未来的分析。

本周重点：
${topItems}

请输出 JSON 数组，每个对象包含：
- trend_key: 趋势简称（如 "RWA合规化加速"）
- summary: 一句话深度总结趋势背后的逻辑
- evidence_count: 基于以上情报，该趋势被引证的次数

注意：只提炼具有长期影响（>1个月）的趋势，不要记录短期新闻。
示例：[{"trend_key":"稳定币监管落地","summary":"SFC 明确了法币稳定币发行商牌照框架，标志着香港进入合规支付新阶段。","evidence_count":3}]`;

  try {
    const text = await callAI([{ role: 'user', content: prompt }], { json: true, temperature: 0.3 });
    if (text) {
      let cleanJson = text;
      if (text.includes('```')) {
        const match = text.match(/```(?:json)?\s*([\s\S]*?)\s*```/);
        if (match) cleanJson = match[1];
      }
      const trends = JSON.parse(cleanJson);
      if (Array.isArray(trends)) {
        for (const trend of trends) {
          await insightDAO.saveInsight(trend);
        }
        console.log(`[Insight] Successfully recorded ${trends.length} new trends.`);
      }
    }
  } catch (e) {
    console.error('[Insight] Trend extraction failed:', e.message);
  }
}

// ═══════════════════════════════════════════════════════════════════════
//  周报（精选模式）
// ═══════════════════════════════════════════════════════════════════════
async function runWeeklyReport(dryRun = false) {
  console.log('[WeeklyReport] Start…');
  const weekStart = getWeekStartBJ();
  const startDate = formatDateBJ(weekStart);
  const endDate   = formatDateBJ(Date.now());
  const rawRows   = await fetchNewsForReport(weekStart);

  // 噪声过滤：仅去除报告噪声源
  let rows = rawRows.filter(r => !isReportNoise(r));

  if (rows.length === 0) {
    console.log('[WeeklyReport] No qualifying news this week.');
    return null;
  }

  console.log(`[WeeklyReport] ${rawRows.length} raw → ${rows.length} after filter`);

  // 补充分类
  rows = await fillMissingCategories(rows);

  // 统计
  const stats = {
    total:      rows.length,
    important:  rows.filter(r => r.alpha_score >= 70).length,
    sources:    new Set(rows.map(r => r.source)).size,
    categories: new Set(rows.filter(r => r.business_category).map(r => r.business_category)).size,
  };

  // AI 周报
  const aiInput   = rows.filter(r => r.alpha_score >= 70 || r.detail?.length > 5);
  const aiSummary = await generateWeeklySummary(aiInput.length >= 5 ? aiInput : rows.slice(0, 80), stats);

  // 竞品矩阵（详细版，仅用于邮件）
  const competitorMatrix = buildCompetitorMatrix(rows);

  // ── 企微版：只发 AI 总结，严格控制长度 ──────────────────────────────────────
  const wecomHeader = `📰 **Web3Watch HK 行业周报 | ${startDate} ~ ${endDate}**\n`;
  const wecomReport = wecomHeader + '\n' + (aiSummary || '本周暂无 AI 总结。');

  // ── 邮件版：完整内容（总结 + 竞品矩阵详情）────────────────────────────────
  let emailReport = `📰 **Web3Watch HK 行业周报 | ${startDate} ~ ${endDate}**\n\n`;
  if (aiSummary) emailReport += aiSummary + '\n\n';
  if (competitorMatrix.trim()) {
    emailReport += `---\n\n🏢 **竞品动态矩阵** | 本周各主要玩家行动汇总\n${competitorMatrix}\n`;
  }
  emailReport += `\n---\n*Web3Watch HK 战略分析引擎 | ${new Date().toLocaleString('zh-CN', { timeZone: 'Asia/Shanghai' })}*`;

  if (dryRun) {
    console.log('\n=== WEEKLY REPORT WeCom (DRY RUN) ===\n', wecomReport, '\n=== END ===\n');
    console.log('\n=== WEEKLY REPORT Email (DRY RUN) ===\n', emailReport, '\n=== END ===\n');
    return emailReport;
  }

  await sendReportToWeCom(wecomReport, '周报');

  // 邮件发送给领导层（SMTP 未配置时自动跳过）
  await sendWeeklyReportEmail(emailReport, startDate, endDate);

  console.log('[WeeklyReport] Done.');

  // 4. 提炼并保存行业趋势（记忆系统）
  if (!dryRun) {
    extractAndSaveTrends(rows).catch(e => console.error('[Insight] background task error:', e));
  }

  return emailReport;
}

module.exports = { runDailyReport, runWeeklyReport };