weflow-cli 1.5.0

package/scripts/chat_report.py

@@ -0,0 +1,397 @@
+#!/usr/bin/env python3
+"""
+微信聊天月报 — 汇总指定联系人聊天记录，DeepSeek AI 分析任务和回复。
+
+用法:
+  python scripts/chat_report.py --month 2026-04
+  python scripts/chat_report.py --month 2026-04 --talker 张三 --talker 李老师
+  python scripts/chat_report.py --month 2026-04 --from-whitelist
+  python scripts/chat_report.py --month 2026-04 --talker 张三 --李老师 --家人群
+  python scripts/chat_report.py --month 2026-04 --no-ai
+
+输出: output/report-2026-04.md
+"""
+
+import sys, os, json, re, time, random, hashlib, argparse
+from datetime import datetime, timezone, timedelta
+from pathlib import Path
+from collections import defaultdict, Counter
+
+try:
+    from sqlcipher3 import dbapi2 as sqlcipher
+except ImportError:
+    print("请安装: pip install sqlcipher3"); sys.exit(1)
+
+from _utils import load_config, decrypt_lock, get_db_config, call_deepseek
+
+OUTPUT_ROOT = 'output'
+AI_DELAY = (2, 4)  # AI 调用间隔 (min, max) 秒
+
+# 任务关键词预筛 — 只有命中这些的联系人才调 AI
+TASK_KEYWORDS = ['帮忙', '做一下', '问个', '帮我', '麻烦', '有空吗', '方便吗',
+                 '能不能', '可以吗', '好不好', '行吗', '对吗', '怎么', '什么时候',
+                 '发给我', '看一下', '改一下', '整理', '准备', '安排', '通知',
+                 '?', '？']
+
+# ====== Database ======
+
+def open_db(db_path, key_hex, salt_hex):
+    raw_key = f"x'{key_hex}{salt_hex}'"
+    conn = sqlcipher.connect(db_path)
+    c = conn.cursor()
+    c.execute(f'PRAGMA key = "{raw_key}";')
+    c.execute("SELECT count(*) FROM sqlite_master")
+    return conn
+
+
+def get_sessions(conn):
+    """Get active chat sessions from Name2Id (NT format)."""
+    c = conn.cursor()
+    try:
+        c.execute("SELECT user_name FROM Name2Id WHERE is_session = 1")
+        sessions = [{'username': r[0]} for r in c.fetchall()]
+        return sessions
+    except:
+        pass
+    # Fallback: scan Msg_ tables
+    c.execute("SELECT name FROM sqlite_master WHERE type='table' AND name LIKE 'Msg_%'")
+    return [{'username': t, 'table': t} for t in [r[0] for r in c.fetchall()]]
+
+
+def get_name_map(contact_db, contact_key, contact_salt):
+    name_map = {}
+    if not contact_db or not contact_key or not os.path.exists(contact_db):
+        return name_map
+    try:
+        conn = open_db(contact_db, contact_key, contact_salt)
+        c = conn.cursor()
+        c.execute("SELECT username, COALESCE(NULLIF(remark,''), NULLIF(nick_name,''), username) FROM contact")
+        for r in c.fetchall(): name_map[r[0]] = r[1]
+        conn.close()
+    except: pass
+    return name_map
+
+
+def get_messages(conn, talker, start_ts, end_ts, sender_map, name_map, own_wxid):
+    """Get text messages for a talker within time range."""
+    tbl = f"Msg_{hashlib.md5(talker.encode()).hexdigest()}"
+    c = conn.cursor()
+    try:
+        c.execute(f'SELECT COUNT(*) FROM sqlite_master WHERE name="{tbl}"')
+        if c.fetchone()[0] == 0: return []
+    except: return []
+
+    c.execute(f'SELECT local_id, local_type, create_time, real_sender_id, message_content FROM "{tbl}" WHERE create_time >= ? AND create_time <= ? ORDER BY create_time ASC', (start_ts, end_ts))
+    rows = c.fetchall()
+
+    msgs = []
+    for row in rows:
+        lt, ct, sid = row[1] or 0, row[2] or 0, row[3] or 0
+        username = sender_map.get(sid, '')
+        is_self = False
+        if own_wxid:
+            is_self = (username == own_wxid)
+            if not is_self:
+                p = own_wxid.rsplit('_', 1)
+                if len(p) == 2 and len(p[1]) == 4 and p[1].isalnum():
+                    is_self = (username == p[0])
+
+        content = row[4] if isinstance(row[4], str) else (f'[{lt}类型]' if lt != 1 else '')
+        sender = '我' if is_self else name_map.get(username, username)
+        msgs.append({'time': ct, 'is_self': is_self, 'sender': sender, 'content': content, 'type': lt})
+    return msgs
+
+
+def has_tasks(msgs):
+    """Quick keyword check: does this conversation likely contain tasks?"""
+    text = ' '.join(m['content'][:100] for m in msgs if not m['is_self'] and m['type'] == 1)
+    return any(kw in text for kw in TASK_KEYWORDS)
+
+
+# ====== AI ======
+
+TALKER_PROMPT = """分析微信聊天记录，识别任务和请求。
+
+格式：[时间] 发送者: 内容
+规则：只关注对方（非"我"）的任务/请求/问句，判断我是否回复、紧急度、完成情况。
+
+返回JSON（不要其他内容）：
+{"tasks": [{"content":"任务摘要","time":"消息时间","replied":true,"reply":"回复","urgency":"高/中/低","status":"完成/进行中/未回复"}], "summary": "总结"}
+
+==MESSAGES==
+"""
+
+SUMMARY_PROMPT = """根据以下联系人分析结果生成月度总结。统计任务数、已/未回复，按紧急度列出未回复项，给出3条行动建议。
+
+格式：
+【任务统计】共 N 项，已回复 M，未回复 K
+【紧急事项】
+- 来源 | 任务 | 逾期
+【建议】
+1. ...
+2. ...
+3. ...
+
+==TALKER_DATA==
+"""
+
+
+# ====== Main ======
+
+def fmt_time(ts): return datetime.fromtimestamp(ts, tz=timezone(timedelta(hours=8))).strftime('%m-%d %H:%M')
+def fmt_date(ts): return datetime.fromtimestamp(ts, tz=timezone(timedelta(hours=8))).strftime('%Y-%m-%d')
+
+
+def main():
+    sys.stdout.reconfigure(encoding='utf-8', errors='replace')
+    p = argparse.ArgumentParser(description='微信聊天月报')
+    p.add_argument('--month'); p.add_argument('--year'); p.add_argument('--start'); p.add_argument('--end')
+    p.add_argument('--talker', action='append', help='指定联系人（可多选）')
+    p.add_argument('--from-whitelist', action='store_true', help='使用白名单中的联系人')
+    p.add_argument('--api-key'); p.add_argument('--no-ai', action='store_true')
+    p.add_argument('--output', '-o', default=OUTPUT_ROOT)
+    args = p.parse_args()
+
+    # Time range
+    tz = timezone(timedelta(hours=8))
+    if args.month:
+        y, m = map(int, args.month.split('-'))
+        sd = datetime(y, m, 1, tzinfo=tz)
+        ed = (datetime(y, m+1, 1, tzinfo=tz) if m < 12 else datetime(y+1, 1, 1, tzinfo=tz)) - timedelta(seconds=1)
+        period = args.month
+    elif args.year:
+        y = int(args.year)
+        sd = datetime(y, 1, 1, tzinfo=tz); ed = datetime(y, 12, 31, 23, 59, 59, tzinfo=tz); period = args.year
+    elif args.start and args.end:
+        sd = datetime.strptime(args.start, '%Y-%m-%d').replace(tzinfo=tz)
+        ed = datetime.strptime(args.end, '%Y-%m-%d').replace(hour=23, minute=59, second=59, tzinfo=tz)
+        period = f"{args.start}~{args.end}"
+    else:
+        now = datetime.now(tz)
+        fm = (now.year, now.month-1) if now.month > 1 else (now.year-1, 12)
+        sd = datetime(fm[0], fm[1], 1, tzinfo=tz)
+        ed = datetime(fm[0], fm[1]+1, 1, tzinfo=tz) - timedelta(seconds=1) if fm[1] < 12 else datetime(fm[0]+1, 1, 1, tzinfo=tz) - timedelta(seconds=1)
+        period = f"{fm[0]}-{fm[1]:02d}"
+
+    s_ts, e_ts = int(sd.timestamp()), int(ed.timestamp())
+
+    api_key = args.api_key or os.environ.get('DEEPSEEK_API_KEY', '')
+    if not args.no_ai and not api_key:
+        print('[ERROR] 需要 DeepSeek API key'); sys.exit(1)
+
+    config = load_config()
+    db = get_db_config(config)
+    if not db['nt_db'] or not db['nt_key']:
+        print('[ERROR] 未配置，先运行: weflow-cli init'); sys.exit(1)
+
+    print(f'=== 微信聊天月报 {period} ===')
+    print(f'时间: {fmt_date(s_ts)} ~ {fmt_date(e_ts)}')
+    print(f'AI: {"关闭" if args.no_ai else "DeepSeek V4"}')
+
+    # Resolve talker filter
+    target_wxids = set()
+    if args.talker:
+        # Need to resolve nicknames to wxids
+        conn = open_db(db['nt_db'], db['nt_key'], db['nt_salt'])
+        name_map = get_name_map(db['contact_db'], db['contact_key'], db['contact_salt'])
+        c = conn.cursor()
+        c.execute("SELECT user_name FROM Name2Id WHERE is_session = 1")
+        all_sessions = {r[0]: name_map.get(r[0], r[0]) for r in c.fetchall()}
+        conn.close()
+        for t in args.talker:
+            t = t.strip()
+            found = False
+            for wxid, name in all_sessions.items():
+                if t in (wxid, name) or t.lower() in name.lower():
+                    target_wxids.add(wxid); found = True; break
+            if not found:
+                # Try as wxid directly
+                target_wxids.add(t)
+        if target_wxids:
+            print(f'选定联系人: {len(target_wxids)} 人')
+    if args.from_whitelist:
+        wl = config.get('whitelist', [])
+        if wl: target_wxids.update(wl); print(f'白名单: {len(wl)} 人')
+
+    # Connect
+    print('\n连接数据库...')
+    conn = open_db(db['nt_db'], db['nt_key'], db['nt_salt'])
+    name_map = get_name_map(db['contact_db'], db['contact_key'], db['contact_salt'])
+    own_wxid = config.get('wxid', '')
+
+    # Get sessions
+    print('获取会话...')
+    sessions = get_sessions(conn)
+    if target_wxids:
+        sessions = [s for s in sessions if s['username'] in target_wxids]
+
+    print(f'活跃会话: {len(sessions)}')
+
+    # Build sender map
+    c = conn.cursor()
+    c.execute("SELECT rowid, user_name FROM Name2Id")
+    sender_map = {rowid: uname for rowid, uname in c.fetchall()}
+
+    # Phase 1: Collect
+    print(f'\n=== Phase 1: 采集消息 ===\n')
+    data = []
+    total_msgs = total_sent = total_recv = 0
+    active_days = set(); hour_dist = [0]*24
+
+    for i, s in enumerate(sessions):
+        wxid = s['username']; name = name_map.get(wxid, wxid)
+        msgs = get_messages(conn, wxid, s_ts, e_ts, sender_map, name_map, own_wxid)
+        if not msgs: continue
+        sent = sum(1 for m in msgs if m['is_self']); recv = len(msgs) - sent
+        days = set(fmt_date(m['time']) for m in msgs)
+        total_msgs += len(msgs); total_sent += sent; total_recv += recv
+        active_days.update(days)
+        for m in msgs: hour_dist[datetime.fromtimestamp(m['time'], tz=tz).hour] += 1
+
+        data.append({'wxid': wxid, 'name': name, 'msgs': msgs, 'sent': sent, 'recv': recv, 'days': len(days)})
+        if (i+1) % 20 == 0: print(f'  {i+1}/{len(sessions)}...')
+
+    data.sort(key=lambda d: d['sent'] + d['recv'], reverse=True)
+    print(f'\n联系人: {len(data)} | 消息: {total_msgs} | 我发: {total_sent} | 收到: {total_recv}')
+
+    # Phase 2: AI
+    all_tasks = []; monthly_summary = ''
+    if not args.no_ai and api_key:
+        print(f'\n=== Phase 2: AI 分析 ===\n')
+        # Pre-filter: only analyze contacts with likely tasks
+        ai_targets = [d for d in data if has_tasks(d['msgs'])]
+        skipped = len(data) - len(ai_targets)
+        if skipped: print(f'预筛跳过 {skipped} 人（无任务关键词）\n')
+
+        for i, d in enumerate(ai_targets):
+            lines = []
+            for m in d['msgs']:
+                if m['type'] != 1: continue
+                lines.append(f"[{fmt_time(m['time'])}] {m['sender']}: {m['content'][:120]}")
+            if not lines: continue
+
+            prompt = TALKER_PROMPT.replace('==MESSAGES==', '\n'.join(lines[-30:]))
+            # V4 reasoning model: ~1 token/chars (CN overhead), need 2x input + 500 output
+            estimated_tokens = max(2000, len(prompt) * 2 + 500)
+            try:
+                resp = call_deepseek(prompt, api_key, max_tokens=estimated_tokens, timeout=120)
+                # Extract JSON from response (may be in code block or raw)
+                json_str = resp
+                code_match = re.search(r'```(?:json)?\s*([\s\S]*?)```', resp)
+                if code_match: json_str = code_match.group(1)
+                m = re.search(r'\{[\s\S]*\}', json_str)
+                if m:
+                    raw_json = m.group()
+                    # Fix common DeepSeek JSON issues
+                    raw_json = re.sub(r',\s*}', '}', raw_json)  # trailing comma
+                    raw_json = re.sub(r',\s*]', ']', raw_json)  # trailing comma in array
+                    analysis = json.loads(raw_json)
+                    d['tasks'] = analysis.get('tasks', [])
+                    d['summary'] = analysis.get('summary', '')
+                    for t in d['tasks']: t['talker'] = d['name']
+                    all_tasks.extend(d['tasks'])
+                    print(f'  [{i+1}/{len(ai_targets)}] {d["name"][:20]:20s} → {len(d["tasks"])} 任务')
+                else:
+                    d['tasks'] = []; d['summary'] = ''
+                    # Save raw response for debugging
+                    with open(f'debug_ai_{i}.json', 'w', encoding='utf-8') as df:
+                        df.write(resp[:2000])
+                    print(f'  [{i+1}/{len(ai_targets)}] {d["name"][:20]:20s} → 解析失败 (已保存debug_ai_{i}.json)')
+            except Exception as e:
+                d['tasks'] = []; d['summary'] = ''
+                print(f'  [{i+1}/{len(ai_targets)}] {d["name"][:20]:20s} → ERR: {e}')
+            time.sleep(AI_DELAY[0] + random.random() * (AI_DELAY[1] - AI_DELAY[0]))
+
+        # Monthly summary
+        print(f'\n=== 月度总结 ===\n')
+        parts = []
+        for d in ai_targets:
+            if d.get('tasks'):
+                parts.append(f"{d['name']}: {json.dumps(d['tasks'], ensure_ascii=False)}")
+        if parts:
+            monthly_summary = call_deepseek(SUMMARY_PROMPT.replace('==TALKER_DATA==', '\n\n'.join(parts)), api_key, max_tokens=2000)
+        print(f'  识别任务: {len(all_tasks)} | 未回复: {sum(1 for t in all_tasks if t.get("status")=="未回复")}')
+
+    # Phase 3: Generate markdown
+    print(f'\n=== Phase 3: 生成报告 ===\n')
+    lines = [f'# 微信聊天月报 — {period}', '',
+             f'> 统计范围: {fmt_date(s_ts)} ~ {fmt_date(e_ts)} | 生成: {datetime.now(tz).strftime("%Y-%m-%d")}',
+             f'> AI: {"DeepSeek V4 Pro" if not args.no_ai else "关闭"}',
+             f'> 联系人: {len(data)} | 消息: {total_msgs} | 活跃天数: {len(active_days)}',
+             '', '---', '', '## 总览', '',
+             '| 指标 | 数值 |', '|------|------|',
+             f'| 活跃联系人 | {len(data)} |', f'| 消息总数 | {total_msgs:,} |',
+             f'| 我发出 | {total_sent:,} ({total_sent/max(total_msgs,1)*100:.0f}%) |',
+             f'| 收到 | {total_recv:,} ({total_recv/max(total_msgs,1)*100:.0f}%) |',
+             f'| 活跃天数 | {len(active_days)} |', '']
+
+    if not args.no_ai and all_tasks:
+        unreplied = [t for t in all_tasks if t.get('status') == '未回复']
+        replied = [t for t in all_tasks if t.get('status') != '未回复']
+        lines += ['## AI 任务分析', '', '> DeepSeek V4 Pro', '']
+        lines += [f'### 🔴 未回复 ({len(unreplied)} 项)', '']
+        if unreplied:
+            lines += ['| # | 来源 | 任务 | 时间 | 紧急度 |', '|---|------|------|------|--------|']
+            for i, t in enumerate(unreplied):
+                u = '🔴' if t.get('urgency') == '高' else ('⚠️' if t.get('urgency') == '中' else '🟢')
+                lines.append(f'| {i+1} | {t.get("talker","")} | {t.get("content","")[:30]} | {t.get("time","")} | {u} {t.get("urgency","")} |')
+        else: lines.append('✅ 无未回复任务')
+        lines.append('')
+        lines += [f'### 🟡 已回复 ({len(replied)} 项)', '']
+        if replied:
+            lines += ['| # | 来源 | 任务 | 时间 | 回复 | 状态 |', '|---|------|------|------|------|------|']
+            for i, t in enumerate(replied):
+                s = '✅' if t.get('status') == '完成' else '⏳'
+                lines.append(f'| {i+1} | {t.get("talker","")} | {t.get("content","")[:25]} | {t.get("time","")} | {t.get("reply","")[:20]} | {s} {t.get("status","")} |')
+        lines.append('')
+        if monthly_summary:
+            lines += ['### 📊 AI 月度总结', '']
+            for line in monthly_summary.strip().split('\n'): lines.append(f'> {line}')
+            lines.append('')
+
+    # Per contact
+    lines += ['## 按联系人', '']
+    for d in data:
+        total = d['sent'] + d['recv']
+        lines += [f'### {d["name"]}（{total} 条）', '',
+                   f'我发 {d["sent"]} | 对方发 {d["recv"]} | 活跃 {d["days"]} 天', '']
+        if d.get('summary'):
+            lines += [f'> {d["summary"]}', '']
+        if d.get('tasks'):
+            lines += ['**任务：**', '']
+            for t in d['tasks']:
+                s = '✅' if t.get('status') == '完成' else ('⏳' if t.get('status') == '进行中' else '❌')
+                lines.append(f'- [{t.get("time","")}] {t.get("content","")[:60]} → {s} {t.get("status","")}')
+            lines.append('')
+        # Recent conversations
+        text_msgs = [m for m in d['msgs'] if m['type'] == 1]
+        if text_msgs:
+            lines += ['**最近对话：**', '']
+            for m in text_msgs[-5:]:
+                lines.append(f'> **{fmt_time(m["time"])} {m["sender"]}:** {m["content"][:100]}')
+            lines.append('')
+
+    # Hour dist
+    lines += ['## 活跃时段', '', '| 时段 | 消息 | 占比 |', '|------|------|------|']
+    for label, s, e in [('00-06',0,6),('06-09',6,9),('09-12',9,12),('12-14',12,14),('14-18',14,18),('18-22',18,22),('22-24',22,24)]:
+        n = sum(hour_dist[s:e])
+        lines.append(f'| {label} | {n} | {n/max(total_msgs,1)*100:.1f}% |')
+    lines.append('')
+
+    # Top 10
+    lines += ['## TOP 10', '', '| 排名 | 联系人 | 消息 | 占比 |', '|------|--------|------|------|']
+    for i, d in enumerate(data[:10]):
+        t = d['sent'] + d['recv']
+        lines.append(f'| {i+1} | {d["name"]} | {t} | {t/max(total_msgs,1)*100:.1f}% |')
+    lines += ['', '*由 weflow-cli report 自动生成*', '']
+
+    # Write
+    out = Path(args.output); out.mkdir(parents=True, exist_ok=True)
+    fp = out / f'report-{period}.md'
+    with open(fp, 'w', encoding='utf-8') as f: f.write('\n'.join(lines))
+    print(f'✓ {fp}')
+    print(f'  联系人: {len(data)} | 消息: {total_msgs}')
+    if not args.no_ai: print(f'  任务: {len(all_tasks)} | 未回复: {sum(1 for t in all_tasks if t.get("status")=="未回复")}')
+
+
+if __name__ == '__main__': main()