"""
统一算法模块 — 全部核心算法的唯一定义处（Single Source of Truth）

包含:
1. compute_recommend   — 统一推荐逻辑（买入/卖出/加仓/观望等，严格遵循suanfa.md）
2. get_kline_data      — 获取K线数据并返回 DataFrame（优先本地DB → 阿里云 → 腾讯 → 麦蕊 → AKShare）
3. fetch_kline_rows    — 获取K线数据并返回 tuple 行列表（用于写入DB同步）
4. get_latest_price    — 获取股票最新价格
5. code_to_market      — 股票代码→市场判断（SH/SZ/BJ）
6. 各 API session 管理
7. compute_bull_stage  — 牛股阶段识别（底部→起爆→确立→加速→补涨）
8. find_bull_stocks    — 从扫描结果中找出潜在牛股

调用方:
- routes/analysis.py   → compute_recommend, get_kline_data
- services/scheduler.py → compute_recommend, get_latest_price
- full_signal_scan.py  → get_kline_data, API sessions
- sync_kline.py        → fetch_kline_rows, API sessions
- routes/market.py     → get_kline_data
"""

import threading
from datetime import datetime, timedelta

import requests
from requests.adapters import HTTPAdapter
from urllib3.util.retry import Retry

from config import Config


# ═══════════════════════════════════════════════
# 1. 股票代码 → 市场 工具函数
# ═══════════════════════════════════════════════

def code_to_market(code):
    """6位股票代码 → 市场代码（SH/SZ/BJ）"""
    if code.startswith(('0', '3')):
        return 'SZ'
    elif code.startswith(('8', '9')):
        return 'BJ'
    else:
        return 'SH'


def is_bj_stock(code):
    """是否是北交所股票"""
    return code.startswith(('8', '9'))


def code_to_ali_symbol(code):
    """6位股票代码 → 阿里云API格式（SH600519 / SZ000001 / BJ920720）"""
    return f'{code_to_market(code)}{code}'


def code_to_tencent_symbol(code):
    """6位股票代码 → 腾讯API格式（sh600519 / sz000001 / bj920720）"""
    return f'{code_to_market(code).lower()}{code}'


# ═══════════════════════════════════════════════
# 2. API Session 管理（线程安全，连接池复用）
# ═══════════════════════════════════════════════

_ali_session = None
_ali_lock = threading.Lock()

_tencent_session = None
_tencent_lock = threading.Lock()


def get_ali_session():
    """获取阿里云API专用 Session（线程安全，单例）"""
    global _ali_session
    if _ali_session is None:
        with _ali_lock:
            if _ali_session is None:
                s = requests.Session()
                retry = Retry(total=2, backoff_factor=0.3,
                              status_forcelist=[500, 502, 503, 504])
                adapter = HTTPAdapter(max_retries=retry,
                                      pool_connections=20, pool_maxsize=20)
                s.mount('https://', adapter)
                s.headers.update({
                    'Authorization': f'APPCODE {Config.ALICLOUD_APPCODE}',
                    'Content-Type': 'application/x-www-form-urlencoded',
                })
                _ali_session = s
    return _ali_session


def get_tencent_session():
    """获取腾讯K线API专用 Session（线程安全，单例）"""
    global _tencent_session
    if _tencent_session is None:
        with _tencent_lock:
            if _tencent_session is None:
                s = requests.Session()
                retry = Retry(total=2, backoff_factor=0.3,
                              status_forcelist=[500, 502, 503, 504])
                adapter = HTTPAdapter(max_retries=retry,
                                      pool_connections=20, pool_maxsize=20)
                s.mount('https://', adapter)
                s.headers.update({
                    'User-Agent': ('Mozilla/5.0 (Windows NT 10.0; Win64; x64) '
                                   'AppleWebKit/537.36'),
                })
                _tencent_session = s
    return _tencent_session


# API URL 常量
ALICLOUD_KLINE_URL = Config.ALICLOUD_KLINE_URL
TENCENT_KLINE_URL = 'https://proxy.finance.qq.com/ifzqgtimg/appstock/app/newfqkline/get'


# ═══════════════════════════════════════════════
# 3. K线数据获取 — DataFrame 格式（供信号检测/分析用）
# ═══════════════════════════════════════════════

def get_kline_data(stock_code, days=120, use_local_db=True):
    """
    获取K线数据，返回 pandas DataFrame (columns: date, open, high, low, close, volume)
    
    数据源优先级: 本地DB → 阿里云API → 腾讯API → 麦蕊API → AKShare
    
    参数:
        stock_code: 6位股票代码
        days: 获取天数
        use_local_db: 是否优先使用本地DB（全景扫描时为True, 实时分析时可为False）
    
    返回:
        DataFrame 或 None
    """
    import pandas as pd

    # 1. 优先从本地数据库读取
    if use_local_db:
        df = _get_kline_from_local_db(stock_code, days)
        if df is not None:
            return df

    # 2. 阿里云K线API（沪深最稳定，北交所可能不支持）
    if not is_bj_stock(stock_code):
        df = _fetch_ali_kline_df(stock_code, days)
        if df is not None and len(df) >= 30:
            return df

    # 3. 腾讯K线API（全市场，含北交所）
    df = _fetch_tencent_kline_df(stock_code, days)
    if df is not None and len(df) >= 30:
        return df

    # 4. 麦蕊API
    df = _fetch_mairui_kline_df(stock_code, days)
    if df is not None and len(df) >= 30:
        return df

    # 5. AKShare
    df = _fetch_akshare_kline_df(stock_code, days)
    if df is not None and len(df) >= 30:
        return df

    return None


def _get_kline_from_local_db(stock_code, days=120):
    """从本地数据库读取K线（最快，毫秒级）"""
    import pandas as pd
    try:
        import psycopg2
        conn = psycopg2.connect(
            host=Config.DB_HOST, port=Config.DB_PORT,
            dbname=Config.DB_NAME, user=Config.DB_USER, password=Config.DB_PASSWORD,
        )
        conn.autocommit = True
        start_date = (datetime.now() - timedelta(days=days)).strftime('%Y-%m-%d')
        with conn.cursor() as cur:
            cur.execute("""
                SELECT trade_date, open, high, low, close, volume
                FROM stock_kline_daily
                WHERE code = %s AND trade_date >= %s
                ORDER BY trade_date
            """, (stock_code, start_date))
            rows = cur.fetchall()
        conn.close()

        if rows and len(rows) >= 30:
            df = pd.DataFrame(rows, columns=['date', 'open', 'high', 'low', 'close', 'volume'])
            df['date'] = df['date'].astype(str)
            for col in ('open', 'high', 'low', 'close', 'volume'):
                df[col] = df[col].astype(float)
            return df
    except Exception:
        pass
    return None


# ---- 线程本地连接（供多线程扫描时使用，避免频繁建连） ----
_thread_local = threading.local()


def get_kline_from_local_db_threaded(stock_code, days=120):
    """多线程扫描专用：使用线程本地连接从本地DB读取K线"""
    import pandas as pd
    try:
        conn = getattr(_thread_local, 'kline_conn', None)
        if conn is None or conn.closed:
            import psycopg2
            conn = psycopg2.connect(
                host=Config.DB_HOST, port=Config.DB_PORT,
                dbname=Config.DB_NAME, user=Config.DB_USER, password=Config.DB_PASSWORD,
            )
            conn.autocommit = True
            _thread_local.kline_conn = conn

        start_date = (datetime.now() - timedelta(days=days)).strftime('%Y-%m-%d')
        with conn.cursor() as cur:
            cur.execute("""
                SELECT trade_date, open, high, low, close, volume
                FROM stock_kline_daily
                WHERE code = %s AND trade_date >= %s
                ORDER BY trade_date
            """, (stock_code, start_date))
            rows = cur.fetchall()

        if rows and len(rows) >= 30:
            df = pd.DataFrame(rows, columns=['date', 'open', 'high', 'low', 'close', 'volume'])
            df['date'] = df['date'].astype(str)
            for col in ('open', 'high', 'low', 'close', 'volume'):
                df[col] = df[col].astype(float)
            return df
    except Exception:
        pass
    return None


def _fetch_ali_kline_df(stock_code, days=120):
    """阿里云K线API → DataFrame"""
    import pandas as pd
    try:
        session = get_ali_session()
        symbol = code_to_ali_symbol(stock_code)
        resp = session.post(ALICLOUD_KLINE_URL, data={
            'symbol': symbol,
            'type': '240',
            'limit': str(min(days, 300)),
            'ma': '5',
        }, timeout=10)
        if resp.status_code == 200:
            data = resp.json()
            if data.get('success') and data.get('data', {}).get('list'):
                records = []
                for item in data['data']['list']:
                    day_str = item.get('day', '')
                    if not day_str or len(day_str) < 10:
                        continue
                    records.append({
                        'date': day_str[:10],
                        'open': float(item.get('open', 0)),
                        'high': float(item.get('high', 0)),
                        'low': float(item.get('low', 0)),
                        'close': float(item.get('close', 0)),
                        'volume': float(item.get('volume', 0)),
                    })
                if records:
                    return pd.DataFrame(records)
    except Exception:
        pass
    return None


def _fetch_tencent_kline_df(stock_code, days=120):
    """腾讯K线API → DataFrame（全市场含北交所）"""
    import pandas as pd
    try:
        session = get_tencent_session()
        symbol = code_to_tencent_symbol(stock_code)
        start_date = (datetime.now() - timedelta(days=days + 30)).strftime('%Y-%m-%d')
        resp = session.get(TENCENT_KLINE_URL, params={
            'param': f'{symbol},day,{start_date},,{min(days, 300)},qfq',
        }, timeout=15)
        if resp.status_code == 200:
            data = resp.json()
            stock_data = data.get('data', {}).get(symbol, {})
            klines = stock_data.get('qfqday') or stock_data.get('day') or []
            if klines:
                records = []
                for item in klines:
                    if len(item) < 6:
                        continue
                    # 腾讯格式: [date, open, close, high, low, volume, ...]
                    records.append({
                        'date': item[0][:10],
                        'open': float(item[1]),
                        'high': float(item[3]),   # high = position 3
                        'low': float(item[4]),    # low = position 4
                        'close': float(item[2]),  # close = position 2
                        'volume': float(item[5]),
                    })
                if records:
                    return pd.DataFrame(records)
    except Exception:
        pass
    return None


def _fetch_mairui_kline_df(stock_code, days=120):
    """麦蕊API → DataFrame"""
    import pandas as pd
    try:
        from services.mairui_api import get_kline
        result = get_kline(stock_code, period='d', days=days, adjust='f')
        if result['success'] and result['data']:
            df = pd.DataFrame(result['data'])
            df.rename(columns={
                'date': 'date', 'open': 'open', 'high': 'high',
                'low': 'low', 'close': 'close', 'volume': 'volume',
            }, inplace=True)
            if len(df) >= 30:
                return df
    except Exception:
        pass
    return None


def _fetch_akshare_kline_df(stock_code, days=120):
    """AKShare → DataFrame (支持自动降级到腾讯数据源)"""
    import pandas as pd
    try:
        from utils.data_fetcher import fetch_stock_hist
        end_date = datetime.now().strftime('%Y%m%d')
        start_date = (datetime.now() - timedelta(days=days)).strftime('%Y%m%d')
        df = fetch_stock_hist(
            stock_code=stock_code, period='daily',
            start_date=start_date, end_date=end_date, adjust='qfq',
        )
        if df is not None and not df.empty:
            df = df.rename(columns={
                '日期': 'date', '开盘': 'open', '最高': 'high',
                '最低': 'low', '收盘': 'close', '成交量': 'volume',
            })
            df = df[['date', 'open', 'high', 'low', 'close', 'volume']]
            return df
    except Exception:
        pass
    return None


# ═══════════════════════════════════════════════
# 4. K线数据获取 — tuple行格式（供 sync_kline.py 写入DB用）
# ═══════════════════════════════════════════════

def fetch_kline_rows(code, days):
    """
    获取K线数据，返回 list of tuple: (code, date, open, high, low, close, volume, amount)
    
    数据源优先级: 阿里云API → 腾讯API → 麦蕊API → AKShare
    北交所(8XX/9XX)直接走腾讯API
    
    供 sync_kline.py 同步到本地数据库使用。
    """
    bj = is_bj_stock(code)

    # 北交所：直接用腾讯API（阿里云/Mairui不支持BJ）
    if bj:
        rows = _fetch_tencent_kline_rows(code, days)
        if rows:
            return rows
        return None

    # 1. 首选：阿里云K线API
    rows = _fetch_ali_kline_rows(code, days)
    if rows:
        return rows

    # 2. 回退：腾讯API
    rows = _fetch_tencent_kline_rows(code, days)
    if rows:
        return rows

    # 3. 回退：Mairui API
    rows = _fetch_mairui_kline_rows(code, days)
    if rows:
        return rows

    # 4. 最后回退：AKShare
    rows = _fetch_akshare_kline_rows(code, days)
    if rows:
        return rows

    return None


def _fetch_ali_kline_rows(code, days):
    """阿里云K线API → tuple rows"""
    try:
        session = get_ali_session()
        symbol = code_to_ali_symbol(code)
        resp = session.post(ALICLOUD_KLINE_URL, data={
            'symbol': symbol,
            'type': '240',
            'limit': str(min(days, 300)),
            'ma': '5',
        }, timeout=10)
        if resp.status_code == 200:
            data = resp.json()
            if data.get('success') and data.get('data', {}).get('list'):
                rows = []
                for item in data['data']['list']:
                    day_str = item.get('day', '')
                    if not day_str or len(day_str) < 10:
                        continue
                    rows.append((
                        code,
                        day_str[:10],
                        float(item.get('open', 0)),
                        float(item.get('high', 0)),
                        float(item.get('low', 0)),
                        float(item.get('close', 0)),
                        int(item.get('volume', 0)),
                        float(item.get('amount', 0)),
                    ))
                if rows:
                    return rows
    except Exception:
        pass
    return None


def _fetch_tencent_kline_rows(code, days):
    """腾讯K线API → tuple rows（全市场含北交所）"""
    try:
        symbol = code_to_tencent_symbol(code)
        session = get_tencent_session()
        start_date = (datetime.now() - timedelta(days=days + 30)).strftime('%Y-%m-%d')
        resp = session.get(TENCENT_KLINE_URL, params={
            'param': f'{symbol},day,{start_date},,{min(days, 300)},qfq',
        }, timeout=15)
        if resp.status_code == 200:
            data = resp.json()
            stock_data = data.get('data', {}).get(symbol, {})
            klines = stock_data.get('qfqday') or stock_data.get('day') or []
            if klines:
                rows = []
                for item in klines:
                    if len(item) < 6:
                        continue
                    date_str = item[0]
                    if not date_str or len(date_str) < 10:
                        continue
                    rows.append((
                        code,
                        date_str[:10],
                        float(item[1]),          # open
                        float(item[3]),          # high (position 3)
                        float(item[4]),          # low (position 4)
                        float(item[2]),          # close (position 2)
                        int(float(item[5])),     # volume
                        float(item[8]) * 10000 if len(item) > 8 and item[8] else 0,
                    ))
                if rows:
                    return rows
    except Exception:
        pass
    return None


def _fetch_mairui_kline_rows(code, days):
    """麦蕊API → tuple rows"""
    try:
        from services.mairui_api import get_kline
        result = get_kline(code, period='d', days=days, adjust='f')
        if result['success'] and result['data']:
            rows = []
            for item in result['data']:
                date_str = item.get('date', '')
                if not date_str:
                    continue
                rows.append((
                    code,
                    date_str,
                    item.get('open', 0),
                    item.get('high', 0),
                    item.get('low', 0),
                    item.get('close', 0),
                    int(item.get('volume', 0)),
                    item.get('amount', 0),
                ))
            if rows:
                return rows
    except Exception:
        pass
    return None


def _fetch_akshare_kline_rows(code, days):
    """AKShare → tuple rows (支持自动降级到腾讯数据源)"""
    try:
        from utils.data_fetcher import fetch_stock_hist
        end_date = datetime.now().strftime('%Y%m%d')
        start_date = (datetime.now() - timedelta(days=days)).strftime('%Y%m%d')
        df = fetch_stock_hist(
            stock_code=code, period='daily',
            start_date=start_date, end_date=end_date, adjust='qfq',
        )
        if df is not None and not df.empty:
            rows = []
            for _, r in df.iterrows():
                rows.append((
                    code,
                    str(r['日期']),
                    float(r['开盘']),
                    float(r['最高']),
                    float(r['最低']),
                    float(r['收盘']),
                    int(r['成交量']),
                    float(r.get('成交额', 0)),
                ))
            if rows:
                return rows
    except Exception:
        pass
    return None


# ═══════════════════════════════════════════════
# 5. 统一推荐算法
# ═══════════════════════════════════════════════

def compute_recommend(signal_status, indicators, triggered_count, is_holding):
    """
    统一推荐逻辑 — 全局唯一定义（严格遵循 suanfa.md 体系最强战法）。
    
    体系最强战法流程（suanfa.md）：
    1. 日线底背离 → 纳入关注范围
    2. 龙抬头出现 → 执行买入操作（实操核心买点）
    3. 真龙/主升浪 → 持有仓位+加仓（不是新买入！）
    4. 不见主升浪 → 不出场
    
    参数:
        signal_status: list[dict] 信号状态列表（来自 signal_detector._check_all_signal_status）
        indicators: dict 最新技术指标（含 macd.dif, macd.dea 等）
        triggered_count: int 触发信号数量
        is_holding: bool 当前是否持仓该股票
    
    返回:
        tuple: (signal_type, display_text, reason, recommend_rate)
        - signal_type: 'buy' | 'sell' | 'watch'
        - display_text: '买入' | '卖出' | '加仓' | '持有' | '关注' | '观察' | '观望'
        - reason: str 推荐理由
        - recommend_rate: int 推荐评分 0-100
    
    使用场景:
        - 全景扫描结果推荐列
        - 策略建议分档
        - 提醒 tab 买卖推荐
        - 模拟交易自动买卖决策
    """
    if not signal_status:
        return ('watch', '观望', '暂无信号数据', 0)

    ss = signal_status
    sig_map = {}
    for s in ss:
        sig_map[s.get('type', '')] = s

    has_main_wave = sig_map.get('main_rising_wave', {}).get('triggered', False)
    has_divergence = sig_map.get('daily_bottom_divergence', {}).get('triggered', False)
    has_dragon = sig_map.get('dragon_head', {}).get('triggered', False)
    has_real_dragon = sig_map.get('true_dragon', {}).get('triggered', False)

    macd = (indicators or {}).get('macd', {})
    dif = macd.get('dif', 0)
    dea = macd.get('dea', 0)

    triggered_signals = [s.get('name', s.get('type', '')) for s in ss if s.get('triggered')]

    # ════════════════════════════════════════════
    # 持仓逻辑（suanfa.md 步骤3-4）
    # ════════════════════════════════════════════
    if is_holding:
        # 卖出条件: MACD死叉 + 无主升浪 → 趋势走弱，不见主升浪则出场
        if dif < dea and not has_main_wave:
            return ('sell', '卖出',
                    f"MACD死叉(DIF={dif:.3f}<DEA={dea:.3f})+主升浪消失", 75)
        # 主升浪 → 加仓（suanfa.md: 主升浪=持有仓位+加仓）
        if has_main_wave:
            return ('buy', '加仓', '主升浪信号 → 加速拉升阶段，建议加仓', 90)
        # 真龙 → 持有（suanfa.md: 真龙=趋势确认，持有）
        if has_real_dragon:
            return ('buy', '持有', '真龙信号 → 趋势确立，继续持有', 70)
        return ('watch', '观望', '持仓中，等待主升浪信号', 50)

    # ════════════════════════════════════════════
    # 非持仓逻辑（suanfa.md 步骤1-2）
    # ════════════════════════════════════════════

    # 最佳买入: 底背离+龙抬头（suanfa.md 步骤1→2 的理想组合）
    # MACD 死叉时降级，避免与持仓卖出信号矛盾
    if has_divergence and has_dragon:
        macd_golden = (dif is None or dea is None or dif >= dea)
        if macd_golden:
            return ('buy', '买入', '日线底背离+龙抬头 → 最佳买入信号', 95)
        return ('watch', '关注',
                f'底背离+龙抬头但MACD死叉(DIF={dif:.3f}<DEA={dea:.3f}) → 等待MACD金叉确认', 65)

    # 核心买入: 龙抬头（suanfa.md: "龙抬头出现 → 执行买入操作"）
    # 但需要 MACD 配合：如果 MACD 死叉则信号冲突，降级为关注
    if has_dragon:
        macd_ok = (dif is None or dea is None or dif >= dea)  # MACD 金叉或无数据
        if has_main_wave and macd_ok:
            return ('buy', '买入', '龙抬头+主升浪 → 强势买入信号', 90)
        if macd_ok:
            return ('buy', '买入', '龙抬头出现 → 短线起爆点，执行买入', 80)
        # MACD死叉 + 龙抬头 → 信号冲突，降级为关注
        return ('watch', '关注',
                f'龙抬头出现但MACD死叉(DIF={dif:.3f}<DEA={dea:.3f}) → 信号冲突，谨慎观望', 55)

    # 主升浪（非持仓）→ 不是新买入！suanfa.md: 主升浪=加速段，属于持有/加仓信号
    if has_main_wave:
        return ('watch', '关注', '主升浪(加速段) → 已过最佳买点，关注回调机会', 75)

    # 真龙 → 关注（趋势刚启动，可跟踪）
    if has_real_dragon:
        return ('watch', '关注', '真龙出现 → 趋势启动，等待龙抬头确认', 65)

    # MACD死叉 → 卖出/回避
    if dif is not None and dea is not None and dif < dea:
        return ('sell', '卖出', f"MACD死叉(DIF={dif:.3f}<DEA={dea:.3f})", 75)

    # 底背离 → 关注（suanfa.md 步骤1: 纳入关注范围）
    if has_divergence:
        return ('watch', '关注', '日线底背离出现 → 纳入关注，等待龙抬头', 60)

    # 其他信号 → 观察
    if triggered_count and triggered_count > 0:
        sigs = '、'.join(triggered_signals[:3])
        return ('watch', '观察', f"触发{triggered_count}个信号: {sigs}", 40)

    return ('watch', '观望', '无核心信号触发', 0)


# ═══════════════════════════════════════════════
# 6. 获取最新价格
# ═══════════════════════════════════════════════

def get_latest_price(stock_code):
    """
    获取股票最新价格（从本地 stock_realtime_price 表）
    
    返回:
        float: 最新价格, 失败返回 0
    """
    try:
        import psycopg2
        conn = psycopg2.connect(
            host=Config.DB_HOST, port=Config.DB_PORT,
            dbname=Config.DB_NAME, user=Config.DB_USER, password=Config.DB_PASSWORD,
        )
        cur = conn.cursor()
        cur.execute("""
            SELECT price FROM stock_realtime_price
            WHERE code = %s AND price > 0
        """, (stock_code,))
        row = cur.fetchone()
        conn.close()
        if row:
            return float(row[0])
    except Exception:
        pass
    return 0


# ═══════════════════════════════════════════════
# 7. 牛股阶段识别（suanfa.md 标准牛股启动信号先后顺序）
# ═══════════════════════════════════════════════

# 标准牛股启动流程（底部→拉升）:
# 阶段1 → 日线底背离/短底背离（跌到底部，停止下跌）
# 阶段2 → 龙抬头（资金进场，短线起爆）
# 阶段3 → 真龙（趋势正式确立）
# 阶段4 → ★主升浪（进入加速段，利润兑现最快）
# 阶段5 → 反弹（中途回调后的补涨信号）
# 补充  → 老鼠仓可在底部任意位置提前出现

BULL_STAGES = {
    1: {'name': '底部探测', 'icon': '', 'color': '#2196F3',
        'desc': '日线底背离/短底背离 → 跌到底部，停止下跌'},
    2: {'name': '资金进场', 'icon': '', 'color': '#4CAF50',
        'desc': '龙抬头 → 资金进场，短线起爆点（最佳买入时机）'},
    3: {'name': '趋势确立', 'icon': '', 'color': '#FF9800',
        'desc': '真龙 → 中期趋势正式确立'},
    4: {'name': '加速拉升', 'icon': '', 'color': '#F44336',
        'desc': '★主升浪 → 进入加速段，利润兑现最快'},
    5: {'name': '回调补涨', 'icon': '', 'color': '#9C27B0',
        'desc': '反弹 → 中途回调后的补涨信号'},
}


def compute_bull_stage(signal_status):
    """
    识别股票在标准牛股启动流程中的阶段。
    
    参数:
        signal_status: list[dict] 信号状态列表
    
    返回:
        dict: {
            'stage': int (0-5, 0=未进入流程),
            'stage_name': str,
            'stage_icon': str,
            'stage_color': str,
            'stage_desc': str,
            'signals_active': list[str],   # 当前活跃的信号名称
            'progress': int (0-100),        # 牛股流程进度百分比
            'next_signal': str,             # 下一个期待的信号
            'investment_advice': str,       # 投资建议
            'has_rat_trading': bool,        # 是否有老鼠仓（提前埋伏信号）
        }
    """
    if not signal_status:
        return {
            'stage': 0, 'stage_name': '观望', 'stage_icon': '',
            'stage_color': '#9E9E9E', 'stage_desc': '无信号触发',
            'signals_active': [], 'progress': 0,
            'next_signal': '等待底背离/短底背离', 'investment_advice': '暂无操作机会',
            'has_rat_trading': False,
        }

    sig_map = {}
    for s in signal_status:
        sig_map[s.get('type', '')] = s

    has_divergence = sig_map.get('daily_bottom_divergence', {}).get('triggered', False)
    has_short_div = sig_map.get('short_bottom_divergence', {}).get('triggered', False)
    has_dragon = sig_map.get('dragon_head', {}).get('triggered', False)
    has_true_dragon = sig_map.get('true_dragon', {}).get('triggered', False)
    has_main_wave = sig_map.get('main_rising_wave', {}).get('triggered', False)
    has_rebound = sig_map.get('rebound', {}).get('triggered', False)
    has_rat = sig_map.get('rat_trading', {}).get('triggered', False)

    signals_active = []
    if has_divergence:
        signals_active.append('日线底背离')
    if has_short_div:
        signals_active.append('短底背离')
    if has_dragon:
        signals_active.append('龙抬头')
    if has_true_dragon:
        signals_active.append('真龙')
    if has_main_wave:
        signals_active.append('主升浪')
    if has_rebound:
        signals_active.append('反弹')
    if has_rat:
        signals_active.append('老鼠仓')

    # 确定阶段（按最高阶段判定）
    stage = 0
    if has_main_wave:
        stage = 4
    elif has_true_dragon:
        stage = 3
    elif has_dragon:
        stage = 2
    elif has_divergence or has_short_div:
        stage = 1
    elif has_rebound:
        stage = 5
    elif has_rat:
        stage = 1  # 老鼠仓归入底部阶段

    if stage == 0:
        return {
            'stage': 0, 'stage_name': '观望', 'stage_icon': '',
            'stage_color': '#9E9E9E', 'stage_desc': '无核心信号触发',
            'signals_active': signals_active, 'progress': 0,
            'next_signal': '等待底背离/短底背离',
            'investment_advice': '暂无操作机会',
            'has_rat_trading': has_rat,
        }

    info = BULL_STAGES[stage]

    # 计算流程进度（越靠后越高）
    # 加分项：多信号叠加说明流程更完整
    base_progress = {1: 20, 2: 45, 3: 65, 4: 85, 5: 50}
    progress = base_progress.get(stage, 0)
    if stage <= 2 and has_divergence:
        progress += 10  # 有底背离做基础更好
    if stage >= 2 and has_dragon:
        progress += 5
    if stage >= 3 and has_true_dragon:
        progress += 5
    if has_rat:
        progress += 5  # 老鼠仓加分
    progress = min(progress, 100)

    # 下一步信号期待
    next_signals = {
        1: '等待龙抬头（资金进场信号）',
        2: '等待真龙（趋势确认信号）',
        3: '等待主升浪（加速拉升信号）',
        4: '持有！不见主升浪消失不出场',
        5: '等待龙抬头/真龙确认趋势',
    }

    # 投资建议
    advices = {
        1: '纳入关注池，等待龙抬头出现后买入',
        2: '最佳买入时机！龙抬头=实操核心买点',
        3: '趋势已确立，可以追入，建议等回调买入',
        4: '已在加速段，持仓者加仓/持有，新入者谨慎追高',
        5: '回调中可关注，但需确认不是假反弹',
    }

    return {
        'stage': stage,
        'stage_name': info['name'],
        'stage_icon': info['icon'],
        'stage_color': info['color'],
        'stage_desc': info['desc'],
        'signals_active': signals_active,
        'progress': progress,
        'next_signal': next_signals.get(stage, ''),
        'investment_advice': advices.get(stage, ''),
        'has_rat_trading': has_rat,
    }


def find_bull_stocks(scan_rows, holding_codes=None):
    """
    从扫描结果中找出潜在牛股，按阶段分组排序。
    
    参数:
        scan_rows: list[dict] 扫描结果列表 (含 code, name, signal_status, indicators, triggered_count)
        holding_codes: set 持仓代码集合
    
    返回:
        dict: {
            'stages': {1: [...], 2: [...], ...},  # 按阶段分组的股票列表
            'summary': {1: count, 2: count, ...},  # 各阶段数量统计
            'total': int,  # 有信号的总数
        }
    """
    if holding_codes is None:
        holding_codes = set()

    stages = {1: [], 2: [], 3: [], 4: [], 5: []}
    summary = {0: 0, 1: 0, 2: 0, 3: 0, 4: 0, 5: 0}

    for row in scan_rows:
        signal_status = row.get('signal_status') or []
        if not signal_status:
            summary[0] += 1
            continue

        # 判断牛股阶段
        bull = compute_bull_stage(signal_status)
        stage = bull['stage']
        summary[stage] += 1

        if stage == 0:
            continue

        # 计算推荐
        is_holding = row.get('code', '') in holding_codes
        st, disp, reason, rate = compute_recommend(
            signal_status, row.get('indicators'),
            row.get('triggered_count'), is_holding,
        )

        item = {
            'code': row.get('code', ''),
            'name': row.get('name', ''),
            'stage': stage,
            'stage_name': bull['stage_name'],
            'stage_icon': bull['stage_icon'],
            'stage_color': bull['stage_color'],
            'signals_active': bull['signals_active'],
            'progress': bull['progress'],
            'next_signal': bull['next_signal'],
            'investment_advice': bull['investment_advice'],
            'has_rat_trading': bull['has_rat_trading'],
            'recommend_type': st,
            'recommend_text': disp,
            'recommend_reason': reason,
            'recommend_rate': rate,
            'is_holding': is_holding,
            'triggered_count': row.get('triggered_count', 0),
        }

        stages[stage].append(item)

    # 每个阶段内按推荐评分降序排序
    for stage_num in stages:
        stages[stage_num].sort(key=lambda x: (-x['recommend_rate'], -x['progress']))

    total = sum(len(v) for v in stages.values())

    return {
        'stages': stages,
        'summary': summary,
        'total': total,
        'stage_info': BULL_STAGES,
    }