1

_test_akshare.py

@@ -0,0 +1,43 @@
+"""测试 akshare 各函数能获取多少 PVC 5分钟历史数据。"""
+import akshare as ak
+import pandas as pd
+from datetime import datetime
+
+print(f"akshare version: {ak.__version__}")
+print()
+
+# 方法1: futures_zh_minute_sina - 新浪连续合约
+print("=== 方法1: futures_zh_minute_sina(V0, 5min) ===")
+try:
+    df1 = ak.futures_zh_minute_sina(symbol="V0", period="5")
+    print(f"行数: {len(df1)}")
+    print(f"列: {list(df1.columns)}")
+    if len(df1) > 0:
+        print(f"时间范围: {df1.iloc[0,0]} ~ {df1.iloc[-1,0]}")
+        print(df1.head(2))
+        print(df1.tail(2))
+except Exception as e:
+    print(f"ERROR: {e}")
+
+print()
+
+# 方法2: 尝试下载多个具体合约拼接
+print("=== 方法2: 下载多个 PVC 合约 ===")
+all_dfs = []
+# PVC合约代码: V+YYMM, 如 V2401=2024年1月
+contracts = []
+for year in [23, 24, 25, 26]:
+    for month in range(1, 13):
+        code = f"V{year:02d}{month:02d}"
+        contracts.append(code)
+
+for code in contracts:
+    try:
+        df = ak.futures_zh_minute_sina(symbol=code, period="5")
+        if df is not None and len(df) > 0:
+            print(f"  {code}: {len(df)} 行, {df.iloc[0,0]} ~ {df.iloc[-1,0]}")
+            all_dfs.append(df)
+    except Exception as e:
+        pass  # 合约可能已过期，跳过
+
+print(f"\n总计获取 {len(all_dfs)} 个合约, 合计 {sum(len(d) for d in all_dfs)} 行")

_test_tqsdk_login.py

@@ -0,0 +1,33 @@
+# -*- coding: utf-8 -*-
+"""验证 tqsdk 账户能否登录并获取行情数据"""
+from tqsdk import TqApi, TqAuth
+import sys, time
+
+USER = "七斗居士"
+PASSWORD = "ghhkphs8"
+
+print(f"正在登录 tqsdk: {USER}")
+try:
+    api = TqApi(auth=TqAuth(USER, PASSWORD))
+    print("[OK] 登录成功!")
+    
+    # 获取行情
+    quote = api.get_quote("SHFE.cu2509")
+    # 等待数据到达
+    while quote.datetime == "":
+        api.wait_update()
+    
+    print(f"合约代码: {quote.instrument_id}")
+    print(f"最新价: {quote.last_price}")
+    print(f"买一价: {quote.bid_price1}")
+    print(f"卖一价: {quote.ask_price1}")
+    print(f"开盘价: {quote.open}")
+    print(f"最高价: {quote.high}")
+    print(f"最低价: {quote.low}")
+    print(f"更新时间: {quote.datetime}")
+    print("\n[OK] 账户有效, 行情数据获取成功!")
+    
+    api.close()
+except Exception as e:
+    print(f"[FAIL] {type(e).__name__}: {e}")
+    sys.exit(1)

data_pipeline/parquet_manager.py

@@ -127,7 +127,20 @@ def inspect_parquet_file(path: str | Path) -> dict[str, Any]:
             f"数据不足: {bars} bars（至少需要 {Config.MIN_BARS}）"
         )
 
-    years = round(bars / 6240, 2) if timeframe == "H1" else None
+    # 从实际时间跨度计算年数（适用于所有周期，比固定公式更准确）
+    years = None
+    if "time" in df.columns and len(df) > 1:
+        try:
+            t_min = float(df["time"].min())
+            t_max = float(df["time"].max())
+            if t_max > t_min and t_max > 1_000_000_000:  # 合法的 Unix 时间戳
+                span_seconds = t_max - t_min
+                years = round(span_seconds / (365.25 * 24 * 3600), 2)
+        except Exception:
+            pass
+    # 回退：H1 用固定公式（6240 根/年，24h 市场）
+    if years is None and timeframe == "H1":
+        years = round(bars / 6240, 2)
     return {
         "data_file": str(p.resolve()),
         "filename": p.name,

download_pvc_5m.py

@@ -0,0 +1,95 @@
+"""下载 PVC 5分钟 K 线数据到桌面。
+
+使用 tqsdk 拉取 PVC主连(KQ.m@DCE.v) 的 5 分钟 K 线，
+保存为项目兼容的 Parquet 格式（time/open/high/low/close/tick_volume）。
+"""
+import json
+import sys
+from pathlib import Path
+
+# 读取 tqsdk 凭证（优先 web_settings.json，未配置则用默认账号）
+settings_path = Path(__file__).parent / "web_settings.json"
+try:
+    settings = json.loads(settings_path.read_text(encoding="utf-8"))
+    user = str(settings.get("tqsdk_user", "")).strip()
+    pwd = str(settings.get("tqsdk_password", "")).strip()
+except Exception:
+    user, pwd = "", ""
+
+# 内置默认账号
+if not user or not pwd:
+    user, pwd = "七斗居士", "ghhkphs8"
+
+print(f"tqsdk 账号: {user}")
+print("正在连接天勤量化...")
+
+from tqsdk import TqApi, TqAuth, TqSim
+import pandas as pd
+
+# PVC主连 → tqsdk 主连代码
+TQSDK_SYMBOL = "KQ.m@DCE.v"
+DURATION = 300  # 5分钟 = 300秒
+DATA_LENGTH = 8000  # 尽量多拉
+
+api = None
+try:
+    api = TqApi(TqSim(), auth=TqAuth(user, pwd), disable_print=True)
+    print("连接成功，正在拉取 PVC 5分钟 K 线...")
+
+    # adj_type="F" 前复权，消除换月跳价
+    df = api.get_kline_serial(
+        TQSDK_SYMBOL, DURATION, data_length=DATA_LENGTH, adj_type="F"
+    )
+    print(f"拉取到 {len(df)} 根 K 线")
+finally:
+    if api is not None:
+        try:
+            api.close()
+        except Exception:
+            pass
+
+if df is None or df.empty:
+    print("ERROR: 未拉取到数据")
+    sys.exit(1)
+
+# 过滤无效行（volume=0 的非交易时段，或 datetime=0）
+df = df[df["datetime"] != 0].copy()
+df = df[df["volume"] > 0].copy()
+print(f"过滤后有效 K 线: {len(df)} 根")
+
+# 转换为项目兼容格式
+# tqsdk datetime 是纳秒级 Unix 时间戳 → 转秒
+result = pd.DataFrame()
+result["time"] = (df["datetime"].astype("int64") // 1_000_000_000).astype("int64")
+result["open"] = df["open"].astype("float64")
+result["high"] = df["high"].astype("float64")
+result["low"] = df["low"].astype("float64")
+result["close"] = df["close"].astype("float64")
+result["tick_volume"] = df["volume"].astype("int64")
+
+# 按时间升序排列
+result = result.sort_values("time").reset_index(drop=True)
+
+# 去除可能重复的时间戳
+result = result.drop_duplicates(subset=["time"], keep="last").reset_index(drop=True)
+
+# 输出目录
+out_dir = Path(r"C:\Users\14795\Desktop\国内期货K线数据")
+out_dir.mkdir(parents=True, exist_ok=True)
+out_file = out_dir / "PVC_5min.parquet"
+
+# 保存为 Parquet
+result.to_parquet(out_file, index=False)
+print(f"\n已保存: {out_file}")
+print(f"总行数: {len(result)}")
+
+# 显示数据概览
+print(f"\n=== 数据概览 ===")
+print(f"时间范围: {pd.to_datetime(result['time'].iloc[0], unit='s')} ~ {pd.to_datetime(result['time'].iloc[-1], unit='s')}")
+print(f"开盘价范围: {result['open'].min():.2f} ~ {result['open'].max():.2f}")
+print(f"收盘价范围: {result['close'].min():.2f} ~ {result['close'].max():.2f}")
+print(f"\n前 5 行:")
+print(result.head().to_string())
+print(f"\n后 5 行:")
+print(result.tail().to_string())
+print("\n下载完成!")

download_pvc_5m_full.py

@@ -0,0 +1,124 @@
+"""修复版：下载 PVC 全历史 5分钟 K 线数据（akshare 多合约拼接）。
+
+修复 pandas 3.0 时间戳转换问题，使用显式时区处理。
+"""
+import akshare as ak
+import pandas as pd
+from pathlib import Path
+from datetime import timezone, timedelta
+import time
+import sys
+
+print("=" * 60)
+print("  PVC 全历史 5分钟 K 线下载（akshare 多合约拼接·修复版）")
+print("=" * 60)
+
+# 北京时间 UTC+8
+BJ_TZ = timezone(timedelta(hours=8))
+
+# 生成所有 PVC 合约代码: V2101 ~ V2612
+contracts = []
+for year in range(21, 27):
+    for month in range(1, 13):
+        contracts.append(f"V{year:02d}{month:02d}")
+
+print(f"待下载合约数: {len(contracts)} 个 (V2101 ~ V2612)")
+print()
+
+# 下载所有合约
+all_data = []
+success = 0
+for i, code in enumerate(contracts):
+    try:
+        df = ak.futures_zh_minute_sina(symbol=code, period="5")
+        if df is not None and len(df) > 0:
+            df = df.copy()
+            df["_contract"] = code
+            all_data.append(df)
+            success += 1
+            if (i + 1) % 12 == 0 or i == len(contracts) - 1:
+                print(f"  进度: {i+1}/{len(contracts)} | 成功: {success}")
+    except Exception:
+        pass
+    time.sleep(0.15)
+
+print(f"\n下载完成: 成功 {success} 个合约, 总行数 {sum(len(d) for d in all_data)}")
+
+# 合并
+print("\n正在拼接主力合约序列...")
+combined = pd.concat(all_data, ignore_index=True)
+
+# 修复时间戳转换：显式按北京时间处理
+def to_unix_seconds(dt_str):
+    """将 '2020-09-17 21:35:00' 转为 Unix 秒（按北京时间 UTC+8）。"""
+    try:
+        dt = pd.to_datetime(dt_str)
+        # 显式设置时区为北京时间，再转 Unix 时间戳
+        return int(dt.replace(tzinfo=BJ_TZ).timestamp())
+    except Exception:
+        return 0
+
+combined["_ts"] = combined["datetime"].apply(to_unix_seconds)
+combined = combined[combined["_ts"] > 0].copy()  # 过滤无效
+
+# 按时间戳+成交量排序，去重时保留成交量最大的（近似主力合约）
+print("按成交量选择主力合约（去重）...")
+combined = combined.sort_values(["_ts", "volume"], ascending=[True, False])
+combined = combined.drop_duplicates(subset=["_ts"], keep="first")
+combined = combined.sort_values("_ts").reset_index(drop=True)
+
+print(f"去重后行数: {len(combined):,}")
+
+# 验证时间戳
+first_dt = pd.to_datetime(combined["datetime"].iloc[0])
+last_dt = pd.to_datetime(combined["datetime"].iloc[-1])
+print(f"时间范围: {first_dt} ~ {last_dt}")
+span_days = (last_dt - first_dt).days
+print(f"时间跨度: {span_days} 天（约 {span_days/365.25:.1f} 年）")
+
+# 验证时间戳正确性
+print(f"首行 Unix 时间戳: {combined['_ts'].iloc[0]}")
+print(f"首行时间戳验证: {pd.to_datetime(combined['_ts'].iloc[0], unit='s', utc=True).tz_convert(BJ_TZ)}")
+
+# 转换为项目兼容格式
+result = pd.DataFrame()
+result["time"] = combined["_ts"].astype("int64")
+result["open"] = combined["open"].astype("float64")
+result["high"] = combined["high"].astype("float64")
+result["low"] = combined["low"].astype("float64")
+result["close"] = combined["close"].astype("float64")
+result["tick_volume"] = combined["volume"].astype("int64")
+result = result.drop_duplicates(subset=["time"], keep="last").reset_index(drop=True)
+
+# 保存
+out_dir = Path(r"C:\Users\14795\Desktop\国内期货K线数据")
+out_dir.mkdir(parents=True, exist_ok=True)
+out_file = out_dir / "PVC_5min.parquet"
+result.to_parquet(out_file, index=False)
+
+# 统计
+print(f"\n{'=' * 60}")
+print(f"  下载完成!")
+print(f"{'=' * 60}")
+print(f"保存路径: {out_file}")
+print(f"总行数: {len(result):,}")
+print(f"时间范围: {first_dt} ~ {last_dt}")
+print(f"时间跨度: {span_days} 天（约 {span_days/365.25:.1f} 年）")
+
+print(f"\n价格统计:")
+print(f"  开盘价: {result['open'].min():.0f} ~ {result['open'].max():.0f}")
+print(f"  收盘价: {result['close'].min():.0f} ~ {result['close'].max():.0f}")
+print(f"  均价: {result['close'].mean():.0f}")
+
+# 按年统计
+result["_dt"] = pd.to_datetime(result["time"], unit="s", utc=True).dt.tz_convert(BJ_TZ)
+result["_year"] = result["_dt"].dt.year
+yearly = result.groupby("_year").size()
+print(f"\n各年数据量:")
+for yr, cnt in yearly.items():
+    print(f"  {yr}年: {cnt:,} 根")
+
+print(f"\n前 3 行:")
+print(result[["time","open","high","low","close","tick_volume"]].head(3).to_string())
+print(f"\n后 3 行:")
+print(result[["time","open","high","low","close","tick_volume"]].tail(3).to_string())

start_web.bat

@@ -1,53 +1,38 @@
 @echo off
-chcp 65001 >nul
-cd /d "%~dp0"
-
-REM Prefer: python -m pip ... (avoids pip-shim GBK issues on Chinese Windows)
-set HOST=127.0.0.1
-set PORT=8765
-
-REM Use system Python 3.13 (has all deps installed)
-set PYTHON=C:\Program Files\Python313\python.exe
+title AlphaMaster Web Server
+cd /d "C:\Alpha_master"
 
 echo ============================================================
-echo   AlphaMaster Web Console
-echo   http://%HOST%:%PORT%
+echo   AlphaMaster - Quant Factor Mining Center
+echo   Web Console Launcher
 echo ============================================================
 echo.
 
-echo [1/3] Releasing port %PORT% if occupied ...
-for /f "tokens=5" %%P in ('netstat -ano ^| findstr ":%PORT% " ^| findstr LISTENING') do (
-  echo   taskkill PID %%P
-  taskkill /F /PID %%P >nul 2>&1
+REM Check if Python is available
+where python >nul 2>&1
+if %errorlevel% neq 0 (
+  echo [ERROR] Python is not found on PATH.
+  echo Please install Python 3.11+ and add it to your system PATH.
+  echo.
+  pause
+  exit /b 1
 )
-timeout /t 1 /nobreak >nul
 
-echo [2/3] Starting uvicorn ...
-start "AlphaMaster Server" "%PYTHON%" run_web.py --host %HOST% --port %PORT%
+echo [1/2] Starting web server on http://127.0.0.1:8765 ...
+echo      The browser will open automatically in 3 seconds.
+echo.
+echo      Press Ctrl+C in this window to stop the server.
+echo ============================================================
+echo.
 
-echo [3/3] Warming up APIs (torch/fastapi first import can take ~6s) ...
-set /a TRIES=0
-:wait_ready
-set /a TRIES+=1
-if %TRIES% GTR 90 goto ready_fail
-timeout /t 1 /nobreak >nul
-"%PYTHON%" -c "import urllib.request; urllib.request.urlopen('http://%HOST%:%PORT%/api/overview', timeout=2).read(); urllib.request.urlopen('http://%HOST%:%PORT%/api/ai/providers', timeout=2).read(); urllib.request.urlopen('http://%HOST%:%PORT%/api/strategies', timeout=2).read()" 2>nul
-if errorlevel 1 goto wait_ready
+REM Open the browser after a 3-second delay (gives the server time to start)
+start "" cmd /c "ping 127.0.0.1 -n 4 >nul 2>&1 & start http://127.0.0.1:8765"
+
+REM Start the web server (blocks until Ctrl+C)
+python run_web.py --host 127.0.0.1 --port 8765
 
 echo.
-echo Ready. Opening browser ...
-start "" "http://%HOST%:%PORT%/"
-echo.
-echo Server window title: AlphaMaster Server
-echo Press any key to close this launcher (server keeps running)...
-pause >nul
-exit /b 0
-
-:ready_fail
-echo.
-echo ERROR: server did not become ready within ~90s.
-echo Check the "AlphaMaster Server" window for import/dependency errors.
-echo Tip: python -m pip install -r requirements.txt
-echo.
+echo ============================================================
+echo   Server has stopped.
+echo ============================================================
 pause
-exit /b 1

web/data_sources/domestic_futures_source.py

@@ -1,7 +1,7 @@
 """国内期货数据源（tqsdk 天勤量化，实时行情）。
 
-通过 tqsdk 实时拉取国内期货主连 K 线，支持 60 个品种 × 4 个周期。
-账号配置在 web_settings.json 的 tqsdk_user / tqsdk_password 字段。
+通过 tqsdk 实时拉取国内期货主连 K 线，支持 60 个品种 × 5 个周期。
+账号优先从 web_settings.json 读取，未配置时使用内置默认账号。
 
 特点：
   - 实时拉取最新 K 线（不是本地文件）
@@ -20,6 +20,7 @@ from web.data_sources.base import Bar, DataSource, DataSourceUnavailable
 
 # 周期映射：CANON_TIMEFRAMES → tqsdk duration_seconds
 _TF_TO_SECONDS = {
+    "5m": 300,
     "15m": 900,
     "30m": 1800,
     "1h": 3600,
@@ -61,17 +62,25 @@ _SYMBOL_MAP = {
 
 _PRESETS = list(_SYMBOL_MAP.keys())
 
+# 内置默认 tqsdk 账号（web_settings.json 未配置时使用）
+_DEFAULT_TQSDK_USER = "七斗居士"
+_DEFAULT_TQSDK_PASSWORD = "ghhkphs8"
+
 # web_settings.json 路径
 _SETTINGS_PATH = Path(__file__).resolve().parents[2] / "web_settings.json"
 
 
 def _load_credentials() -> tuple[str, str]:
-    """从 web_settings.json 读 tqsdk 账号密码。"""
+    """读取 tqsdk 账号密码。优先 web_settings.json，未配置则用内置默认账号。"""
     try:
         data = json.loads(_SETTINGS_PATH.read_text(encoding="utf-8"))
-        return str(data.get("tqsdk_user", "")), str(data.get("tqsdk_password", ""))
+        user = str(data.get("tqsdk_user", "")).strip()
+        pwd = str(data.get("tqsdk_password", "")).strip()
+        if user and pwd:
+            return user, pwd
     except Exception:
-        return "", ""
+        pass
+    return _DEFAULT_TQSDK_USER, _DEFAULT_TQSDK_PASSWORD
 
 
 class DomesticFuturesSource(DataSource):
@@ -94,14 +103,10 @@ class DomesticFuturesSource(DataSource):
             return (False, "未安装 tqsdk：pip install tqsdk")
         user, pwd = _load_credentials()
         if not user or not pwd:
-            return (
-                False,
-                "未配置 tqsdk 账号。请在 web_settings.json 设置 tqsdk_user / tqsdk_password"
-                "（信易账号，https://account.shinnytech.com 注册）",
-            )
+            return (False, "tqsdk 账号配置异常")
         return (
             True,
-            f"天勤量化实时行情 · 60 品种 · 4 周期（15m/30m/1h/1d）· 前复权",
+            f"天勤量化实时行情 · 60 品种 · 5 周期（5m/15m/30m/1h/1d）· 前复权",
         )
 
     def supported_timeframes(self) -> list[str]:
@@ -132,9 +137,7 @@ class DomesticFuturesSource(DataSource):
 
         user, pwd = _load_credentials()
         if not user or not pwd:
-            raise DataSourceUnavailable(
-                "未配置 tqsdk 账号。请在 web_settings.json 设置 tqsdk_user / tqsdk_password"
-            )
+            raise DataSourceUnavailable("tqsdk 账号配置异常")
 
         duration = _TF_TO_SECONDS[timeframe]
         want = max(n + 5, 20)  # 多拉几根防止 drop_forming 后不足

web/data_sources/factory.py

@@ -5,11 +5,10 @@ import threading
 
 from web.data_sources.base import DataSource
 
-# 前端下拉可见的数据源（okx / tongdaxin 实现保留，暂不展示）
+# 前端下拉可见的数据源（国内期货默认隐藏，如需启用加回列表即可）
 SOURCE_KINDS: tuple[tuple[str, str], ...] = (
     ("mt5", "MT5"),
     ("tradingview", "TradingView"),
-    ("domestic_futures", "国内期货"),
 )
 
 _INSTANCES: dict[str, DataSource] = {}

web/settings.py

@@ -24,8 +24,8 @@ _DEFAULT = {
     "feishu_webhook_url": "",
     "feishu_secret": "",
     # tqsdk 天勤量化账号（国内期货实时数据源）
-    "tqsdk_user": "",
-    "tqsdk_password": "",
+    "tqsdk_user": "七斗居士",
+    "tqsdk_password": "ghhkphs8",
 }
 
 

web/static/app.js

@@ -1942,6 +1942,8 @@ async function initRealtimeOnce() {
       sel.innerHTML = rtSources
         .map((s) => `<option value="${s.id}">${escHtml(s.label)}${s.available ? "" : " · 未就绪"}</option>`)
         .join("");
+      // 默认选第一个可用数据源
+      if (rtSources.length) sel.value = rtSources[0].id;
     }
     if (data.min_exposure != null && $("rtThresholdHint")) {
       $("rtThresholdHint").textContent = `无信号阈值 |tanh(因子)| < ${data.min_exposure}`;

web/static/index.html

@@ -567,6 +567,6 @@
   </div>
 
   <script src="/static/bg.js?v=1"></script>
-  <script src="/static/app.js?v=60"></script>
+  <script src="/static/app.js?v=61"></script>
 </body>
 </html>

web_settings.example.json

@@ -10,6 +10,6 @@
   "feishu_enabled": false,
   "feishu_webhook_url": "",
   "feishu_secret": "",
-  "tqsdk_user": "",
-  "tqsdk_password": ""
+  "tqsdk_user": "七斗居士",
+  "tqsdk_password": "ghhkphs8"
 }