aria-code 0.167.0__py3-none-win_amd64.whl

aria_code/tools/data_analysis_tools.py

@@ -0,0 +1,831 @@
+"""
+data_analysis_tools.py — 数据分析增强层
+==========================================
+提供：
+  - Excel 导出 (多 Sheet + 内嵌图表)
+  - DuckDB SQL 接口（本地内存数据仓库）
+  - 价格预警系统 (~/.arthera/alerts.json)
+  - 多资产组合回测 (相关性矩阵 + 权重收益)
+  - 相关性热力图数据
+  - 自定义因子表达式 DSL
+
+依赖（可选）：
+    pip install openpyxl duckdb pandas numpy yfinance
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import os
+from datetime import datetime, timedelta
+from pathlib import Path
+from typing import Any, Dict, List, Optional, Tuple
+from aria_code.packages.aria_core.paths import aria_home
+
+logger = logging.getLogger(__name__)
+
+# ── Optional imports ──────────────────────────────────────────────────────────
+
+try:
+    import pandas as pd
+    _HAS_PD = True
+except ImportError:
+    _HAS_PD = False
+
+try:
+    import numpy as np
+    _HAS_NP = True
+except ImportError:
+    _HAS_NP = False
+
+try:
+    import yfinance as yf
+    _HAS_YF = True
+except ImportError:
+    _HAS_YF = False
+
+try:
+    import duckdb
+    _HAS_DUCK = True
+except ImportError:
+    _HAS_DUCK = False
+
+try:
+    import openpyxl
+    from openpyxl.styles import PatternFill, Font, Alignment, Border, Side
+    from openpyxl.chart import LineChart, Reference
+    from openpyxl.utils import get_column_letter
+    _HAS_OPENPYXL = True
+except ImportError:
+    _HAS_OPENPYXL = False
+
+ALERTS_PATH = aria_home() / "alerts.json"
+EXPORT_DIR  = aria_home() / "exports"
+
+
+# ── Excel 导出 ────────────────────────────────────────────────────────────────
+
+def export_to_excel(params: dict) -> dict:
+    """
+    将数据导出到格式化 Excel 文件（多 Sheet）。
+
+    参数：
+      data:      dict，键=Sheet名，值=list[dict] 行数据
+      filename:  输出文件名（默认 aria_export_<timestamp>.xlsx）
+      add_chart: 是否为含 '价格'/'close'/'收盘' 列的 Sheet 添加折线图
+    """
+    if not _HAS_OPENPYXL:
+        return {"success": False, "error": "openpyxl 未安装，请运行: pip install openpyxl"}
+    if not _HAS_PD:
+        return {"success": False, "error": "pandas 未安装"}
+
+    data = params.get("data", {})
+    if not data:
+        return {"success": False, "error": "data 参数不能为空"}
+
+    EXPORT_DIR.mkdir(parents=True, exist_ok=True)
+    ts = datetime.now().strftime("%Y%m%d_%H%M%S")
+    filename = params.get("filename") or f"aria_export_{ts}.xlsx"
+    if not filename.endswith(".xlsx"):
+        filename += ".xlsx"
+    out_path = EXPORT_DIR / filename
+    add_chart = bool(params.get("add_chart", True))
+
+    wb = openpyxl.Workbook()
+    wb.remove(wb.active)  # remove default empty sheet
+
+    # Style constants
+    header_fill  = PatternFill("solid", fgColor="1F4E79")
+    header_font  = Font(color="FFFFFF", bold=True, size=10)
+    alt_fill     = PatternFill("solid", fgColor="EBF5FB")
+    center_align = Alignment(horizontal="center", vertical="center")
+    thin_border  = Border(
+        bottom=Side(style="thin", color="BDC3C7"),
+    )
+
+    for sheet_name, rows in data.items():
+        if not rows:
+            continue
+        ws = wb.create_sheet(title=str(sheet_name)[:31])  # Excel max 31 chars
+        df = pd.DataFrame(rows)
+        cols = list(df.columns)
+
+        # Header row
+        for ci, col in enumerate(cols, 1):
+            cell = ws.cell(row=1, column=ci, value=str(col))
+            cell.fill   = header_fill
+            cell.font   = header_font
+            cell.alignment = center_align
+
+        # Data rows
+        price_col_idx = None
+        for ri, row_dict in enumerate(rows, 2):
+            for ci, col in enumerate(cols, 1):
+                val = row_dict.get(col)
+                cell = ws.cell(row=ri, column=ci, value=val)
+                if ri % 2 == 0:
+                    cell.fill = alt_fill
+                cell.border = thin_border
+                # Auto-detect price column
+                col_lower = col.lower()
+                if price_col_idx is None and any(k in col_lower for k in
+                        ["价格","close","收盘","price","last","最新"]):
+                    price_col_idx = ci
+
+        # Auto-column width
+        for ci, col in enumerate(cols, 1):
+            max_len = max(len(str(col)),
+                          *(len(str(r.get(col,""))) for r in rows[:50]))
+            ws.column_dimensions[get_column_letter(ci)].width = min(max_len + 3, 40)
+
+        # Freeze header
+        ws.freeze_panes = "A2"
+
+        # Add line chart if price column detected
+        if add_chart and price_col_idx and len(rows) >= 3:
+            chart = LineChart()
+            chart.title  = f"{sheet_name} 价格走势"
+            chart.style  = 10
+            chart.y_axis.title = "价格"
+            chart.x_axis.title = "时间"
+            data_ref = Reference(ws, min_col=price_col_idx,
+                                  min_row=1, max_row=len(rows) + 1)
+            chart.add_data(data_ref, titles_from_data=True)
+            chart.width  = 20
+            chart.height = 12
+            ws.add_chart(chart, f"A{len(rows) + 4}")
+
+    # Summary sheet
+    ws_sum = wb.create_sheet(title="📊 汇总", index=0)
+    ws_sum["A1"] = "Aria Code 数据导出报告"
+    ws_sum["A1"].font = Font(bold=True, size=14, color="1F4E79")
+    ws_sum["A3"] = "导出时间："
+    ws_sum["B3"] = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
+    ws_sum["A4"] = "Sheet 数量："
+    ws_sum["B4"] = len(data)
+    ws_sum["A5"] = "总行数："
+    ws_sum["B5"] = sum(len(v) for v in data.values())
+    ws_sum["A7"] = "包含 Sheet："
+    for i, sname in enumerate(data.keys(), 8):
+        ws_sum[f"A{i}"] = f"  • {sname}"
+        ws_sum[f"B{i}"] = f"{len(data[sname])} 行"
+
+    wb.save(str(out_path))
+    return {
+        "success":   True,
+        "path":      str(out_path),
+        "sheets":    list(data.keys()),
+        "total_rows": sum(len(v) for v in data.values()),
+        "filename":  filename,
+    }
+
+
+# ── DuckDB SQL 接口 ───────────────────────────────────────────────────────────
+
+# Persistent in-memory DuckDB connection for the session
+_duck_conn: Optional[Any] = None
+_duck_tables: Dict[str, bool] = {}
+
+
+def _get_duck_conn():
+    global _duck_conn
+    if _duck_conn is None:
+        _duck_conn = duckdb.connect(":memory:")
+    return _duck_conn
+
+
+def sql_query(params: dict) -> dict:
+    """
+    在内存 DuckDB 中执行 SQL 查询。
+
+    参数：
+      query:   SQL 语句
+      load:    可选，dict {table_name: list[dict]} — 在查询前加载数据
+      limit:   结果行数限制（默认 500）
+    """
+    if not _HAS_DUCK:
+        return {"success": False,
+                "error": "duckdb 未安装，请运行: pip install duckdb"}
+    if not _HAS_PD:
+        return {"success": False, "error": "pandas 未安装"}
+
+    query = str(params.get("query", "")).strip()
+    if not query:
+        return {"success": False, "error": "query 不能为空"}
+
+    conn  = _get_duck_conn()
+    limit = int(params.get("limit", 500))
+    load  = params.get("load", {})
+
+    # Load tables
+    for tname, rows in (load or {}).items():
+        if rows:
+            df = pd.DataFrame(rows)
+            conn.register(tname, df)
+            _duck_tables[tname] = True
+
+    # Safety check: block destructive operations
+    q_upper = query.upper().lstrip()
+    if any(q_upper.startswith(kw) for kw in ("DROP ", "DELETE ", "TRUNCATE ")):
+        return {"success": False,
+                "error": "安全限制：不允许 DROP/DELETE/TRUNCATE 操作"}
+
+    # Auto-add LIMIT if SELECT without one
+    if q_upper.startswith("SELECT") and "LIMIT" not in q_upper:
+        query = f"{query.rstrip(';')} LIMIT {limit}"
+
+    try:
+        result = conn.execute(query).fetchdf()
+        records = result.to_dict("records")
+        cols    = list(result.columns)
+        return {
+            "success":  True,
+            "columns":  cols,
+            "rows":     records,
+            "row_count": len(records),
+            "tables_loaded": list(_duck_tables.keys()),
+        }
+    except Exception as e:
+        return {"success": False, "error": str(e), "query": query}
+
+
+def sql_list_tables(params: dict = None) -> dict:
+    """列出内存 DuckDB 中已加载的表。"""
+    if not _HAS_DUCK:
+        return {"success": False, "error": "duckdb 未安装"}
+    conn = _get_duck_conn()
+    try:
+        df = conn.execute("SHOW TABLES").fetchdf()
+        tables = df["name"].tolist() if "name" in df.columns else []
+        return {"success": True, "tables": tables}
+    except Exception as e:
+        return {"success": False, "error": str(e)}
+
+
+# ── 价格预警系统 ──────────────────────────────────────────────────────────────
+
+def _load_alerts() -> List[dict]:
+    if ALERTS_PATH.exists():
+        try:
+            return json.loads(ALERTS_PATH.read_text(encoding="utf-8"))
+        except Exception:
+            pass
+    return []
+
+
+def _save_alerts(alerts: list) -> None:
+    ALERTS_PATH.parent.mkdir(parents=True, exist_ok=True)
+    ALERTS_PATH.write_text(json.dumps(alerts, ensure_ascii=False, indent=2),
+                            encoding="utf-8")
+
+
+def add_price_alert(params: dict) -> dict:
+    """
+    添加价格预警。
+
+    参数：
+      symbol:     标的代码（如 "AAPL" / "600519"）
+      condition:  触发条件 ("gt" / "lt" / "cross_up" / "cross_down")
+      price:      触发价格
+      note:       备注（可选）
+    """
+    symbol    = str(params.get("symbol", "")).upper().strip()
+    condition = str(params.get("condition", "gt")).lower()
+    price     = float(params.get("price", 0))
+    note      = str(params.get("note", ""))
+
+    if not symbol:
+        return {"success": False, "error": "symbol 不能为空"}
+    if condition not in ("gt", "lt", "cross_up", "cross_down"):
+        return {"success": False,
+                "error": "condition 必须是 gt/lt/cross_up/cross_down"}
+    if price <= 0:
+        return {"success": False, "error": "price 必须 > 0"}
+
+    alerts = _load_alerts()
+    alert_id = f"{symbol}_{condition}_{price}_{int(datetime.now().timestamp())}"
+    new_alert = {
+        "id":        alert_id,
+        "symbol":    symbol,
+        "condition": condition,
+        "price":     price,
+        "note":      note,
+        "created_at": datetime.now().isoformat(),
+        "triggered": False,
+    }
+    alerts.append(new_alert)
+    _save_alerts(alerts)
+
+    cond_label = {"gt": ">", "lt": "<",
+                  "cross_up": "向上突破", "cross_down": "向下跌破"}[condition]
+    return {
+        "success":  True,
+        "alert_id": alert_id,
+        "message":  f"已设置预警：{symbol} 价格 {cond_label} {price}",
+        "total_alerts": len(alerts),
+    }
+
+
+def list_price_alerts(params: dict = None) -> dict:
+    """列出所有价格预警。"""
+    alerts = _load_alerts()
+    active  = [a for a in alerts if not a.get("triggered")]
+    done    = [a for a in alerts if a.get("triggered")]
+    return {
+        "success":        True,
+        "active_alerts":  active,
+        "triggered_alerts": done,
+        "total":          len(alerts),
+    }
+
+
+def delete_price_alert(params: dict) -> dict:
+    """删除指定 ID 的价格预警。"""
+    alert_id = str(params.get("alert_id", ""))
+    if not alert_id:
+        return {"success": False, "error": "alert_id 不能为空"}
+    alerts = _load_alerts()
+    before = len(alerts)
+    alerts = [a for a in alerts if a.get("id") != alert_id]
+    if len(alerts) == before:
+        return {"success": False, "error": f"未找到预警 {alert_id}"}
+    _save_alerts(alerts)
+    return {"success": True, "deleted_id": alert_id, "remaining": len(alerts)}
+
+
+def check_alerts(params: dict = None) -> dict:
+    """
+    检查当前所有未触发预警的状态（需联网获取实时价格）。
+    返回已触发的预警列表。
+    """
+    if not _HAS_YF:
+        return {"success": False, "error": "yfinance 未安装，无法获取实时价格"}
+
+    alerts  = _load_alerts()
+    active  = [a for a in alerts if not a.get("triggered")]
+    if not active:
+        return {"success": True, "triggered": [], "message": "无活跃预警"}
+
+    # Batch price fetch
+    symbols = list(set(a.get("symbol", "") for a in active if a.get("symbol")))
+    prices: Dict[str, float] = {}
+    try:
+        tickers = yf.Tickers(" ".join(symbols))
+        for sym in symbols:
+            try:
+                info = tickers.tickers[sym].fast_info
+                px = getattr(info, "last_price", None) or \
+                     getattr(info, "regularMarketPrice", None)
+                if px:
+                    prices[sym] = float(px)
+            except Exception:
+                pass
+    except Exception as e:
+        logger.debug("Alert price fetch failed: %s", e)
+
+    triggered_now = []
+    for alert in active:
+        sym  = alert.get("symbol")
+        cond = alert.get("condition")
+        tgt  = alert.get("price")
+        if not sym or not cond or tgt is None:
+            continue
+        cur  = prices.get(sym)
+        if cur is None:
+            continue
+        hit = False
+        if cond == "gt" and cur > tgt:
+            hit = True
+        elif cond == "lt" and cur < tgt:
+            hit = True
+        elif cond == "cross_up" and cur >= tgt:
+            hit = True
+        elif cond == "cross_down" and cur <= tgt:
+            hit = True
+        if hit:
+            alert["triggered"]    = True
+            alert["triggered_at"] = datetime.now().isoformat()
+            alert["triggered_price"] = cur
+            triggered_now.append(alert)
+
+    _save_alerts(alerts)
+
+    # Push notifications for newly triggered alerts
+    if triggered_now:
+        try:
+            from notification_tools import send_alert_notification
+            for _alrt in triggered_now:
+                send_alert_notification(_alrt)
+        except Exception as _ne:
+            logger.debug("Notification dispatch failed: %s", _ne)
+
+    return {
+        "success":   True,
+        "triggered": triggered_now,
+        "checked":   len(active),
+        "prices":    prices,
+    }
+
+
+# ── 相关性矩阵 ────────────────────────────────────────────────────────────────
+
+def calc_correlation_matrix(params: dict) -> dict:
+    """
+    计算多资产收益率相关性矩阵。
+
+    参数：
+      symbols:   list[str]，如 ["AAPL","MSFT","TSLA","SPY"]
+      period:    历史区间 ("1y" / "2y" / "6mo"，默认 "1y")
+      interval:  频率 ("1d" / "1wk"，默认 "1d")
+    """
+    if not _HAS_YF:
+        return {"success": False, "error": "yfinance 未安装"}
+    if not _HAS_PD or not _HAS_NP:
+        return {"success": False, "error": "pandas / numpy 未安装"}
+
+    symbols  = params.get("symbols", [])
+    period   = str(params.get("period", "1y"))
+    interval = str(params.get("interval", "1d"))
+
+    if len(symbols) < 2:
+        return {"success": False, "error": "至少需要 2 个标的"}
+    if len(symbols) > 20:
+        symbols = symbols[:20]
+
+    try:
+        raw = yf.download(symbols, period=period, interval=interval,
+                          progress=False, auto_adjust=True)
+        if raw.empty:
+            return {"success": False, "error": "下载数据为空"}
+
+        if isinstance(raw.columns, pd.MultiIndex):
+            closes = raw["Close"]
+        else:
+            closes = raw
+
+        closes = closes.dropna(how="all", axis=1)
+        rets   = closes.pct_change().dropna()
+        corr   = rets.corr()
+
+        # Convert to serializable format
+        corr_dict = {}
+        for sym in corr.columns:
+            corr_dict[sym] = {
+                other: round(float(corr.loc[sym, other]), 4)
+                for other in corr.columns
+            }
+
+        # Summary stats
+        stats = {}
+        for sym in closes.columns:
+            s = closes[sym].dropna()
+            r = rets[sym].dropna() if sym in rets.columns else pd.Series(dtype=float)
+            if s.empty:
+                continue
+            stats[sym] = {
+                "return_total": round((s.iloc[-1]/s.iloc[0] - 1)*100, 2),
+                "volatility":   round(_annual_vol(r) * 100, 2) if len(r) > 1 else None,
+                "sharpe":       _calc_sharpe(r),
+                "max_drawdown": round(_max_drawdown(s) * 100, 2),
+            }
+
+        return {
+            "success":    True,
+            "symbols":    list(closes.columns),
+            "period":     period,
+            "interval":   interval,
+            "corr_matrix": corr_dict,
+            "stats":       stats,
+            "data_points": len(rets),
+        }
+    except Exception as e:
+        return {"success": False, "error": str(e)}
+
+
+# ── 多资产组合回测 ─────────────────────────────────────────────────────────────
+
+def portfolio_backtest(params: dict) -> dict:
+    """
+    多资产组合历史回测。
+
+    参数：
+      symbols:   list[str]，如 ["AAPL","MSFT","GOOG"]
+      weights:   list[float]，权重（自动归一化）；为空则等权
+      period:    "1y" / "2y" / "3y" / "5y"（默认 "2y"）
+      benchmark: 基准标的（默认 "SPY"）
+      rebalance: 再平衡频率 ("monthly" / "quarterly" / "none"，默认 "monthly")
+    """
+    if not _HAS_YF or not _HAS_PD or not _HAS_NP:
+        return {"success": False, "error": "需要 yfinance + pandas + numpy"}
+
+    symbols   = params.get("symbols", [])
+    weights_r = params.get("weights", [])
+    period    = str(params.get("period", "2y"))
+    benchmark = str(params.get("benchmark", "SPY"))
+    rebalance = str(params.get("rebalance", "monthly"))
+
+    if not symbols:
+        return {"success": False, "error": "symbols 不能为空"}
+    if len(symbols) > 15:
+        symbols = symbols[:15]
+
+    # Normalize weights
+    if weights_r and len(weights_r) == len(symbols):
+        ws = [float(w) for w in weights_r]
+        total = sum(ws)
+        weights = [w / total for w in ws]
+    else:
+        weights = [1 / len(symbols)] * len(symbols)
+
+    all_syms = list(dict.fromkeys(symbols + [benchmark]))
+    try:
+        raw = yf.download(all_syms, period=period, progress=False, auto_adjust=True)
+        if raw.empty:
+            return {"success": False, "error": "数据下载失败"}
+
+        closes = raw["Close"] if isinstance(raw.columns, pd.MultiIndex) else raw
+        closes = closes.dropna(how="all", axis=1)
+
+        # Align symbols with available data
+        avail = [s for s in symbols if s in closes.columns]
+        if not avail:
+            return {"success": False, "error": "所有标的数据均不可用"}
+        if len(avail) < len(symbols):
+            missing = [s for s in symbols if s not in closes.columns]
+            weights = [weights[i] for i, s in enumerate(symbols) if s in closes.columns]
+            wt_total = sum(weights)
+            weights = [w / wt_total for w in weights]
+            symbols = avail
+
+        port_data = closes[symbols].ffill().dropna()
+        if benchmark in closes.columns:
+            bench_data = closes[benchmark].ffill().dropna()
+        else:
+            bench_data = None
+
+        # Rebalance
+        if rebalance == "monthly":
+            freq = "MS"
+        elif rebalance == "quarterly":
+            freq = "QS"
+        else:
+            freq = None
+
+        # Compute portfolio returns
+        if freq and len(port_data) > 0:
+            port_rets = port_data.pct_change().dropna()
+            # Monthly rebalancing: compound weighted returns per period
+            port_monthly_rets = (port_rets + 1).resample(freq).prod() - 1
+            port_rets_w = (port_monthly_rets[symbols] * weights).sum(axis=1)
+            # Expand back to daily for metrics
+            port_cum = (1 + port_rets_w).cumprod()
+        else:
+            port_rets = port_data.pct_change().dropna()
+            port_rets_w = (port_rets[symbols] * weights).sum(axis=1)
+            port_cum = (1 + port_rets_w).cumprod()
+
+        total_return = float(port_cum.iloc[-1] - 1) * 100
+        annual_vol   = _annual_vol(port_rets_w) * 100
+        sharpe       = _calc_sharpe(port_rets_w)
+        max_dd       = round(_max_drawdown(port_cum) * 100, 2)
+
+        # Benchmark
+        bench_stats = None
+        if bench_data is not None:
+            bench_rets = bench_data.pct_change().dropna()
+            b_cum = (1 + bench_rets).cumprod()
+            bench_stats = {
+                "symbol":     benchmark,
+                "total_return": round((float(b_cum.iloc[-1]) - 1)*100, 2),
+                "volatility": round(_annual_vol(bench_rets) * 100, 2),
+                "sharpe":     _calc_sharpe(bench_rets),
+                "max_drawdown": round(_max_drawdown(b_cum) * 100, 2),
+            }
+
+        # Allocation breakdown
+        allocation = [{"symbol": s, "weight_pct": round(w*100, 1),
+                       "weight_abs": w} for s, w in zip(symbols, weights)]
+
+        # Performance curve (monthly points)
+        if freq:
+            curve_pts = [{"date": str(d.date()), "value": round(float(v), 4)}
+                         for d, v in port_cum.items()]
+        else:
+            curve_pts = [{"date": str(d.date()), "value": round(float(v), 4)}
+                         for d, v in port_cum.iloc[::5].items()]  # every 5 days
+
+        return {
+            "success":     True,
+            "symbols":     symbols,
+            "weights":     weights,
+            "allocation":  allocation,
+            "period":      period,
+            "rebalance":   rebalance,
+            "portfolio": {
+                "total_return_pct": round(total_return, 2),
+                "annual_vol_pct":   round(annual_vol, 2),
+                "sharpe_ratio":     sharpe,
+                "max_drawdown_pct": max_dd,
+                "calmar_ratio":     round(total_return / abs(max_dd), 2) if max_dd != 0 else None,
+            },
+            "benchmark":   bench_stats,
+            "curve":       curve_pts[-60:],  # last 60 points for chart
+            "data_points": len(port_rets_w),
+        }
+    except Exception as e:
+        logger.exception("Portfolio backtest error")
+        return {"success": False, "error": str(e)}
+
+
+# ── 自定义因子 DSL ─────────────────────────────────────────────────────────────
+
+def eval_custom_factor(params: dict) -> dict:
+    """
+    计算自定义因子表达式。
+
+    参数：
+      symbol:  标的代码
+      period:  历史区间（默认 "1y"）
+      expr:    因子表达式字符串（Python 数学语法，可使用变量）
+               可用变量：close, open, high, low, volume, returns
+               可用函数：sma(n), ema(n), std(n), rsi(n), atr(n)
+               示例：
+                 "(close - sma(20)) / std(20)"     → Z-Score
+                 "sma(5) / sma(20) - 1"            → 短期动量
+                 "volume / sma(20, volume)"         → 量比
+    """
+    if not _HAS_YF or not _HAS_PD or not _HAS_NP:
+        return {"success": False, "error": "需要 yfinance + pandas + numpy"}
+
+    symbol = str(params.get("symbol", "")).upper()
+    period = str(params.get("period", "1y"))
+    expr   = str(params.get("expr", "")).strip()
+
+    if not symbol or not expr:
+        return {"success": False, "error": "symbol 和 expr 均为必填"}
+
+    # Security: block imports and exec
+    forbidden = ["import", "__", "exec(", "eval(", "open(", "os.", "sys."]
+    for tok in forbidden:
+        if tok in expr:
+            return {"success": False, "error": f"表达式包含禁止关键字: {tok}"}
+
+    try:
+        tkr = yf.Ticker(symbol)
+        hist = tkr.history(period=period, auto_adjust=True)
+        if hist.empty:
+            return {"success": False, "error": f"无法获取 {symbol} 历史数据"}
+
+        close  = hist["Close"].astype(float)
+        open_  = hist["Open"].astype(float)
+        high   = hist["High"].astype(float)
+        low    = hist["Low"].astype(float)
+        volume = hist["Volume"].astype(float)
+        returns = close.pct_change()
+
+        def sma(n, series=None):
+            s = series if series is not None else close
+            return s.rolling(n).mean()
+
+        def ema(n, series=None):
+            s = series if series is not None else close
+            return s.ewm(span=n, adjust=False).mean()
+
+        def std(n, series=None):
+            s = series if series is not None else close
+            return s.rolling(n).std()
+
+        def rsi(n=14):
+            delta = close.diff()
+            gain  = delta.clip(lower=0).rolling(n).mean()
+            loss  = (-delta.clip(upper=0)).rolling(n).mean()
+            rs    = gain / loss
+            return 100 - (100 / (1 + rs))
+
+        def atr(n=14):
+            tr = pd.concat([
+                high - low,
+                (high - close.shift()).abs(),
+                (low  - close.shift()).abs(),
+            ], axis=1).max(axis=1)
+            return tr.rolling(n).mean()
+
+        ns = dict(
+            close=close, open=open_, high=high, low=low,
+            volume=volume, returns=returns,
+            sma=sma, ema=ema, std=std, rsi=rsi, atr=atr,
+            pd=pd, np=np, abs=abs, min=min, max=max,
+        )
+
+        result_series = eval(expr, {"__builtins__": {}}, ns)  # noqa: S307
+        if hasattr(result_series, "dropna"):
+            result_series = result_series.dropna()
+
+        latest_val = float(result_series.iloc[-1]) if len(result_series) > 0 else None
+
+        # Last 20 values for chart
+        tail = result_series.tail(20)
+        series_pts = [{"date": str(d.date()), "value": round(float(v), 6)}
+                      for d, v in tail.items()]
+
+        return {
+            "success":   True,
+            "symbol":    symbol,
+            "expr":      expr,
+            "latest":    latest_val,
+            "series":    series_pts,
+            "data_len":  len(result_series),
+        }
+    except Exception as e:
+        return {"success": False, "error": f"表达式计算失败: {e}"}
+
+
+# ── 通用数据导入 ───────────────────────────────────────────────────────────────
+
+def load_csv_data(params: dict) -> dict:
+    """
+    从 CSV 文件加载数据到内存 DuckDB 表。
+
+    参数：
+      path:       CSV 文件路径
+      table_name: 目标表名（默认由文件名推导）
+      encoding:   编码（默认 utf-8）
+    """
+    if not _HAS_PD:
+        return {"success": False, "error": "pandas 未安装"}
+
+    path = Path(str(params.get("path", ""))).expanduser()
+    if not path.exists():
+        return {"success": False, "error": f"文件不存在: {path}"}
+    if path.suffix.lower() not in (".csv", ".tsv", ".txt"):
+        return {"success": False, "error": "仅支持 CSV/TSV 文件"}
+
+    enc   = params.get("encoding", "utf-8")
+    tname = str(params.get("table_name", path.stem))
+
+    try:
+        df = pd.read_csv(str(path), encoding=enc)
+        if _HAS_DUCK:
+            conn = _get_duck_conn()
+            conn.register(tname, df)
+            _duck_tables[tname] = True
+
+        return {
+            "success":    True,
+            "table_name": tname,
+            "rows":       len(df),
+            "columns":    list(df.columns),
+            "sample":     df.head(3).to_dict("records"),
+            "path":       str(path),
+        }
+    except Exception as e:
+        return {"success": False, "error": str(e)}
+
+
+# ── Private Helpers ───────────────────────────────────────────────────────────
+
+def _periods_per_year(rets) -> float:
+    """Observations a year in this return series, read from its index.
+
+    Monthly-rebalanced portfolio returns are monthly, so scaling them by
+    sqrt(252) overstated volatility and Sharpe about 4.6x. An index that is not
+    dates falls back to 252 (daily), the old assumption.
+    """
+    from aria_code.domain import performance_stats as stats
+
+    try:
+        idx = rets.index
+        # N dated returns span N - 1 intervals.
+        found = stats.periods_per_year_from_dates(len(rets) - 1, idx[0], idx[-1])
+    except Exception:
+        found = None
+    return found if found else float(stats.FALLBACK_PERIODS_PER_YEAR)
+
+
+def _annual_vol(rets) -> float:
+    return float(rets.std() * np.sqrt(_periods_per_year(rets)))
+
+
+def _calc_sharpe(rets, rf_daily: float = 0.0) -> Optional[float]:
+    if not _HAS_NP or not hasattr(rets, "__len__") or len(rets) < 2:
+        return None
+    try:
+        excess = rets - rf_daily
+        if float(excess.std()) == 0:
+            return None
+        return round(float(excess.mean() / excess.std() * np.sqrt(_periods_per_year(rets))), 3)
+    except Exception:
+        return None
+
+
+def _max_drawdown(cum_series) -> float:
+    if not _HAS_NP or not _HAS_PD:
+        return 0.0
+    try:
+        s = pd.Series(cum_series)
+        roll_max = s.cummax()
+        drawdown = (s - roll_max) / roll_max
+        return float(drawdown.min())
+    except Exception:
+        return 0.0