Automated Replenishment Decision Engine — 备货决策自动化引擎：从规则到智能补货

"""
Automated Replenishment Decision Engine
备货决策自动化引擎：预测→优化→执行一体化
"""
import numpy as np
from dataclasses import dataclass
from typing import Optional


@dataclass
class SKUProfile:
    sku_id: str
    current_stock: float       # 当前库存（件）
    daily_demand_mean: float   # 日均需求
    daily_demand_std: float    # 需求标准差
    lead_time_days: int        # 供货周期
    unit_cost: float           # 单位成本（美元）
    min_order_qty: int         # 最小起订量
    holding_cost_rate: float = 0.02  # 日持货成本率
    stockout_penalty: float = 15.0   # 缺货惩罚（$/件）


@dataclass
class ReplenishmentDecision:
    sku_id: str
    recommended_qty: int
    order_date: str
    estimated_arrival: str
    risk_score: float
    auto_execute: bool
    rationale: str
    financial_impact: dict


def compute_safety_stock(sku: SKUProfile, service_level: float = 0.95) -> float:
    """计算安全库存"""
    from scipy import stats
    z = stats.norm.ppf(service_level)
    # 考虑需求和交期的双重不确定性
    safety_stock = z * sku.daily_demand_std * np.sqrt(sku.lead_time_days)
    return max(0, safety_stock)


def compute_optimal_order_qty(sku: SKUProfile, forecast_horizon: int = 60) -> int:
    """EOQ + 安全库存的最优补货量"""
    safety_stock = compute_safety_stock(sku)
    reorder_point = sku.daily_demand_mean * sku.lead_time_days + safety_stock
    
    # 判断是否需要补货
    days_of_stock = sku.current_stock / max(sku.daily_demand_mean, 0.1)
    if days_of_stock > sku.lead_time_days + 7:
        return 0  # 库存充足，无需补货
    
    # 目标库存水位（forecast_horizon 天的需求 + 安全库存）
    target_stock = sku.daily_demand_mean * forecast_horizon + safety_stock
    order_qty = max(0, target_stock - sku.current_stock)
    
    # 向上取整到 MOQ 的整数倍
    order_qty = max(sku.min_order_qty, 
                    int(np.ceil(order_qty / sku.min_order_qty)) * sku.min_order_qty)