import pandas as pd
import random
import math

work_order_excel="data/excel/6万工单记录.xlsx"

soft_row_data={
    "博微配网计价通D3":{"基本功能":[], "高级功能":[]},
    "储能C1软件":{"基本功能":[], "高级功能":[]},
    "西藏计价通Z1":{"基本功能":[], "高级功能":[]},
    "技改检修工程计价通T1":{"基本功能":[], "高级功能":[]},
    "检修清单计价通T1":{"基本功能":[], "高级功能":[]},
    "电力建设计价通软件":{"基本功能":[], "高级功能":[]},
}

df = pd.read_excel(work_order_excel)

for idx, row in df.iterrows():
    if pd.isna(row["产品线"]):
        continue

    if "博微配网计价通D3" in row["产品线"]:
        soft_row_data["博微配网计价通D3"][row["问题类型"]].append((idx, row))
    elif "博微电力建设计价通软件" in row["产品线"]:
        soft_row_data["电力建设计价通软件"][row["问题类型"]].append((idx, row))
    elif "新能源系列" in row["产品线"] and "博微新型储能电站建设计价通C1软件" in row["产品名称"]:
        soft_row_data["储能C1软件"][row["问题类型"]].append((idx, row))
    elif "博微西藏计价通Z1" in row["产品线"]:
        soft_row_data["西藏计价通Z1"][row["问题类型"]].append((idx, row))
    elif "博微技改检修计价通T1软件" in row["产品线"] and "技改检修计价通T1软件-概预算" in row["产品名称"]:
        soft_row_data["技改检修工程计价通T1"][row["问题类型"]].append((idx, row))
    elif "博微技改检修计价通T1软件" in row["产品线"] and "技改检修计价通T1软件-清单" in row["产品名称"]:
        soft_row_data["检修清单计价通T1"][row["问题类型"]].append((idx, row))

# 计算每个软件和功能类型的数据量
total_count = 0
counts = {}
for software, types in soft_row_data.items():
    counts[software] = {}
    for type_name, rows in types.items():
        counts[software][type_name] = len(rows)
        total_count += len(rows)

print(f"原始数据总量: {total_count}条")
for software, types in counts.items():
    print(f"{software}: 基本功能 {types['基本功能']}条, 高级功能 {types['高级功能']}条")

# 计算均衡提取的数量
total_target = 2000
categories_count = sum(len(types) for types in soft_row_data.values())
per_category_target = math.ceil(total_target / categories_count)

# 均衡提取数据
balanced_data = []
extracted_counts = {}
extracted_indices = set()  # 使用集合存储已提取数据的索引

for software, types in soft_row_data.items():
    extracted_counts[software] = {}
    
    for type_name, rows in types.items():
        # 如果数据量不足，全部提取；否则随机抽取目标数量
        if len(rows) <= per_category_target:
            extracted = rows
        else:
            extracted = random.sample(rows, per_category_target)
        
        extracted_counts[software][type_name] = len(extracted)
        for idx, row in extracted:
            extracted_indices.add(idx)  # 记录已提取数据的索引
            balanced_data.append(row)

# 数据量不足2000时，从剩余数据中补充
remaining_target = total_target - len(balanced_data)
if remaining_target > 0:
    # 收集所有未被选中的数据
    remaining_data = []
    for software, types in soft_row_data.items():
        for type_name, rows in types.items():
            # 添加未被选中的数据
            for idx, row in rows:
                if idx not in extracted_indices:
                    remaining_data.append(row)
    
    # 如果剩余数据足够，随机抽取补充
    if len(remaining_data) >= remaining_target:
        additional_data = random.sample(remaining_data, remaining_target)
    else:
        additional_data = remaining_data
    
    balanced_data.extend(additional_data)

# 输出结果
print(f"\n均衡提取后数据总量: {len(balanced_data)}条")
for software, types in extracted_counts.items():
    print(f"{software}: 基本功能 {types['基本功能']}条, 高级功能 {types['高级功能']}条")

# 将均衡提取的数据转换为DataFrame并保存
balanced_df = pd.DataFrame(balanced_data)
balanced_df.to_excel("data/excel/均衡提取2000条工单.xlsx", index=False)
print(f"\n已将均衡提取的{len(balanced_data)}条数据保存至'data/excel/均衡提取2000条工单.xlsx'")