SBrainCO/db/pipeline/import_monthly_data.py

#!/usr/bin/env python3
"""
通用月度数据导入脚本
支持: 库存成本、菜品销售明细、菜品成本BOM、营业费用、中央厨房、配送明细、采购货品
用法: python3 import_monthly_data.py --dataset inventory --month 2026-04-01 --file <path> --db bill_query_test
"""
import argparse
import hashlib
import os
import sys
import openpyxl
import pandas as pd
import psycopg2
from psycopg2.extras import execute_values
from datetime import datetime

DB_CONFIG = {
    'host': 'localhost',
    'port': 5432,
    'user': 'freedak',
    'password': '',
}

def file_sha256(filepath):
    h = hashlib.sha256()
    with open(filepath, 'rb') as f:
        for chunk in iter(lambda: f.read(8192), b''):
            h.update(chunk)
    return h.hexdigest()

def to_num(val):
    if val is None or val == '':
        return None
    if isinstance(val, float) and pd.isna(val):
        return None
    try:
        f = float(val)
        import math
        if math.isinf(f) or math.isnan(f):
            return None
        return f
    except (ValueError, TypeError):
        return None

def to_text(val):
    if val is None:
        return None
    if isinstance(val, float) and pd.isna(val):
        return None
    s = str(val).strip()
    if s.lower() == 'nan' or s.lower() == 'none':
        return None
    return s if s else None

def to_date(val):
    """Convert value to date string YYYY-MM-DD or None"""
    if val is None:
        return None
    if isinstance(val, float) and pd.isna(val):
        return None
    if isinstance(val, (datetime,)):
        return val.strftime('%Y-%m-%d')
    s = str(val).strip()
    if not s or s.lower() == 'nan' or s.lower() == 'none':
        return None
    # Try pandas Timestamp
    try:
        ts = pd.Timestamp(s)
        if pd.isna(ts):
            return None
        return ts.strftime('%Y-%m-%d')
    except (ValueError, TypeError):
        return s  # Return as-is, let DB handle it

def clean_percent(val):
    """Convert percentage value: string '81.23%' -> 81.23, float 0.8123 -> 81.23"""
    if val is None:
        return None
    if isinstance(val, float) and pd.isna(val):
        return None
    if isinstance(val, str):
        s = val.strip().replace('%', '')
        if not s or s.lower() == 'nan':
            return None
        try:
            return float(s)
        except ValueError:
            return None
    num = float(val)
    return num * 100 if abs(num) <= 1 else num

def connect(dbname='bill_query'):
    cfg = DB_CONFIG.copy()
    cfg['dbname'] = dbname
    return psycopg2.connect(**cfg)

# ============================================================
# 库存成本导入
# ============================================================
def import_inventory_cost(conn, filepath, report_month):
    """库存倒挤成本导入"""
    print(f"\n=== 库存成本导入 ===")
    print(f"文件: {filepath}")
    print(f"月份: {report_month}")
    
    sha = file_sha256(filepath)
    source_file = os.path.basename(filepath)
    
    wb = openpyxl.load_workbook(filepath, read_only=True, data_only=True)
    ws = wb.active
    
    # Row 2: group headers, Row 3: sub-headers, Row 4+: data
    rows = list(ws.iter_rows(min_row=4, values_only=True))
    data_rows = [r for r in rows if r[4] is not None and str(r[4]).strip()]
    print(f"数据行数: {len(data_rows)}")
    
    cur = conn.cursor()
    # Check if already imported
    cur.execute(
        "SELECT 1 FROM public.inventory_cost_import_log WHERE source_file = %s AND report_month = %s",
        (source_file, report_month)
    )
    if cur.fetchone():
        print("已导入过，跳过")
        wb.close()
        return
    
    # Insert import log
    cur.execute(
        """INSERT INTO public.inventory_cost_import_log (source_file, report_month, source_rows, imported_rows, status)
           VALUES (%s, %s, %s, %s, 'importing')""",
        (source_file, report_month, len(data_rows), len(data_rows))
    )
    
    # Batch insert
    insert_sql = """INSERT INTO public.inventory_cost_records (
        report_month, source_file, source_row,
        brand, region, cost_unit_name, cost_unit_code, item_name, specification,
        major_category, minor_category, finance_category, item_code, unit,
        opening_quantity, opening_unit_cost, opening_amount, opening_tax, opening_tax_inclusive,
        purchase_quantity, purchase_unit_cost, purchase_amount, purchase_tax, purchase_tax_inclusive,
        ending_quantity, ending_unit_cost, ending_amount, ending_tax, ending_tax_inclusive,
        consumption_quantity, consumption_unit_cost, consumption_amount, consumption_tax, consumption_tax_inclusive,
        conversion_precision_amount, conversion_precision_tax_inclusive,
        return_loss_amount, return_loss_tax_inclusive
    ) VALUES %s"""
    
    batch = []
    batch_size = 500
    for idx, r in enumerate(data_rows):
        source_row = idx + 4
        consumption_amount = to_num(r[28])
        vals = (
            report_month, source_file, source_row,
            to_text(r[0]), to_text(r[1]), to_text(r[2]), to_text(r[3]),
            to_text(r[4]), to_text(r[5]), to_text(r[6]), to_text(r[7]), to_text(r[8]),
            to_text(r[9]), to_text(r[10]),
            to_num(r[11]), to_num(r[12]), to_num(r[13]), to_num(r[14]), to_num(r[15]),
            to_num(r[16]), to_num(r[17]), to_num(r[18]), to_num(r[19]), to_num(r[20]),
            to_num(r[21]), to_num(r[22]), to_num(r[23]), to_num(r[24]), to_num(r[25]),
            to_num(r[26]), to_num(r[27]), consumption_amount, to_num(r[29]), to_num(r[30]),
            to_num(r[31]), to_num(r[32]),
            to_num(r[33]), to_num(r[34]),
        )
        batch.append(vals)
        if len(batch) >= batch_size:
            execute_values(cur, insert_sql, batch, page_size=500)
            conn.commit()
            print(f"  已导入 {idx + 1}/{len(data_rows)} 行")
            batch = []
    
    if batch:
        execute_values(cur, insert_sql, batch, page_size=500)
        conn.commit()
        print(f"  已导入 {len(data_rows)}/{len(data_rows)} 行")
    
    # Update log status
    cur.execute(
        "UPDATE public.inventory_cost_import_log SET status = 'completed' WHERE source_file = %s AND report_month = %s",
        (source_file, report_month)
    )
    conn.commit()
    wb.close()
    print(f"库存成本导入完成: {len(data_rows)} 行")

# ============================================================
# 菜品销售明细导入
# ============================================================
def import_dish_sales(conn, filepath, report_month, source_file_label=None):
    """菜品销售明细导入（支持单文件或目录）"""
    print(f"\n=== 菜品销售明细导入 ===")
    
    files = []
    if os.path.isdir(filepath):
        for f in sorted(os.listdir(filepath)):
            if f.endswith('.xlsx') and not f.startswith('.'):
                files.append(os.path.join(filepath, f))
    else:
        files.append(filepath)
    
    print(f"文件数: {len(files)}")
    
    cur = conn.cursor()
    total_imported = 0
    
    for file_idx, fpath in enumerate(files):
        fname = os.path.basename(fpath)
        print(f"\n[{file_idx+1}/{len(files)}] {fname}")
        
        file_size = os.path.getsize(fpath)
        
        # Check if already imported
        cur.execute(
            "SELECT 1 FROM public.dish_sales_import_log WHERE source_file = %s",
            (fname,)
        )
        if cur.fetchone():
            print(f"  已导入过, 跳过")
            continue
        
        # Use pandas to read Excel (handles read_only issues with these files)
        # Header is on row 3 (0-indexed: header=2)
        df = pd.read_excel(fpath, header=2, engine='openpyxl')
        # Drop rows where first column is empty
        df = df.dropna(subset=[df.columns[0]])
        data_rows = df.values.tolist()
        headers = list(df.columns)
        print(f"  数据行数: {len(data_rows)}")
        
        if not data_rows:
            continue
        
        # Insert import log
        cur.execute(
            """INSERT INTO public.dish_sales_import_log (source_file, file_size_bytes, data_rows, status)
               VALUES (%s, %s, %s, 'importing')""",
            (fname, file_size, len(data_rows))
        )
        
        # Map headers to columns by name
        col_map = {}
        for i, h in enumerate(headers):
            if h and str(h).strip():
                col_map[str(h).strip()] = i
        
        # Build column mapping based on known header names
        field_mapping = {
            '门店编码': 'store_code',
            '门店名称': 'store_name',
            '门店': 'store_name',
            '账单号': 'bill_no',
            '订单号': 'bill_no',
            '人数': 'guest_count',
            '开台时间': 'opened_at',
            '结账时间': 'closed_at',
            '下单时间': 'ordered_at',
            '菜品名称': 'dish_name',
            '品项名称': 'dish_name',
            '分类': 'category_level1',
            '菜品分类': 'category_level1',
            '出品部门': 'production_department',
            '销量': 'sales_quantity',
            '金额': 'gross_amount',
            '实收金额': 'received_amount',
            '实收': 'received_amount',
            '单价': 'unit_price',
            '单位': 'unit',
            '业务类型': 'business_type',
            '桌台/取餐号': 'table_or_pickup_no',
            '台号': 'table_or_pickup_no',
            '做法': 'preparation_method',
        }
        
        # Determine actual columns
        cols_to_insert = {}
        for h_name, db_col in field_mapping.items():
            if h_name in col_map:
                cols_to_insert[db_col] = col_map[h_name]
        
        # source_file + source_row + mapped columns
        all_cols = ['source_file', 'source_row'] + list(cols_to_insert.keys())
        col_names = ','.join(all_cols)
        
        numeric_cols = {'guest_count', 'sales_quantity', 'gross_amount', 'received_amount', 'unit_price'}
        
        batch = []
        for idx, r in enumerate(data_rows):
            source_row = idx + 4  # data starts at row 4 (header=3)
            vals = [fname, source_row]
            for db_col, excel_idx in cols_to_insert.items():
                v = r[excel_idx] if excel_idx < len(r) else None
                if db_col in numeric_cols:
                    vals.append(to_num(v))
                else:
                    vals.append(to_text(v))
            batch.append(tuple(vals))
            if len(batch) >= 500:
                execute_values(cur, f"INSERT INTO public.dish_sales_details ({col_names}) VALUES %s", batch, page_size=500)
                conn.commit()
                print(f"  已导入 {idx + 1}/{len(data_rows)} 行")
                batch = []
        
        if batch:
            execute_values(cur, f"INSERT INTO public.dish_sales_details ({col_names}) VALUES %s", batch, page_size=500)
            conn.commit()
            print(f"  已导入 {len(data_rows)}/{len(data_rows)} 行")
        
        total_imported += len(data_rows)
        
        # Update log status
        cur.execute(
            "UPDATE public.dish_sales_import_log SET status = 'completed' WHERE source_file = %s",
            (fname,)
        )
        conn.commit()
    
    print(f"\n菜品销售明细导入完成: 共 {total_imported} 行")

# ============================================================
# 菜品成本/BOM导入
# ============================================================
def import_dish_cost(conn, filepath, report_month):
    """菜品成本分析导入：同时导入summary和detail两张表
    Excel结构：Row0=标题, Row1=主表头, Row2=子表头(原料区), Row3+=数据
    每个菜品有多行原料明细，菜品列(col0-12)只在首行填充"""
    print(f"\n=== 菜品成本/BOM导入 ===")
    print(f"文件: {filepath}")
    
    sha = file_sha256(filepath)
    source_file = os.path.basename(filepath)
    
    cur = conn.cursor()
    cur.execute(
        "SELECT 1 FROM public.dish_cost_analysis_import_log WHERE source_file = %s AND file_sha256 = %s",
        (source_file, sha)
    )
    if cur.fetchone():
        print("已导入过，跳过")
        return
    
    # Read with pandas
    raw_df = pd.read_excel(filepath, header=None, engine='openpyxl')
    print(f"  总行数: {len(raw_df)}")
    
    # Headers: Row 1 (main) + Row 2 (sub for material detail cols 13+)
    main_headers = [str(v).strip() if pd.notna(v) else '' for v in raw_df.iloc[1]]
    sub_headers = [str(v).strip() if pd.notna(v) else '' for v in raw_df.iloc[2]]
    
    # Build combined headers
    headers = []
    for i in range(len(main_headers)):
        m = main_headers[i]
        s = sub_headers[i]
        if m and s and m != s:
            headers.append(f"{m}_{s}")
        elif m:
            headers.append(m)
        elif s:
            headers.append(s)
        else:
            headers.append(f'col_{i}')
    
    data_start = 3
    df = raw_df.iloc[data_start:].copy()
    
    # Identify summary rows: col 0 originally non-empty (first row of each dish)
    # Detail rows: col 13 non-empty (material name)
    summary_mask = df[0].notna() & (df[0].astype(str).str.strip() != '') & (df[0].astype(str).str.strip() != 'nan')
    detail_mask = df[13].notna() & (df[13].astype(str).str.strip() != '') & (df[13].astype(str).str.strip() != 'nan')
    
    summary_df = df[summary_mask].copy()
    detail_df = df[detail_mask].copy()
    
    # Filter out 合计 rows
    summary_df = summary_df[~summary_df[0].astype(str).str.strip().isin(['合计', '总计'])]
    
    print(f"  菜品汇总行数: {len(summary_df)}")
    print(f"  原料明细行数: {len(detail_df)}")
    
    if len(summary_df) == 0:
        print("无菜品汇总数据，跳过")
        return
    
    # Calculate totals for log
    sales_amount_total = to_num(summary_df[7].sum()) if 7 < len(summary_df.columns) else 0
    theoretical_cost_total = to_num(summary_df[8].sum()) if 8 < len(summary_df.columns) else 0
    actual_cost_total = to_num(summary_df[9].sum()) if 9 < len(summary_df.columns) else 0
    cost_variance_total = to_num(summary_df[12].sum()) if 12 < len(summary_df.columns) else 0
    
    # Parse report period from filename or use report_month
    import re
    date_match = re.search(r'(\d{4})年(\d{1,2})月', source_file)
    if date_match:
        year, month = date_match.groups()
        period_start = f"{year}-{int(month):02d}-01"
        period_end = (pd.Timestamp(period_start) + pd.offsets.MonthEnd(0)).strftime('%Y-%m-%d')
    else:
        period_start = report_month
        period_end = (pd.Timestamp(report_month) + pd.offsets.MonthEnd(0)).strftime('%Y-%m-%d')
    
    # Insert import log
    cur.execute(
        """INSERT INTO public.dish_cost_analysis_import_log 
           (source_file, file_sha256, sheet_name, workbook_data_rows, 
            dish_summary_rows, material_detail_rows,
            sales_amount_total, theoretical_cost_total, actual_cost_total, cost_variance_total,
            report_period_start, report_period_end, import_status, report_month)
           VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, 'loading', %s) RETURNING import_id""",
        (source_file, sha, 'Worksheet', len(df), len(summary_df), len(detail_df),
         sales_amount_total, theoretical_cost_total, actual_cost_total, cost_variance_total,
         period_start, period_end, report_month)
    )
    import_id = cur.fetchone()[0]
    
    # Import summary rows
    summary_mapping = {
        'dish_name': 0,       # 菜品名称
        'dish_code': 1,       # 菜品编码
        'price': 2,           # 售价
        'category_level1': 3, # 菜品大类
        'category_level2': 4, # 菜品小类
        'dish_unit': 5,       # 菜品单位
        'sales_quantity': 6,  # 销售数量
        'sales_amount': 7,    # 销售金额(元)
        'theoretical_cost': 8,  # 理论成本(元)
        'actual_cost': 9,       # 实际成本(元)
        'theoretical_margin_rate_pct': 10,  # 理论毛利率
        'actual_margin_rate_pct': 11,       # 实际毛利率
        'cost_variance_amount': 12,         # 成本差异金额(元)
    }
    
    summary_cols = ['import_id', 'source_row', 'dish_sequence'] + list(summary_mapping.keys())
    summary_col_names = ','.join(summary_cols)
    
    # Fields that need clean_percent vs to_num vs to_text
    summary_percent_fields = {'theoretical_margin_rate_pct', 'actual_margin_rate_pct'}
    summary_text_fields = {'dish_name', 'dish_code', 'category_level1', 'category_level2', 'dish_unit'}
    
    batch = []
    for idx, (row_idx, r) in enumerate(summary_df.iterrows()):
        source_row = int(row_idx) + data_start + 1  # 1-indexed Excel row
        vals = [import_id, source_row, idx + 1]  # dish_sequence
        for db_col, excel_idx in summary_mapping.items():
            v = r[excel_idx] if excel_idx < len(r) else None
            if v is None or (isinstance(v, float) and pd.isna(v)):
                vals.append(None)
            elif db_col in summary_percent_fields:
                vals.append(clean_percent(v))
            elif db_col in summary_text_fields:
                vals.append(to_text(v))
            else:
                vals.append(to_num(v))
        batch.append(tuple(vals))
        if len(batch) >= 500:
            execute_values(cur, f"INSERT INTO public.dish_cost_analysis_summary ({summary_col_names}) VALUES %s", batch, page_size=500)
            conn.commit()
            print(f"  Summary: 已导入 {idx + 1}/{len(summary_df)} 行")
            batch = []
    
    if batch:
        execute_values(cur, f"INSERT INTO public.dish_cost_analysis_summary ({summary_col_names}) VALUES %s", batch, page_size=500)
        conn.commit()
        print(f"  Summary: 已导入 {len(summary_df)}/{len(summary_df)} 行")
    
    # Get summary_ids for detail linking
    cur.execute("SELECT summary_id, source_row FROM public.dish_cost_analysis_summary WHERE import_id = %s ORDER BY source_row", (import_id,))
    summary_id_map = {r[1]: r[0] for r in cur.fetchall()}
    
    # Import detail rows
    # Detail columns: col13=原料, col14=单位, col15=规格, col16=属性, 
    # col17=理论数量, col18=理论金额, col19=实际数量, col20=实际金额,
    # col21=损耗数量, col22=损耗金额, col23=数量量比
    detail_mapping = {
        'material_name': 13,
        'material_unit': 14,
        'specification': 15,
        'material_type': 16,
        'theoretical_quantity': 17,
        'theoretical_amount': 18,
        'actual_quantity': 19,
        'actual_amount': 20,
        'loss_quantity': 21,
        'loss_amount': 22,
        'loss_quantity_rate_pct': 23,
    }
    
    detail_cols = ['import_id', 'summary_id', 'source_row', 'material_sequence'] + list(detail_mapping.keys()) + ['theoretical_quantity_per_dish', 'actual_quantity_per_dish']
    detail_col_names = ','.join(detail_cols)
    
    # For each detail row, find the parent summary's source_row
    # The parent dish's source_row is the first non-empty col0 at or before this row
    # We already forward-filled col0, so we can match by looking up the original row
    # Actually, we need to find which summary row each detail row belongs to
    # Since we forward-filled, the dish_name (col0) tells us which dish it belongs to
    
    # Build a map: dish_name -> summary source_row
    dish_to_summary_row = {}
    for idx, (row_idx, r) in enumerate(summary_df.iterrows()):
        source_row = int(row_idx) + data_start + 1
        dish_name = str(r[0]).strip() if pd.notna(r[0]) else ''
        if dish_name:
            dish_to_summary_row[dish_name] = source_row
    
    # For detail rows, find parent dish by looking backward in original data for nearest non-empty col0
    # Build a sorted list of summary source_rows for binary search
    summary_source_rows = sorted(summary_id_map.keys())
    
    batch = []
    mat_seq = 0
    current_dish = None
    for idx, (row_idx, r) in enumerate(detail_df.iterrows()):
        source_row = int(row_idx) + data_start + 1
        excel_row_idx = int(row_idx)  # 0-indexed in raw_df
        
        # Find parent dish: look backward from current row for nearest non-empty col0
        parent_dish_name = None
        for back_idx in range(excel_row_idx, data_start - 1, -1):
            val = raw_df.iloc[back_idx, 0]
            if pd.notna(val) and str(val).strip() and str(val).strip() not in ('合计', '总计', 'nan'):
                parent_dish_name = str(val).strip()
                break
        
        if parent_dish_name != current_dish:
            current_dish = parent_dish_name
            mat_seq = 0
        mat_seq += 1
        
        # Find summary_id
        parent_source_row = dish_to_summary_row.get(parent_dish_name)
        summary_id = summary_id_map.get(parent_source_row) if parent_source_row else None
        
        vals = [import_id, summary_id, source_row, mat_seq]
        # Get sales_quantity from parent summary for per_dish calculation
        parent_sales_qty = None
        if parent_source_row:
            for _, srow in summary_df.iterrows():
                if int(srow.name) + data_start + 1 == parent_source_row:
                    parent_sales_qty = to_num(srow[6]) if pd.notna(srow[6]) else None
                    break
        
        for db_col, excel_idx in detail_mapping.items():
            v = r[excel_idx] if excel_idx < len(r) else None
            if v is None or (isinstance(v, float) and pd.isna(v)):
                vals.append(None)
            elif db_col == 'loss_quantity_rate_pct':
                vals.append(clean_percent(v))
            elif db_col in ('material_name', 'material_unit', 'specification', 'material_type'):
                vals.append(to_text(v))
            else:
                vals.append(to_num(v))
        
        # Calculate per_dish quantities
        theoretical_qty = vals[-len(detail_mapping) + 6]  # theoretical_quantity in vals
        actual_qty = vals[-len(detail_mapping) + 8]  # actual_quantity in vals
        # Actually let's just get them directly
        theoretical_qty = to_num(r[17]) if pd.notna(r[17]) else None
        actual_qty = to_num(r[19]) if pd.notna(r[19]) else None
        if theoretical_qty is not None and parent_sales_qty not in (None, 0):
            vals.append(theoretical_qty / parent_sales_qty)
        else:
            vals.append(None)
        if actual_qty is not None and parent_sales_qty not in (None, 0):
            vals.append(actual_qty / parent_sales_qty)
        else:
            vals.append(None)
        batch.append(tuple(vals))
        if len(batch) >= 500:
            execute_values(cur, f"INSERT INTO public.dish_cost_analysis_material_detail ({detail_col_names}) VALUES %s", batch, page_size=500)
            conn.commit()
            print(f"  Detail: 已导入 {idx + 1}/{len(detail_df)} 行")
            batch = []
    
    if batch:
        execute_values(cur, f"INSERT INTO public.dish_cost_analysis_material_detail ({detail_col_names}) VALUES %s", batch, page_size=500)
        conn.commit()
        print(f"  Detail: 已导入 {len(detail_df)}/{len(detail_df)} 行")
    
    # Update log status
    cur.execute(
        "UPDATE public.dish_cost_analysis_import_log SET import_status = 'success' WHERE import_id = %s",
        (import_id,)
    )
    conn.commit()
    print(f"菜品成本导入完成: {len(summary_df)} 菜品, {len(detail_df)} 原料明细")

# ============================================================
# 营业费用导入
# ============================================================
def import_operating_expense(conn, filepath, report_month, expense_type='operating'):
    """营业费用导入：宽表转长表，每行=科目×门店"""
    print(f"\n=== 营业费用导入 ({expense_type}) ===")
    print(f"文件: {filepath}")
    
    sha = file_sha256(filepath)
    source_file = os.path.basename(filepath)
    
    cur = conn.cursor()
    cur.execute(
        "SELECT 1 FROM public.operating_expense_import_log WHERE source_file = %s AND file_sha256 = %s",
        (source_file, sha)
    )
    if cur.fetchone():
        print("已导入过，跳过")
        return
    
    # Read .xls file with pandas
    raw_df = pd.read_excel(filepath, header=None, engine='xlrd')
    print(f"  总行数: {len(raw_df)}, 列数: {len(raw_df.columns)}")
    
    # Row 0: headers (科目编码, 科目名称, 统计方式, 方向3, 合计金额, 01 大钟寺金额, ...)
    headers = [str(v).strip() if pd.notna(v) else '' for v in raw_df.iloc[0]]
    
    # Data starts from row 1
    df = raw_df.iloc[1:].copy()
    
    # Filter out summary rows (col 0 = '费用科目' or empty)
    df = df[df[0].notna() & (df[0].astype(str).str.strip() != '') & (df[0].astype(str).str.strip() != '费用科目')]
    
    print(f"  科目行数: {len(df)}")
    
    # Parse store columns: col 5+ are store columns (skip col 0-4: code, name, direction, direction3, total)
    # Extract store code and name from header
    store_cols = []
    for ci in range(5, len(headers)):
        h = headers[ci]
        if not h:
            continue
        # Parse "01 大钟寺金额" -> code="01", name="大钟寺"
        import re
        m = re.match(r'([A-Z]\d+)\s+(.+?)金额$', h)
        if m:
            store_cols.append((ci, m.group(1), m.group(2)))
        else:
            # Keep as-is with full header
            store_cols.append((ci, None, h))
    
    print(f"  门店列数: {len(store_cols)}")
    
    # Calculate total amount for log
    total_amount = 0
    for _, r in df.iterrows():
        v = to_num(r[4]) if pd.notna(r[4]) else 0  # col 4 = 合计金额
        total_amount += v if v else 0
    
    # Insert import log
    cur.execute(
        """INSERT INTO public.operating_expense_import_log 
           (report_month, source_file, source_sheet, file_sha256, source_total_amount, imported_record_count)
           VALUES (%s, %s, %s, %s, %s, 0) RETURNING import_id""",
        (report_month, source_file, 'Worksheet', sha, total_amount)
    )
    import_id = cur.fetchone()[0]
    
    # Insert records: long format (one row per account × store)
    insert_cols = ['import_id', 'report_month', 'source_file', 'source_sheet',
                   'source_row', 'source_column', 'account_code', 'account_name',
                   'accounting_direction', 'is_summary_account', 'parent_account_code',
                   'cost_unit_source_header', 'cost_unit_source_code',
                   'cost_unit_source_name', 'amount']
    col_names = ','.join(insert_cols)
    
    batch = []
    total_records = 0
    
    for row_idx, (_, r) in enumerate(df.iterrows()):
        account_code = to_text(r[0])
        account_name = to_text(r[1])
        direction = to_text(r[2])  # 统计方式 (借方/贷方)
        excel_row = int(r.name) + 1  # 1-indexed Excel row
        
        # Determine if summary account: 7-digit codes (5032401) are sub-accounts of 5-digit (50324)
        is_summary = False
        parent_code = None
        if account_code and len(account_code) == 7:
            parent_code = account_code[:5]
            is_summary = False
        elif account_code and len(account_code) == 5:
            is_summary = True
        
        for col_idx, store_code, store_name in store_cols:
            amount = to_num(r[col_idx]) if col_idx < len(r) and pd.notna(r[col_idx]) else None
            if amount is None:
                continue  # Skip NULL amounts, but keep 0 amounts
            
            vals = (
                import_id,
                report_month,
                source_file,
                'Worksheet',
                excel_row,
                col_idx + 1,  # 1-indexed column
                account_code,
                account_name,
                direction,
                is_summary,
                parent_code,
                headers[col_idx],  # full header text
                store_code,
                store_name,
                amount,
            )
            batch.append(vals)
            total_records += 1
            
            if len(batch) >= 500:
                execute_values(cur, f"INSERT INTO public.operating_expense_records ({col_names}) VALUES %s", batch, page_size=500)
                conn.commit()
                print(f"  已导入 {total_records} 条记录")
                batch = []
    
    if batch:
        execute_values(cur, f"INSERT INTO public.operating_expense_records ({col_names}) VALUES %s", batch, page_size=500)
        conn.commit()
        print(f"  已导入 {total_records} 条记录")
    
    # Update log
    cur.execute(
        "UPDATE public.operating_expense_import_log SET imported_record_count = %s WHERE import_id = %s",
        (total_records, import_id)
    )
    conn.commit()
    print(f"营业费用导入完成: {total_records} 条记录")

# ============================================================
# 中央厨房导入 (4表)
# ============================================================
def import_central_kitchen(conn, filepath, report_month, ck_type='finished'):
    """中央厨房导入: finished/recipe/material/processing
    使用pandas读取，精确映射列，raw_data存储原始行JSON"""
    import json
    type_names = {
        'finished': '完工入库',
        'recipe': '配方耗用',
        'material': '材料日报',
        'processing': '加工单价',
    }
    table_map = {
        'finished': 'central_kitchen_finished_receipt',
        'recipe': 'central_kitchen_recipe_consumption',
        'material': 'central_kitchen_material_daily',
        'processing': 'central_kitchen_processing_cost',
    }
    table_name = table_map[ck_type]
    
    print(f"\n=== 中央厨房-{type_names.get(ck_type, ck_type)} 导入 ===")
    print(f"文件: {filepath}")
    
    sha = file_sha256(filepath)
    source_file = os.path.basename(filepath)
    
    cur = conn.cursor()
    cur.execute(
        "SELECT 1 FROM public.central_kitchen_import_log WHERE source_file = %s AND file_sha256 = %s",
        (source_file, sha)
    )
    if cur.fetchone():
        print("已导入过，跳过")
        return
    
    # Header row varies by type:
    # finished: row 2 (0-indexed) with sub-header row 3
    # recipe: row 0
    # material: row 0 with sub-header row 1
    # processing: row 2
    header_rows = {
        'finished': 2,  # 0-indexed, merged header rows 2+3
        'recipe': 0,
        'material': 1,  # skip row 0 (group header), use row 1
        'processing': 2,
    }
    hr = header_rows[ck_type]
    
    # Read raw without header to get all rows
    raw_df = pd.read_excel(filepath, header=None, engine='openpyxl')
    
    # Build headers from the appropriate row(s)
    if ck_type == 'finished':
        # Row 2 has group headers, row 3 has sub-headers
        # Don't fill forward - empty cells stay empty (merged cells)
        h1 = raw_df.iloc[2].fillna('').astype(str).str.strip()
        h2 = raw_df.iloc[3].fillna('').astype(str).str.strip()
        headers = []
        for i in range(len(h1)):
            g = h1[i]
            s = h2[i]
            if g and s:
                headers.append(f"{g}_{s}")
            elif g:
                headers.append(g)
            elif s:
                headers.append(s)
            else:
                headers.append(f'col_{i}')  # Placeholder for empty headers
        data_start = 4
    elif ck_type == 'material':
        # Row 0 has group headers, row 1 has sub-headers
        h1 = raw_df.iloc[0].fillna('').astype(str).str.strip()
        h2 = raw_df.iloc[1].fillna('').astype(str).str.strip()
        headers = []
        current_group = ''
        for i in range(len(h1)):
            g = h1[i] if h1[i] else current_group
            if g:
                current_group = g
            s = h2[i] if h2[i] else ''
            headers.append(f"{g}_{s}" if s else g)
        data_start = 2
    else:
        # processing: row 2 (main) + row 3 (sub) — like finished
        h1 = raw_df.iloc[hr].fillna('').astype(str).str.strip()
        h2 = raw_df.iloc[hr + 1].fillna('').astype(str).str.strip()
        headers = []
        for i in range(len(h1)):
            g = h1[i]
            s = h2[i]
            if g and s and g != s:
                headers.append(f"{g}_{s}")
            elif g and g != 'nan':
                headers.append(g)
            elif s and s != 'nan':
                headers.append(s)
            else:
                headers.append(f'col_{i}')
        data_start = hr + 2
    
    # Extract data rows
    df = raw_df.iloc[data_start:]
    df = df.dropna(subset=[df.columns[0]])
    # Filter out 合计/total rows by checking all string cells for '合计'
    mask = pd.Series([True] * len(df), index=df.index)
    for ci in range(min(15, df.shape[1])):
        mask = mask & ~df[ci].astype(str).str.strip().isin(['合计', '总计', 'Total', 'total'])
    # Filter out footer rows (e.g. "制表人:xxx")
    mask = mask & ~df[0].astype(str).str.strip().str.startswith('制表人')
    mask = mask & ~df[0].astype(str).str.strip().str.startswith('打印')
    mask = mask & ~df[0].astype(str).str.strip().str.startswith('第')
    df = df[mask]
    data_rows = df.values.tolist()
    print(f"数据行数: {len(data_rows)}")
    
    if not data_rows:
        return
    
    # Insert import log
    cur.execute(
        """INSERT INTO public.central_kitchen_import_log 
           (report_month, dataset_type, source_file, source_sheet, file_sha256, 
            workbook_rows, imported_rows, excluded_rows, status)
           VALUES (%s, %s, %s, %s, %s, %s, %s, 0, 'loading') RETURNING import_id""",
        (report_month, ck_type, source_file, 'Worksheet', sha, len(data_rows), len(data_rows))
    )
    import_id = cur.fetchone()[0]
    
    # Get DB columns (exclude record_id, import_id, source_row, raw_data)
    cur.execute(f"SELECT column_name FROM information_schema.columns WHERE table_name='{table_name}' ORDER BY ordinal_position")
    db_cols = [r[0] for r in cur.fetchall()]
    insertable_cols = [c for c in db_cols if c not in ('record_id', 'import_id', 'source_row', 'raw_data')]
    
    # Build header name -> excel index map (keep first occurrence)
    col_map = {}
    for i, h in enumerate(headers):
        if h and str(h).strip():
            key = str(h).strip()
            if key not in col_map:
                col_map[key] = i
    
    # Type-specific column mappings (db_col -> list of possible header names)
    mappings = {
        'finished': {
            'receipt_date': ['入库日期'],
            'workshop': ['col_1'],
            'processing_warehouse': ['加工仓库'],
            'recipe_name': ['配方名称'],
            'product_name': ['货品名称'],
            'specification': ['规格'],
            'unit': ['库存单位'],
            'return_quantity': ['退库数量'],
            'return_amount': ['退库金额'],
            'inbound_avg_unit_price': ['平均单价'],
            'inbound_quantity': ['入库数量'],
            'inbound_amount': ['入库金额'],
            'raw_material_cost': ['加工利润分析_原料成本', '原料成本'],
            'source_fee_cost': ['费用成本'],
            'source_processing_profit': ['增值金额（加工利润）'],
            'source_margin_rate': ['毛利率'],
            'source_cost_rate': ['成本率'],
            'theoretical_inbound_quantity': ['加工达成率分析_理论入库数量', '理论入库数量'],
            'actual_inbound_quantity': ['实际入库数量'],
            'inbound_difference_quantity': ['差异数量'],
            'achievement_rate': ['加工达成率'],
            'standard_expected_quantity': ['加工应产量分析_标准应产量', '标准应产量'],
            'average_expected_quantity': ['平均应产量'],
            'expected_quantity_variance_rate': ['应产量差异量比'],
            'water_cost': ['水费'],
            'electricity_cost': ['电费'],
            'gas_cost': ['燃气费'],
            'labor_cost': ['人工费'],
            'other_cost': ['其他费用'],
            'detailed_fee_total': ['费用合计'],
        },
        'recipe': {
            'business_date': ['日期'],
            'processing_warehouse': ['加工仓库'],
            'finance_category_major': ['财务大类'],
            'finance_category_minor': ['财务小类'],
            'category_major': ['所属大类'],
            'category_minor': ['所属小类'],
            'item_id': ['货品ID'],
            'item_code': ['编码'],
            'item_name': ['货品名称'],
            'specification': ['规格'],
            'unit': ['库存单位'],
            'unit_price_excl_tax': ['未税单价'],
            'unit_price': ['单价'],
            'recipe_name': ['配方名称'],
            'standard_usage': ['标准用量'],
            'actual_average_usage': ['实际平均用量'],
            'planned_output_quantity': ['计划加工数量'],
            'actual_output_quantity': ['实际加工数量'],
            'output_unit': ['加工数量库存单位'],
            'theoretical_quantity': ['理论用量'],
            'net_quantity': ['净料用量'],
            'theoretical_amount': ['理论金额'],
            'issue_quantity': ['领用数量'],
            'issue_amount': ['领用金额'],
            'source_inventory_cost_quantity': ['成本数量（进销存）'],
            'source_inventory_cost_amount': ['成本金额（进销存）'],
            'actual_vs_theory_quantity': ['实际与理论差异数量'],
            'actual_vs_theory_amount': ['实际与理论差异金额'],
            'actual_vs_theory_rate': ['实际差异量比'],
            'issue_vs_theory_quantity': ['领用与理论差异数量'],
            'issue_vs_theory_amount': ['领用与理论差异金额'],
            'issue_vs_theory_rate': ['领用差异量比'],
            'actual_yield': ['实际出成率'],
            'recipe_yield': ['配方出成率'],
            'yield_difference': ['出成率差异'],
        },
        'material': {
            'business_date': ['日期'],
            'finance_category_major': ['财务大类'],
            'finance_category_minor': ['财务小类'],
            'category_major': ['所属大类'],
            'category_minor': ['所属小类'],
            'item_id': ['货品ID'],
            'item_code': ['编码'],
            'item_name': ['货品名称'],
            'specification': ['规格'],
            'unit': ['库存单位'],
        },
        'processing': {
            'report_month': ['report_month'],  # Will be set from parameter
            'item_id': ['货品ID'],
            'recipe_name': ['配方名称'],
            'product_code': ['货品编码'],
            'product_name': ['货品名称'],
            'category_major': ['所属大类'],
            'category_minor': ['所属小类'],
            'finance_category_major': ['财务大类'],
            'finance_category_minor': ['财务小类'],
            'specification': ['规格'],
            'unit': ['单位'],
            'inbound_quantity': ['入库数量'],
            'inbound_avg_unit_price': ['入库均价'],
            'theoretical_unit_cost': ['理论单价'],
            'standard_unit_cost': ['标准单价'],
            'actual_unit_cost': ['实际单价'],
            'theoretical_cost': ['理论成本'],
            'standard_cost': ['标准成本'],
            'actual_cost': ['实际成本'],
        },
    }
    
    field_mapping = mappings[ck_type]
    cols_to_insert = {}
    fixed_values = {}  # Values not from Excel (e.g. report_month)
    for db_col, h_names in field_mapping.items():
        if db_col not in insertable_cols:
            continue
        found = False
        for h_name in h_names:
            if h_name in col_map:
                cols_to_insert[db_col] = col_map[h_name]
                found = True
                break
        if not found and db_col == 'report_month':
            fixed_values['report_month'] = report_month
    
    # Determine numeric columns from DB schema
    cur.execute(f"SELECT column_name, data_type FROM information_schema.columns WHERE table_name='{table_name}'")
    col_types = {r[0]: r[1] for r in cur.fetchall()}
    numeric_db_cols = {c for c, t in col_types.items() if t in ('numeric', 'integer', 'bigint', 'double precision', 'real')}
    date_db_cols = {c for c, t in col_types.items() if t == 'date'}
    
    all_cols = ['import_id', 'source_row'] + list(cols_to_insert.keys()) + list(fixed_values.keys()) + ['raw_data']
    col_names = ','.join(all_cols)
    
    batch = []
    for idx, r in enumerate(data_rows):
        source_row = idx + data_start + 1  # 1-indexed Excel row
        vals = [import_id, source_row]
        for db_col, excel_idx in cols_to_insert.items():
            v = r[excel_idx] if excel_idx < len(r) else None
            if db_col in numeric_db_cols:
                vals.append(to_num(v) if v is not None else None)
            elif db_col in date_db_cols:
                vals.append(to_date(v))
            else:
                vals.append(to_text(v))
        # Fixed values (not from Excel)
        for fv_col in fixed_values:
            if fv_col in date_db_cols:
                vals.append(to_date(fixed_values[fv_col]))
            else:
                vals.append(fixed_values[fv_col])
        # raw_data: store entire row as JSON
        raw_dict = {}
        for hi, hv in enumerate(headers):
            if hv and hi < len(r):
                rv = r[hi]
                if pd.notna(rv):
                    raw_dict[str(hv)] = str(rv) if not isinstance(rv, (int, float)) else rv
        vals.append(json.dumps(raw_dict, ensure_ascii=False))
        batch.append(tuple(vals))
        if len(batch) >= 500:
            execute_values(cur, f"INSERT INTO public.{table_name} ({col_names}) VALUES %s", batch, page_size=500)
            conn.commit()
            print(f"  已导入 {idx + 1}/{len(data_rows)} 行")
            batch = []
    
    if batch:
        execute_values(cur, f"INSERT INTO public.{table_name} ({col_names}) VALUES %s", batch, page_size=500)
        conn.commit()
        print(f"  已导入 {len(data_rows)}/{len(data_rows)} 行")
    
    # Update log status
    cur.execute(
        "UPDATE public.central_kitchen_import_log SET status = 'success', completed_at = now() WHERE import_id = %s",
        (import_id,)
    )
    conn.commit()
    print(f"中央厨房{type_names[ck_type]}导入完成: {len(data_rows)} 行")

# ============================================================
# 配送明细导入
# ============================================================
def import_distribution(conn, filepath, report_month):
    """配送明细导入（支持单文件或目录）"""
    import json
    print(f"\n=== 配送明细导入 ===")
    
    files = []
    if os.path.isdir(filepath):
        for f in sorted(os.listdir(filepath)):
            if f.endswith('.xlsx') and not f.startswith('.'):
                files.append(os.path.join(filepath, f))
    else:
        files.append(filepath)
    
    print(f"文件数: {len(files)}")
    
    cur = conn.cursor()
    total_imported = 0
    
    for file_idx, fpath in enumerate(files):
        fname = os.path.basename(fpath)
        print(f"\n[{file_idx+1}/{len(files)}] {fname}")
        
        sha = file_sha256(fpath)
        
        cur.execute(
            "SELECT 1 FROM public.distribution_import_log WHERE source_file = %s AND file_sha256 = %s",
            (fname, sha)
        )
        if cur.fetchone():
            print("  已导入过, 跳过")
            continue
        
        # Read with pandas, header on row 0
        raw_df = pd.read_excel(fpath, header=None, engine='openpyxl')
        headers = [str(v).strip() if pd.notna(v) else '' for v in raw_df.iloc[0]]
        df = raw_df.iloc[1:]
        df = df.dropna(subset=[df.columns[0]])
        data_rows = df.values.tolist()
        print(f"  数据行数: {len(data_rows)}")
        
        if not data_rows:
            continue
        
        # Parse period from filename or use report_month
        import re
        date_match = re.search(r'(\d{4})年(\d{1,2})月(\d{1,2})日.*?(\d{1,2})日', fname)
        if date_match:
            year, month, start_day, end_day = date_match.groups()
            period_start = f"{year}-{int(month):02d}-{int(start_day):02d}"
            period_end = f"{year}-{int(month):02d}-{int(end_day):02d}"
        else:
            # Use full month
            period_start = report_month
            # Last day of month
            rm_date = pd.Timestamp(report_month)
            period_end = (rm_date + pd.offsets.MonthEnd(0)).strftime('%Y-%m-%d')
        
        cur.execute(
            """INSERT INTO public.distribution_import_log 
               (source_file, source_sheet, file_sha256, workbook_rows, summary_rows_excluded, 
                imported_rows, report_month, period_start, period_end, status)
               VALUES (%s, %s, %s, %s, 0, %s, %s, %s, %s, 'loading') RETURNING import_id""",
            (fname, 'Worksheet', sha, len(data_rows), len(data_rows), report_month, period_start, period_end)
        )
        import_id = cur.fetchone()[0]
        
        # Get DB columns
        cur.execute("SELECT column_name, data_type FROM information_schema.columns WHERE table_name='distribution_detail_records' ORDER BY ordinal_position")
        db_col_types = {r[0]: r[1] for r in cur.fetchall()}
        insertable_cols = [c for c in db_col_types if c not in ('record_id', 'import_id', 'source_row', 'raw_data')]
        
        # Build header map
        col_map = {}
        for i, h in enumerate(headers):
            if h and str(h).strip():
                col_map[str(h).strip()] = i
        
        # Known mappings for distribution based on actual Excel headers
        field_mapping = {
            'business_date': ['日期', '业务日期'],
            'distribution_center_code': ['配送中心编码'],
            'distribution_center_name': ['配送中心'],
            'distribution_center_company': ['配送中心所属公司'],
            'store_company': ['门店所属公司'],
            'store_type': ['门店类型'],
            'store_brand': ['门店品牌'],
            'region_level1': ['一级区域'],
            'terminal_region': ['末级区域'],
            'store_code': ['门店编码'],
            'store_name': ['门店'],
            'store_tax_entity': ['门店财税主体'],
            'original_batch_unit_price': ['原始批次单价'],
            'cost_avg_unit_price': ['成本平均单价'],
            'cost_total_amount': ['成本总金额'],
            'team': ['班组'],
            'cost_excl_tax_amount': ['成本未含税金额'],
            'cost_tax_amount': ['成本税额'],
            'cost_excl_tax_unit_price': ['成本未含税单价'],
            'outbound_tax_amount': ['配出税额'],
            'distribution_sales_tax_rate': ['配送销售税点'],
            'route': ['路线'],
            'profit_amount': ['利润'],
            'profit_excl_tax_amount': ['未含税利润', '利润未含税'],
            'distribution_type': ['配送类型'],
            'logistics_attribute': ['物流属性'],
            'item_id': ['货品ID'],
            'item_name': ['货品名称', '货品'],
            'item_alias': ['货品别名'],
            'specification': ['规格'],
            'item_purpose': ['货品用途', '用途'],
            'unit': ['单位'],
            'normal_quantity': ['正常品数量', '正常数量'],
            'gift_quantity': ['赠品数量'],
            'total_quantity': ['总数量', '合计数量'],
            'outbound_avg_unit_price': ['配出平均单价', '配出均价'],
            'outbound_total_amount': ['配出总金额', '配出金额'],
            'outbound_margin_rate': ['配出毛利率'],
            'outbound_excl_tax_unit_price': ['配出未含税单价'],
            'outbound_excl_tax_amount': ['配出未含税金额'],
            'document_no': ['单号'],
            'requisition_no': ['要货单号', '申请单号'],
            'payment_method': ['支付方式', '结算方式'],
            'value_added_coefficient': ['增值系数'],
            'audit_time': ['审核时间'],
            'inbound_batch': ['入库批次'],
            'inbound_batch_tax_rate': ['入库批次税点', '入库批次税率'],
            'custom_batch_code': ['自定义批次码', '自定义批次编码'],
            'invoice_type': ['发票种类', '发票类型'],
            'production_time': ['生产时间'],
            'shelf_life': ['保质期'],
            'supplier_batch': ['供应商批次'],
            'supplier_code': ['供应商编码'],
            'supplier_category': ['供应商类别', '供应商分类'],
            'supplier_name': ['供应商'],
            'supplier_contact': ['供应商联系人'],
            'supplier_phone': ['供应商联系电话', '供应商电话'],
            'supplier_address': ['供应商地址'],
            'from_warehouse': ['调出仓库', '发货仓库'],
            'to_warehouse': ['调入仓库', '收货仓库'],
            'item_code': ['货品编码'],
            'item_barcode': ['货品条形码', '货品条码'],
            'minor_category': ['所属小类', '小类'],
            'major_category': ['所属大类', '大类'],
            'internal_brand': ['内部品牌'],
            'finance_category': ['财务分类'],
            'created_time': ['制单时间', '创建时间'],
            'creator': ['制单人', '创建人'],
            'auditor': ['审核人'],
            'arrival_time': ['到货时间'],
            'document_remark': ['单据备注'],
            'document_item_remark': ['单据项备注'],
            'reason_remark': ['原因备注'],
            'business_date': ['业务日期'],
            'contact_address': ['联系地址'],
            'contact_phone': ['联系电话'],
            'contact_person': ['联系人'],
        }
        
        cols_to_insert = {}
        for db_col, h_names in field_mapping.items():
            if db_col not in insertable_cols:
                continue
            for h_name in h_names:
                if h_name in col_map:
                    cols_to_insert[db_col] = col_map[h_name]
                    break
        
        numeric_db_cols = {c for c, t in db_col_types.items() if t in ('numeric', 'integer', 'bigint', 'double precision', 'real')}
        date_db_cols = {c for c, t in db_col_types.items() if t == 'date'}
        
        all_cols = ['import_id', 'source_row', 'report_month', 'source_file', 'source_sheet'] + list(cols_to_insert.keys())
        col_names = ','.join(all_cols)
        
        batch = []
        for idx, r in enumerate(data_rows):
            source_row = idx + 2  # 1-indexed, data starts at row 2
            vals = [import_id, source_row, report_month, fname, 'Worksheet']
            for db_col, excel_idx in cols_to_insert.items():
                v = r[excel_idx] if excel_idx < len(r) else None
                if db_col in numeric_db_cols:
                    vals.append(to_num(v) if v is not None else None)
                elif db_col in date_db_cols:
                    vals.append(to_date(v))
                else:
                    vals.append(to_text(v))
            batch.append(tuple(vals))
            if len(batch) >= 500:
                execute_values(cur, f"INSERT INTO public.distribution_detail_records ({col_names}) VALUES %s", batch, page_size=500)
                conn.commit()
                print(f"  已导入 {idx + 1}/{len(data_rows)} 行")
                batch = []
        
        if batch:
            execute_values(cur, f"INSERT INTO public.distribution_detail_records ({col_names}) VALUES %s", batch, page_size=500)
            conn.commit()
            print(f"  已导入 {len(data_rows)}/{len(data_rows)} 行")
        
        total_imported += len(data_rows)
        
        # Update log status
        cur.execute(
            "UPDATE public.distribution_import_log SET status = 'success', completed_at = now() WHERE import_id = %s",
            (import_id,)
        )
        conn.commit()
    
    print(f"\n配送明细导入完成: 共 {total_imported} 行")

# ============================================================
# Main
# ============================================================
def main():
    parser = argparse.ArgumentParser(description='月度数据导入工具')
    parser.add_argument('--dataset', required=True, 
                        choices=['inventory', 'dish_sales', 'dish_cost', 'operating_expense', 
                                 'central_kitchen', 'distribution'],
                        help='数据域')
    parser.add_argument('--month', required=True, help='报告月份 (YYYY-MM-01)')
    parser.add_argument('--file', required=True, help='文件路径或目录')
    parser.add_argument('--db', default='bill_query', help='目标数据库 (默认bill_query)')
    parser.add_argument('--ck-type', default='finished', 
                        choices=['finished', 'recipe', 'material', 'processing'],
                        help='中央厨房子类型')
    parser.add_argument('--expense-type', default='operating',
                        choices=['operating', 'management'],
                        help='费用类型')
    
    args = parser.parse_args()
    
    conn = connect(args.db)
    conn.autocommit = False
    
    try:
        if args.dataset == 'inventory':
            import_inventory_cost(conn, args.file, args.month)
        elif args.dataset == 'dish_sales':
            import_dish_sales(conn, args.file, args.month)
        elif args.dataset == 'dish_cost':
            import_dish_cost(conn, args.file, args.month)
        elif args.dataset == 'operating_expense':
            import_operating_expense(conn, args.file, args.month, args.expense_type)
        elif args.dataset == 'central_kitchen':
            import_central_kitchen(conn, args.file, args.month, args.ck_type)
        elif args.dataset == 'distribution':
            import_distribution(conn, args.file, args.month)
        
        print("\n=== 导入完成 ===")
    except Exception as e:
        conn.rollback()
        print(f"错误: {e}", file=sys.stderr)
        raise
    finally:
        conn.close()

if __name__ == '__main__':
    main()