crossborder/anhui/gov_commodity_anhui_import_export.py

import re
from pathlib import Path

import pandas as pd

from com.zf.crawl import base_country_code
from com.zf.crawl import base_mysql

CUSTOM_COMMODITY_REPLACEMENTS = {
    '家具': '家具及其零件',
    '眼镜': '眼镜及其零件',
}
# 需要保留中文括号及内容的商品关键词
PRESERVE_PARENTHESES_KEYWORDS = {
    '汽车（包括底盘）',
}

def clean_commodity_name(name, preserve_keywords=None):
    """
    自定义清洗商品名称逻辑，支持条件保留中文括号内容

    :param name: 商品名称字符串
    :param preserve_keywords: 需要保留括号的关键词集合
    :return: 清洗后的商品名称
    """
    name = str(name).strip()

    # 去除非必要符号
    name = re.sub(r'[#*]', '', name)

    # 判断是否需要保留括号内容
    if preserve_keywords:
        for keyword in preserve_keywords:
            if keyword == name:
                # 匹配到关键词时，不移除括号内容
                return name

    # 默认移除中文括号及内容
    name = re.sub(r'（[^）]*）', '', name)
    return name.strip()

def process_folder(path):
    file_paths = list(Path(path).glob('*'))
    if not file_paths:
        print("未找到任何文件")
        return
    year, month = base_country_code.extract_year_month_from_path(path)

    import_df = pd.DataFrame()
    export_df = pd.DataFrame()
    for file in file_paths:
        file_path = Path(path) / file
        df = pd.read_excel(file_path, header=None).iloc[6:]
        value_index = 1 if month == 2 else 3
        if "进口商品总值" in file.name:
            temp_df = df[[0, value_index]].rename(columns={0: 'commodity', value_index: 'import'})
            temp_df['commodity'] = (
                temp_df['commodity']
                .astype(str)
                .apply(lambda x: clean_commodity_name(x, preserve_keywords=PRESERVE_PARENTHESES_KEYWORDS))
                .replace(CUSTOM_COMMODITY_REPLACEMENTS, regex=False)
            )
            temp_df['import'] = pd.to_numeric(temp_df['import'].replace('--', 0), errors='coerce')
            # 去重 commodity 列，保留第一个出现的行
            temp_df = temp_df.drop_duplicates(subset=['commodity'], keep='first')
            import_df = pd.concat([import_df, temp_df])

        elif "出口商品总值" in file.name:
            temp_df = df[[0, value_index]].rename(columns={0: 'commodity', value_index: 'export'})
            temp_df['commodity'] = (
                temp_df['commodity']
                .astype(str)
                .apply(lambda x: clean_commodity_name(x, preserve_keywords=PRESERVE_PARENTHESES_KEYWORDS))
                .replace(CUSTOM_COMMODITY_REPLACEMENTS, regex=False)
            )
            temp_df['export'] = pd.to_numeric(temp_df['export'].replace('--', 0), errors='coerce')
            temp_df = temp_df.drop_duplicates(subset=['commodity'], keep='first')
            export_df = pd.concat([export_df, temp_df])

    save_to_database(import_df, export_df, year, month)

def save_to_database(import_df, export_df, year, month):
    # 合并数据（使用outer join保留所有商品）
    merged_df = pd.merge(
        import_df.groupby('commodity')['import'].sum().reset_index(),
        export_df.groupby('commodity')['export'].sum().reset_index(),
        on='commodity',
        how='outer'
    )

    year_month = f'{year}-{month:02d}'
    processed_commodities = set()
    sql_arr = []
    sql_arr_copy = []
    try:
        for _, row in merged_df.iterrows():
            commodity_name = str(row['commodity'])
            if commodity_name == '肉类' or commodity_name == '其他' or commodity_name == '干鲜瓜果' or commodity_name == '钟表':
                print(f'{commodity_name} 商品不存在，跳过')
                continue
            commodity_code, commodity_name_fix = base_mysql.get_commodity_id(commodity_name)
            if not commodity_code:
                print(f"未找到商品名称 '{commodity_name}' 对应的 ID")
                continue
            if not commodity_name_fix or commodity_name_fix in processed_commodities:
                continue

            monthly_import = round(row['import'] * 10000, 4)
            monthly_export = round(row['export'] * 10000, 4)
            monthly_total = round(monthly_import + monthly_export, 4)

            if month == 2:
                year_month_2 = f'{year}-01'
                monthly_import = round(monthly_import / 2, 4)
                monthly_export = round(monthly_export / 2, 4)
                monthly_total = round(monthly_import + monthly_export, 4)
                sql = (f"INSERT INTO t_yujin_crossborder_prov_commodity_trade "
                       f"(crossborder_year, crossborder_year_month, prov_code, prov_name, commodity_code, commodity_name, monthly_total, monthly_export, monthly_import, create_time) VALUES "
                       f"('{year}', '{year_month_2}', '340000', '安徽省', '{commodity_code}', '{commodity_name_fix}', {monthly_total}, {monthly_export}, {monthly_import}, now());")
                sql_arr_copy.append(sql)

            sql = (f"INSERT INTO t_yujin_crossborder_prov_commodity_trade "
                   f"(crossborder_year, crossborder_year_month, prov_code, prov_name, commodity_code, commodity_name, monthly_total, monthly_export, monthly_import, create_time) VALUES "
                   f"('{year}', '{year_month}', '340000', '安徽省', '{commodity_code}', '{commodity_name_fix}', {monthly_total}, {monthly_export}, {monthly_import}, now());")
            sql_arr.append(sql)

            processed_commodities.add(commodity_name_fix)
            # print(f'{commodity_name} -> {commodity_name_fix}')

    except Exception as e:
        print(f"{year_month} prov_commodity_trade 生成 SQL 文件时发生异常: {str(e)}")

    print(f"√ {year_month} prov_commodity_trade 成功生成 SQL 文件 size {len(sql_arr)} ")
    # 解析完后生成sql文件批量入库
    base_mysql.bulk_insert(sql_arr)
    if month == 2:
        print(f"√ {year_month} prov_commodity_trade copy 成功生成 SQL 文件 size {len(sql_arr_copy)} ")
        base_mysql.bulk_insert(sql_arr_copy)
    print(f"√ {year_month} prov_commodity_trade SQL 存表完成!\n")

def hierarchical_traversal(root_path):
    """分层遍历：省份->年份->月目录"""
    root = Path(root_path)
    # 获取所有年份目录
    year_dirs = [
        item for item in root.iterdir()
        if item.is_dir() and base_country_code.YEAR_PATTERN.match(item.name)
    ]

    # 按年倒序
    for year_dir in sorted(year_dirs, key=lambda x: x.name, reverse=True):
        # 构造完整的路径：download/shandong/2025/03
        print(f"\n年份：{year_dir.name} | 省份：jiangsu")

        # 提取月份目录
        month_dirs = []
        for item in year_dir.iterdir():
            if item.is_dir() and base_country_code.MONTH_PATTERN.match(item.name):
                month_dirs.append({
                    "path": item,
                    "month": int(item.name)
                })
        # 按月倒序输出
        if month_dirs:
            for md in sorted(month_dirs, key=lambda x: x["month"], reverse=True):
                print(f"  月份：{md['month']:02d} | 路径：{md['path']}")
                process_folder(md['path'])

if __name__ == '__main__':
    hierarchical_traversal(base_country_code.download_dir)

    # root = Path(base_country_code.download_dir)/'2025'/'04'
    # process_folder(root)
    print("安徽合肥海关类章所有文件处理完成！")