cdr_analysis/美团/地区表转化为省份.py

import csv

# 尝试使用ANSI编码打开文件
try:
    with open('手机地区表.txt', 'r', encoding='ansi') as infile:
        lines = infile.readlines()
except UnicodeDecodeError:
    # 如果ANSI失败，尝试使用gbk编码
    with open('手机地区表.txt', 'r', encoding='gbk') as infile:
        lines = infile.readlines()

# 创建CSV输出文件
with open('手机地区表.csv', 'w', encoding='utf-8', newline='') as outfile:
    writer = csv.writer(outfile)

    # 处理每一行数据
    for line in lines:
        line = line.strip()
        if not line:
            continue

        # 使用|分隔符拆分字段
        parts = line.split('|')

        # 检查是否为标题行
        if parts[0] == "手机前缀" or len(parts) < 3:
            writer.writerow(parts)  # 标题行保持不变
            continue

        # 从地区信息中提取省份
        location = parts[2]

        # 处理特殊省份名称
        if location.startswith(("内蒙古", "黑龙江")):
            province = location[:3]
        else:
            province = location[:2]

        # 更新地区字段为省份名称
        parts[2] = province

        # 将修改后的行写入CSV
        writer.writerow(parts)

print("转换完成，已生成：手机地区表.csv")