# coding=utf-8
import os, re, sys


def use_reg(input_file, output_file):
    print '\n正则表达式示例：匹配 IP 地址：192.168.[0-9]{1,3}.[0-9]{1,3}，匹配域名：[^\s]*\.test\.com'
    reg = None
    while reg is None or reg == '':
        reg = str(raw_input('→ 请输入正则表达式：'))

    print '\n→ 把正则表达式【匹配到的字符串】输出到文件，请输入 1'
    print '→ 把正则表达式【匹配到的行】输出到文件，请输入 2'
    print '→ 把正则表达式【匹配不到的行】输出到文件，请输入 3'
    is_next_input = True
    is_fisrt_line = True
    after_line = None
    operation_code = '-1'
    while is_next_input:
        is_next_input = False
        operation_code = raw_input('→ 请输入数字：')
        if not (operation_code == '1' or operation_code == '2' or operation_code == '3'):
            is_next_input = True
            print '无效字符，请重新输入！\n'

    for line in input_file.readlines():
        line = re.sub(r'[\r\n]', '', line)
        match_obj = re.search(reg, line)
        if operation_code == '1':
            after_line = match_obj.group() if match_obj else None
        elif operation_code == '2':
            after_line = line if match_obj else None
        elif operation_code == '3':
            after_line = None if match_obj else line

        if after_line:
            if is_fisrt_line:
                is_fisrt_line = False
                output_file.write(after_line)
            else:
                output_file.write('\n' + after_line)


def delete_duplicate_lines(input_file, output_file):
    lines_set = set()
    is_fisrt_line = True
    for line in input_file.readlines():
        line = re.sub(r'[\r\n]', '', line)
        if line not in lines_set:
            lines_set.add(line)
            if is_fisrt_line:
                is_fisrt_line = False
                output_file.write(line)
            else:
                output_file.write('\n' + line)


def main():
    print '当前 Python 版本号为：' + str(sys.version) + '\n'
    # print '当前 Python 路径为：' + str(sys.path) + '\n'
    input_txt = "input.txt"
    output_txt = "output.txt"

    # 如果 output.txt 文件存在，则删除
    if os.path.exists(output_txt):
        os.remove(output_txt)
        print('output.txt 文件已自动删除！\n')

    output_file = open(output_txt, 'a')
    input_file = open(input_txt, 'r')

    print '提示：输入文件为 input.txt ，输出文件为 output.txt\n'
    print '→ 使用正则表达式请输入 1'
    print '→ 删除重复的行请输入 2'
    print '→ 直接退出请按 0'
    is_next_input = True
    operation_code = '-1'
    while is_next_input:
        is_next_input = False
        operation_code = raw_input('→ 请输入数字：')
        if operation_code == '1':
            use_reg(input_file, output_file)
        elif operation_code == '2':
            delete_duplicate_lines(input_file, output_file)
        elif operation_code != '0':
            is_next_input = True
            print '无效字符，请重新输入！\n'

    input_file.close()
    output_file.flush()
    output_file.close()

    print '\n已退出' if operation_code == '0' else '\n处理完成，请查看 output.txt 文件！（下次执行该脚本时 output.txt 文件会自动删除）'


if __name__ == "__main__":
    main()