#!/usr/bin/python
# -*- coding: UTF-8 -*-
"""
@author:Awen
@file:task_test.py
@time:2023/07/04
@function：Python實(shí)現(xiàn)在word中批量查找關(guān)鍵字，如果關(guān)鍵字在word中，則將文件移出到新建的文件夾中
"""
import os
import shutil
from docx import Document
from docx.opc.exceptions import PackageNotFoundError
def search_keyword_in_word(keyword, file_path):
    # 若報錯，則跳過，有的文件無法讀取成功
    try:
        document = Document(file_path)
        found_paragraphs = []
        for paragraph in document.paragraphs:
            if keyword in paragraph.text:
                found_paragraphs.append(paragraph.text)
        return found_paragraphs
    except PackageNotFoundError:
        print(f"文件 '{file_path}' 未找到。")
        pass
# 原文件夾路徑
source_file_path = 'D:\Pycharmproject2023\code_test_project\shan_test\正樣本終版\\'
#  搜索的關(guān)鍵字
keyword = '求職意向'
for filename in os.listdir(source_file_path):
    try:
        # 創(chuàng)建新文件夾，用于找到相應(yīng)文件之后將其移入到該文件夾中
        new_directory = os.path.join(source_file_path, "data123")
        os.makedirs(new_directory, exist_ok=True)
        if filename.endswith('.docx'):
            print(filename)
            result = search_keyword_in_word(keyword, source_file_path+filename)
            for paragraph in result:
                print(paragraph)
                # 如果求職意向中包含java字樣，則是任務(wù)目標(biāo)文件
                if "java" in paragraph.lower():
                    new_file_path = os.path.join(new_directory, filename)
                    shutil.move(source_file_path+filename, new_file_path)
                    print(f"已移動文件: {filename}")
                    # # 刪除原文件夾中的文件
                    # os.remove(source_file_path+filename)
                    # print(f"已刪除文件: {filename}")
    except Exception as e:
        print(f"處理文件 '{filename}' 時出現(xiàn)錯誤: {str(e)}")
        pass