python批量替换word文字

python批量替换word文字

代码

"""
        本项目特点：
            1、获取word中数据,批量替换word中文字！

"""
import os
from docx import Document


def replace_word(doc, old_word, new_word):
    """
    定义批量替换文字的函数
    :param doc: 要替换的文档
    :param old: 被替换的文字
    :param new: 替换后的文字
    :return:
    """
    #ss = False
    for p in doc.paragraphs:  # 遍历文档段落

        for run in p.runs:  # 遍历段落的字块
            #if "一般公共预算当年拨款结构情况" in run.text:
                #ss = True
            #elif "一般公共预算当年拨款具体使用情况" in run.text:
                #ss = False
            #if ss:
                #if "（类）" not in run.text:
            if old_word in run.text:
                run.text = run.text.replace(old_word, new_word)

    # 遍历文档的表格， 替换表格里的要替换的文字
    # for table in doc.tables:
    #     for row in table.rows:
    #         for cell in row.cells:
    #             cell.text = cell.text.replace(old_word, new_word)


data_text = [
    {"old": "一般公共服务支出", "new": "一般公共服务（类）支出"},
    {"old": "外交支出", "new": "外交（类）支出"},
    {"old": "国防支出", "new": "国防（类）支出"},
    {"old": "公共安全支出", "new": "公共安全（类）支出"},
    {"old": "教育支出", "new": "教育（类）支出"},
    {"old": "科学技术支出", "new": "科学技术（类）支出"},
    {"old": "文化旅游体育与传媒支出", "new": "文化旅游体育与传媒（类）支出"},
    {"old": "社会保障和就业支出", "new": "社会保障和就业（类）支出"},
    {"old": "社会保险基金支出", "new": "社会保险基金（类）支出"},
    {"old": "卫生健康支出", "new": "卫生健康（类）支出"},
    {"old": "节能环保支出", "new": "节能环保（类）支出"},
    {"old": "城乡社区支出", "new": "城乡社区（类）支出"},
    {"old": "农林水支出", "new": "农林水（类）支出"},
    {"old": "交通运输支出", "new": "交通运输（类）支出"},
    {"old": "资源勘探信息等支出", "new": "资源勘探信息等（类）支出"},
    {"old": "商业服务业等支出", "new": "商业服务业等（类）支出"},
    {"old": "金融支出", "new": "金融（类）支出"},
    {"old": "援助其他地区支出", "new": "援助其他地区（类）支出"},
    {"old": "自然资源海洋气象等支出", "new": "自然资源海洋气象等（类）支出"},
    {"old": "住房保障支出", "new": "住房保障（类）支出"},
    {"old": "粮油物资储备支出", "new": "粮油物资储备（类）支出"},
    {"old": "国有资本经营预算支出", "new": "国有资本经营预算（类）支出"},
    {"old": "灾害防治及应急管理支出", "new": "灾害防治及应急管理（类）支出"},
    {"old": "其他支出", "new": "其他（类）支出"},
    {"old": "转移性支出", "new": "转移性（类）支出"},
    {"old": "债务还本支出", "new": "债务还本（类）支出"},
    {"old": "债务付息支出", "new": "债务付息（类）支出"},
    {"old": "债务发行费用支出", "new": "债务发行费用（类）支出"},
    {"old": "抗疫特别国债还本支出", "new": "抗疫特别国债还本（类）支出"},
]
for i in range(100):
    try:
        path = input(str(i + 1) + ':请输入文件夹地址：').replace('\\', '/')

        for file_name in os.listdir(path):

            print(file_name)
            doc_path = path + '/' + file_name
            doc = Document(doc_path)
            for data in data_text:
                replace_word(doc, data["old"], data["new"])
            doc.save(file_name)
        print("================   完成个文件 =================")
        input('=========================================')
    except Exception as E:
        print('错误：', '1.文件夹里的文件不是”.docx“类型文件！  2.文件地址有无（”D:\”）')

Gitee地址

https://gitee.com/hayratjan点击打开海哥Gitee

微信赞赏