Python高效实现Excel转TXT文本

from spire.xls import *
from spire.xls.common import *

workbook = Workbook()
workbook.LoadFromFile("示例.xlsx")

sheet = workbook.Worksheets[0]
sheet.SaveToFile("output.txt", "\t", Encoding.get_UTF8())

from spire.xls import *
from spire.xls.common import *

def excel_sheets_to_txt(input_file, output_folder, delimiter="\t"):
  
    # 创建Workbook对象
    workbook = Workbook()
    # 加载Excel文件
    workbook.LoadFromFile(input_file)
        
    # 遍历所有工作表
    for i in range(workbook.Worksheets.Count):
        sheet = workbook.Worksheets[i]
            
        # 构建输出文件路径
        output_file = f"sheet_{i+1}_{sheet.Name}.txt"
        output_path = os.path.join(output_folder, output_file)
            
        # 将工作表内容保存为TXT文件
        sheet.SaveToFile(output_path, delimiter, Encoding.get_UTF8())

import openpyxl

def xlsx_to_txt(file_path):
    # 加载Excel工作簿
    workbook = openpyxl.load_workbook(file_path)
    # 假设数据在第一个工作表中
    sheet = workbook.active

    # 遍历每一行
    for row_num, row in enumerate(sheet.iter_rows(values_only=True), start=1):
        if row_num > 1:  # 假设第一行是标题行，从第二行开始处理
            # 检查row[1]是否存在，以防有些行可能缺失数据
            if row[1] is not None:
                # 将row[1]的值转换为字符串并写入以row[0]命名的txt文件
                with open(f"{row[0]}.txt", "w", encoding="utf-8") as txt_file:
                    txt_file.write(str(row[1]))

# 假设我们有一个名为`data.xlsx`的Excel文件                    
xlsx_to_txt('data.xlsx')

import pandas as pd
import re
import codecs
#将excel转化为txt文件
def exceltotxt(excel_dir, txt_dir):
with codecs.open(txt_dir, 'w', 'utf-8') as f:
neg=pd.read_excel(excel_dir, header=none, index=none)
f.write(neg.to_string())
#去除记录行首的数字和空格
def del_linehead_number_speace(orig_txt_dir,saveas_txt_dir):
with open(orig_txt_dir,'r+') as f, open(saveas_txt_dir,'r+') as fw:
lines = f.readlines()
print(len(lines)) #行数
texts = [re.sub(r'(\d)+(\s)+','',lines[num]) for num in range(len(lines)) ]
texts = list(set(texts)) #去重如果要保留重复记录注释该行
line_num = len(texts)
# for num in range(line_num): #查看转化后的文本
# print(texts[num])
fw.writelines(texts)
exceltotxt('./data/neg.xls', './data/neg_temp.txt')
del_linehead_number_speace('./data/neg_temp.txt','./data/neg.txt')