dataTools/init.py

from jiexiExcel import write_data_to_file  # 导入特定的函数
from checkBad import beginCheckBadFun# excel文件夹路径
from formatData import formatAlpaca# excel文件夹路径
from merge import merge_json_files # 合并所有为一个训练脚本
# import asyncio
# 清洗数据函数
def formatDataFun ():
      
  input_directory = '/Users/yushanghui/hongshantianping/git/dataTools/book/迭代修改/'
  # 把excel文件转成json文件 并保存到指定路径
  output_file = '/Users/yushanghui/hongshantianping/git/dataTools/book/json/diedaiEdit.json'
  # 通过checkBad.py 函数进行过滤指定输出文件夹
  out_file = '/Users/yushanghui/hongshantianping/git/dataTools/book/jsonOut/diedaiEdit'

  alpaca_file = '/Users/yushanghui/hongshantianping/git/dataTools/book/alpaca/diedaiEdit.json' 
 
  write_data_to_file(input_directory,output_file)
  beginCheckBadFun(output_file,out_file)
  formatAlpaca(out_file,alpaca_file)

async def mergedFun ():
  
  input_folder1 = './book/alpaca'  # 输入文件夹路径
  output_file1 = './book/all/merged.json'  # 输出文件名

  # 第一次执行 merge_json_files
  await merge_json_files(input_folder1, output_file1)
  
if __name__ == "__main__": 
  # asyncio.run(mergedFun())
  formatDataFun()