import pdfkit
import os
import easygui
import pypandoc

filepath = easygui.diropenbox()

config = pdfkit.configuration(wkhtmltopdf=r'wkhtmltopdf.exe')


#判断文件大小
for curDir, dirs, files in os.walk(filepath):
    for file in files:
        # print(os.path.join(curDir, file))
        if os.path.getsize(os.path.join(curDir, file)) == 0 :
            os.remove(os.path.join(curDir, file))
        else:
            prefix = os.path.splitext(os.path.join(curDir, file))
            os.rename(os.path.join(curDir, file),os.path.join(curDir, prefix[0]+".html"))
            #转为PDF
            pdfkit.from_file(os.path.join(curDir, file), os.path.join(curDir, prefix[0]+".pdf"), configuration=config)
            #转为doc
            pypandoc.convert_file(os.path.join(curDir, file), 'docx', outputfile=os.path.join(curDir, prefix[0]+".doc"))