temp_code/python/merge_file_py/merge_file.py

# -*- coding: utf-8 -*-

import os
import PyPDF2
import re
import shutil

def main():
    path = './'

    pdf_list = []

    all_file_list = []

    for file_name in os.listdir(path):
        if '.pdf' in file_name:
            pdf_list.append(file_name)
        if '.pdf' in file_name or '.xlsx' in file_name:
            all_file_list.append(file_name)

    if not pdf_list:
        return -1

    for pdf_name in pdf_list:
        pdffile = open('./' + pdf_name, 'rb')
        pdfreader = PyPDF2.PdfFileReader(pdffile)
        page0 = pdfreader.getPage(0)
        pdf_text = page0.extractText()

        code = re.findall('款号：(.*?)颜色', pdf_text)
        if code:
            code = code[0]
        folder_name = './' + code
        if not os.path.exists(folder_name):
            os.mkdir(folder_name)
        
        target = './' + folder_name + '/' + pdf_name
        source = './' + pdf_name
        shutil.copyfile(source, target)

        excel_name = pdf_name.split('.')[0] + '.xlsx'
        excel_source = './' + excel_name
        excel_target = './' + folder_name + '/' + excel_name
        shutil.copyfile(excel_source, excel_target)

        pdffile.close

    for f in all_file_list:
        os.remove(f)

main()
print('ok')