MonotaroReceiptRename.py

#-----------------------------------------------------------
# 20230901 
# Monotaroの領収書のpdfのファイル名を"注文日付＿領収金額＿Monotaro.pdf"とする
#-----------------------------------------------------------

import glob,os
from pdfminer.high_level import extract_text
from datetime import datetime


# 文字列抽出（検索文字列、開始文字、終了文字）
def extraction_string(input_string,start_string,end_string):
  start_index = input_string.find(start_string)
  end_index = input_string.find(end_string, start_index)

  if start_index != -1 and end_index != -1:
    extracted_string = input_string[start_index + len(start_string):end_index]
    print("抽出された文字列:", extracted_string)
    return extracted_string
  else:
    print("指定された文字列が見つかりませんでした。")
    return None



input_list=glob.glob("*.pdf")
print(input_list)


for pdf in input_list:

  if "領収証の発行" in pdf: #まだ、ファイル名の変換がされていない場合の処理
    input_string  = extract_text(pdf)
#    print(input_string)

    accounting_day=extraction_string(input_string,"領収⽇︓","\n")
    accounting_amount_money=extraction_string(input_string,"￥","\n")
#    print(accounting_day,accounting_amount_money)

#    accounting_day=accounting_day.replace(" ","")
    accounting_amount_money= accounting_amount_money.replace(",","")
    accounting_amount_money= accounting_amount_money.replace("-","")
    accounting_amount_money= accounting_amount_money.zfill(7)  #百万まで左0詰め


    accounting_day=accounting_day.replace("年","")
    accounting_day=accounting_day.replace("⽉","")
    accounting_day=accounting_day.replace("⽇","")

    print(accounting_day+"_"+accounting_amount_money+"_Monotaro.pdf")
    new_filename = accounting_day+"_"+accounting_amount_money+"_Monotaro.pdf"

  # ファイル名を変更
    os.rename(pdf, new_filename)