计算某目录下电子发票总金额

fapiao.py

#!/usr/bin/env python
#coding:utf-8

#sudo yum install poppler-utils

import os
import re
import subprocess
from decimal import *


def scrape_text(src):
    return subprocess.check_output(['pdftotext', src, '-', '-q'])

def main():
    all_value = Decimal(0)
    all_nums = set()
    for src in os.listdir('.'):
        if os.path.isfile(src) and src.endswith('.pdf'):

	    text = scrape_text(src)
            
            num = None
            o = re.search('发票号码\s*(：|:)\s*(\d+)', text, re.M)
            if o is not None:
                num = o.group(2)
                if num in all_nums:
                    print src, 'already exists'
                    continue
                else:
                    all_nums.add(num)
            else:
                print src, 'error get tax num --------------'
                continue

            o = re.search('([(]|（)小写([)]|）)\s*(¥|￥)?\s*(\d+[.]\d+)', text, re.M)
            if o is None:
                print src, 'error get tax value --------------'
                continue

            m = Decimal(o.group(4))
            print src, num, m
            all_value += m

    print 'total is', all_value
    raw_input('plz input enter to close this')
            
if __name__ == "__main__":
    main()