pids_manipulator.py

#!/usr/bin/env python

import re
import sys


def read_files():
    '''Open and read log file for searching content files'''
    with open(sys.argv[1]) as f:
        content = f.read()
        search1 = re.findall(r'(\w+)(\.)(jpg|pdf|gif)', content)
        search2 = re.findall(r'(\w+)(\/)(index.html)', content)
        return (search1, search2)


def extract_pids(search1, search2):
    '''Extract PIDs from content file name also merging and removing duplications'''
    ext1 = tuple(pids1[0] for pids1 in search1)
    ext2 = tuple(pids2[0] for pids2 in search2)
    ext = set(ext1 + ext2)
    return ext


def store_pids(ext):
    '''Store PIDs in file'''
    for out in ext:
        with open('PIDs.log', 'a') as result:
            result.write(out+'\n')

def main():
    matches = read_files()
    pids = extract_pids(*matches)
    store_pids(pids)


if __name__ == '__main__':
    main()