splitter.py

# -*- coding: utf-8 -*-
"""
Created on Tue Oct 18 19:35:04 2022

@author: Arman Hossain
"""
import re
def split_cpe(cpe):
    '''
    cpe::  cpe:2.3:o:juniper:junos:16.1:r1:*:*:*:*:*:*
           0   1  2    3      4     5   6 7 8 9 10 11,12 -->(index)
    return (cpe,2.3,o,juniper,junos,16.1,r1,*,*,*,*,*,*)
    '''
    
    # https://cpe.mitre.org/specification/
    
    if(not isinstance(cpe, str)):
        print("--- cpe1: ",cpe, " ----")
        print("Error from spliter.py split_cpe(): cpe syntax is not string")
        return ("cpe_idx1","cpe_idx2","cpe_idx3","cpe_idx4","cpe_idx5","cpe_idx6","cpe_idx7","cpe_idx8","cpe_idx9","cpe_idx10","cpe_idx11","cpe_idx12")
    
    cpe = cpe.replace("\:","-")
    cpe = re.sub("[-]+","-",cpe)
    '''
        parsing needed for this type of cpe [here net\:\ part is vendor and net\:\:netmask part is product]
        cpe:2.3:a:net\:\:netmask_project:net\:\:netmask:*:*:*:*:*:perl:*:* 
    '''
    splited = cpe.split(':')
    
    if(len(splited)!=13):
        print("cpe2: ",cpe)
        print("Error from spliter.py split_cpe(): cpe syntax is not recognized")
    return splited


def split_date(date):
    
    '''
    date: 2019-01-15T21:29Z
    return: (2019,01,15,T21,29Z)
    '''
    if(not isinstance(date, str)):
        print("--- date1: ",date," ----")
        print("Error from spliter.py date(): date syntax is not recognized")
        return (0000,00,00,"T00","00Z")
    # sdate = dates[0]
    tz = date.split("T")
    
    time = tz[0].split("-")
    if len(time)!=3:
        print("--- date2: ",date," ----")
        print("Error from spliter.py date(): date syntax is not recognized")
        return (0000,00,00,"T00","00Z")
    yy = int(time[0])
    mm = int(time[1])
    dd = int(time[2])
    if(len(tz)!=2 or mm>12):
        print("--- date3: ",date," ----")
        print("Error from spliter.py date(): date syntax is not recognized")
        return (yy,mm,dd,"T00","00Z")
    
    time =  tz[1].split(":")
    if len(time)!=2:
        print("--- date4: ",date," ----")
        print("Error from spliter.py date(): date syntax is not recognized")
        return (yy,mm,dd,"T00","00Z")
    return (yy,mm,dd,"T"+time[0],time[1])