getRightCol.py

"""
This file takes the column closest to the right and performs a set analysis on it 
"""

from typing import List, Set

def getRightCol(lines: List[str]) -> List[str]:
    """
    get the rightest column of a kind of tables

    lines: the data like [
                "0.01    0.000           2        18         9 access           "
                "0.00    0.000025           1        25        22 readlink      "
                "0.00    0.000019           0        96           mmap          "
                "0.00 0.000017           1        12           stat             "
            ]

    return: the item in rightest columns [ "access", "readlink", "mmap", "stat" ]
    """
    return [line.split()[-1] for line in lines]

def get_intersection(list_a: List[str], list_b: List[str]) -> Set[str]: 
    """
    get the item both in a and b
    """
    return set(list_a) & set(list_b) 
    
def get_difference(list_a: List[str], list_b: List[str]) -> Set[str]: 
    """
    get the item which existed in a but not b
    """
    return set(list_a) - set(list_b) 


if __name__ == "__main__":
    subscriber = None
    publisher = None

    with open("p_out", "r") as File:
        publisher = File.readlines()        

    with open("s_out", "r") as File:
        subscriber = File.readlines()        

    assert(publisher != None and subscriber != None)
    pub_list = getRightCol(publisher)
    sub_list = getRightCol(subscriber)


    intersection = get_intersection(pub_list, sub_list)
    print(intersection)
    diff = get_difference(pub_list, sub_list)
    print(diff)
    diff = get_difference(sub_list, pub_list)
    print(diff)