Parsing PITest mutations.csv

parse_mutation_info.py

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

import os.path, sys, sets


def reformat(line):
    return replace_slash_dot(
        split_line(
            remove_first_char(
                remove_newline(
                    line))))

def remove_newline(line):
    return line.rstrip('\n')

def remove_first_char(line):
    return line[1:]

def split_line(line):
    return line.split('(')[0]

def replace_slash_dot(line):
    return line.replace("/",".")
    

def main(mutations_file):
    assert os.path.exists('functions.txt')
    assert os.path.exists(mutations_file)

    lst_fns = [reformat(line) for line in open('functions.txt')]
    set_fns = sets.Set(lst_fns)

    lst_mutations = [remove_newline(line) for line in open(mutations_file)]
    dict_mutants = {};
    for mutant in lst_mutations:
        arr = mutant.split(",")
        fn = '.'.join([arr[1],arr[3]]) 
        if fn in set_fns:
            if fn in dict_mutants:
                dict_mutants[fn].add(mutant)
            else:
                dict_mutants[fn] = sets.Set([mutant])


    f= open("mutations.csv","w+")

    for fn, mutants in dict_mutants.items():
        size = 0.0
        counter = 0.0
        size = len(mutants)
 
        for mutant in mutants:
            arr = mutant.split(',')
            if arr[6] != 'none':
                counter = counter + 1  

        f.write("{}\t{:.2f}\n".format(fn.replace(".","/"),counter/size)) 

    for fn in set_fns:
        if fn not in dict_mutants:
            f.write("{}\t0\n".format(fn.replace(".","/")))
    
    f.close()

if __name__ == '__main__':
    mutations_file = sys.argv[1]
    main(mutations_file)