hashing.py

"""
    Ayni dizinde bulunan algoritmalar_2014_ogrenci_listesi.csv
    dosyasini en alt satirda ki;
    differentValues(hashing(dosya_adi.csv)) seklinde arguman olarak verin
    Cakismayan sayisi = 106, cakisan sayisi = 41 seklincde cikacaktir.
    Alaattin USTA([email protected])
"""
import csv
import time
import hashlib

def hashing(csv_file):
    start_time = time.time()
    reverse_alphabet = "ZYXWVUTSRQPONMLKJIHGFEDCBA"
    hash_table = dict()
    adjusted_data = ""
    index = 0
    open_file = open(csv_file)

    for row in csv.reader(open_file):
        changer = 0
        hashingResultValue = 0
        for col in row:
           for character in col:
               hashingResultValue +=abs(hash(character)) % (10**8) + \
               (ord(character) * 119) + reverse_alphabet.find(character) * (146**3)
               changer += 1
        hash_table[index] = hashingResultValue % 147
        index += 1
    elapsed_time = time.time()- start_time
    print "Gecen sure:", elapsed_time
    return hash_table

def differentValues(hash_table):
    hashLenght = len(hash_table)
    for i in range(hashLenght):
        for j in range(hashLenght):
            if not (i == j):
                if (hash_table[i] == hash_table[j]):
                    hash_table[j] = "overlap"
    differentValues = 0
    overlapValues = 0
    for i in range(hashLenght):
        if(hash_table[i] == "overlap"):
            overlapValues += 1
        else:
            differentValues += 1
    print "Cakismayanlar sayisi:", differentValues
    print "Cakisan sayisi:", overlapValues

differentValues(hashing("algoritmalar_2014_ogrenci_listesi.csv"))