Python Dictionary File Maker v0.2

[Fi8sVrs]

#!/usr/bin/env python
#Python Dictionary File Maker v0.2
#Written By: Absolution

import os, string, time, hashlib, math,sys
from itertools import product, chain

bench = 0

def UI():
    print "Dictionary Maker v0.2"
    print "Written By: Absolution"
    print '''
        1 - Make Dictionary
        2 - Benchmark System
        3 - Exit
        '''
    ui = raw_input("What would you like to do? => ")
    if ui == '1':
        DictInfo()
    elif ui == '2':
        Benchmark()
    elif ui == '3':
        exit()

def DictInfo():
    print '''
        1 - Lowercase Letters
        2 - Uppercase Letters
        3 - Numbers
        4 - Uppercase & Lowercase
        5 - Uppercase, Lowercase & Numbers
        6 - Uppercase, Lowercase & Punctuation
        7 - Uppercase, Lowercase, Numbers & Punctuation
        8 - Enter your own charset
    '''
    charset = raw_input("Which charset to use? => ")
    maxfilesize = raw_input("Enter max size of each file (MB) => ")
    minlength = raw_input("Enter min length => ")
    maxlength = raw_input("Enter max length => ")
    maxlength = int(maxlength) + 1
    md5 = raw_input("Include MD5 Hash? (Y or N) => ")
    MakeDict(charset, maxfilesize, minlength, maxlength, md5)

def MakeDict(charset, maxfilesize, minlength, maxlength, md5):
    global bench
    #Load charset for dictionary maker to use
    if charset == '1':
        letters = string.ascii_lowercase
    elif charset == '2':
        letters = string.ascii_uppercase
    elif charset == '3':
        letters = string.digits
    elif charset == '4':
        letters = string.ascii_letters
    elif charset == '5':
        letters = string.ascii_letters + string.digits
    elif charset == '6':
        letters = string.ascii_letters + string.punctuation
    elif charset == '7':
        letters = string.ascii_letters + string.digits + string.punctuation
    elif charset == '8':
        letters = raw_input("Enter characters to use => ")
    else:
        print 'No valid char set was chosen.'
        DictInfo()

    combinations = 0
    totalsize = 0

    for z in range(int(minlength),int(maxlength)):
        combs = math.pow(int(len(letters)), z)
        combinations = int(combinations) + int(combs)
        size = int(combs) * z
        totalsize = int(totalsize) + int(size)

    wspace = int(combinations) * 2
    totalsize = int(totalsize) + int(wspace)

    if md5 == "Y":
        md5calc = int(combinations) * 34
        totalsize = int(totalsize) + int(md5calc)

    if totalsize < 1000:
        sizeappend = "bytes"
    elif totalsize > 1000 and totalsize < 1000000:
        sizeappend = "kB"
        totalsize = int(totalsize) / 1000
    elif totalsize > 1000000 and totalsize < 1000000000:
        sizeappend = "MB"
        totalsize = int(totalsize) / 1000000
    elif totalsize > 1000000000 and totalsize < 1000000000000:
        sizeappend = "GB"
        totalsize = int(totalsize) / 1000000000
    elif totalsize > 1000000000000:
        sizeappend = "TB"
        totalsize = int(totalsize) / 1000000000000


    #Creates variable for filename
    y = 1 #Dictionary iteration
    filename = 'dict' #Change name to whatever you want
    extension = '.txt' #Change extension to whatever you want
    dictfile = (filename + str(y) + str(extension)) #File name variable
    print ("Writing dictionary that contains the charset %s, with a min length of %s characters and a max length of %s characters." % (letters, minlength, str(int(maxlength) -1)))
    print ("You have selected %i different combinations, the output file size will be about %i %s" % (combinations, totalsize, sizeappend))
    if bench > 0:
        estimatedtime = int(combinations) / int(bench)
        estimatedtime = int(estimatedtime) / 60
        print("It is estimated that the dictionary creation will take %i minutes to complete" %(estimatedtime))
    startstop = raw_input("Are you sure you wish to continue? (Y or N) => ")
    if startstop == "N":
        exit()
    starttime = time.time()

    #Create dictionary words
    try:
        f = open(dictfile,'w') #Initially opens file
        for x in xrange(int(minlength),int(maxlength)):
                my_chain = product(letters, repeat=x)
                print ("Writing %i letter words..." % (x))
                for char_set in my_chain:
                    word = string.join(char_set) #Join letters together in list
                    word = string.replace(word, ' ', '') #Strip out whitespace
                    f.write(word + '\n') #Writes current word to file
                    if md5 == 'Y':
                        hash = hashlib.md5(word).hexdigest()
                        f.write(hash + '\n')

    #When each word is written to file, check filesize
                    filesize = os.path.getsize(dictfile) #Get current filesize
                    filesize = int(filesize)/1048576 #Convert to MB
                    if int(filesize) >= int(maxfilesize):
                        print ('%s has reached %i MB. Closing file and making a new one' % (dictfile, filesize))
                        print ("Last written word was " + str(word))
                        f.close() #Close current file
                        y = int(y) + 1 #Move to next file number
                        dictfile = (filename + str(y) + str(extension)) #Update filename variable
                        print ('Opening new dictionary file called ' + dictfile)
                        f = open(dictfile,'w') #Open new file for writing
    except KeyboardInterrupt or StandardError or TypeError as error:
        f.close()
        print ('Process interrupted...' + str(error))
    #When done:
    f.close() #Finished so close the file
    finishtime = time.time()
    totaltime = int(finishtime) - int(starttime)
    print ("Finished!\n %i dictionary file(s) were created and your files are located in folder %s" %(y, os.getcwd()))
    print ('Time Taken: %s seconds' % (str(totaltime)))
    UI()

def Benchmark():
    global bench
    benchfile = "benchmark.txt"
    print("Creating benchmark test file...")
    benchstart = time.time()
    try:
        f = open(benchfile,'w')
        for x in xrange(1,5):
                    chain = product('abcdefghijklmnopqrstuvwxyz', repeat=x)
                    print ("Writing %i letter words..." % (x))
                    for char_set in chain:
                        word = string.join(char_set) #Join letters together in list
                        word = string.replace(word, ' ', '') #Strip out whitespace
                        f.write(word + '\n')
        f.close()
        benchstop = time.time()
        benchtime = int(benchstop) - int(benchstart)
        bench = 475254 / int(benchtime)
    except ZeroDivisionError or IOError or ArithmeticError as error:
        print "There was an error, trying again."
        Benchmark()
    print("Your computer has been benched at %i words per second" %(bench))
    print("Removing benchmark test file...")
    os.remove(benchfile)
    UI()

UI()

mirror

scr