In [1]:
def readFile( filename ):
f = open( filename )
text = f.read()
f.close()
return text
def writeFile( filename, text ):
f = open( filename, "w" )
f.write( text )
f.close()
def filterText( text ):
output = ""
wasSpace = False
for ch in text.upper():
if ch >= 'A' and ch <= 'Z':
# je to zaciatok dalsieho slova?
if wasSpace:
output = output + " "
wasSpace= False;
output = output + ch
else:
wasSpace = True;
return output
In [2]:
message = readFile('TXT/spolu_ascii.txt')
print(message[:500])
In [3]:
message = filterText(message)
print(message[:500])
In [4]:
alphabet = "ABCDEFGHIJKLMNOPQRSTUVWXYZ "
def countChars(message):
msg = message.upper()
multiplicities = [ msg.count(i) for i in alphabet ]
return multiplicities
def getProbabilities(message):
multiplicities = countChars(message)
total = sum(multiplicities)
probabilities = [ float(n) / float(total) for n in multiplicities ]
return probabilities
def printProbabilities(probabilities):
for i in range(0, len(alphabet)):
print("%c\t%.4f" % (alphabet[i], probabilities[i]))
In [5]:
print(countChars(message))
In [6]:
probabilities = getProbabilities(message)
printProbabilities(probabilities)
In [7]:
def sortProbabilities(probabilities):
array = []
for i in range(0, len(alphabet)):
tuple = (probabilities[i], alphabet[i])
array.append(tuple)
array.sort(reverse=True)
return array
In [8]:
print(sortProbabilities(probabilities))
In [9]:
import matplotlib
import numpy as np
import matplotlib.pyplot as plt
def showProbabilities(probabilities):
index = np.arange(len(alphabet))
yy = [y for y in probabilities]
plt.bar(index, yy, 0.8)
plt.xticks(index, [letter for letter in alphabet])
plt.title('Probabilities')
plt.show()
In [10]:
showProbabilities(probabilities)
In [11]:
def spocitajVyskyt(text, pocetZnakov=2):
pocetnosti = {}
for i in range(len(text)-pocetZnakov):
s = text[i:i+pocetZnakov]
pocetnosti[s] = pocetnosti.get(s, 0) + 1
return pocetnosti
def spocitajDvojice(text):
pocet = {}
for i in range(len(text)-1):
s = text[i:i+2]
pocet[s] = pocet.get(s, 0) + 1
return pocet
In [12]:
dvojice = spocitajDvojice(message)
print(sorted(dvojice.items(), key=lambda it: it[1], reverse=True))
In [13]:
def spocitajTrojice(text):
pocet = {}
for i in range(len(text)-2):
s = text[i:i+3]
pocet[s] = pocet.get(s, 0) + 1
return pocet
In [14]:
trojice = spocitajTrojice(message)
In [15]:
len(trojice)
Out[15]:
In [16]:
27**3
Out[16]:
In [17]:
def spocitajStvorice(text):
pocet = {}
for i in range(len(text)-3):
s = text[i:i+4]
pocet[s] = pocet.get(s, 0) + 1
return pocet
In [18]:
stvorice = spocitajStvorice(message)
In [19]:
print(len(stvorice))
In [20]:
len(stvorice)/27**4
Out[20]:
In [21]:
def spocitajPatice(text):
pocet = {}
for i in range(len(text)-4):
s = text[i:i+5]
pocet[s] = pocet.get(s, 0) + 1
return pocet
patice = spocitajPatice(message)
print(len(patice))
In [22]:
from math import log
def spocitajEntropiu(ntice):
entropy = 0.0
n = sum(ntice.values())
for i in ntice.values():
p = i / n
entropy = entropy - p * log(p, 2)
return entropy
In [23]:
print(spocitajEntropiu(dvojice))
print(spocitajEntropiu(trojice))
print(spocitajEntropiu(stvorice))
print(spocitajEntropiu(patice))
print(spocitajEntropiu(trojice)-spocitajEntropiu(dvojice))
print(spocitajEntropiu(stvorice)-spocitajEntropiu(trojice))
print(spocitajEntropiu(patice)-spocitajEntropiu(stvorice))
In [24]:
for i in range(2,20):
ntice = spocitajVyskyt(message, i)
print(i,len(ntice),spocitajEntropiu(ntice))
In [25]:
xx = [item for item in dvojice.items() if item[0].startswith('B')]
print(sorted(xx, key=lambda it: it[1], reverse=True))
In [26]:
yy = [item for item in trojice.items() if item[0].startswith('BO')]
print(sorted(yy, key=lambda it: it[1], reverse=True))
In [27]:
def getIndexOfCoincidence(message):
multiplicities = countChars(message)
n = sum(multiplicities)
return float(sum([ ni * (ni-1) for ni in multiplicities ])) / float((n * (n - 1)))
In [28]:
print(getIndexOfCoincidence(message))
In [29]:
def ctoi( c ):
return ord(c) - 65
def itoc( i ):
return chr(i + 65)
def caesarEncrypt( plainText, key ):
output = ""
for ch in plainText.upper():
if ch >= 'A' and ch <= 'Z':
output = output + itoc( (ctoi(ch) + key) % 26 )
else:
output = output + ch
return output
def caesarDecrypt( cipherText, key ):
decipherKey = 26 - key
return caesarEncrypt( cipherText, decipherKey)
In [30]:
encrypted = filterText(readFile('TXT/viegenere_encrypted.txt'))
print(encrypted)
In [31]:
print(getIndexOfCoincidence(encrypted))
In [32]:
prob2 = getProbabilities(encrypted)
printProbabilities(prob2)
In [33]:
showProbabilities(prob2)
In [34]:
def showProbabilities2(probabilities1, probabilities2):
index = np.arange(len(alphabet))
xx = [x for x in alphabet]
yy1 = [y for y in probabilities1]
yy2 = [y for y in probabilities2]
plt.bar(index, yy1, 0.35)
plt.bar(index + 0.4, yy2, 0.35)
plt.xticks(index, [letter for letter in alphabet])
plt.title('Probabilities')
plt.show()
In [35]:
showProbabilities2(probabilities,prob2)
In [36]:
pom1 = probabilities.copy()
pom1.sort(reverse=True)
pom2 = prob2.copy()
pom2.sort(reverse=True)
showProbabilities2(pom1,pom2)
In [37]:
decrypted = caesarDecrypt(encrypted, 3)
prob2 = getProbabilities(decrypted)
showProbabilities2(probabilities,prob2)
In [38]:
print(decrypted)
In [ ]: