Laboratory 8 Simple Ciphers

Computer Science CS134 
Laboratory 8
Simple Ciphers (due Thursday at 5pm)
Objective. To use inheritance to quickly construct a new character substitution cipher.
In this lab, we'll experiment with the use of inheritance to help us build classes that support a variety
of cipher mechanisms. A cipher is an approach to transforming the characters of a readable plaintext
message into an encoded message. It's necessary, of course, for a successful cipher technique to provide a
decoding method that allows the recipient to recover the plaintext.
The ciphers we'll implement this week range from the simple to the complex, but every cipher technique,
coder, supports two main methods: coder.encode(sourceString) and coder.decode(sourceString).
We'll provide these in a base class, Cipher, that we will extend several new classes that support solid
cipher implementations. This approach is important for two reasons. First, utilities that make use of
ciphers, for example tools for reading and writing encoded  les, will be more  exible if all our cipher
techniques support the same methods. We can quickly substitute more or less secure approaches by a
simple substitution of subclasses of Cipher that are used in a consistent manner. Second, when we
realize that several classes will share the same interface with the outside world, there is a promise that
code can be shared. Thoughtful sharing of code through a central Cipher implementation will make new
implementations very simple.
Code Review. Let's download and review this week's code.
Download the repository in the usual manner, using your name instead of 22xyz3:
git clone https://evolene.cs.williams.edu/cs134-labs/22xyz3/lab08.git ~/cs134/lab08
We'll focus, mainly, on one  le in this repository: cipher.py. Browse through this code as we discuss it
below.
In this module we have de ned a new class, Cipher. This class is not meant to be directly used. Instead,
it describes the interface common to all the cipher mechanisms we'll implement. Again: the interface is
simply the methods that we promise our users will be available. This includes, among others, the methods
we expect them to use (like encode(sourceString) and decode(sourceString)). The Cipher class is
also a convenient place for us to  nd public and private methods we would like to write once and use in
many subclasses.
Here are the highlights of the Cipher class:
 An empty slots variable. Users will not able to hang unwanted attributes o  Cipher or any of
its subclasses.
 A simple init (self) method. This basic class does not have any state, so its initializer does
nothing. Later, if we want to share implementations of state common to all our ciphers, we can add
them as state variables (in slots ) and initialize them here.
 The encode(self,sourceString) method. Ideally, this method will accept a plaintext string
(sourceString) and return its encoded equivalent. The implementation here simply returns the
string, unmodi ed.
 The decode(self,sourceString) method. Similarly, this method would normally reverse the encoding, returning the plaintext. Again, this implementation does nothing.
 The a2i(self,c) method. This converts an alphabet letter c to its alphabet code (in range(26)).
For example, calling self. a2i(’d’) (or self. a2i(’D’)) within a cipher returns the integer 3.
 The methods i2a(self,i) (and i2A(self,i)) reverse this process, converting alphabet codes to
lowercase (and uppercase) letters.
 The rotate(self,c,n) \rotates" the letter c forward through (and around) the alphabet by n
letters. For example, self. rotate(’I’,-1) returns 'H'. It does not modify non-letters.
 The normalize(self,sourceString) method. This method extracts the alphabetic characters
from sourceString and converts them to uppercase. Many of our ciphers do not extend to nonletters, so spacing and punctuation make the cipher more prone to attack. We'll use normalize to
generate a \canonical form" for message text.
As we gain more experience, we may  nd there are other methods that could be shared among cipher
implementations; we'll place them here. We may also  nd improvements in the public interface. If that
happens, we'll enforce those changes here.
We've also included a CaesarCipher class here, as well. It's declared:
class CaesarCipher(Cipher):
...
This is how we indicate that CaesarCipher is an extension or a subtype of Cipher. Because of that
relationship, all the features of Cipher are inherited by CaesarCipher. If we wish, we can reimplement
or override implementations of methods that must be specialized for this particular Cipher.
Note these specializations:
 The CaesarCipher declares a state variable, self. n. This keeps track of the degree of forward
rotation used for encoding. (Decoding, obviously shifts backward by negating self. n.)
 This state variable is initialized in init (self,n), with an integer passed in the constructor.
Before the initializer does anything, it calls its super-class's initializer. This is accomplished with:
super().__init__()
This ensures that any part of the state that Cipher is responsible for is initialized before we initialize
any state in the subclass. In most cases, we don't have to specify an initializer; the default is to
inherit and call the Cipher initializer. When we extend state, we must be sure to explicitly call the
superclass initializer.
 It provides encode(self,sourceString) and decode(self,sourceString) methods that apply the
same rotation to every element of the normalized plaintext. The removal of spaces and punctuation
greatly improves its security.
 There is a repr (self) method that describes how this CaesarCipher can be constructed.
We can exercise our CaesarCipher from within interactive Python as follows:
>>> from cipher import CaesarCipher
>>> coder = CaesarCipher(1)
>>> coder.encode('H.A.L.')
'IBM'
>>> code = open('story.txt').read()
>>> code[:26]
'GJBDGHPNHIWPILWTCEGTHXSTCI'
>>> coder2 = CaesarCipher(15)
>>> coder2.decode(code[:26])
'RUMORSAYSTHATWHENPRESIDENT'
Our work this week will be the development of new ciphers which extend Cipher and its subclasses.
Required Tasks. This week, we would like you to create and test two new cipher classes, for a full 10
points.
1. In cipher.py build a new class, Rot13(), that is a subclass of CaesarCipher(n).
(a) This class's parameterless initializer does one thing: it calls its superclass initializer with a  xed
value of n, 13.
(b) Since Rot13 is not meant to be secure, override the normalize(sourceString) method so that
it does not remove punctuation or change case. This allows us to preserve the punctuation in
cases where it's important:
>>> code = open('reform.txt').read()[:46]
>>> coder = Rot13()
>>> plain = coder.decode(code)
'A Plan for the Improvement of English Spelling'
>>> coder.encode(plain)
'N Cyna sbe gur Vzcebirzrag bs Ratyvfu Fcryyvat'
(c) Demonstrate the functionality of Rot13 by writing some new simple doc-tests on its initializer.
(d) Write an appropriate repr (self) method.
(e) Notice how you inherited encode and decode. You need not write these methods because they
were already written in CaesarCipher.
2. In cipher.py, build a new subclass of Cipher, Vigenere.
(a) This cipher's initializer takes a string, key, normalized to all uppercase letters. This key is saved
in a state variable and serves to determine the shift amount during encoding and decoding.
(b) Provide an encode(self,sourceString) method. Normalize sourceString. At position i
of the normalized plaintext, rotate the letter using letter i of the key (repeating the key, as
necessary): the alphabet code associated with the key letter is interpreted as the amount to
rotate the plaintext letter. Using 'WILLIAMS' as the key, the  rst letter of 'PURPLECOWS' is
shifted 22 (the alphabet code of W) spaces in the alphabet to L. Any A in the key causes the
corresponding letter to not shift (i.e., stay the same):
plaintext P U R P L E C O W S
key W I L L I A M S W I
shift 22 8 11 11 8 0 12 18 22 8
encoded L C C A T E O G S A
(c) Override the decode(self,sourceString) method to decode your cipher.
(d) Override the repr (self) method.
(e) Thoroughly test your class. At the very least, it should act as follows:
>>> vCoder = Vigenere('Williams')
>>> vCoder.encode('purple cows')
'LCCATEOGSA'
>>> vCoder.decode('lccateogsa')
'PURPLECOWS'
>>> code = open('cia.txt').read()[:51]
>>> coder = Vigenere('SANBORN')
>>> code
'KAACCIAKKEZDKBKIFBGTHDPGVFVYGCNUSUBFTUFUIBMNQTCWPAA'
>>> coder.decode(code)
'SANBORNSKRYPTOSISASCULPTURELOCATEDONTHEGROUNDSOFCIA'
>>> coder
Vigenere('SANBORN')
3. Review and document your code, adding appropriate symbols to the all method.
4. Sign the honor code. Add, commit, and push your changes to cipher.py.
?
The following is a fun little project with another coded text to decode. This part of the lab is for
independent investigation and will not be graded.
A really tough cipher. The Enigma Machine1 was a physical coding machine that generated very
complex substitution ciphers. The operators would set a set of three rotors to their \zero" position
(indicated by a 3-letter key-of-the-day). They would then press a letter on the keyboard. The signal
would get substituted or scrambled as they went through the rotors. Eventually, the signal reached a
display where the encoded character was read. The internal racheting mechanism would then spin the
rotors one notch, e ectively changing the encoding of the next character pressed. It was a devastating
device.
In the scrambler module, we've built a scrambler(c,spin,key) function that simulates the action of
the Enigma's rotors on character c, after spin characters from the beginning of the message to be encoded
with key. You can play with the scrambler by hand:
>>> scrambler('A',0,start='EPH')
'H'
What is amazing is the Enigma is a mirror encoder . The same device, set up in the same state, fed the
encoded text will regenerate the plaintext!
>>> scrambler('H',0,start='EPH')
'A'
To build your own Enigma-like machine, here are the steps:
1. Declare a new subclass of Cipher called Enigma.
2. You'll need two state variables that keep track of the key-of-the-day, and the current value of spin.
Keeping the value of spin as state outside of the encode method will allow you to encode (or decode)
a long message using several calls to encode. You may  nd it useful to have a reset(self) method
that allows you to reset spin before each batch of encoding or decoding.
3. The encode method (decode is identical) takes a normalized string. It then calls the scrambler
with each character of the plaintext, spin, and the key. After each operation spin is incremented.
Here is a basic test of functionality:
>>> coder = Enigma('TUR')
>>> code = open('turing.txt').read()[:35]
>>> code
'EFBFKPRLJNLFGYFEPMQZXPFMWSKKLROTIHG'
>>> coder.decode(code)
'TURINGWASTHEFATHEROFCOMPUTERSCIENCE'
Can you translate the rest of the message?
Congratulations!
1
https://bletchleypark.org.uk/our-story/the-challenge/enigma