Enigma Machine in Python

A didactic Python implementation of the Enigma machine for use in cryptography courses. The codebase is intentionally kept small and readable: every component maps directly to a physical part of the historical machine, and every design decision is explained in the source comments.

This repository is structured as a self-contained learning environment:

• A faithful, well-documented implementation of the Wehrmacht/Luftwaffe Enigma I
• Runnable examples showing encryption, decryption, and key configuration
• A ready-to-use cryptanalysis challenge for student assessment

Repository layout

enigma-classroom/
├── enigma.py # Core machine implementation
├── enigma_solver.py # Crib attack on Enigma
├── examples/
│ └── demo.py # Usage examples (encrypt, decrypt, custom keys)
├── docs/
│ └── enigma_architecture.md # Deep-dive on machine internals
│ └── sample_challenge.md # Student cryptanalysis challenge (coursework)
├── .gitignore
├── LICENSE # Apache License 2.0
└── README.md

Quick start

No dependencies beyond the Python standard library are required.

git clone https://github.com/angelborroy/enigma-classroom.git
cd enigma-classroom
python enigma.py # runs the built-in demo
python examples/demo.py # extended usage examples

Machine architecture

Keyboard → Plugboard → Rotor R → Rotor M → Rotor L → Reflector
↓
Lamp ← Plugboard ← Rotor R ← Rotor M ← Rotor L ←──────┘

The signal travels right-to-left through three rotors, bounces off the reflector, and returns left-to-right through the same rotors. This symmetric design means that encryption and decryption are the same operation, provided the machine is reset to the same starting position.

Key components

Component	Class	Role
Plugboard (Steckerbrett)	Plugboard	Swaps up to 13 letter pairs before and after the rotors
Rotor (Walze)	Rotor	Stepping substitution cipher; advances on every keypress
Reflector (Umkehrwalze)	Reflector	Fixed self-reciprocal wiring; makes the circuit symmetric
Machine	EnigmaMachine	Orchestrates stepping (incl. double-step) and signal routing

Key space

Parameter	Options	Combinations
Rotor selection & order (choose 3 from 5)	—	60
Starting positions	A–Z per rotor	$$26^3 = 17 576$$
Ring settings	A–Z per rotor	$$26^3 = 17 576$$
Plugboard (10 pairs)	—	$$≈ 1.5 × 10^{14}$$
Reflector choice	B or C	2
Total		$$≈ 10^{18}$$

Usage

Basic encryption / decryption

fromenigmaimportEnigmaMachineenigma=EnigmaMachine(
rotor_names= ["I", "II", "III"], # left -> rightpositions= ['A', 'B', 'C'],
ring_settings= ['A', 'A', 'A'],
plugboard= [('A', 'Z'), ('B', 'Y')],
reflector="B",
)
ciphertext=enigma.encrypt("HELLO WORLD")
# Reset to the same starting position and decryptenigma.reset(['A', 'B', 'C'])
plaintext=enigma.encrypt(ciphertext)

Available rotors and reflectors

Rotors: I, II, III, IV, V
Reflectors: B, C

All wiring data is taken from the historical Wehrmacht/Luftwaffe Enigma I specification.

Cryptanalysis workshop — breaking the challenge

enigma_solver.py is a guided starting point for solving the cryptanalysis challenge in docs/sample_challenge.md. It implements the three-phase crib attack that Bletchley Park pioneered, using enigma.py from this repository.

How the attack works

A crib is a fragment of known (or strongly guessed) plaintext. Because the first word of the intercepted message is WEATHER, each ciphertext letter at that position satisfies the equation

stecker[ c_i ] = C_i( stecker[ p_i ] )

where C_i is the rotor core at offset i and stecker is the plugboard. Seven such equations over-constrain the plugboard enough that almost every offset triple can be discarded in microseconds via backtracking.

Phase	What it does	Key technique
0	Pre-compute all 17,576 rotor cores (no plugboard, ring = AAA)	Indexed lookup table
1	For each offset triple, solve the 7 crib equations	Constraint backtracking
2	Decrypt the message prefix; keep candidates that look like English	Dictionary filter
3	Complete the plugboard; rank survivors	N-gram scoring

Getting started

No installation required. Copy enigma_solver.py into the repository root alongside enigma.py and run it:

git clone https://github.com/angelborroy/enigma-python.git
cd enigma-python
python enigma_solver.py

The script will immediately pre-compute the core table (Phase 0) and then stop with a NotImplementedError at the first function you need to write.

What you need to implement

The skeleton runs out of the box and fails at the first incomplete function. Work through the TODO markers in order:

#	Function	Difficulty	Concept practised
1	assign(a, b) inside solve_plugboard	★☆☆	Plugboard as involution
2	looks_like_english()	★☆☆	Heuristic language detection
3	build_core_table() with real ring settings	★★☆	Effective offset = position − ring
4	Correct rotor stepping inside phase1()	★★☆	Double-stepping anomaly
5	complete_plugboard() with hill-climbing	★★★	Local-search optimisation
6	ngram_score() with real log-probabilities	★★★	Statistical language modelling

Tip — start small. Implement assign() first (10 lines). Once Phase 1 produces candidates, move to Phase 2, and so on. Each phase is independently testable.

Verifying your solution

When the correct key is recovered, re-encrypting the recovered plaintext must reproduce the original ciphertext exactly:

fromenigmaimportEnigmaMachinemachine=EnigmaMachine(
rotor_names= ["I", "IV", "III"],
positions= [...], # recovered positionsring_settings= [...], # recovered ring settingsplugboard= [...], # recovered plugboard pairsreflector="B",
)
plaintext=machine.encrypt(CIPHERTEXT)
machine.reset([...]) # same positionsassertmachine.encrypt(plaintext) ==CIPHERTEXTprint(plaintext) # WEATHER ...