Quick Start for Contributors

This guide provides instructions for contributing to OCP.

---

Contribution to Operators

1. Adding New Operators

• Define a new operator class in operators/operators.py.
• Implement the operator's functionality and modeling methods in alignment with existing operators.

2. Extending Existing Operators

• Add new Python and C implementations to optimize performance in generate_implementation().
• Add new modeling techniques (e.g., "MILP", "SAT") by specifying model_type in generate_model()
• Add new modeling versions (e.g., "diff", "truncated_diff", "linear") for various cryptanalysis techniques by specifying model_version in generate_model().

---

Contribution to Primitives

1. Adding New Primitives

• Create a new_primitive.py file in primitives/.
• Define a new operator class in new_primitive.py and implement its structure in alignment with existing primitives.

2. Extending Existing Primitives

• Add new representation modes by specifying represent_mode in the corresponding primitive class to enhance flexibility.

---

Contribution to Cryptanalysis

1. Adding New Cryptanalysis Techniques(e.g., Linear Cryptanalysis)

• Implement modeling methods for the new cryptanalysis technique in operators.py.
• Define the new attack method in attacks/attacks.py, for example, def linear_attacks(), following the structure of existing diff_attacks(). This includes:
  • Constructing MILP or SAT models for the attack.
  • Solving models using available solvers.
  • Obtaining and analyzing results.

---

Contribution to Automated Analysis Techniques

1. Adding Automated Analysis Techniques

• Extend model_type in operators.py to support additional techniques (e.g., CP).
• Integrate the new techniques into attacks/attacks.py, such as extending model_type in diff_attack().

2. Allowing flexible solver selection

• Extend support support for additional solvers and enable user-defined solver configurations in solving/solving.py.
• Currently, Gurobi is used for MILP models, and CryptoMinisat for SAT models.