The Leo Program Runner is a Python package that bridges the gap between Leo programs and the Python environment. It allows for seamless interaction with Leo programs and significantly simplifies the development process. By dynamically generating Python classes that serve as representatives of the loaded Leo programs, along with their associated structs and functions, it brings the power of Leo programs to the Python ecosystem.
The Leo Program Runner operates by parsing the source code of a Leo program and dynamically constructing a Python class based on its contents. The structs and records defined in the Leo program are represented as member variables of the dynamically-generated Python class, while the transition functions are transformed into Python methods.
In the import sections of Leo programs, the naming of structs and records will be slightly altered. For instance, board.leo/board_state.record in Leo will be translated into the board__board_state attribute of the auto-generated class.
The Leo Program Runner also offers automatic conversion between Python and Leo data types. When you call a method (representing a transition function) of the Python class, it takes care of converting the provided arguments into a format suitable for Leo programs, executes the computation, and then converts the results back into a Python-friendly format. This ensures a seamless development experience, making the interaction with a Leo program feel just like working with a Python package.
To obtain reproducible results with the Leo Program Runner, it is important to provide consistent input data and parameters. This means that you should:
- Use the same versions of Python and any dependencies. We recommend Python 3.8 or later.
- Use the same input data each time. You should avoid relying on data that may change over time, such as live web data or random numbers, unless you specifically want to test how the program handles changing data.
If you encounter any inconsistencies in your results, please first check these factors.
The Leo Program Runner has a significant impact on the Python ZkML tooling ecosystem. It provides a novel approach to interacting with Leo programs within the Python environment, which greatly simplifies the development process. The automatic conversion between Python and Leo data types makes the tool highly usable for machine learning developers and practitioners.
Furthermore, the Leo Program Runner is designed with extensibility in mind. While it currently supports importing Leo programs from a specific directory, it could be extended in the future to support importing from the web or other sources. As the Leo language continues to evolve, the Leo Program Runner will also adapt and update, with new releases tracking changes and enhancements in the Leo language.
You can easily install the package using pip:
pip install git+https://github.com/snowtigersoft/run_leo.gitThe Leo Program Runner supports the creation of Aleo accounts. You can create a new account by simply calling Account.new().
from run_leo import Account
acc = Account.new()
# Save to a file
acc.save('/path/to/file')
# Load from a file
acc2 = Account.load('/path/to/file')Here's an example of how you can use the Leo Program Runner:
from run_leo import import_leo_program, Account
# Import the Leo program from a specific directory
leo_program = import_leo_program('./leo_examples/tictactoe')
# set the account use for leo program
acc = Account.new()
leo_program.set_account(acc)
# Call the available functions in the Leo program
leo_program.call_functions()The import_leo_program() function requires the path to the Leo program's directory as its argument. The returned object will have methods and attributes that correspond to the structs and functions defined in the Leo program.
In the examples directory, there is an interactive program play_tic_tac_toe.py that lets you experience playing Tic Tac Toe. Players 1 and 2 take turns to place their markers, and the Leo program determines who wins.
You can also find more example Leo programs in the tests/leo_examples directory. These examples are taken from the AleoHQ/workshop repository. If you want to experience running a Leo program through Python, just execute the test_leo_program.py file.
Please note that this package assumes that there is a main.leo file in the src directory of the Leo program and that the path provided should point to the Leo program's root directory. If you encounter any issues while using this package, please ensure that your Leo program adheres to the expected directory structure and file naming conventions.