maxmalysh / tiny-py-interpreter

Python interpreter written in Python. Uses ANTLR4. 🐍

TinyPy Interpreter

About

TinyPy is an interpreter of a small Python subset I have written as a coursework.

Installation

This project uses ANTLR4 as a parser generator. To run the interpreter, you will need to install ANTLR4 Python3 runtime and ANTLR itself.

Please note, that 4.5.2 runtime has a bug which results in a dramatic performance dropdown. At the moment this text was written, pypi had an older version, so it's recommended to install ANTLR4 runtime manually.

Step-by-step instruction:

1. Install ANTLR4

2. Install ANTLR4 Python3 runtime:

   1. git clone https://github.com/antlr/antlr4
   2. cd antlr4/runtime/Python3
   3. python3 setup.py install

   It's also possible to use pip, package name is antlr4-python3-runtime. Be aware of the bug described above.

3. Generate parser

   1. cd tiny-py-interpreter/tinypy
   2. antlr4 -visitor parser/TinyPy.g4

4. Install tinypy: pip3 install .

5. Try to launch some tests: python3 setup.py test

tiny-py-interpreter's Issues

Add installation scripts for Unix and Windows

Should use these scripts for travis and appveyor.

Fix break / continue statement parsing

Add exceptions (to the parser) for break / continue outside of loops and return outside of function definition. Grammar allows it, so we should check this manually. See how Python3 does this.

Fix error handling for parser in the file input mode.

Don't evaluate input if error recovery occurred. When reading from file, there should be no attempt to evaluate the parse tree, if minor errors were encountered.

Use proper unit testing framework

Currently, we have some tests for the parser (which have numeric names) and for the AST. This solution is far from perfect.

Add support for collections

Add lists, dictionaries, sets. For loops, "in" keyword.

Add support for Python 3.4

Add tests for runtime exceptions

Should check for runtime exceptions (e.g. ZeroDivisionError).

Don't print control flow mark while in interactive mode

Wrong precedence for is not / is operators

The following expression:
3 is not 4
is parsed as
3 is (not 4) because of the wrong precedence in parser grammar.

Fix unicode / Cyrillic support

File input currently doesn't allow to use unicode. Manually reading from the stdin in the shell mode breaks unicode support too.

Setup script

Should add setup script.