GitXplorerGitXplorer
J

lib2toast

public
2 stars
1 forks
4 issues

Commits

List of commits on branch main.
Verified
3278eccc4205b6fb21d579cfae1152401bf91955

Fix crash on nested async comprehension (#16)

JJelleZijlstra committed 5 months ago
Verified
081b6a4df6447eaf82179653747bb7be4e4bc353

Fix line range with semicolons (#15)

JJelleZijlstra committed 5 months ago
Verified
ce0baca8a8579d412ccbc329ec54f3fc30699445

Fix crash on semicolon at end of line (#14)

JJelleZijlstra committed 6 months ago
Verified
06b62cb401c730598eee518fe174dd6d194d656a

Fix crash on some triple-quoted f-strings (#13)

JJelleZijlstra committed 6 months ago
Verified
5d72a7c65e792bbe66d48b2cf3814a112bd3f277

Fix a(b).c = d (#12)

JJelleZijlstra committed 6 months ago
Verified
91be95bb45bdf8f51a5719db6d84ae41601f8e01

Make it easier to add trailers (#11)

JJelleZijlstra committed 6 months ago

README

The README file for this repository.

🍞 lib2toast

This library converts a lib2to3 concrete syntax tree (CST) to a standard Python AST.

Potential use cases include:

  • Parsing Python code with less dependence on the Python version
  • Extending or modifying the Python grammar in order to experiment with new features or to create a Python-like dialect

Usage

This library is still at an early stage and the API may change.

  • lib2toast.api.compile(code, *, grammar=..., compiler=...): Compile a string of code to an AST. This AST can then be compiled to a Python code object or executed with the built-in compile() and exec() functions. By default, this uses a grammar that covers all syntax that is accepted by the latest version of Python, plus some additional syntax. Pass a custom grammar to use different syntax. You can use lib2toast.api.load_grammar to load a grammar object from a file. If you do this, you'll usually also want to pass a custom compiler object by subclassing lib2toast.compile.Compiler.
  • lib2toast.api.run(code, *, filename=..., grammar=..., compiler=...): Compiles code and then immediately executes it.
  • lib2toast.api.load_grammar(path, *, async_keywords=True): Load a grammar file from a path. If async_keywords is True, treats async as a keyword as in Python 3.7+.

There is also a command-line interface: python -m lib2toast -c code runs code after parsing it using lib2toast.

Showcase

The command-line interface shows that lib2toast supports parsing some new syntax in older Python versions:

$ python3.9 -m lib2toast -c 'print(f"{"x"}")'
x

This is new syntax introduced in Python 3.12 by PEP 701.

It also supports some (not all) Python 2 syntax that was removed in Python 3:

$ python3.9 -m lib2toast -c 'print(1 <> 2)'
True

The test suite shows some examples of syntactic variants of Python parsed with lib2toast. For example:

dataclass(frozen=True) C:
    x: int
    y: int = 0

Implementation

lib2toast is implemented on top of blib2to3, the fork of lib2to3 maintained by the Black project in order to parse and format Python code. It originates from lib2to3, a tool shipped with earlier Python 3 versions to support converting between Python 2 and 3 code.

The core part of the implementation is a tool that converts Python code to an AST. This makes it easy to test for correctness: just run Python's built-in ast.parse and assert that it produces the same tree, including line and column numbers. So far I have tested the compiler on lib2toast's own code as well as some of Black's code (the Black test cases were especially helpful), but there are probably more bugs.

Python version support

This library supports Python versions 3.9 and up.

Python 3.8 is unsupported because it is about to reach the end of its support period, the AST structure is quite different between 3.8 and 3.9, and I don't have a use case for 3.8.

In the future I plan to support all supported upstream versions of Python.

Contributing

Contributions to this project are welcome, including ideas for new ways to use the core functionality of the library.

Check the "Issues" tab for potential areas to contribute.

Release notes

Version 0.1.0 (August 6, 2024)

  • Fix miscompilation of Unicode identifiers; they are now NFKC-normalized
  • Fix crash on augmented assignment with an unparenthesized tuple on the right-hand side
  • Fix crash on nested async comprehensions
  • Fix incorrect line ranges in suites containing statements ending in a semicolon
  • Fix crash on statement ending in a semicolon
  • Fix crash on certain triple-quoted f-strings
  • Fix miscompilation of calls with arguments on the left-hand side of an assignment
  • Make it easier to create a dialect with more kinds of trailers
  • Adjust some type annotations to make subclassing easier

Version 0.0.1 (July 1, 2024)

Initial release.