ocaml-tree-sitter-core

Generate OCaml parsers based on tree-sitter grammars. This project was originally intended for semgrep. This repository contains the code for generating code for tree-sitter grammars but does not contain grammars for specific languages other than tests.

The ocaml-tree-sitter repositories are:

• ocaml-tree-sitter-core: this repo; provides the code generator that takes a tree-sitter grammar and produces an OCaml library from it.
• ocaml-tree-sitter-languages: community repository that has scripts for building and publishing OCaml libraries for parsing a variety of programming languages.
• ocaml-tree-sitter-semgrep: same as ocaml-tree-sitter-languages but extends each language with constructs specific to semgrep patterns.

Installation

Installing ocaml-tree-sitter alone isn't of great use at the moment. It is useful as part of the flow for generating OCaml parsers from tree-sitter grammars, which is done from the lang/ folder in ocaml-tree-sitter-semgrep. See the Contributing instructions below.

Contributing

Before you get started, make sure you're ok with signing the CLA which will be needed before we accept your pull request. The goal is only to allow future relicensing without having to track down any past contributor, if such need were to arise. Note that the current license is LGPL and any contribution made today will remain available under that license no matter what.

Development setup

1. Install opam.
2. Install ocaml dev tools for your favorite editor: typically opam install merlin + some plugin for your editor.
3. Install pre-commit with pip3 install pre-commit and run pre-commit install to set up the pre-commit hook. This will re-indent code in a consistent fashion each time you call git commit.
4. Check out the extra instructions for MacOS.

For building or rebuilding everything after big changes:

# Install OCaml dependencies (if not already installed)
$ opam install --deps-only .

# Build and test (automatically sets up tree-sitter on first build)
$ dune build
$ dune runtest

tree-sitter version

The default tree-sitter version to use is in the tree-sitter-version file.

Under the default configuration used for local development purposes, the version being actually used is stored in the file tree-sitter-version. To use a different version, edit that file; dune will download and build the corresponding tree-sitter sources on the next build.

IMPORTANT Note that the selection in tree-sitter-version affects the tree-sitter runtime that the ocaml-tree-sitter code generator and its OCaml bindings are compiled against, as well as the CLI used to generate parsers.

It does not determine which tree-sitter version grammars are built with. In ocaml-tree-sitter-semgrep, every grammar is built by the tree-sitter version specified by the lang/languages-<version> lists. See the "tree-sitter versions (per language)" section of the ocaml-tree-sitter-semgrep README.

Because the C API surface we rely on is stable across these versions, a single build of the ocaml-tree-sitter generator works with grammars generated for any of the supported versions.

Documentation

We have limited documentation which is mostly targeted at early contributors. It's growing organically based on demand, so don't hesitate to file an issue explaining what you're trying to do.

License

ocaml-tree-sitter is free software with contributors from multiple organizations. The project is driven by Semgrep.

• OCaml code developed specifically for this project is distributed under the terms of the GNU LGPL 2.1.
• The OCaml bindings to tree-sitter's C API were created by Bryan Phelps as part of the reason-tree-sitter project.
• The tree-sitter grammars for major programming languages are external projects. Each comes with its own license.