Kotlin Grammar for Tree-sitter

Gitar’s Kotlin grammar for Tree-sitter began as a fork of tree-sitter-kotlin. Our goal is to provide a Kotlin grammar tailored for program analysis, not just syntax highlighting. This requires correctness and adherence to the Kotlin compiler. Over time, our grammar has diverged so much from the fork that it effectively stands as a new grammar.

• The original repo is based on the Kotlin specification grammar, which only loosely reflects the compiler’s actual grammar. Our grammar more closely matches the compiler.
• We handle newline characters directly in the grammar rules to take advantage of full parser context. See our blog post for details on newline handling in Kotlin.
• Some lexer hacks are still needed for newline handling — for example, distinguishing newlines after binary operators (to avoid conflicts with operator precedence) and ignoring newlines inside brackets.
• For program analysis, whitespace and comments often need to be properly attached to nodes. To support this, we’ve split some rules (such as class_declaration) so that whitespace and comments become part of the intended node.

Setup

To run the grammar, install the tree-sitter-cli. From the project root:

• tree-sitter generate → Generates the parser.c file. (Note: the file is currently quite large — a known Tree-sitter issue. We’re working on ways to reduce the size.)
• tree-sitter test → Runs all tests in the test/corpus directory.
• tree-sitter build --wasm → Builds the WebAssembly (wasm) file, which you can use to run the local playground with tree-sitter play.