gotreesitter

Pure-Go tree-sitter runtime. No CGo, no C toolchain. It cross-compiles to any GOOS/GOARCH target Go supports, including wasip1.

The widely used Go bindings (smacker/go-tree-sitter and tree-sitter/go-tree-sitter) wrap the C runtime through CGo. That needs a C cross-toolchain per target, breaks go install for downstream users without a C compiler, and hides bugs from go test -race. gotreesitter has no C dependency: the parser, lexer, query engine, incremental reparsing, arena allocator, external scanners, and tree cursor are all implemented in Go. The grammar blob is the only input.

Docs site: https://gotreesitter.m31labs.dev. Try it in the browser: https://gotreesitter.m31labs.dev/playground.

Install

go get github.com/odvcencio/gotreesitter

gotreesitter loads the same parse-table format that tree-sitter's C runtime uses. ts2go extracts grammar tables from upstream parser.c files, compresses them into binary blobs, and deserializes them on first use. 206 grammars ship in the registry.

The current release is v0.56.0. It includes YAML syntax-error fixes, recovery fixes, bounded scanner checkpoint memory, certified incremental reuse for large edits, and opt-in declaration, signature and call-argument facts for Go, Python, JavaScript, TypeScript and TSX. See CHANGELOG.md for the release scope for shipped history and docs/v1-design.md for the v1 plan.

Quick start

import (
    "fmt"

    "github.com/odvcencio/gotreesitter"
    "github.com/odvcencio/gotreesitter/grammars"
)

func main() {
    lang := grammars.GoLanguage()
    parser := gotreesitter.NewParser(lang)

    tree, _ := parser.Parse([]byte("package main\n\nfunc main() {}\n"))
    fmt.Println(tree.RootNode().SExpr(lang))
}

grammars.DetectLanguage("main.go") resolves a filename to the matching LangEntry.

To detect the language and parse in one call, use grammars.ParseFile. The caller must release the returned tree:

bt, err := grammars.ParseFile("main.go", src)
if err != nil {
    return err
}
defer bt.Release()

Each grammar is also a standalone package, so a binary that needs one language does not import the whole catalog:

import "github.com/odvcencio/gotreesitter/grammars/typescript"

lang := typescript.Language()

Features

• 206 grammars, all producing error-free trees on smoke samples; 119 ship a hand-written Go external scanner.
• Incremental reparsing that reuses unchanged tree content by reference, plus a no-edit fast path with zero allocations.
• Queries with the full S-expression pattern language and all standard tree-sitter predicates and directives.
• Typed query codegen (cmd/tsquery) that generates Go structs and match helpers from .scm files.
• Injection parsing for multi-language documents (HTML+JS+CSS, Markdown+code fences, Vue/Svelte).
• Highlighting, tagging, and file outlines built on the same tags-query captures.
• UTF-16 input and editor coordinates, so editor integrations do not hand-convert offsets.
• Source rewriting that produces InputEdit records ready for incremental reparse.
• WebAssembly/browser runtime with both a blob-loading target and an in-browser grammargen target. See the WebAssembly guide.
• Build-tag-selected grammar embedding (external blobs, a curated core set, or a hand-picked subset) for smaller binaries.

Agent skill

Agents working with gotreesitter should use the using-gotreesitter skill.

Documentation

Topic	Where
Parsing, queries, injections, incremental reparse, UTF-16, cursor, highlighting, tagging, outlines	docs/api-guide.md
Runtime architecture (parser, lexer, scanners, arena, query engine, grammar loading)	docs/architecture.md
Build tags and environment variables	docs/build-tags.md
Supported languages, query feature matrix, adding a language	docs/languages.md
Building, testing, and sending a pull request	CONTRIBUTING.md
Benchmarks and methodology	BENCH.md
v1 plan	docs/v1-design.md, docs/roadmap.md
Docs site and playground	https://gotreesitter.m31labs.dev, https://gotreesitter.m31labs.dev/playground
Root package file map and ownership	docs/repository-map.md
Root-package file group map by subsystem	docs/package-layout.md
Result-compatibility tier (parser_result_*.go)	docs/compat-tier.md
Adding a grammar outside this repo	docs/authoring-languages.md
External scanner certification and fallback	docs/external-scanners.md
Release process	docs/releasing.md
Full changelog	CHANGELOG.md, docs/changelog/

License

MIT. The default grammars import currently includes four copyleft grammars: caddy, disassembly and jq (GPL-3.0) and nim (MPL-2.0). Build with -tags gotreesitter_no_copyleft to leave them out. See docs/licensing.md.