tree-sitter-powershell

A tree-sitter grammar for PowerShell, with Rust bindings and a highlighting query for .ps1, .psm1, and .psd1 files.

The grammar covers commands, expressions, pipelines, redirection, control flow, functions, classes, enums, types, and string interpolation. PowerShell 7 features include pipeline chains, ternary and null-coalescing operators, null-conditional access, generic method calls, and clean blocks. Tests use PowerShell 7.5.4 as the reference parser. Standard DSC resources and legacy workflow syntax are also represented. Parser conformance describes the test coverage, semantic boundaries, and known differences.

Rust usage

The Rust crate is woodruffw-trough-tree-sitter-powershell. It requires Rust 1.90 or later and a C compiler. Add it from crates.io:

[dependencies]
tree-sitter = "0.27.1"
woodruffw-trough-tree-sitter-powershell = "0.1.6"

let mut parser = tree_sitter::Parser::new();
parser
    .set_language(&woodruffw_trough_tree_sitter_powershell::LANGUAGE.into())
    .unwrap();
let tree = parser
    .parse("Get-ChildItem | Where-Object { $_.Length -gt 0 }", None)
    .unwrap();
assert!(!tree.root_node().has_error());

The crate exports LANGUAGE, NODE_TYPES, and HIGHLIGHTS_QUERY. Generated parser sources are included, so building the Rust crate does not require the tree-sitter CLI.

Syntax and trees

The tree preserves command arguments and parameters, expression grouping, member and index access, declarations, and interpolation. It supports error recovery and incremental edits for editor use. PowerShell handles name resolution and execution.

Microsoft's language specification describes PowerShell 3.0. Current syntax is checked against Microsoft's documentation, the PowerShell parser source, and Parser::ParseInput in the pinned runtime. See sources and tree design for references and examples.

Tests include explicit corpus trees, focused oracle cases, generated token matrices, and complete files from PowerShell and independent projects. The oracle parses their source without running the scripts; checks compare corresponding tree and AST structure as well as syntax acceptance.

Contributing

CONTRIBUTING.md covers setup, formatting, regeneration, and tests. Grammar generation uses tree-sitter's embedded QuickJS backend. Rust is the only binding provided by this repository.