tree-sitter-iec61131-3-st

A tree-sitter grammar for IEC 61131-3 Structured Text (ST) — the standard programming language for industrial PLCs. Standard-compliant first; vendor dialects (Beckhoff TwinCAT, Codesys, B&R Automation Studio, Siemens TIA, Rockwell) are deferred to separate dialect grammars that extend this base.

About the name — IEC 61131 is the umbrella PLC-programming standard. Part 3 (IEC 61131-3) defines the programming languages: ST (Structured Text), LD (Ladder Diagram), FBD (Function Block Diagram), IL (Instruction List, deprecated), and SFC (Sequential Function Chart). This repo covers ST only — the -3-st suffix encodes both: Part 3 of the standard, ST language specifically.

Features

• IEC 61131-3 (3rd edition, 2013) Structured Text — POUs, all VAR blocks, every elementary / derived / generic type, every operator with correct precedence, every statement, full OOP (METHOD / PROPERTY / EXTENDS / IMPLEMENTS / INTERFACE), namespaces, configuration / resource / task.
• Case-insensitive keywords (IF, if, If all parse as the same keyword) implemented at the lexer level.
• Error-tolerant: produces a useful tree even on partial or broken input — usable in editors during typing.
• Dialect-extensible: the base grammar exposes named hidden rules (_declaration, _statement, _expression, _type_specifier, _var_block) so dialect grammars can add vendor-specific constructs via grammar(base, {…}) without forking. See EXTENDING.md.
• Editor-ready queries: highlights.scm, locals.scm, tags.scm, folds.scm, indents.scm, injections.scm. Standard tree-sitter capture vocabulary.
• Bindings for Node, Rust, Python, Go.

Quick demo

FUNCTION_BLOCK PID
VAR_INPUT
    setpoint, process_var : REAL;
    Kp, Ki, Kd            : REAL;
END_VAR
VAR_OUTPUT
    output : REAL;
END_VAR
VAR
    error, prev_error, integral : REAL;
END_VAR

error := setpoint - process_var;
integral := integral + error;
output := Kp * error + Ki * integral + Kd * (error - prev_error);
prev_error := error;
END_FUNCTION_BLOCK

Parsing this with tree-sitter parse produces a clean tree with function_block_declaration → var_input / var_output / var_block → assignments with binary_expression operands at the right precedence.

Install

Node

npm install tree-sitter tree-sitter-iec61131-3-st

Rust

# Cargo.toml
[dependencies]
tree-sitter = "0.25"
tree-sitter-iec61131-3-st = "0.0"

Python

pip install tree-sitter tree-sitter-iec61131-3-st

import tree_sitter, tree_sitter_iec61131_3_st
language = tree_sitter.Language(tree_sitter_iec61131_3_st.language())
parser = tree_sitter.Parser(language)
tree = parser.parse(b"PROGRAM Hello END_PROGRAM")

Go

import (
    sitter "github.com/tree-sitter/go-tree-sitter"
    iec61131_3_st "github.com/HeytalePazguato/tree-sitter-iec61131-3-st/bindings/go"
)

Editor setup

Neovim with nvim-treesitter

require('nvim-treesitter.configs').setup {
  ensure_installed = { 'iec61131_3_st' },   -- once published; pre-publish, install from local path
  highlight = { enable = true },
  indent    = { enable = true },
  fold      = { enable = true },
}

For a local development install before the parser is on the npm/CDN registry, add to your init.lua:

local parser_config = require'nvim-treesitter.parsers'.get_parser_configs()
parser_config.iec61131_3_st = {
  install_info = {
    url = 'https://github.com/HeytalePazguato/tree-sitter-iec61131-3-st',
    files = { 'src/parser.c' },
    branch = 'main',
  },
  filetype = 'st',
}

Helix

languages.toml:

[[language]]
name = "iec61131-3-st"
scope = "source.iec61131-3.st"
file-types = ["st", "iecst"]
roots = []
comment-token = "//"
indent = { tab-width = 4, unit = "    " }

[[grammar]]
name = "iec61131_3_st"
source = { git = "https://github.com/HeytalePazguato/tree-sitter-iec61131-3-st", rev = "main" }

Zed

Zed picks up tree-sitter grammars from extensions; see Zed's docs for the current recommended packaging path.

VSCode

VSCode does not natively use tree-sitter for grammar parsing — its highlighting comes from TextMate grammars and its semantic tokens come from language servers. A future companion repo will provide a VSCode extension that loads this grammar via the vscode-tree-sitter integration.

What's covered, what's not

Implemented in v0.0.1:

• POU declarations: PROGRAM, FUNCTION (with return type), FUNCTION_BLOCK, INTERFACE, TYPE, NAMESPACE, CONFIGURATION, RESOURCE.
• All VAR_* block kinds with CONSTANT / RETAIN / NON_RETAIN qualifiers, AT %{I,Q,M}{X,B,W,D,L} direct addresses, initial values.
• All elementary types, generic ANY_* types, STRING(N) / WSTRING(N), ARRAY [a..b, …] OF, structures, enumerations, subranges, POINTER TO, REF_TO.
• All literals: integers (plain, 2#…, 8#…, 16#…, with _ separators), reals with exponent, strings with $ escapes, T#…, D#…, TOD#…, DT#…, typed-prefixed (INT#42, REAL#3.14, etc).
• All statements: assignment (:=), reference assignment (REF=), function/method calls with positional and named (:= / =>) arguments, IF/ELSIF/ELSE/END_IF, CASE with single, list, range values + ELSE, FOR … TO … BY … DO … END_FOR, WHILE, REPEAT, EXIT, CONTINUE, RETURN.
• All operators with IEC 61131-3 §6.6.5 precedence: parentheses, calls, indexing […,…], member access, dereference ^, ADR(), unary - / + / NOT, right-associative **, * / / / MOD, + / -, comparisons, equality, AND / &, XOR, OR.
• OOP (3rd edition): METHOD / END_METHOD, PROPERTY with GET / SET accessor bodies, EXTENDS, IMPLEMENTS, INTERFACE, ABSTRACT / FINAL / OVERRIDE, PUBLIC / PRIVATE / PROTECTED / INTERNAL, THIS / SUPER.
• Comments ((* … *), //) and pragmas ({ … }, opaque body).

Out of scope for v0.0.1:

• Vendor dialect extensions (TwinCAT __VERSION, Codesys structured pragmas, B&R ACTION, etc.) — those will live in dialect repos that extend this grammar.
• Other IEC 61131-3 languages — Ladder Diagram, Function Block Diagram, Instruction List, Sequential Function Chart.
• Type checking, symbol resolution, code generation, formatting — this is a parser, not a compiler.

Performance

The CI benchmark parses a synthetic ~10 000-line ST file (200× the combined PID + conveyor + state-machine examples) and fails the build if the parse exceeds 200 ms on a ubuntu-latest runner. Typical run is well under that.

Development

# Install tree-sitter-cli and a C compiler.
npm install -g tree-sitter-cli

# Generate the parser (writes src/parser.c).
tree-sitter generate

# Run the corpus.
tree-sitter test

# Parse a single file.
tree-sitter parse examples/blink.st

See the project's BLUEPRINT.md for branch / release conventions: develop → release/<version> → main, semver from a single VERSION file.

Roadmap

Future repos that will extend this base grammar:

• tree-sitter-iec61131-3-st-twincat — Beckhoff TwinCAT 3 (TwinCAT-specific pragmas, S= / R= set/reset, OR_ELSE / AND_THEN short-circuit operators, conditional compilation, ACTION blocks).
• tree-sitter-iec61131-3-st-codesys — Codesys 3 (attribute pragmas with structured contents, action / transition blocks).
• tree-sitter-iec61131-3-st-br — B&R Automation Studio (ACTION, task-specific extensions).
• tree-sitter-iec61131-3-st-siemens — Siemens TIA Portal SCL.
• tree-sitter-iec61131-3-st-rockwell — Rockwell Studio 5000 ST.

Pull requests welcome — see CONTRIBUTING.md and EXTENDING.md.

Acknowledgments

Patterns, organization, and corpus references from prior work in the ecosystem (all MIT or otherwise permissively licensed):

• tmatijevich/tree-sitter-structured-text — B&R-leaning partial grammar; this project's literal regex shape and basic precedence layout drew from it.
• teunreyniers/tree-sitter-structured-text — generic ST grammar; informed the named-precedence-table approach and VAR-block factoring.
• klauer/blark — Lark-based TwinCAT parser; the most comprehensive open-source IEC 61131-3 grammar. Used as a reference for the rule organization, the OOP/extension surface, and which features are truly TwinCAT-only vs. standard.

License

MIT — see LICENSE.