tree-sitter-cfml

Tree-sitter grammars for ColdFusion Markup Language (CFML).

There are three grammars: two for CFML in .cfc/.cfm and .cfs files, and one for SQL inside <cfquery> (embedded dialect).

Grammar	Scope	File types	Description
`cfml`	`source.cfml`	`.cfc`, `.cfm`	ColdFusion components and template files - CFScript, tag-based components, and HTML with embedded CF tags
`cfscript`	`source.cfscript`	`.cfs`	Pure CFScript files
`cfquery`	`source.cfquery`	(embedded)	SQL inside `<cfquery>` bodies (including QueryExecute-style usage), with `#hash#` interpolation and CF tags in the body

Playground

Browser demo: cfmleditor.github.io/tree-sitter-cfml

Installation

Node.js

npm install @cfmleditor/tree-sitter-cfml

const {
  cfml,
  cfscript,
  cfquery,
} = require("@cfmleditor/tree-sitter-cfml");
const Parser = require("tree-sitter");

const parser = new Parser();
parser.setLanguage(cfml.language);

const tree = parser.parse("<cfif condition>#value#</cfif>");
console.log(tree.rootNode.toString());

Rust

[dependencies]
tree-sitter = "0.25"
tree-sitter-cfml = "0.26.34"

The tree-sitter crate should be 0.25+ so the ABI matches the generated parsers (see LANGUAGE_VERSION in cf*/src/parser.c).

use tree_sitter_cfml::LANGUAGE_CFML;

let mut parser = tree_sitter::Parser::new();
parser.set_language(&LANGUAGE_CFML.into())
    .expect("Error loading CFML grammar");
// LANGUAGE_CFSCRIPT, LANGUAGE_CFQUERY load the same way.

Python

pip install tree-sitter-cfml

import tree_sitter_cfml as ts_cfml
from tree_sitter import Language, Parser

# cfml for .cfc and .cfm files
parser = Parser(Language(ts_cfml.language_cfml()))
tree = parser.parse(b'<cfif x GT 0>#x#</cfif>')

# cfscript for .cfs pure script files
parser = Parser(Language(ts_cfml.language_cfscript()))

# cfquery SQL dialect (embedded)
parser = Parser(Language(ts_cfml.language_cfquery()))

Go

import (
    tree_sitter_cfml "github.com/cfmleditor/tree-sitter-cfml/bindings/go"
    sitter "github.com/tree-sitter/go-tree-sitter"
)

// cfml for .cfc and .cfm files
parser := sitter.NewParser()
parser.SetLanguage(sitter.NewLanguage(tree_sitter_cfml.LanguageCfml()))

// cfscript for .cfs pure script files
parser.SetLanguage(sitter.NewLanguage(tree_sitter_cfml.LanguageCfscript()))

// cfquery SQL dialect (embedded)
parser.SetLanguage(sitter.NewLanguage(tree_sitter_cfml.LanguageCfquery()))

Java

Needs JDK 23+ — the binding is built on the Foreign Function & Memory API, through jtreesitter.

<dependency>
  <groupId>io.github.cfmleditor</groupId>
  <artifactId>tree-sitter-cfml</artifactId>
  <version>0.26.31</version>
</dependency>

import io.github.cfmleditor.jtreesitter.cfml.TreeSitterCfml;
import io.github.cfmleditor.jtreesitter.cfscript.TreeSitterCfscript;
import io.github.cfmleditor.jtreesitter.cfquery.TreeSitterCfquery;
import io.github.treesitter.jtreesitter.Language;
import io.github.treesitter.jtreesitter.Parser;

// cfml for .cfc and .cfm files
try (var parser = new Parser(new Language(TreeSitterCfml.language()))) {
    var tree = parser.parse("<cfif x GT 0>#x#</cfif>").orElseThrow();
}

// cfscript for .cfs pure script files
var cfscript = new Language(TreeSitterCfscript.language());

// cfquery SQL dialect (embedded)
var cfquery = new Language(TreeSitterCfquery.language());

Unlike the other bindings, this one does not compile the C for you. It loads four shared libraries at runtime — libtree-sitter plus one per grammar — so they have to be somewhere the loader looks (LD_LIBRARY_PATH, java.library.path, or a system library directory). Either install them:

make && sudo make install    # libtree-sitter-{cfml,cfscript,cfquery}

or, from a checkout, build all four (including the tree-sitter runtime, pinned by package-lock.json) into build/native/:

npm install && npm run build:native
java --enable-native-access=ALL-UNNAMED -Djava.library.path=build/native …

Run the binding's own tests with mvn test — it points java.library.path at both locations.

The artifact is not on Maven Central yet; see the note on publish-maven in .github/workflows/release.yml.

Development

Each dialect has grammar.js, generated C under src/, corpus tests under test/corpus/, and queries under queries/. Shared scanner code is under common/. The multi-grammar CLI and playground config is tree-sitter.json. Upstream docs: Creating parsers, CLI.

Setup

git clone https://github.com/cfmleditor/tree-sitter-cfml.git
cd tree-sitter-cfml

Use Node >=18 and <24 (package.json engines). Optional: .nvmrc with nvm / fnm (nvm use).

npm install

That installs dependencies, builds the Node native addon (node-gyp-build), and runs postinstall, which downloads the tree-sitter CLI binary into node_modules/tree-sitter-cli/ when needed. Repo npm scripts do not require a global tree-sitter on PATH.

npm test          # all three grammars
npm run lint      # ESLint
npm run build     # regenerate parsers + rebuild native addon (after grammar edits)

Windows

Put GCC from MinGW-w64 on your PATH (gcc / g++). This repo uses GCC for npm test (tree-sitter compile), the Node native binding, Python extension builds, and Go CGO — not MSVC.

Standalone toolchain (recommended): install WinLibs with winget, then add the extracted mingw64\bin directory (contains gcc.exe) to your user PATH, open a new terminal, and verify:

gcc --version

Example package (UCRT, POSIX threads):

winget install BrechtSanders.WinLibs.POSIX.UCRT

The installer path varies by machine; locate mingw64\bin under the WinLibs folder (or under %LOCALAPPDATA%\Microsoft\WinGet\Packages\ after install) and add that bin to PATH.

Alternatively: MSYS2 with pacman -S mingw-w64-x86_64-gcc, then prepend msys64\mingw64\bin to PATH (or develop from an MSYS2 MinGW64 shell).

If node-gyp still picks Visual Studio instead of MinGW, set CC / CXX to your MinGW gcc / g++ for npm install / npm rebuild, or keep MinGW’s bin before MSVC entries on PATH.

macOS

Install the Xcode command-line tools:

xcode-select --install

Homebrew

brew install gcc

Linux

Install a C/C++ toolchain (for example build-essential on Debian/Ubuntu, gcc / clang plus development headers on other distributions).

CI

CI (.github/workflows/ci.yml): npm install, npm test, npm run lint on Ubuntu, macOS, and Windows. It does not run npm run build; generated cf*/src/ files are committed.

Tree-sitter CLI

Scripts use scripts/tree-sitter-cli.cjs (node node_modules/tree-sitter-cli/cli.js). A global tree-sitter-cli install is optional. If the binary is missing after install:

node scripts/ensure-tree-sitter-cli-binary.js

From a dialect directory (after npm install at repo root):

cd cfml
node ../node_modules/tree-sitter-cli/cli.js test
node ../node_modules/tree-sitter-cli/cli.js generate
node ../node_modules/tree-sitter-cli/cli.js parse path/to/file.cfc

Dependency versions

Pinned in package.json / tree-sitter.json; approximate roles:

Role	Package	Version
Native binding (peer / dev)	`tree-sitter`	`0.25.0`
Parser CLI	`tree-sitter-cli`	`0.26.8`
Native addon	`node-addon-api`	`^8.3.0`
Native addon	`node-gyp-build`	`^4.8.4`
Prebuild	`prebuildify`	`^6.0.1`
Runtime	Node.js	`>=18` `<24`
Java binding (`pom.xml`)	`jtreesitter`	`0.26.1`
Java binding	JDK	`>=23`

CFML engines

Corpus and behavior are checked mainly against Lucee. Overlapping Adobe ColdFusion syntax should still parse in a reasonable way. Avoid Adobe-only or Lucee-only assumptions in examples or grammar design where portable CFML is enough.

Building

After changing common/define-grammar.js or a grammar.js:

npm run build

On Unix, make generate at the repo root works if tree-sitter is on your PATH; otherwise use npm run build.

tree-sitter generate may warn about “unnecessary conflicts” (expressions vs _property_name, cfscript declaration / primary_expression, cfquery hash rules, etc.). Those come from common/define-grammar.js. If npm run build and npm test succeed, the warnings can be ignored.

Testing and helpers

See Setup for npm test, npm run lint, and npm run build.

npm run testbindings  # Node binding smoke test
npm run probe         # real-world construct probes (test/probes/)
npm run build:native && mvn test  # Java binding smoke test (JDK 23+)

One grammar only: run test via the CLI from that dialect’s directory (above), or npm test for all three.

Real-world corpus: npm run corpus:fetch shallow-clones ~25 public CFML projects into a gitignored corpus/, npm run scan corpus reports every ERROR/MISSING node, and npm run corpus:report clusters those into distinct failure sites. See CORPUS.md for the current results and the known gaps they turned up.

Parse a file: from the dialect folder (e.g. cfml for .cfc), use the parse subcommand with the same node ../node_modules/.../cli.js pattern.

Playground / WebAssembly (WASM)

npm start — playground at repo root (tree-sitter.json). Run npm run prestart first if WASM is stale.
npm run prestart — tree-sitter build --wasm
npm run playground — playground in each of cfml/, cfscript/, cfquery/
npm run docswasm — writes docs/tree-sitter-{cfml,cfscript,cfquery}.wasm for docs/ (e.g. GitHub Pages)

Releasing

npm run release -- 0.26.18
npm run release -- 0.26.18 --user=ghedwards  # optional: switch gh auth before push

The release script (scripts/release.js) will:

Validate version format and ensure it's greater than current
Ensure the working tree is clean and local branch is not behind remote
Verify tag v<version> doesn't already exist
Verify CHANGELOG.md has a ## [<version>] or ## [Unreleased] entry with notes
Update the version in package.json, Cargo.toml, pyproject.toml, tree-sitter.json and pom.xml
Run npm run build (regenerate parsers)
Run npm run lint (ESLint)
Run npm test (all three grammars)
Run npm run docswasm (rebuild playground WASM)
Run npm run install (rebuild native addon)
Commit all changes and create a v<version> tag (prompted)
Push commit and tag (prompted)

Once the tag is pushed, the GitHub Release workflow (.github/workflows/release.yml) will automatically publish to npm, PyPI, crates.io, and create a GitHub Release with the changelog notes. Maven Central is wired up but disabled — see publish-maven in that workflow.

Grammar structure

Shared rules: common/define-grammar.js. External scanner: common/scanner.h (implicit end tags, CF tag names, hash expressions, raw text).

common/
  define-grammar.js
  scanner.h
  tag.h

cfml/          # .cfc, .cfm
  grammar.js
  src/         # generated
  queries/

cfscript/      # .cfs
  grammar.js
  src/
  queries/

cfquery/       # embedded SQL
  grammar.js
  src/
  queries/

Queries

Grammar	Highlights	Indents	Injections	Tags
`cfml`	yes	yes	yes	yes
`cfscript`	yes	no	no	yes
`cfquery`	yes	no	no	yes

Contributing

See CONTRIBUTING.md.

Security

See SECURITY.md.

Agent and AI assistant guidance

See AGENTS.md.

License

MIT

tree-sitter-cfml 0.26.34