dali-lang (Foundation v0)

dali-lang is the first foundation of the .dali/.dl audio DSL for web/Electron audio processing.

npm package

You can use dali-lang independently from the main app via npm:

npm install aurivo-dali-lang

CLI usage after install:

npx dali dali-lang/examples/web-bass-enhancer.dali /tmp/web-bass-enhancer.generated.js
npx dali dali-lang/examples/web-eq32-reference.dl /tmp/web-eq32-reference.generated.js --hardened

One-step editor setup (VS Code syntax extension):

npx dali setup

This installs aurivo.dali-language for VS Code and enables .dali/.dl highlighting.

Publish package:

cd dali-lang
npm login
npm publish --access public

Automated publish (GitHub Actions):

1. Add repository secret: NPM_TOKEN (npm automation token).
2. Bump dali-lang/package.json version.
3. Create tag in format dali-lang-v<version> and push:

git tag dali-lang-v0.1.4
git push origin dali-lang-v0.1.4

VS Code extension publish (GitHub Actions):

1. Add repository secret: VSCE_PAT (Visual Studio Marketplace personal access token).
2. Bump dali-lang/editors/vscode/package.json version.
3. Create tag in format dali-vscode-v<version> and push:

git tag dali-vscode-v0.1.4
git push origin dali-vscode-v0.1.4

Quality gates for portability + setup:

• .github/workflows/dali-multi-arch-ci.yml:
  • native smoke: ubuntu-24.04, macos-14, windows-2025
  • portable arch smoke: linux/amd64 + linux/arm64 (QEMU + Docker)
• .github/workflows/dali-setup-smoke.yml:
  • verifies dali setup --skip-editor-install across OS matrix
• scripts/dali-security-suite.js:
  • malicious corpus regression (unknown effect, unsafe capability, invalid unit, out-of-range latency, insecure task URL, hardened missing capability)

Recommended merge flow to main:

git checkout release_v2_0_12_clean
git pull --ff-only linux release_v2_0_12_clean
git checkout main
git pull --ff-only linux main
git merge --no-ff release_v2_0_12_clean -m "chore(dali): integrate security+ci hardening"
git push linux main

Capability policy file:

• dali-lang/spec/capability-policy.json
• Default model is deny-by-default capability enforcement.

Current scope:

• Parse .dali and .dl preset files.
• Compile one preset into a JavaScript module that builds a Web Audio graph.
• Support baseline effects: preamp, low_shelf, peaking, high_shelf, compressor, limiter.
• Security stage-1 active: strict whitelist validation for input/output targets, allowed effects, allowed effect params, and quality keys.
• Security stage-2 active: numeric range + unit validation for critical DSP params (gain/freq/q/attack/release/latency/sample_rate).
• Security stage-3 active: source size/line limits + complexity limits (preset/effect/param counts) for DoS resistance.
• Security stage-4 active: runtime sandbox guards in generated modules (AudioNode type checks, same-context isolation, safe connect/disconnect limits).
• Security stage-5 active: cryptographic signature infrastructure (ED25519 sign/verify + compile-time optional signature verification).
• Diagnostics improved: compiler/lint errors now include source-range code frames when line/column is available.

Quick start

Compile sample preset:

node dali-lang/src/cli.js dali-lang/examples/web-bass-enhancer.dali dali-lang/examples/web-bass-enhancer.generated.js

Compile 32-band web EQ reference preset:

node dali-lang/src/cli.js dali-lang/examples/web-eq32-reference.dl dali-lang/examples/web-eq32-reference.generated.js

Compile all .dali/.dl presets in one command:

npm run -s dali:compile

This scans dali-lang/examples recursively and regenerates all .generated.js files. Compiler now supports baseline .dl v2 blocks (engine, chain, quality) and eq32 band(...) = ... syntax. Strict mode is default in bulk compile. Use --no-strict only for temporary migration/debug.

Compile with AudioWorklet backend foundation:

npm run -s dali:compile:worklet
# or single file:
node dali-lang/src/cli.js dali-lang/examples/web-bass-enhancer.dl /tmp/web-bass-enhancer.worklet.generated.js --backend audioworklet

Compile with WASM skeleton target:

npm run -s dali:compile:wasm
# or single file:
node dali-lang/src/cli.js dali-lang/examples/web-bass-enhancer.dl /tmp/web-bass-enhancer.wasm.generated.js --target wasm

Worklet backend now runs baseline DSP chain inside the processor for:

• preamp, low_shelf, peaking, high_shelf, compressor, limiter

audioworklet backend output exposes an async builder (await preset.buildGraph(...)).

Lint preset files (validation + quality hints):

npm run -s dali:lint -- dali-lang/examples/web-bass-enhancer.dl dali-lang/examples/web-eq32-reference.dl

Task runtime dry-run (permissions + task planning, no external execution):

npm run -s dali:run -- dali-lang/examples/web-smart-task-reference.dl --dry-run
node dali-lang/src/cli.js run dali-lang/examples/web-smart-task-reference.dl --dry-run --json

Task runtime stub execution (sandboxed mock adapter, still no external system call):

node dali-lang/src/cli.js run dali-lang/examples/web-smart-task-reference.dl --no-dry-run --execute-stub
node dali-lang/src/cli.js run dali-lang/examples/web-smart-task-reference.dl --no-dry-run --execute-stub --json

Generate IR (with hash-based cache):

npm run -s dali:ir -- dali-lang/examples/web-smart-task-reference.dl
node dali-lang/src/cli.js ir dali-lang/examples/web-smart-task-reference.dl --json

Machine-readable JSON diagnostics:

node dali-lang/src/cli.js dali-lang/examples/web-bass-enhancer.dl /tmp/out.generated.js --json
node scripts/dali-lint.js --json dali-lang/examples/web-bass-enhancer.dl
node scripts/compile-dali-presets.js --json

LSP-compatible diagnostics payload:

npm run -s dali:lint:lsp -- dali-lang/examples/web-bass-enhancer.dl

VS Code tasks (Task: Run Task):

• DALI: Lint Current File
• DALI: Lint Examples
• DALI: Compile All
• DALI: Compile Worklet
• DALI: Compile WASM
• DALI: Security Tests
• DALI: Run Dry-Run
• DALI: Run Stub Exec
• DALI: IR Current File

These tasks are defined in .vscode/tasks.json and include problem matchers for DALI lint/compile errors.

Sign a preset file:

npm run -s dali:sign -- dali-lang/examples/web-bass-enhancer.dl --private-key dali-lang/keys/dali-ed25519.private.pem --key-id dali-main

Verify a preset signature:

npm run -s dali:verify-signature -- dali-lang/examples/web-bass-enhancer.dl --public-key dali-lang/keys/dali-ed25519.public.pem

Compile with signature verification enabled:

npm run -s dali:compile -- --verify-signatures --public-key dali-lang/keys/dali-ed25519.public.pem

Require every preset to be signed:

npm run -s dali:compile -- --require-signatures --public-key dali-lang/keys/dali-ed25519.public.pem

Run automated security tests (malformed corpus + fuzz + signature tamper):

npm run -s dali:test:security

Hardened compile mode:

node dali-lang/src/cli.js dali-lang/examples/web-eq32-reference.dl /tmp/out.generated.js --hardened

Strict mode (fail on any compile error):

node scripts/compile-dali-presets.js --strict

Install Linux file type + D icon for .dali/.dl files:

npm run -s dali:install:filetype:linux

Install Kate syntax highlighting for .dali/.dl:

npm run -s dali:install:kate:linux

Use VS Code syntax highlighting (local extension files):

# open this folder in VS Code and run Extension Development Host (F5)
dali-lang/editors/vscode

Full Linux runtime build (DALI + native addon + visualizer + runtime libs):

npm run -s build:runtime:linux

This single command executes:

• DALI lint + strict compile + security tests
• native addon rebuild (aurivo_audio.node)
• projectM visualizer CMake build and copy to native-dist/linux
• runtime library copy/check + Linux artifact verification
• syntax checks for main.js, preload.js, renderer.js

Load generated module in your web audio pipeline:

const preset = require('./dali-lang/examples/web-bass-enhancer.generated.js');
const handle = await preset.buildGraphSafe(audioContext, sourceNode, audioContext.destination);

buildGraphSafe is always async-safe and works for both webaudio and audioworklet backends.

Aurivo web engine integration (current):

• Web 32-band effect state is initialized from .dl source file: dali-lang/examples/web-eq32-reference.dl.
• Renderer parses preamp + peaking bands from that file and applies them to WebAudio.
• Secure web mode whitelist: only input web;, output speakers;, and chain effects preamp, peaking, limiter are accepted.

.dl v2 style (richer syntax) is also supported by web runtime:

preset "Web EQ32 Reference v2" {
  engine {
    input: web;
    output: speakers;
    safety: strict;
  }
  chain {
    preamp { gain: -1.5db; }
    eq32 {
      band(80hz) = 1.5db;
      band(125hz) = 2.0db;
    }
    limiter { ceiling: -0.8db; }
  }
}

Why this architecture

.dali is designed as a DSL layer. Real-time processing still runs on Web Audio nodes (and later can target AudioWorklet/WASM backend).

Important: zero latency is not physically possible on real hardware. The goal is ultra-low and stable latency.

Roadmap file: dali-lang/ROADMAP.md

Next milestones

1. Add validation + diagnostics with source ranges.
2. Add AudioWorklet backend target.
3. Add WASM DSP backend target.
4. Add adaptive loudness and bass-preserving protection blocks.