slhaf-hub

Kotlin-based dynamic script host with HTTP APIs, root/sub token auth, script metadata, timeout control, and companion CLI/TUI tools.

Language:

• English: README.md
• 中文: README.zh-CN.md

Features

• Dynamic script loading from scripts/*.hub.kts without restarting host
• Root/Sub token authorization model
• Metadata in script comments (@desc, @timeout, @param)
• Compiled script cache (reuses compiled artifacts when script file is unchanged)
• Script CRUD + run + metadata APIs
• Subtoken management APIs
• Run concurrency limit (--max-run-concurrency)
• Script timeout (default 10s, script-level override)

Requirements

• JDK 17+
• Gradle (or Gradle wrapper)

Quick Start

Clone project:

git clone https://github.com/slhafzjw/slhaf-hub.git
cd slhaf-hub

Server

1) Run from terminal (Gradle)

./gradlew runWeb --args='--host=0.0.0.0 --port=8080 --scripts-dir=./scripts'

2) Run with Docker

docker build -t slhaf-hub:latest .
docker run --rm -p 8080:8080 \
  -v "$(pwd)/scripts:/app/scripts" \
  -e SLHAF_HUB_TOKEN=your-token \
  -e MAX_RUN_CONCURRENCY=8 \
  slhaf-hub:latest

3) Run with Docker Compose

# optional: export SLHAF_HUB_TOKEN=your-token
# optional: export SLHAF_HUB_PORT=8080
# optional: export MAX_RUN_CONCURRENCY=8
docker compose up -d --build

Health check:

curl http://127.0.0.1:8080/health

Clients

CLI

kotlin tools/slhaf-hub-cli.kts --base-url=http://127.0.0.1:8080 --token-file=./scripts/.host-api-token list
kotlin tools/slhaf-hub-cli.kts --token-file=./scripts/.host-api-token type
kotlin tools/slhaf-hub-cli.kts --token-file=./scripts/.host-api-token run hello --arg=name=Alice --arg=upper=true

TUI

kotlin tools/slhaf-hub-tui.kts --base-url=http://127.0.0.1:8080 --token-file=./scripts/.host-api-token

CLI/TUI env vars:

• SLHAF_HUB_BASE_URL
• SLHAF_HUB_TOKEN
• SLHAF_HUB_TOKEN_FILE

Auth Model

Auth headers:

• Authorization: Bearer <token> (recommended)
• X-Host-Token: <token>

Token source priority:

1. SLHAF_HUB_TOKEN env var
2. scripts/.host-api-token
3. Auto-generated token saved to scripts/.host-api-token

Token types:

• root: full access
• sub: access to /health, /type, filtered /scripts, and allowed-script /meta/{script} + /run/{script}

Script Metadata (*.hub.kts)

Example:

// @desc: Demo greeting API
// @timeout: 10s
// @param: name | required=false | default=world | desc=Name to greet

val args: Array<String> = emptyArray()
val kv = args.mapNotNull {
    val i = it.indexOf('=')
    if (i <= 0) null else it.substring(0, i) to it.substring(i + 1)
}.toMap()

println("hello " + (kv["name"] ?: "world"))

Fields:

• @desc: <text>
• @timeout: <value>
• @param: <name> | required=true|false | default=<value> | desc=<text>

Timeout formats:

• 500ms, 10s, 1m, or plain integer seconds (10)

Default timeout:

• 10s

Metadata Validation

POST /scripts/{script} and PUT /scripts/{script} validate metadata before saving.

On validation failure, server returns 400 Bad Request with:

• line-based reason details
• valid metadata examples

API Summary

Public:

• GET /health

Authenticated:

• GET /type
• GET /scripts
• GET /meta/{script}
• GET /run/{script}?k=v
• POST /run/{script}?k=v

Root only:

• GET /scripts/{script}
• POST /scripts/{script}
• PUT /scripts/{script}
• DELETE /scripts/{script}
• GET /subtokens
• GET /subtokens/{name}
• POST /subtokens/{name}
• PUT /subtokens/{name}
• DELETE /subtokens/{name}

Common statuses:

• 200, 201, 400, 401, 403, 404, 408 (timeout)

Runtime Controls

Run concurrency:

• arg: --max-run-concurrency=<N>
• env (compose): MAX_RUN_CONCURRENCY
• default: number of available processors

The limit only applies to /run/* endpoints.

Testing

Run:

./gradlew test

Current automated coverage focuses on WebHost APIs:

• auth behavior
• script CRUD/meta/run
• metadata validation responses
• subtoken permission filtering
• run timeout behavior

Project Layout

• src/main/kotlin/work/slhaf/hub: host implementation
• src/test/kotlin/work/slhaf/hub: automated tests
• scripts: runtime scripts and token/subtoken storage
• tools: standalone CLI/TUI scripts
• Dockerfile, docker-compose.yml: container deployment